Urbanisation and State Formation in the Ancient

Sahara and Beyond

The themes of sedentarisation, urbanisation and state formation are

fundamental ones in the archaeology of many diverse parts of the
world. Hitherto these concepts have been little explored in relation to
early societies of the Saharan zone. Discussion of these issues in relation
to the precocious civilisations that bordered the vast North African
desert has rarely considered the possibility that they were interconnected
by long-range contacts and knowledge networks. The orthodox opinion
of many of the key oasis zones within the Sahara is that they were not
created before the early Medieval period and the Islamic conquest of
Mediterranean North Africa. Major claims of this volume are that the
ultimate origins of oasis settlements in many parts of the Sahara were
considerably earlier, that by the first millennium AD some of these oasis
settlements were of a size and complexity to merit the categorisation
‘towns’ and that a few exceptional examples were focal centres within
proto-states or early state-level societies.

martin sterry is Honorary Research Fellow at Durham University.

His research on the archaeology of the Sahara and North Africa makes
particular use of GIS and remote sensing. He has undertaken fieldwork
on various projects in Italy, Britain, Libya and most recently southern
Morocco, where he is co-director of the Middle Draa Project. He has
published many articles on the Libyan Fazzan, Saharan trade, urbani-
sation and oasis settlements.
david j. mattingly is Professor of Roman Archaeology at the
University of Leicester. He has worked in the Sahara for forty years
and is the author of many books and articles related to Saharan archae-
ology, such as Farming the Desert (2 vols, 1996), which won the James
R. Wiseman book award of the American Institute of Archaeology, and
The Archaeology of Fazzan series (4 vols, 2003–2013). He was the
principal investigator of the European Research Council-funded Trans-
SAHARA Project (2011–2017) which created the groundwork for this
volume, and he is the overall series editor of Trans-Saharan Archaeology,
in which this is the third of four projected volumes.

Published online by Cambridge University Press

 the trans-saharan archaeology series

Series Editor
D. J. Mattingly

Trade in the Ancient Sahara and Beyond

Edited by D. J. Mattingly, V. Leitch, C. N. Duckworth, A. Cuénod, M. Sterry
and F. Cole
Burials, Migration and Identity in the Ancient Sahara and Beyond
Edited by M. C. Gatto, D. J. Mattingly, N. Ray and M. Sterry
Urbanisation and State Formation in the Ancient Sahara and Beyond
Edited by M. Sterry and D. J. Mattingly

Forthcoming:
Mobile Technologies in the Ancient Sahara and Beyond
Edited by C. N. Duckworth, A. Cuénod and D. J. Mattingly

Published online by Cambridge University Press

 Urbanisation and State Formation in
the Ancient Sahara and Beyond

Edited by

martin sterry
Durham University

david j. mattingly
University of Leicester

Trans-Saharan Archaeology Volume 3

Series Editor:
David J. Mattingly

The Society for Libyan Studies

Published online by Cambridge University Press

 University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre,
New Delhi – 110025, India
79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781108494441
DOI: 10.1017/9781108637978
© Cambridge University Press 2020
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2020
Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall
A catalogue record for this publication is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Names: Sterry, Martin, editor. | Mattingly, D. J., editor.
Title: Urbanisation and state formation in the ancient Sahara and beyond / edited by
M. Sterry, Durham University, D.J. Mattingly, University of Leicester.
Description: Cambridge, United Kingdom ; New York, NY : Cambridge University Press,
2020. | Series: Trans-Saharan archaeology ; volume 3 | Includes bibliographical references.
Identifiers: LCCN 2019039763 (print) | LCCN 2019039764 (ebook) | ISBN 9781108494441
(hardback) | ISBN 9781108637978 (ebook)
Subjects: LCSH: Cities and towns – Africa, Sub-Saharan – History – To 1500. | Human
settlements – Africa, Sub-Saharan – History – To 1500. | City-states – Africa, Sub-Saharan –
History – To 1500. | Africa, Sub-Saharan – History. | Africa, North – History.
Classification: LCC HT114 .U728 2020 (print) | LCC HT114 (ebook) | DDC 307.760967–dc23
LC record available at https://lccn.loc.gov/2019039763
LC ebook record available at https://lccn.loc.gov/2019039764
ISBN 978-1-108-49444-1 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.

Published online by Cambridge University Press

 Urbanisation and State Formation in the Ancient
Sahara and Beyond

The themes of sedentarisation, urbanisation and state formation are

fundamental ones in the archaeology of many diverse parts of the
world. Hitherto these concepts have been little explored in relation to
early societies of the Saharan zone. Discussion of these issues in relation
to the precocious civilisations that bordered the vast North African
desert has rarely considered the possibility that they were interconnected
by long-range contacts and knowledge networks. The orthodox opinion
of many of the key oasis zones within the Sahara is that they were not
created before the early Medieval period and the Islamic conquest of
Mediterranean North Africa. Major claims of this volume are that the
ultimate origins of oasis settlements in many parts of the Sahara were
considerably earlier, that by the first millennium AD some of these oasis
settlements were of a size and complexity to merit the categorisation
‘towns’ and that a few exceptional examples were focal centres within
proto-states or early state-level societies.

martin sterry is Honorary Research Fellow at Durham University.

His research on the archaeology of the Sahara and North Africa makes
particular use of GIS and remote sensing. He has undertaken fieldwork
on various projects in Italy, Britain, Libya and most recently southern
Morocco, where he is co-director of the Middle Draa Project. He has
published many articles on the Libyan Fazzan, Saharan trade, urbani-
sation and oasis settlements.
david j. mattingly is Professor of Roman Archaeology at the
University of Leicester. He has worked in the Sahara for forty years
and is the author of many books and articles related to Saharan archae-
ology, such as Farming the Desert (2 vols, 1996), which won the James
R. Wiseman book award of the American Institute of Archaeology, and
The Archaeology of Fazzan series (4 vols, 2003–2013). He was the
principal investigator of the European Research Council-funded Trans-
SAHARA Project (2011–2017) which created the groundwork for this
volume, and he is the overall series editor of Trans-Saharan Archaeology,
in which this is the third of four projected volumes.

Published online by Cambridge University Press

 the trans-saharan archaeology series

Series Editor
D. J. Mattingly

Trade in the Ancient Sahara and Beyond

Edited by D. J. Mattingly, V. Leitch, C. N. Duckworth, A. Cuénod, M. Sterry
and F. Cole
Burials, Migration and Identity in the Ancient Sahara and Beyond
Edited by M. C. Gatto, D. J. Mattingly, N. Ray and M. Sterry
Urbanisation and State Formation in the Ancient Sahara and Beyond
Edited by M. Sterry and D. J. Mattingly

Forthcoming:
Mobile Technologies in the Ancient Sahara and Beyond
Edited by C. N. Duckworth, A. Cuénod and D. J. Mattingly

Published online by Cambridge University Press

 Urbanisation and State Formation in
the Ancient Sahara and Beyond

Edited by

martin sterry
Durham University

david j. mattingly
University of Leicester

Trans-Saharan Archaeology Volume 3

Series Editor:
David J. Mattingly

The Society for Libyan Studies

Published online by Cambridge University Press

 University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre,
New Delhi – 110025, India
79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781108494441
DOI: 10.1017/9781108637978
© Cambridge University Press 2020
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2020
Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall
A catalogue record for this publication is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Names: Sterry, Martin, editor. | Mattingly, D. J., editor.
Title: Urbanisation and state formation in the ancient Sahara and beyond / edited by
M. Sterry, Durham University, D.J. Mattingly, University of Leicester.
Description: Cambridge, United Kingdom ; New York, NY : Cambridge University Press,
2020. | Series: Trans-Saharan archaeology ; volume 3 | Includes bibliographical references.
Identifiers: LCCN 2019039763 (print) | LCCN 2019039764 (ebook) | ISBN 9781108494441
(hardback) | ISBN 9781108637978 (ebook)
Subjects: LCSH: Cities and towns – Africa, Sub-Saharan – History – To 1500. | Human
settlements – Africa, Sub-Saharan – History – To 1500. | City-states – Africa, Sub-Saharan –
History – To 1500. | Africa, Sub-Saharan – History. | Africa, North – History.
Classification: LCC HT114 .U728 2020 (print) | LCC HT114 (ebook) | DDC 307.760967–dc23
LC record available at https://lccn.loc.gov/2019039763
LC ebook record available at https://lccn.loc.gov/2019039764
ISBN 978-1-108-49444-1 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.

Published online by Cambridge University Press

 Urbanisation and State Formation in the Ancient
Sahara and Beyond

The themes of sedentarisation, urbanisation and state formation are

fundamental ones in the archaeology of many diverse parts of the
world. Hitherto these concepts have been little explored in relation to
early societies of the Saharan zone. Discussion of these issues in relation
to the precocious civilisations that bordered the vast North African
desert has rarely considered the possibility that they were interconnected
by long-range contacts and knowledge networks. The orthodox opinion
of many of the key oasis zones within the Sahara is that they were not
created before the early Medieval period and the Islamic conquest of
Mediterranean North Africa. Major claims of this volume are that the
ultimate origins of oasis settlements in many parts of the Sahara were
considerably earlier, that by the first millennium AD some of these oasis
settlements were of a size and complexity to merit the categorisation
‘towns’ and that a few exceptional examples were focal centres within
proto-states or early state-level societies.

martin sterry is Honorary Research Fellow at Durham University.

His research on the archaeology of the Sahara and North Africa makes
particular use of GIS and remote sensing. He has undertaken fieldwork
on various projects in Italy, Britain, Libya and most recently southern
Morocco, where he is co-director of the Middle Draa Project. He has
published many articles on the Libyan Fazzan, Saharan trade, urbani-
sation and oasis settlements.
david j. mattingly is Professor of Roman Archaeology at the
University of Leicester. He has worked in the Sahara for forty years
and is the author of many books and articles related to Saharan archae-
ology, such as Farming the Desert (2 vols, 1996), which won the James
R. Wiseman book award of the American Institute of Archaeology, and
The Archaeology of Fazzan series (4 vols, 2003–2013). He was the
principal investigator of the European Research Council-funded Trans-
SAHARA Project (2011–2017) which created the groundwork for this
volume, and he is the overall series editor of Trans-Saharan Archaeology,
in which this is the third of four projected volumes.

Published online by Cambridge University Press

 the trans-saharan archaeology series

Series Editor
D. J. Mattingly

Trade in the Ancient Sahara and Beyond

Edited by D. J. Mattingly, V. Leitch, C. N. Duckworth, A. Cuénod, M. Sterry
and F. Cole
Burials, Migration and Identity in the Ancient Sahara and Beyond
Edited by M. C. Gatto, D. J. Mattingly, N. Ray and M. Sterry
Urbanisation and State Formation in the Ancient Sahara and Beyond
Edited by M. Sterry and D. J. Mattingly

Forthcoming:
Mobile Technologies in the Ancient Sahara and Beyond
Edited by C. N. Duckworth, A. Cuénod and D. J. Mattingly

Published online by Cambridge University Press

 Urbanisation and State Formation in
the Ancient Sahara and Beyond

Edited by

martin sterry
Durham University

david j. mattingly
University of Leicester

Trans-Saharan Archaeology Volume 3

Series Editor:
David J. Mattingly

The Society for Libyan Studies

Published online by Cambridge University Press

 University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre,
New Delhi – 110025, India
79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781108494441
DOI: 10.1017/9781108637978
© Cambridge University Press 2020
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2020
Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall
A catalogue record for this publication is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Names: Sterry, Martin, editor. | Mattingly, D. J., editor.
Title: Urbanisation and state formation in the ancient Sahara and beyond / edited by
M. Sterry, Durham University, D.J. Mattingly, University of Leicester.
Description: Cambridge, United Kingdom ; New York, NY : Cambridge University Press,
2020. | Series: Trans-Saharan archaeology ; volume 3 | Includes bibliographical references.
Identifiers: LCCN 2019039763 (print) | LCCN 2019039764 (ebook) | ISBN 9781108494441
(hardback) | ISBN 9781108637978 (ebook)
Subjects: LCSH: Cities and towns – Africa, Sub-Saharan – History – To 1500. | Human
settlements – Africa, Sub-Saharan – History – To 1500. | City-states – Africa, Sub-Saharan –
History – To 1500. | Africa, Sub-Saharan – History. | Africa, North – History.
Classification: LCC HT114 .U728 2020 (print) | LCC HT114 (ebook) | DDC 307.760967–dc23
LC record available at https://lccn.loc.gov/2019039763
LC ebook record available at https://lccn.loc.gov/2019039764
ISBN 978-1-108-49444-1 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.

Published online by Cambridge University Press

 Urbanisation and State Formation in the Ancient
Sahara and Beyond

The themes of sedentarisation, urbanisation and state formation are

fundamental ones in the archaeology of many diverse parts of the
world. Hitherto these concepts have been little explored in relation to
early societies of the Saharan zone. Discussion of these issues in relation
to the precocious civilisations that bordered the vast North African
desert has rarely considered the possibility that they were interconnected
by long-range contacts and knowledge networks. The orthodox opinion
of many of the key oasis zones within the Sahara is that they were not
created before the early Medieval period and the Islamic conquest of
Mediterranean North Africa. Major claims of this volume are that the
ultimate origins of oasis settlements in many parts of the Sahara were
considerably earlier, that by the first millennium AD some of these oasis
settlements were of a size and complexity to merit the categorisation
‘towns’ and that a few exceptional examples were focal centres within
proto-states or early state-level societies.

martin sterry is Honorary Research Fellow at Durham University.

His research on the archaeology of the Sahara and North Africa makes
particular use of GIS and remote sensing. He has undertaken fieldwork
on various projects in Italy, Britain, Libya and most recently southern
Morocco, where he is co-director of the Middle Draa Project. He has
published many articles on the Libyan Fazzan, Saharan trade, urbani-
sation and oasis settlements.
david j. mattingly is Professor of Roman Archaeology at the
University of Leicester. He has worked in the Sahara for forty years
and is the author of many books and articles related to Saharan archae-
ology, such as Farming the Desert (2 vols, 1996), which won the James
R. Wiseman book award of the American Institute of Archaeology, and
The Archaeology of Fazzan series (4 vols, 2003–2013). He was the
principal investigator of the European Research Council-funded Trans-
SAHARA Project (2011–2017) which created the groundwork for this
volume, and he is the overall series editor of Trans-Saharan Archaeology,
in which this is the third of four projected volumes.

Published online by Cambridge University Press

 the trans-saharan archaeology series

Series Editor
D. J. Mattingly

Trade in the Ancient Sahara and Beyond

Edited by D. J. Mattingly, V. Leitch, C. N. Duckworth, A. Cuénod, M. Sterry
and F. Cole
Burials, Migration and Identity in the Ancient Sahara and Beyond
Edited by M. C. Gatto, D. J. Mattingly, N. Ray and M. Sterry
Urbanisation and State Formation in the Ancient Sahara and Beyond
Edited by M. Sterry and D. J. Mattingly

Forthcoming:
Mobile Technologies in the Ancient Sahara and Beyond
Edited by C. N. Duckworth, A. Cuénod and D. J. Mattingly

Published online by Cambridge University Press

 Urbanisation and State Formation in
the Ancient Sahara and Beyond

Edited by

martin sterry
Durham University

david j. mattingly
University of Leicester

Trans-Saharan Archaeology Volume 3

Series Editor:
David J. Mattingly

The Society for Libyan Studies

Published online by Cambridge University Press

 University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre,
New Delhi – 110025, India
79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781108494441
DOI: 10.1017/9781108637978
© Cambridge University Press 2020
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2020
Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall
A catalogue record for this publication is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Names: Sterry, Martin, editor. | Mattingly, D. J., editor.
Title: Urbanisation and state formation in the ancient Sahara and beyond / edited by
M. Sterry, Durham University, D.J. Mattingly, University of Leicester.
Description: Cambridge, United Kingdom ; New York, NY : Cambridge University Press,
2020. | Series: Trans-Saharan archaeology ; volume 3 | Includes bibliographical references.
Identifiers: LCCN 2019039763 (print) | LCCN 2019039764 (ebook) | ISBN 9781108494441
(hardback) | ISBN 9781108637978 (ebook)
Subjects: LCSH: Cities and towns – Africa, Sub-Saharan – History – To 1500. | Human
settlements – Africa, Sub-Saharan – History – To 1500. | City-states – Africa, Sub-Saharan –
History – To 1500. | Africa, Sub-Saharan – History. | Africa, North – History.
Classification: LCC HT114 .U728 2020 (print) | LCC HT114 (ebook) | DDC 307.760967–dc23
LC record available at https://lccn.loc.gov/2019039763
LC ebook record available at https://lccn.loc.gov/2019039764
ISBN 978-1-108-49444-1 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.

Published online by Cambridge University Press

 Contents

List of Figures [page viii]

List of Tables [xv]
List of Contributors [xvi]
Preface [xix]
david j. mattingly

part i introduction [1]

1. Introduction to the Themes of Sedentarisation,
Urbanisation and State Formation in the Ancient Sahara
and Beyond [3]
david j. mattingly and martin sterry

part ii oasis origins in the sahara: a

region-by-region survey [51]
2. Garamantian Oasis Settlements in Fazzan [53]
david j. mattingly, stefania merlo, lucia mori
and martin sterry

3. Pre-Islamic Oasis Settlements in the Eastern Sahara [112]

david j. mattingly, martin sterry, louise rayne
and muftah al-haddad

4. The Urbanisation of Egypt’s Western Desert under Roman

Rule [147]
anna lucille boozer

5. Pre-Islamic Oasis Settlements in the Northern

Sahara [187]
david j. mattingly, martin sterry, muftah al-
haddad and pol trousset

6. Pre-Islamic Oasis Settlements in the North-Western

Sahara [239]
martin sterry, david j. mattingly and youssef
bokbot

Published online by Cambridge University Press

 vi Contents

7. Pre-Islamic Oasis Settlements in the Southern Sahara [277]

martin sterry and david j. mattingly

8. Discussion: Sedentarisation and Urbanisation in the

Sahara [330]
martin sterry and david j. mattingly

part iii neighbours and comparanda [357]

9. Early States and Urban Forms in the Middle Nile [359]
david n. edwards

10. Mediterranean Urbanisation in North Africa: Greek, Punic

and Roman Models [396]
andrew i. wilson

11. Numidian State Formation in the Tunisian High Tell [438]

joan sanmartí, nabil kallala, maria carme
belarte, joan ramon, francisco josé cantero,
dani ló pez, marta portillo and sílvia
valenzuela

12. The Origins of Urbanisation and Structured Political

Power in Morocco: Indigenous Phenomenon or Foreign
Colonisation? [476]
youssef bokbot

13. Architecture and Settlement Growth on the Southern Edge

of the Sahara: Timing and Possible Implications for
Interactions with the North [498]
kevin c. macdonald

14. Long-Distance Exchange and Urban Trajectories in the

First Millennium AD: Case Studies from the Middle Niger
and Middle Senegal River Valleys [521]
susan keech mcintosh

15. First Millennia BC/AD Fortified Settlements at Lake Chad:

Implications for the Origins of Urbanisation and State
Formation in Sub-Saharan Africa [564]
carlos magnavita

16. At the Dawn of Sijilmasa: New Historical Focus on the

Process of Emergence of a Saharan State and a Caravan
City [594]
chloé capel

Published online by Cambridge University Press

 Contents vii

17. The Early Islamic Trans-Saharan Market Towns of West

Africa [621]
sam nixon

18. Urbanisation, Inequality and Political Authority in the

Sahara [667]
judith scheele

part iv concluding discussion [693]

19. State Formation in the Sahara and Beyond [695]
david j. mattingly and martin sterry

Index [722]

Published online by Cambridge University Press

 Figures

1.1. Map of the principal oasis groups and areas of modern

vegetation in hyper-arid and arid areas of the Sahara.
[page 6]
1.2. The ‘archetypal’ oasis? Lake Umm al-Ma in the Ubari Sand
Sea, Libya. [7]
1.3. Major routes across the Sahara in relation to rainfall
data. [9]
1.4. Examples of irrigation regimes. [13]
1.5. Distribution of different irrigation technologies across the
Sahara. [16]
1.6. The distribution of Proto-East-West Amazigh language
c.500 BC. [20]
1.7. Places discussed in Chapters 2–8. [31]
1.8. Places discussed in Chapters 9–18. [33]
2.1. Fazzan showing main regions and sites discussed in the
text. [58]
2.2. Garamantian settlement (and probable Garamantian
settlement) in the Wadi al-Ajal. [70]
2.3. Comparative plans of Garamantian hillfort sites in Wadi
al-Ajal. [72]
2.4. Hypothetical development sequence in the Taqallit
landscape. [74]
2.5. Examples of Garamantian village settlements from the Wadi
al-Ajal: a) ELH003; b) GBD001; c) GER002; d) FJJ056. [75]
2.6. Distribution of Garamantian oasis sites in the Murzuq
depressions. [77]
2.7. Possible Garamantian urban centres: a) Old Jarma; b) Qasr
ash-Sharraba. [79]
2.8. Detailed mapping of the Garamantian fortified settlements
and their associated gardens in the Zizaw area. [80]
2.9. Comparative plans of fortified sites (qsur) in the Murzuq
depression. [81]

Published online by Cambridge University Press

 List of Figures ix

2.10. Detail of possible Garamantian centre at Zuwila. [82]

2.11. Garamantian sites in eastern Fazzan. [83]
2.12. Garamantian sites recorded in the Wadi ash-Shati. [85]
2.13. Garamantian sites recorded in the Wadi Tanzzuft and Wadi
Awis. [91]
2.14. Comparative plans of Garamantian sites in the Tanzzuft
area. [94]
2.15. Detail of abandoned oasis settlement and gardens in the
Wadi Hikma area. [98]
3.1. Western Desert showing main regions and sites discussed in
the text. [113]
3.2. Eastern Libya showing main regions and sites discussed in
the text. [114]
3.3. Sites in al-Jufra oasis. [131]
3.4. The settlement and associated field-system of Busi. [133]
3.5. Comparative plans of larger settlements in al-Jufra
oasis. [134]
3.6. Foggaras and settlements in al-Jufra oasis. [135]
4.1. Map of Egypt. [149]
4.2. The Great Oasis Map. [150]
4.3. Plan of Trimithis (Amheida). [162]
5.1. South Tunisia and Tripolitania, main regions and sites
discussed in the text. [191]
5.2. Development of the Ghadamis oasis from Roman to
Present. [197]
5.3. Nefzaoua and surrounding oases main regions and sites
discussed in the text. [202]
5.4. Photograph of Roman remains at the main spring of
Telmine/Turris Tamelleni. [203]
5.5. Areas of Roman materials in the vicinity of Telmine/Turris
Tamelleni. [204]
5.6. Eastern Algeria, main regions and sites discussed in the
text. [213]
5.7. The headquarters building in the centre of the fort and vicus
of Gemellae. [216]
5.8. Satellite image of Sedrata. [222]
5.9. Comparative plans of Oasis forts and settlements (Bu Nijim,
al-Qurayyat al-Gharbia, Ras al-Ain, Negrine, Badias,
Gemellae). [226]

Published online by Cambridge University Press

 x List of Figures

6.1. North-west Sahara, main regions and sites discussed in the

text. [241]
6.2. Protohistoric settlements and funerary zones in the Wadi
Draa. [260]
6.3. Comparative plans of protohistoric sites in the Wadi
Draa. [261]
6.4. Satellite image of Tamdult. [264]
6.5. Distribution of fortified sites in the Gurara, Tuwat and
Tidikelt oases. [266]
7.1. Eastern Sahara, main regions and sites discussed in the
text. [299]
7.2. Sites in the Kawar Oasis. [304]
7.3. Satellite image of Gezabi. [305]
7.4. Mali and Niger, main regions and sites discussed in
the text. [309]
7.5. Mauritania sites discussed in the text. [317]
8.1. Spread of urbanisation across the Trans-Saharan
region. [342]
9.1. Plan of Kerma ‘town’ c.2300–1450 BC. [364]
9.2. Massive mudbrick temple, c.2000 BC – the ‘Defuffa’ – in the
centre of Kerma. [365]
9.3. Plan of colonial Egyptian ‘temple-town’, combining
substantial temple complex and residential quarter. [367]
9.4. General map of central Meroitic territories. [370]
9.5. Meroe ‘royal city’ and environs. [374]
9.6. Examples of regularly designed ‘palace’ structures: Meroe,
Muweis and Wad ben Naqa and Jebel Barkal. [376]
9.7. Map of planned enclosed settlement and later ‘suburbs’ to
south, with pottery kilns and iron-working slag heaps. [378]
9.8. Musawwarat es Sofra – royal pilgrimage centre – a periodic
royal centre? [380]
9.9. Hinterland settlement of Naqa – palace complexes and
temples. [382]
9.10. Jebel Qeili inscription. [383]
9.11. Examples of throne dais, emphasising subjugation – mainly
fragmentary finds from Meroitic palace centres. [384]
10.1. Carthage – Punic houses on the Byrsa hill, destroyed in the
Third Punic War, 146 BC. [399]
10.2. Kerkouane – interior of Punic house with basin and bath
tub. [401]

Published online by Cambridge University Press

 List of Figures xi

10.3. Plan of the Greek city of Euesperides (Benghazi). [402]

10.4. Cyrene – Temple of Zeus (sixth century BC), built with
massive columns in the Doric order. [403]
10.5. Plan of Roman Carthage showing the main public buildings
and known elements of water-supply infrastructure. [409]
10.6. Plan of Roman Timgad, showing fulleries (shaded) and other
workshops. [412]
10.7. Timgad – the view along the decumanus maximus towards
the so-called ‘Arch of Trajan’. [414]
10.8. The Severan nymphaeum (monumental fountain) at Lepcis
Magna. [417]
10.9. Aerial view of Sabratha, looking west. [419]
10.10. Satellite image of the Roman fort of Bu Nijim and its
surrounding vicus or settlement. [426]
10.11. Sufetula (Sbeïtla): Late antique building with two oil presses,
attached to the church of Saints Gervasius and
Tryphon. [427]
10.12. Sabratha – late antique graves in a street by the church to the
north of the theatre. [428]
10.13. Sullecthum (Salakta) – late defensive enclosure, belonging
probably to the Vandal period. [430]
11.1. Map of the late third-century BC polities in the Central and
Eastern Maghrib. [440]
11.2. Northern Tunisia and the location of Althiburos. [448]
11.3. Schematic plan of the capitol area and location of the
excavation zones. [449]
11.4. Section of sectors 3–4a in excavation zone 2 showing the
stratigraphic sequence of the Numidian period. [451]
11.5. Schematic Early Numidian 3 constructions in excavation
zone 2. [453]
11.6. Punic-type cistern of the Middle Numidian period in
excavation zone 2. [454]
11.7. Defensive wall seen from the south-west. [454]
11.8. Percentage of charcoal taxa per phase, from a diachronic
perspective. [456]
11.9. 1) Percentage of cultivated plants (PC), wild plants (PS) and
other types (AU) over the total of the remains; 2) Percentage
of different types of cultures over the total number of
individuals; 3) and 4) Percentage of the total number of

Published online by Cambridge University Press

 xii List of Figures

cultivated and wild plants; 5) and 6) estimation of the

distribution of frequencies. [457]
11.10. Relative frequency of the three main faunal taxa all over the
occupation. The vertical lines indicate twice the standard
deviation (confidence level of 95 per cent). [458]
11.11. View of monument 53 during the excavation. [461]
11.12. Chamber of monument 53. [462]
11.13. Plan of monument 42. [464]
11.14. Plan of monument 647. [465]
11.15. Chamber of monument 647. [466]
11.16. Plan of the first phase of monument 53 with the two
‘antennas’ or ‘arms’. [467]
12.1. Map of sites mentioned in this chapter. [477]
12.2. Ceramics from Lixus. [481]
12.3. Sword from the Loukkos. [482]
12.4. The ‘allée couverte’ tomb of al Quantara. [483]
12.5. a) and b) Mogador: hand-made pottery. [484]
12.6. Kach Kouch plateau, overlooking the lower Oued Laou
valley. [485]
12.7. a) Kach Kouch: storage structures; b) Kach Kouch:
archaeological structures. [486]
12.8. Kach Kouch, vase decorated with ‘graffito’. [487]
12.9. Aerial view of the Mzora tumulus. [491]
12.10. Monument in the form of a dwelling under the mound of Sidi
Slimane. [493]
13.1. Map of regions and key sites discussed in this chapter. [499]
13.2. Plan of Tichitt ‘Village 72’. [502]
13.3. Plan of the Tagant site T150. [504]
13.4. Linked round structures with coursed earth walls from
Tongo Maaré Diabel, Horizon I, Unit B, AD 500–650. [507]
13.5. Loaf-shaped mudbrick from Tongo Maaré Diabel, Horizon
II, AD 650–750. [507]
13.6. Plan of excavations at Unit(s) A-B-C, Tongo Maaré Diabel,
Horizon II, AD 650–750. [508]
13.7. Chart showing the size distribution (ha) of individual
settlement mounds surveyed in the Méma. [512]
13.8. Chart showing the size distribution (ha) by aggregated
settlement mound clusters surveyed in the Méma. [512]
14.1. Map of sites, regions and major trade routes from the tenth
century. [522]

Published online by Cambridge University Press

 List of Figures xiii

14.2. Change in land use zones from 1600 to 1850 as reconstructed

from historical documents. [526]
14.3. Timeline showing excavation sequences discussed in the
text. [527]
14.4. Excavation and augur coring locations on Jenné-jeno,
showing the depth of deposits and phase chronology for the
excavation units. [530]
14.5. White-on-red geometric pottery links the Lakes Region (left),
Jenné-jeno (centre), and Kumbi Saleh (right). [532]
14.6. Clustered mounds around Jenné-jeno and Djenné. [533]
14.7. The location of Gao Ancien and Gao Saney. [537]
14.8. Middle Senegal study area with inventoried sites and
geomorphological features. [543]
14.9. Distribution of first-millennium BC copper in Mauritania
extending to Walaldé. [545]
14.10. Ceramic styles on the Middle Senegal from AD 600 to
1100. [549]
15.1. Map showing the location of sites discussed in the
text. [569]
15.2. Magnetogram of Zilum, with indication of features
discussed. [571]
15.3. Map showing the location of Iron Age sites in the Gajiganna
area. [575]
15.4. Geophysical survey of Dorota. [576]
15.5. Magnetogram of Zubo showing the location of features
discussed. [579]
15.6. Aerial view of Goulfei. [585]
16.1. Map of Tafilalt today. [606]
16.2. Aerial view of a salt extraction site along Wadi Ziz at the place
called Tamellaht. [608]
16.3. Bed and riverbanks of Wadi Ziz on the northern side of
Tafilalt plain. [610]
16.4. Jabal Afilal and its pre-Islamic settlement on the top of a hill,
view from north-east. [613]
16.5. Aerial view of Jabal Afilal, surrounded by cliffs and
overlooking Wadi Ziz. [614]
16.6. The northern wall of the Jabal Afilal settlement. [614]
17.1. Map showing Trans-Saharan trade routes to West Africa and
localities referred to in the text. [622]
17.2. Aerial photograph of the site of Tagdaoust. [627]

Published online by Cambridge University Press

 xiv List of Figures

17.3. Plan of the excavated area at Tagdaoust, Mauritania, Building

Phase 6. [628]
17.4. Plan of the central area of the urban ruins of Kumbi
Saleh. [632]
17.5. Excavated buildings at Kumbi Saleh. [633]
17.6. Map showing the Essouk-Tadmakka ruins in relation to the
Wadi Essouk and surrounding cliffs, and also illustrating
excavation locations. [637]
17.7. East–west view across a portion of the central Essouk-
Tadmakka. [638]
17.8. Plan of selected structures present on the surface of
Essouk-Tadmakka. [639]
17.9. Essouk-Tadmakka: looking down into excavation unit
Ek-A. [640]
17.10. Structural remains detected during large-scale exposure at
Gao Ancien, including pillared schist building. [644]
17.11. Close-up of example of architectural construction in early
Islamic Gao Ancien. [645]

Published online by Cambridge University Press

 Tables

2.1. Protohistoric and historic radiocarbon dates from the Central

Sahara (Fazzan) [page 59]
2.2. List of qsur and fortified settlements surveyed in the Wadi
ash-Shati [86]
2.3. The distribution of site types between the different regions of
Fazzan [102]
3.1. Protohistoric and historic radiocarbon dates from the Eastern
Sahara [117]
5.1. Protohistoric and historic radiocarbon dates from the
Northern Sahara [189]
6.1. Protohistoric and historic radiocarbon dates from the
Western Sahara [242]
7.1. Protohistoric and historic radiocarbon dates from the
Southern Sahara [279]
10.1. Areas of selected large and medium-sized cities in Roman
North Africa [415]
14.1. Imported materials excavated from Jenné-jeno [531]
19.1. States and proto-states in the Sahara [707]

Published online by Cambridge University Press

 Contributors

Muftah al-Haddad is Professor at the University of Azzaytuna, Tarhuna,

Libya.
Youssef Bokbot is Professor at the National Institute for Archaeological
Sciences and Heritage (INSAP), Morocco.
Anna Lucille Boozer is Associate Professor at Baruch College, New York.
Maria Carme Belarte is Research Professor at ICREA, Barcelona (Spain),
and researcher at the Catalan Institute of Classical Archaeology
(ICAC), Tarragona (Spain).
Chloé Capel is Post-Doctoral Researcher at the Université Paris I.
David N. Edwards is Lecturer in the School of Archaeology and Ancient
History, University of Leicester.
Francisco José Cantero is Research Associate at the University of
Barcelona.
Nabil Kallala is Emeritus Professor at the University of Tunis and the
former President of the Institut national du patrimoine (INP),
Tunisia.
Dani López is a carpologist at ArqueoVitis SCCL and Research Associate
at the University of Barcelona.
Kevin C. MacDonald is Professor of African Archaeology at the Institute
of Archaeology, London.
Susan Keech McIntosh is Herbert S. Autrey Professor of Anthropology at
Rice University, Houston, Texas.
Carlos Magnavita is Research Fellow in the Frobenius Institute, Goethe-
University Frankfurt.
David J. Mattingly is Professor in the School of Archaeology and Ancient
History, University of Leicester.
Stefania Merlo is Senior Lecturer at Witwatersrand University, South
Africa.
Lucia Mori is Assistant Professor in the Department of Classics, Sapienza
University of Rome.
Sam Nixon is Curator and Head of Africa Section at the British Museum.
Marta Portillo is EU Horizon 2020 Marie Sklodowska-Curie Fellow,
Department of Archaeology, University of Reading.

Published online by Cambridge University Press

 List of Contributors xvii

Joan Ramon is Research Associate at the University of Barcelona.

Louise Rayne is Research Associate on the Endangered Archaeology
Project in the School of Archaeology and Ancient History,
University of Leicester.
Joan Sanmartí is Professor in the Department of History and Archaeology,
University of Barcelona.
Judith Scheele is Director of Studies at the École des hautes études en
sciences sociales (EHESS), Marseille.
Martin Sterry is Honorary Research Fellow in the Department of
Archaeology, Durham University.
Pol Trousset is Emeritus Researcher of the Centre Camille Julian,
Aix-en-Provence.
Sílvia Valenzuela is Principal Researcher ERC-StG ZooMWest, Consejo
Superior de Investigaciones Científicas (CSIC), Barcelona.
Andrew Wilson is Professor of Archaeology of the Roman Empire,
University of Oxford.

Published online by Cambridge University Press

Published online by Cambridge University Press
 Preface

When I was working on my PhD thesis on the Roman province of

Tripolitania (north-west Libya) in the early 1980s, I became intrigued by
a desert people who inhabited Fazzan, the area of the Central Sahara to the
south of Tripolitania. This was my first introduction to the Garamantes.
They were regularly mentioned in the ancient Greek and Roman sources,
though seldom in complimentary terms – for the most part being depicted
as nomadic and uncouth barbarians.1 However, some pioneering archae-
ological work in the 1930s and then again in the 1960s–1970s had revealed
their physical traces to be considerably more sophisticated than would be
assumed on the basis of the literary stereotypes.2
This volume arises out of my subsequent direct engagement across more
than 20 years now with the archaeology of Fazzan. In 1996, I was given the
chance to renew field research in what were effectively the Garamantian
heartlands. Following an initial scoping visit that year, I directed the Fazzan
Project across six years, carrying out excavations and survey around the
capital of the Garamantes at Garama (Old Jarma), with an emphasis on
tracing evidence for their settlements, but also mapping other archaeological
features including cemeteries and irrigation systems.3 A notable result of this
work was the clear demonstration of the sophisticated and substantial net-
work of oasis farming settlements that lay at the heart of the Garamantian
territory. Rather than being ‘nomadic barbarians’, the Garamantes now
appear to have been predominantly sedentary oasis farmers, living in sub-
stantial permanent and complex settlements of mudbrick buildings. That is
not to say that the Garamantes did not also incorporate pastoral elements, as
will be further discussed at various points below, but simply to highlight the
unexpected density and sophistication of sedentary oasis settlements. There

1
See in particular, Mattingly 2003, 79–81; 2011, 34–37 on the concept of ‘progressive
barbarisation’ imposed by ancient authors as a factor of distance from the Mediterranean.
2
Ayoub 1967; Daniels 1968; 1970; 1971; 1989; Pace et al. 1951.
3
There were five seasons of fieldwork (1997–2001) and a finds study season (2002). The results
are now fully published as Mattingly 2003; 2007; 2010; 2013 (now free to download from the
Society for Libyan Studies website). Funding for the Fazzan Project came primarily from the
Society for Libyan Studies, the Leverhulme Trust, the British Academy and the Arts and
Humanities Research Council.

https://doi.org/10.1017/9781108637978.001 Published online by Cambridge University Press

 xx Preface

is strong evidence to identify the top level of their settlement hierarchy as

‘urban’ in character and their overall society as an early Saharan state.4 The
prime aim of this volume is to evaluate those claims in a broader geogra-
phical and chronological framework.
My work on the Garamantes has subsequently evolved through a series
of further projects. Between 2007 and 2011, I directed the Desert
Migrations Project, with a particular focus on Garamantian burials and
funerary traditions.5 The increasing availability of high resolution satellite
imagery opened a new avenue of research in 2011, the Peopling the Desert
project, which extended research on the Garamantes to another of the
major oasis bands in Fazzan, the Murzuq depression.6 Another survey on
the oasis of Ghadamis in Western Libya was cut short by the Libyan civil
war in 2011 and it has so far been impossible to complete that work. The
Trans-SAHARA Project (2011–2017) marked a further evolution of this
body of work, seeking to place the Garamantes in their Saharan context
and to address the wider implications of the results obtained in the earlier
work.7 As part of the Trans-SAHARA project, although unable to return to
Libya for fieldwork, we continued to work closely with Libyan colleagues –
who followed up with ground visits to sites we identified through satellite
image analysis. This has contributed to a number of specific studies of
historic oasis clusters that are reported on below. We also commenced
a new phase of work on early oases in the Wadi Draa area of southern
Morocco (see below, Chapter 6). The widening of our field of investigation
has been hugely beneficial to our thinking about the Garamantes.
One of the major obstacles hindering understanding of the Sahara
through history is that the study of the desert and the neighbouring
zones of North Africa, the Nile Valley, Sudan and West Africa has tended
to be compartmentalised into chronologically or regionally specific inves-
tigations. Broader synthesis across the vast Trans-Saharan zone has been
lacking. The term ‘Trans-Saharan’ should be understood in the context of
this book as referring to the connected spaces of the Sahara and its eastern,
northern and southern peripheries. The Sahara has often been likened to
a great sea and no sea can be understood without reference to its adjacent

4
Mattingly 2013, 530–34; Mattingly and Sterry 2013.
5
Five planned seasons of fieldwork were completed by 2011, but the scheduled study season could
not take place in 2012 because of the Libyan civil war. Interim reports have been published in
Libyan Studies from 2007 to 2011, Mattingly et al. 2007; 2008; 2009; 2010a; 2010b; 2011a.
Funding for the Desert Migrations Project came primarily from the Society for Libyan Studies.
6
Sterry and Mattingly 2011; 2013; Sterry et al. 2012. The Peopling the Desert Project was funded
by the Leverhulme Trust.
7
The Trans-SAHARA project was funded by the European Research Council (grant no. 269418).

https://doi.org/10.1017/9781108637978.001 Published online by Cambridge University Press

 Preface xxi

shore-lands. The idea of Trans-Saharan perspectives on historical devel-

opments thus shares much in common with recent studies of the
Mediterranean, which have stressed the importance of connectivity and
supra-regional influences.8
The work of the Trans-SAHARA project was organised around a series
of four workgroups, each one supported by early career post-doctoral
research associates and each dealing with a discrete group of themes:
trade; migration, burial practice and identity; mobile technologies; urba-
nisation and state formation. As a key element of the work programme,
a workshop was held at Leicester for each of the workgroups, to which
international scholars working on neighbouring areas of the Trans-
Saharan zone were invited. From the outset, these workshops were con-
ceived as offering a chance to engage a group of leading experts in the field
in a high-level debate about the implications of the new information on the
Garamantes for studies of the wider Trans-Saharan world. Papers were
commissioned for an intended series of agenda-setting volumes on Trans-
Saharan Archaeology and pre-circulated so that the workshop sessions
focused entirely on discussion of their content. The volumes in this series
are thus unusual edited books in that each one has at its core an extended
and detailed presentation of the key results of the Trans-SAHARA research
team’s work, combined with the comparative perspectives of invited exter-
nal experts. As the Cambridge University Press reviewers of the volumes
have noted, in the interests of promoting debate we also invited critique
and contradiction from these external specialists. We think that adds to the
special character of the resulting books, integrating new evidence with
a broad overview of the state of the field and combining agenda-setting
ideas with different perspectives.
This third volume in the resulting series of four, thus presents some of the
key work of the Trans-SAHARA team and an international pool of collabora-
tors on the themes of urbanisation and state formation. The territorial expanse
of the Trans-Saharan zone is vast and, given the hostile climate and environ-
ment of the Sahara across the last 5,000 years, it is perhaps unsurprising that
scholarly research has become regionally segmented. Archaeologists have
most commonly self-identified with one of the great civilisations bordering
the Sahara: the Classical or Medieval Maghrib, the Nilotic civilisations or the
precocious polities of West Africa. Saharan historical archaeologists have been
fewer in number, vastly outnumbered by prehistorians (and especially the

8
Abulafia 2011; Broodbank 2013; Horden and Purcell 2000. See Lichtenberger 2016 for the
explicit comparison of Mediterranean and Sahara.

https://doi.org/10.1017/9781108637978.001 Published online by Cambridge University Press

 xxii Preface

devotees of rock art). The Trans-Saharan archaeology series seeks to explore

the interconnections across this zone in new ways, bringing together archae-
ologists, anthropologists and historians from different regions, varied aca-
demic traditions and multiple time periods and cultural phases.9 The volumes
are designed to reassess traditional assumptions about the history and archae-
ology of the zone, to present and assess alternative hypotheses and to set
a fresh agenda for future studies. The Sahara has for too long been seen as
a vast empty space, separating and keeping apart areas of precocious state
formation and urbanisation along the Nile valley, in the Mediterranean zone
of Libya and the Maghrib, in Sub-Saharan territories around Lake Chad and
the West African Niger Bend area. Recent archaeological studies have started
to cast doubt on this for certain parts of the Sahara and it seems an appropriate
time to review the larger picture of pre-Islamic and early Islamic develop-
ments across this entire zone.
As we are asking our readers to often step outside their core areas of
knowledge and expertise to engage with material from other parts of the Trans-
Saharan zone, place names and their mapping have exercised us all. Systems of
transliteration and spelling of place names across the Trans-Saharan region
vary enormously and the same site can be presented in several distinct ways.
We have tried to impose a measure of consistency in the transliteration of
names, following the practice I adopted for the Archaeology of Fazzan series.
However, for ease of recognition some exceptions have been allowed for sites
whose canonical spelling is so well established in the literature. We trust that
the maps provided will prove helpful with the identification of places named in
the text, but hope that readers will share our sense of being on a journey of
discovery as they read the following contributions. Most of the site mapping on
satellite imagery is the work of Martin Sterry. Thanks are also due to Mike
Hawkes for the production or revision of many other line drawings in the
volume.
David Mattingly

References

Abulafia, D. 2011. The Great Sea: A Human History of the Mediterranean. London:
Allen Lane.

9
This book is a prime output of an Advanced Grant (269418) awarded by the European Research
Council, the Trans-SAHARA Project (principal investigator David Mattingly at the University
of Leicester) 2011–2017.

https://doi.org/10.1017/9781108637978.001 Published online by Cambridge University Press

 Preface xxiii

Ayoub, M.S. 1967. Excavations in Germa between 1962 and 1966. Tripoli: Ministry
of Education.
Broodbank, C. 2013. The Making of the Middle Sea. A History of the Mediterranean
from the Beginning to the Emergence of the Classical World. London: Thames and
Hudson.
Daniels, C.M. 1968. Garamantian excavations: Zinchecra 1965–67. Libya Antiqua
5: 113–94.
Daniels, C.M. 1970. The Garamantes of Southern Libya. London: Oleander.
Daniels, C.M. 1971. The Garamantes of Fezzan. In F.F. Gadallah (ed.), Libya in
History. Benghazi: University of Libya, 261–87.
Daniels, C.M. 1989. Excavation and fieldwork amongst the Garamantes. Libyan
Studies 20: 45–61.
Horden, P. and Purcell, N. 2000. The Corrupting Sea: A Study of Mediterranean
History. Oxford: Blackwell.
Lichtenberger, A. 2016. ‘Sea without water’ – Conceptualising the Sahara and the
Mediterranean. In M. Dabag, D. Haller, N. Jaspert and A. Lichtenberger (eds),
New Horizons. Mediterranean Research in the 21st Century. Paderborn:
Ferdinand Schoningh, 267–83.
Mattingly, D.J. (ed.). 2003. The Archaeology of Fazzan. Volume 1, Synthesis.
London: Society for Libyan Studies, Department of Antiquities.
Mattingly, D.J. (ed.). 2007. The Archaeology of Fazzan. Volume 2, Site Gazetteer,
Pottery and Other Survey Finds. London: Society for Libyan Studies, Department
of Antiquities.
Mattingly, D.J. (ed.). 2010. The Archaeology of Fazzan. Volume 3, Excavations
Carried out by C.M. Daniels. London: Society for Libyan Studies, Department of
Antiquities.
Mattingly, D.J., 2011. Imperialism, Power and Identity Experiencing the Roman
Empire. Princeton: Princeton University Press.
Mattingly, D.J. (ed.). 2013. The Archaeology of Fazzan. Volume 4, Survey and
Excavations at Old Jarma (Ancient Garama) Carried out by C.M. Daniels
(1962–69) and the Fazzan Project (1997–2001). London: Society for Libyan
Studies, Department of Antiquities.
Mattingly, D.J. and Sterry, M. 2013. The first towns in the Central Sahara.
Antiquity 87.366: 503–18.
Mattingly, D.J., Lahr, M., Armitage, S., Barton, H., Dore, J., Drake, N., Foley, R.,
Merlo, S., Salem, M., Stock, J. and White, K. 2007. Desert Migrations: People,
environment and culture in the Libyan Sahara. Libyan Studies 38: 115–56.
Mattingly, D.J., Dore, J. and Lahr, M. (with contributions by others) 2008. DMP II:
2008 fieldwork on burials and identity in the Wadi al-Ajal. Libyan Studies 39:
223–62.
Mattingly, D.J., Lahr, M. and Wilson, A. 2009. DMP V: Investigations in 2009 of
cemeteries and related sites on the west side of the Taqallit promontory. Libyan
Studies 40: 95–131.

https://doi.org/10.1017/9781108637978.001 Published online by Cambridge University Press

 xxiv Preface

Mattingly, D.J., Abduli, H., Aburgheba, H., Ahmed, M., Ali Ahmed Esmaia, M.,
Baker, S., Cole, F., Fenwick, C., Gonzalez Rodriguez, M., Hobson, M., Khalaf, N.,
Lahr, M., Leitch, V., Moussa, F., Nikita, E., Parker, D., Radini, A., Ray, N., Savage,
T., Sterry, M. and Schörle, K. 2010a. DMP IX: Summary report on the fourth
season of excavations of the Burials and Identity team. Libyan Studies 41: 89–104.
Mattingly, D.J., Al-Aghab, S., Ahmed, M., Moussa, F., Sterry, M. and Wilson, A.I.
2010b. DMP X: Survey and landscape conservation issues around the Taqallit
headland. Libyan Studies 41: 105–32.
Mattingly, D.J., Abduli, H., Ahmed, M., Cole, F., Fenwick, C., Gonzalez
Rodriguez, M., Fothergill, B.T., Hobson, M., Khalaf, N., Lahr, M., Moussa, F.,
Nikita, E., Nikolaus, J., Radini, A., Ray, N., Savage, T., Sterry, M. and Wilson, A.I.
2011. DMP XII: Excavations and survey of the so-called Garamantian Royal
Cemetery (GSC030–031). Libyan Studies 42: 89–102.
Pace, B., Sergi, S. and Caputo, G. 1951. Scavi sahariani. Monumenti Antichi 41:
150–549.
Sterry, M. and Mattingly, D.J. 2011. DMP XIII: Reconnaissance survey of archae-
ological sites in the Murzuq area. Libyan Studies 42: 103–16.
Sterry, M. and Mattingly, D.J. 2013. Desert Migrations Project XVII: Further AMS
dates for historic settlements from Fazzan, South-West Libya. Libyan Studies 44:
127–40.
Sterry, M., Mattingly, D.J. and Higham, T. 2012. Desert Migrations Project XVI:
Radiocarbon dates from the Murzuq region, Southern Libya. Libyan Studies 43:
137–47.

https://doi.org/10.1017/9781108637978.001 Published online by Cambridge University Press

 part i

Introduction

Published online by Cambridge University Press

Published online by Cambridge University Press
 1 Introduction to the Themes of Sedentarisation,
Urbanisation and State Formation in the Ancient
Sahara and Beyond
david j. mattingly and martin sterry

Introduction

This volume explores a series of linked themes that have wide relevance in
world archaeology: sedentarisation, urbanisation and state formation. In
this opening chapter we review some of the key background to recent
debate on these themes and identify some of the Saharan particularities
which complicate the application of models developed elsewhere.
A dominant discourse on the Sahara throughout history has been the
idea of a ‘nomad menace’, coupled with a persistent emphasis on the
Sahara as largely uninhabited and uninhabitable.1 It is true that pastoralism
has at all times been a key mode of life and mobile populations have
underpinned the development of networks variously used for trade and
raiding.2 Yet the lifestyle and inter-relations of mobile peoples of the
historic Sahara, such as the Tuareg, have always been contingent to
a greater or lesser extent on the existence of sedentary communities, both
within the Sahara and at its fringes. A second dominant discourse of the
modern era has been the assumption of dependence of African societies on
exogenous contact and colonisation in order to achieve social evolution.3
Both of these discourses need to be challenged and re-evaluated in the light
of recent advances in archaeological knowledge.
A third key issue concerns the chronology of key developments within
the historic Saharan world – the emergence of trade and networks of

1
Rachet 1970 for an extreme example of the tendency.
2
Cf. inter alia, Nicolaisen and Nicolaisen 1997; Shaw 1983.
3
Gsell 1972a; 1972b, 1–11 for a classic characterisation of North Africa and the impact of
Phoenician colonisation. Even in more recent appraisals, there remains a reluctance among
modern scholars to recognise the degree of sedentarism present in Maghrib and Sahara prior to
the coming of Carthage and Rome, Desanges 1980. See also Mattingly 2011a; 2016, for a fuller
analysis. 3

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 4 David J. Mattingly and Martin Sterry

connectivity, the movements of people and the cultural connections

between them (including shared elements of a common Saharan cultural
koine), the technological inventions and transfers within the zone and the
emergence of sedentary oasis communities. These have commonly been
viewed as only really taking off after the Islamic conquest of northern
Africa. A prime contention of the Trans-SAHARA project is that the
ultimate origins must be sought much earlier in the Protohistoric period.
This volume follows already published monographs dealing with early
trade in the Sahara4 and burials, migration and identity.5 The trade volume
highlighted a much greater level of Saharan connectivity and inter-regional
contacts in the pre-Islamic era. While incontrovertible evidence for com-
merce remains elusive (and impossible to quantify), this is not least because
so much Saharan trade has involved organic items or high value commod-
ities like gold and ivory that are under-represented in the archaeological
record. However, archaeometrical analyses now provide us with
a ‘smoking gun’ effect, connecting, for example, some Sub-Saharan metal
artefacts with Mediterranean metal sources. Similarly, the review of burial
practices and the isotopic signatures obtained from human remains high-
lighted a high degree of mobility and motion in the pre-Islamic Sahara. The
final Trans-Saharan Archaeology volume covers questions relating to
mobile technologies and likewise supports the conclusion of the other
volumes that the Sahara was a much more populated and developed
space in the Protohistoric period.6 This volume completes the thematic
review, by focusing on settlements and what they contribute to our under-
standing of Saharan societies. Each volume stands on its own, of course,
but they are also part of an interlocking meta-analysis and frequent refer-
ence will be made in the following pages to contributions to the other
volumes in the series.
The first part of this book focuses on the evidence for the early devel-
opment of oases in different parts of the Sahara, as well as considering to
what extent the largest or most complex oasis settlements merit identifica-
tion as ‘urban’. In this volume we talk of a Protohistoric period of Saharan
history and archaeology with proposed dates of c.1000 BC–650 AD.7

4
Mattingly et al. 2017a. 5 Gatto et al. 2019. 6 Duckworth et al. Forthcoming.
7
We prefer a more defined Protohistoric period which sits between the Pastoral/Late Neolithic
periods and the Medieval period. It is characterised by a range of societal and technological
changes (e.g. metalworking, cereal agriculture and sedentary settlements). We generally avoid
pre-Islamic as the uncritical use of this term ignores the varied and important processes by
which different communities chose to convert (or not) to Islam during the Medieval period. We
acknowledge, however, the longstanding use of the terms ‘pre-Islamic’ and ‘Islamic’ which many
of our authors choose to keep. These can be helpful for distinguishing specific practices that are

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 5

In a concluding discussion at the end of the first section (Chapter 8), we

draw together some ideas about what sort of models of urbanism are
applicable to the ancient Saharan context.
The second section of the book then turns to a series of regional case
studies from lands bordering the Sahara to look at the urban development
of neighbouring societies and evidence of early state formation. The
evidence of early Islamic urbanisation and oasis formation is also reviewed
in detail, not least because much previous work on Saharan oases has
focused on sites of this period.8 A final concluding chapter returns to the
nature of political and social complexity in desert environments and the
vexed question of whether it is justifiable to identify states in the Sahara
itself in antiquity. As a result of our detailed work on their heartlands in
southern Libya, the Garamantes are a prime subject of interest, but the
contributors to this volume also consider evidence from other times or
places for complex levels of social organisation.

Sedentarisation and the Creation of Oases

In the Sahara proper, sedentarisation is synonymous with the emergence of

oases (Fig. 1.1). The oasis is a fundamental aspect of the Sahara, engrained in
public perceptions of the desert.9 One of the most enduring popular percep-
tions of the oasis is the perennial lake in the midst of sand dunes ringed by
a thin band of vegetation – the equivalent of small uninhabited islands for
ship-wrecked mariners. Though such extraordinary locations do exist out-
side mirages (Fig. 1.2), they are rare and exotic exceptions. Oases are more
commonly well frequented places with more vegetation and accessible water,
if often less available as surface lakes and streams than the archetypal
image.10 The derivation of the term ‘oasis’ (ouhat) is very ancient, going
back to Pharaonic times and seeming to denote the specific location of the
closest oases to the Nile. The term in ancient Egyptian also meant a ‘large
open cauldron’, which correlates with the typical physical setting of those

antithetical to Muslim practices such as the construction of burial cairns (although even here
the division is perhaps not as binary as is often suggested).
8
See in particular, Capel, Chapter 16; MacDonald, Chapter 13; McIntosh, Chapter 14; Nixon,
Chapter 17, this volume. Also recent publications by Aillet et al. 2017 (Sedrata); Messier and
Miller 2015 (Sijilmasa).
9
For conventional accounts of the Sahara and its oases, see inter alia, Gautier 1970; Laureano
1991; Sèbe 1989; Sèbe and Sèbe 2003; Villiers and Hirtle 2002.
10
On the definition of oases, see now Purdue et al. 2018b, especially 12–13.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 6 David J. Mattingly and Martin Sterry

Figure 1.1. Map of the principal oasis groups and areas of modern vegetation (as
identified from a MODIS NDVI) in hyper-arid and arid areas of the Sahara.

oases within depressions.11 Despite, or perhaps because of, the early origin of
the term, oases have no universal agreed definition. This is not normally
a problem, they are so distinct from surrounding desert that a ‘know it when
you see it’ attitude works for almost all case studies. However, when
encompassing a view across the whole of the Sahara and ranging into
areas of pre-desert, Sahel and river valleys we require something more
robust if we are to achieve consistency. Published definitions include
‘a fertile green spot in a desert waste, especially a sandy desert’,12 ‘an area
in the midst of a desert which is made fertile by the presence of water’,13 and
‘an area within a desert region where there is sufficient water to sustain
animal and plant life throughout the year’.14 Although fairly vague, these
definitions share in common an interest in the potential for fertile plant
growth and a corresponding availability of water, but there is no attempt to
distinguish between naturally occurring and man-made phenomena. The
definition of desert is also problematic as this is a botanical term suggesting
therefore that oases are defined by pockets of non-desert plants (of which
the most recognisable is undoubtedly the date palm). Biomes combining
similar plants, animals and climates provide one route to distinguishing
desert and oasis, so, for example, we can identify Saharan montane xeric

11 12 13
Vallogia 2004, 25. Stone 1967, 211–68. Perrin and Mitchell 1967.
14
Goudie et al. 1994.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 7

Figure 1.2. The ‘archetypal’ oasis? Lake Umm al-Ma in the Ubari Sand Sea, Libya.

woodlands, but it is hard to gather a satisfactory level of data for the entirety
of the Sahara and this broad brush approach passes over many of the locally
specific oasis environments that may be a square kilometre or less in area.
Nor does this approach encompass the subtleties and connections that make
up an oasis. Instead we prefer a multi-dimensional approach that identifies
different types of oases through a number of factors.

Oasis Vegetation
In keeping with botanical definitions the first factor in an oasis is fertile
vegetation. In the classic form, oases have dense groves of date palms which

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 8 David J. Mattingly and Martin Sterry

provide the shade for fruit trees, cereals and other crops. These palmeries
have such a dramatic effect that a climatic ‘oasis effect’ has been noted
wherein the overall temperature of the oasis is lower during the day and
higher at night and there is a higher humidity around the palms.15 While
palms can naturally wind pollinate and propagate there are distinct advan-
tages to human cultivation: higher numbers of fruit bearing female plants
and faster growing cuttings instead of seedlings. Oases in which date palms
form the keystone species are therefore almost all anthropic to greater or
lesser degree. Other vegetation forms are possible, particularly in the
mountains where pools of standing water (guelta) or seasonal streams
and rivers can form. These can have quite different combinations of
tamarisk, shrubs and grasses to palm oasis and are far more reliant on
the slightly higher, but less predictable rainfall that affects Saharan high-
lands – a wadi that turns green with vegetation for one month in an
exceptional year may be dry for several years after. Recently, the develop-
ment of centre pivot (‘crop circle’) irrigation wherein vegetation is watered
with sprinklers on a rotating boom has created a new form of oasis based
on monoculture or a limited range of crops.
Moreover, vegetation cannot be the only determinant as examples exist
of oases with little or no cultivation such as modern al-Khalil on the
Malian-Algerian frontier or the salt mines of Tawdenni and Taghaza also
in Mali.16 In these cases food and sometimes fresh water were imported to
feed the inhabitants. While this is extreme, even for the Sahara, the move-
ment of food to support oasis populations is common and probably has
a long history.17 This is a theme to which we return below.

Saharan Climate
The lack of water is a key determinant of deserts. The 200 mm isohyet
has long been used as a crude indicator between desert and non-desert
(Fig. 1.3), but this is a highly mobile boundary and the vegetation it
supports can expand or shrink hundreds of kilometres in a few years,
for instance encompassing or excluding the major oases of Mauritania
and Mali. The northern border is less mobile due to the rain shadow
caused by the Atlas mountains that distinguishes well the start of oases
in the Maghrib. The United Nations Environment Programme defini-
tion of desert relies on aridity rather than just precipitation. The centre
15
For example, Potchter et al. 2008.
16
Scheele 2012; see Sterry and Mattingly, Chapter 7, this volume.
17
Scheele 2012; Wilson 2012.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 9

Figure 1.3. Major routes across the Sahara in relation to rainfall data.

of the Sahara can be classified as hyperarid and the northern and

southern borders as arid. While the hyperarid zone is a good descriptor
for the location of oases in the south, the northern oases sit fully within
the arid zone which stretches into and over the Atlas and Aures
mountains. The difference between the measures of precipitation and
aridity is partly a reflection of the much higher winter temperatures
that are experienced closer to the equator and the nature of the very
different weather patterns between the southern fringe of the
Mediterranean and the northern fringe of the tropics.
While it remains difficult to find a suitable climatic description for the
present, the problems are even greater in the past. Localised studies in Fazzan,
the Niger Bend and Ennedi have allowed the construction of detailed climate
histories showing that the changing nature of these biomes with types and
extent of vegetation directly linked to the water histories of these places.18
A few words are necessary at the outset concerning changes to the
climate and environment of the Sahara in the past.19 At various times in
prehistory the Sahara has oscillated between wet and arid phases. The
concept of a green Sahara is now well appreciated in relation to the pluvial

18
Cremaschi 1998; Lutz and Lutz 1995.
19
For some of the most recent syntheses on the subject, see: Brooks et al. 2005; Cremaschi and
Zerboni 2011; Kuper and Kröpelin 2006; Leveau 2018, especially 19–43; Mattingly 2003, 37–74,
327–46 with reviews of earlier literature.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 10 David J. Mattingly and Martin Sterry

phases, which created substantial river systems and vast lakes.20 The last
significant wet phase was in the Early-Middle Holocene period, broadly
10,000–3500 BC. During this period, the wide availability of water in the
form of seasonal rivers, small lakes and a high water table supported
Saharan connectivity and mobility.21 As a general trend, mobile human
communities of hunter-gatherers adapted to herding of domesticated
animals – primarily cattle.22 Although there is evidence for periodic cli-
matic oscillations already within the Early-Mid Holocene phase, with
a major abrupt arid spell recorded at around 6200 BC, it is apparent that
with the Late Holocene, at c.3500 BC, there was a significant step in
climatic change, which marked the start of the modern hyper-arid phase
in the Sahara. Minor climatic oscillations are still recorded in some parts of
the Sahara, such as certain of the mountain massifs, which received some-
what higher rainfall than the region as a whole, but the human experience
of, and interaction with, the Sahara over the last 5,000 years has concerned
a harsh desert environment that imposes limitations on settlement, move-
ment and lifestyles. That is not to say, of course, that the desert denies long-
range movements and contacts, but that these have necessarily become
more focused along axes where water is more readily available in the form
of springs and a high water table. There has been progressive decline in
water availability in the Sahara as non-renewable sub-surface water sources
have been diminished by natural and anthropogenic action and this has
had implications for both Saharan populations and the ease of
movement.23
Interpolating the climatic data is not straightforward. In the north, the
climate of the Neolithic humid phase or the ‘Green Sahara’ became pro-
gressively drier from 7000 BP/5000 BC with areas of desert expanding from
the north-east. Palaeolakes in the Nubian Sahara appear to have dried up
by around 3500 BP/2000 BC, marking the end of transhumant cattle
herding in the Wadi Howar.24 The palaeo-oasis of the Wadi Tanzzuft
gradually contracted from 5000 BP/3000 BC until reaching something
close to its current form around 2000 BP (first century AD), before con-
tracting again in the last 25 years as a result of demographic pressure.25 The
drying of the Sahara has been heterogeneous, and in the case of West Africa
may have occurred through two abrupt phases of desiccation at

20
deMenocal and Tierney 2012; Larrasoaña et al. 2013; see also www.greensahara-
leverhulme.com/ [last accessed 2 September 2019].
21
Drake et al. 2011; Manning and Timpson 2014. 22 di Lernia 2013.
23
Cremaschi and Zerboni 2009; Drake et al. 2004. 24 Kuper and Kröpelin 2006.
25
Cremaschi 2006.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 11

c.4000–3500 BP/2000–1500 BC and c.2500 BP/500 BC.26 While the limits

of the Sahara may have been more stable over the last 2,000 years there
have been oscillations that affected all or parts of the region. In the Middle
Niger, precipitation has oscillated between dry and wet phases with numer-
ous severe droughts interspersed with floods.27 The Middle Niger dry
phase of 300 BC–AD 300 coincides with a slight increase in rainfall in
Fazzan between 400 BC and AD 250.28 But it must be emphasised that after
3000 BC, the minor fluctuations in rainfall have not changed the funda-
mental reality of the Sahara as a desert environment in which cultivation
has generally been dependent on subterranean water sources or by stream
flow from (sometimes distant) rainfall zones.29
The oases of the Sahara must be recognised as an artificial phenomenon,
created in the context of this Holocene aridification phase that extends
down to present times. After the climate changed decisively around 5,000
years ago, we should first consider the significance of refugia, select
environments within the desert that offered enhanced possibilities of
supporting human or animal life and vegetation.30 Examples of refugia in
a drying out Sahara include the relict lakes of once larger bodies of water
(gradually declining in size) or mountain massifs that received slightly
higher rainfall, feeding water holes and providing seasonal vegetation. It is
perfectly logical that people and animals will have congregated around
such locations as the climate became more hostile. But as Purdue et al.
recognise there are important distinctions between refugia and oases:
In anthropology, refugia are commonly regarded as isolated ecological niches
where past populations retracted (allopatry) during hyperarid periods, while
oases are commonly regarded as water-rich spaces in arid landscapes modified
by humans (typically through cultivation and irrigation) in the form of artificial
niches.31

The importance of human actions in transforming refugia into oases is

important, while, equally, it must be stressed that many refugia did not become
oases. For example, most of the relict lakes of the third millennium BC
eventually dried up and people were obliged to move on.32 In recent years

26
Kuhlmann et al. 2004; Shanahan et al. 2006. 27 Nicholson 1979.
28
Compare McIntosh, S. 1995, 9–11 with Cremaschi 2003, 11–12.
29
Leveau 2018, demonstrates that even minor fluctuations for the desert margins and the
Maghrib proper during Classical antiquity had potentially more profound implications for
those zones.
30
Purdue et al. 2018b, 9–12. 31 Purdue et al. 2018b, 9.
32
See Gauthier and Gauthier Forthcoming for an important study that tracks human activity
following shrinking lake margins in Chad.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 12 David J. Mattingly and Martin Sterry

some scholars have used Niche Construction Theory (NCT) as a conceptual

framework that bridges between deterministic environmental and anthropo-
centric explanations.33 It is also very obviously suited to the fact that oases are
essentially constructed niches within constrained environmental settings and
this closely correlates with our approach in this volume.

Oasis Water-Sources
Just as the form of vegetation is variable, so too is the water that makes up
the other component of an oasis. Several principal forms of hydraulic
regime can be identified (Fig. 1.4):34

Perennial Rivers
Active rivers with year-round flow are extremely rare in the Sahara. The
exceptional instances are the two great rivers, the Nile and the Niger, with
the Nile running right across the Sahara, while the Niger penetrates and runs
along its southern margin before turning south again. Less celebrated is the
Wadi Draa in Morocco, which is the only perennial river to flow into the
Sahara from the north. It runs south-east for 200 km out of the High Atlas
ranges, before turning abruptly westwards towards the Atlantic, though only in
years of exceptional flood does water penetrate all the way to the Ocean. The
Senegal is also worth noting as it currently delimitates the southern extent of
the Sahara on the Atlantic coast. The exploitation of rivers for irrigation relies
on the ability to divert or lift water from the main channel onto adjacent land
(Fig. 1.4a). This may take the form of casual exploitation of periodic flood
events or to engineered systems of barrages and diversion canals, linked to
networks of channels (Fig. 1.4b) to distribute water in a controlled manner.

Seasonal Wadis
There are a number of important rivers that flow only seasonally or on an
exceptional basis dependent on sporadic rainfall, sometimes resulting in
spectacular flash floods. Some of the most important of these are wadi
systems that flow from the Atlas ranges of Morocco and western Algeria,
fed by seasonal rains and snow melt. The most celebrated examples are the
Wadi Ziz (Sis) and the Wadi Rheris (Gheris) that feed the south-eastern
33
Purdue et al. 2018a, various papers and Purdue et al. 2018b, 17–19.
34
For an overview, see also Wilson et al. Forthcoming. The analysis here builds on classic
geographical studies of Saharan hydrology, Capot-Rey 1953; Gautier 1970.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 13

Figure 1.4. Examples of irrigation regimes: a) diversion dam in the Wadi Draa; b)
Medieval irrigation canal Wadi Draa; c) seasonal runoff irrigation, Bani Walid, Libyan
pre-desert; d) conical mounds marking vertical shafts of multiple foggaras, Tafilalat,
Morocco; e) ancient artesian spring header basin, Ghadamis; f) animal driven well
(dalw), Fazzan; g) oasis gardens and distribution channels, Wadi Draa (photos a-e), g):
D. Mattingly; photo f): C. Daniels.

Moroccan oasis of Tafilalat, of which Sijilmasa was the celebrated early

Medieval capital. Another important seasonal wadi system c.150 km east of
Sijilmasa comprised the Wadi Gir/Saoura, which in exceptional years
carries floodwaters deep into the Sahara to augment the aquifer that
feeds the Tuwat oases. In eastern Algeria there are numerous small wadi
systems on the south side of Aurés Mountains. Beyond the Jabal in
Tripolitania are the seasonal wadis of the Libyan Valleys with the

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 14 David J. Mattingly and Martin Sterry

Zemzem and the Sofeggin the largest (Fig. 1.4c). The exploitation methods
for non-perennial streams include diversion of floodwaters onto adjacent
terraces as with perennial rivers, though also commonly extend to cultiva-
tion of the dry river beds once the floodwaters have subsided. Floodwater
or runoff farming varies in intensity, depending on the relative predict-
ability or regularity of rain. At one extreme it is marked by adventitious
scratch cultivation of wadis where floods have occurred, at the other it
involves the construction of a complex infrastructure of walls to control the
flow of water, limit erosion and gullying, along with water-storage features,
like cisterns.35 The former may relate to transhumant pastoral populations,
the latter normally indicate sedentary farmers. Sedentary floodwater farm-
ing represents a fundamentally different hydrological approach to irriga-
tion compared to oases that are dependent on groundwater sources.36

Springs
After perennial rivers, the most valuable water resource in a desert environ-
ment is a perennial spring. Two major types of spring need to be differentiated,
artesian and non-artesian sources. Artesian springs are those that reach the
surface from deep water deposits under pressure. These can be prolific and
long-lasting water sources and, once the springhead is contained and linked to
a network of distribution channels, can potentially irrigate large areas (Fig.
1.4e). Non-artesian springs generally extract from higher perched water tables,
have a less abundant flow and because of the non-renewable nature of the
groundwater in much of the Sahara tend to have a more limited life span.
There is a particularly important group of artesian springs in the northern
Algerian/Tunisian Sahara (oases of the Wadi Rhir, Jarid, Nefzaoua). Artesian
springs have often been improved by human enterprise, not only in construct-
ing header tanks and canals, but also in digging additional deep shafts to tap
the artesian waters (though such work is extremely hazardous and difficult).
Spring-fed lakes exist in some sand seas, but the nature of such interdune
depressions constrains the ability to use the water to irrigate wide areas.

Wells
Groundwater in areas lacking springs is most commonly tapped by the
construction of wells. Where the water table lay close to the surface
35
See Barker et al. 1996a; 1996b for the classic investigation of Saharan floodwater farming. Also
in Cyrenaica, on the north-eastern edge of the Sahara there are many seasonal wadis.
36
Mattingly 2004b.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 15

(at a depth of at most a few metres) the most common traditional water
lifting device is the shaduf or balance beam well, in which
a counterbalanced beam allows a bucket to be repeatedly dipped into the
well and emptied into a distribution channel. Where water lies at greater
depth, the mechanical effort of raising it is commensurately higher, invol-
ving either human or animal power (dalw wells) to draw up the water (Fig.
1.4f).37 Irrigation that depends on wells and buckets tends to have a lower
yield in water than that relating to systems exploiting flowing water and
require far more labour on a day to day basis.

Groundwater Catchments
In the Suf oasis of the Great Eastern Erg a high water table beneath dunes
has been exploited by the laborious mechanism of digging down in the
interdune depressions to create micro-catchments where palms and other
crops are irrigated by the groundwater.38 This exploits in an extreme way
a key characteristic of many oases. It is an underappreciated fact that in
many oases the date palms are not watered directly, but rather, with their
deeper root systems, are sustained by the generally higher groundwater
level of the oasis. The main irrigation efforts are directed towards watering
small garden plots where cereals and legumes are intensively cultivated.
A serious decline in the level of the water table is often advertised by
a catastrophic decline in the health of the date palms, as we witnessed
first hand in Fazzan between 1996 and 2011.39

Foggaras
The foggara is the Saharan variant of a technology known in Iran as the
qanat. This is a distinctive form of irrigation, with high initial costs (and
recurrent maintenance demands). It exploits a difference in level that is
sometimes found between the water table beneath a valley or basin floor
and surrounding hills or escarpments. Where the water table at the side of
a valley is at a higher absolute level above the sea than the centre of the
valley, a low gradient underground channel can be dug to carry water from
a mother well to a point towards the centre of the valley where the channel
breaks the surface, effectively forming an artificial spring. Foggara con-
struction involves the digging of long lines of shafts (which can be spotted
by distinctive donut-shaped rings of spoil at the surface) to allow regular
37
See Wilson and Mattingly 2003, 266–70 for a discussion of the main well types.
38
Battailon 1955. 39 Mattingly 2013, 31, with figure 2.6.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 16 David J. Mattingly and Martin Sterry

access to the subterranean channel for the purposes of initial tunnelling

and recurrent maintenance (Fig. 1.4d). The earliest foggaras in the Eastern
Sahara can be dated to the mid first millennium BC, with the Garamantian
heartlands of the Central Sahara containing some of best-documented and
most extensive evidence for Protohistoric foggara construction.40 They are
also attested in many other parts of the Sahara, though the chronology of
introduction of the technology within different regions is not generally
closely dated. Foggaras were particularly important as a means of over-
coming a shortage of local springs.
The scarce water generated by the different forms of irrigation described
above is generally distributed by carefully measured volume or time alloca-
tion along minor channels into small garden plots, where cereals and other
annual crops are cultivated shaded by palms and other fruit trees
(Fig. 1.4g).

Distribution of Different Irrigation Techniques

As the detailed case studies that are presented in Chapters 2–7 show,
Saharan regions made use of the different hydraulic resources available

Figure 1.5. Distribution of different irrigation technologies across the Sahara: rainfall
runoff (R), springs (A), wells (W), underground irrigation channels/foggaras (F), canals (C).

40
Wilson 2005; Wilson and Mattingly 2003.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 17

or suitable to the local topography and used a range of technologies for

exploiting them. The overall distribution pattern of irrigation technol-
ogies in the Sahara is a complex mosaic, dependent on local hydro-
logical resources, topography and social factors (Fig. 1.5). There was
thus no uniform blueprint for creating oases in the Sahara, as the
background hydraulic conditions varied so much. While the most
abundant and most accessible sources of water (like rivers and springs)
were favoured locations for early oasis development, every locality and
each water source presented its own challenges and the costs of oasis
agriculture were always high, both at start up and recurrently. Simple
diffusionist models cannot explain the spread of the oasis, but at the
same time the importance of the acquisition of the technological know-
how that underpinned oasis farming should not be underestimated.
This issue of mobile technologies is the focus of a separate volume in
this series.41 The important point for us here is that Saharan oases were
enormously varied in their evolution, with many showing evidence of
the addition of secondary irrigation works to supplement the primary
hydraulic regime at some point. Such diversification could have been
driven by population growth and a need to expand the cultivated area,
or equally may have been a response to a declining water table impact-
ing the primary irrigation system.

Oasis Networks
So far we have considered oases as zones of fertile vegetation within arid
environments, with variability in the nature of both parts. However, one of
the key differences between an oasis and an isolated well could be considered
to be the presence of people on a permanent, semi-permanent or seasonal
basis at oases, while wells sometimes receive only periodic visits. If we see
oases as concentrations of people within a desert this opens up different
dimensions that help to explain the variability we see archaeologically.
There is a common idea that oases were miraculously productive loca-
tions. The Roman writer Pliny the Elder is typical in his laudatory account
of the fecundity of the oasis of Tacape on the Tunisian coast – extolling the
multiple harvests produced by several layers of crops from palms, to grapes
to cereals) grown in gardens with a premium price tag attached to such
plots of land.42 However, an important aspect of oasis agriculture is that its
high capital investment and running costs were often not economically

41 42
Duckworth et al. Forthcoming. Pliny, Natural History 18.188.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 18 David J. Mattingly and Martin Sterry

sustainable solely from the products of the land. Oases were ‘miraculous’ in
that they represented humanly created or humanly enhanced vegetated
niches within the desert wastes, but they were expensive to create and run.
As Scheele has argued persuasively, oases were uneconomic without refer-
ence to other activities and networks in the desert environment – including
symbiotic relations with pastoral groups and trade, creating value systems
that could subsidise and sustain the elevated costs of oasis farming.43
Scheele thus conceives oases as networks. Many, perhaps most, oases
make little economic sense, as the costs of developing and maintaining
hyper-intensive agriculture are not met by the potential crop returns.44
This is certainly true of the early modern period as has been demonstrated
with the malnutrition and starvation caused by the blocking of trade, and
hence food to the Tuwat oases in the early twentieth century.45 Mid-
twentieth-century records of Ghadamis demonstrate that the town’s gar-
dens could barely produce half the food needed for its small population
(the rest had to be imported from Tripoli) and poverty was a commonplace
of other oases under Italian colonial rule.46 It is an open question to what
extent this was true of earlier periods, but in any case, high investment
costs for the development and maintenance of oases should be considered
the rule in all oases.
There are three key points arising from this. Firstly, as we shall see
also with cities, oases must always have existed as networks, rather than
as isolated individual sites. The linkages to other points in the network
were needed in order to cover initial outlay and growth and when
necessary also to provide support for a struggling community. The
domestication of key pack animals like donkeys, horses, mules and
camels has been another crucial factor in facilitating the navigation of
the arid spaces of the Sahara.47 All the beasts of burden mentioned
above were present in the Sahara by the later first millennium BC,
though the importance of the camel increased over time with the
progressive drop in water tables increasing the distance between and
the delivery capacity of wells on Saharan trails.48 It is precisely because
of such constraints on movement and habitation that the Sahara is such
an interesting theatre in which to explore themes related to human
connectivity across space.
Secondly, the creation of new oases should be viewed as a conscious
expansion of that network, with commercial contacts being a driving force.

43
Scheele 2010. 44 Scheele 2012, 28–36; see also Pascon 1984, 9. 45 Scheele 2010.
46
Eldblom 1968; Scarin 1934. 47 Lichtenberger 2016, 269. 48 Mattingly et al. Forthcoming.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 19

Here of course there is a strong convergence with the conclusions of the

Trans-Saharan trade volume.49
Thirdly, the rationale for creating oases was not solely dependent on
fertile potential (that is, linked to the main botanical and climatic defini-
tion of oases), but rather reflected wider social and economic benefits
perceived for the network. Thus, a number of oases are linked to salt-
mining and others have a crucial role as staging posts along trade routes. In
extreme cases this can lead to oasis settlements without an oasis, or rather
with the oasis cultivation lying tens or even hundreds of kilometres distant.
As oases are intertwined between climatic, hydrological, botanical, techno-
logical, economic and social histories, it is perhaps no surprise that archae-
ological coverage is so patchy by region and period. Research into Saharan
prehistory has tended to concentrate on studying the decline of pastoral
communities in a drying environment, but while shrinking grasslands may
in some cases have had the potential to become oases, in most cases the long-
term water sources were lacking. The evidence of Late Neolithic communities
(c.3000–1000 BC) tends to be best preserved precisely in the locations least
suited for continuing human activity. Hence these studies have tended to paint
a picture of decline into oblivion rather than evolution towards oasis land-
scapes. Meanwhile, studies that have looked at the recent and Medieval history
of specific oases, may struggle to understand the nature of Protohistoric oases
as the earliest activity is masked by later palimpsest oasis infrastructure and the
logic of the modern oasis is sometimes distant from ancient climates, econo-
mies and communities. The Protohistoric period is thus often obscured, with
important consequences for our understanding of the beginnings of Saharan
agriculture, the evolution of oasis communities and the uptake of mobile
technologies such as metallurgy and irrigation strategies.50

Past Oases and Their Populations

As we shall see, the earliest oases in the Eastern Sahara date back to the
third millennium BC. Yet the ultimate origins of the vast majority of
Saharan oases and the identity of their first populations are unknown, or
assumed but unproven. The default verdict, sometimes with the support-
ing ‘evidence’ of foundation myths, for many oases in the Sahara is that
they were created no earlier than the Medieval period.51 However, there is
49
Mattingly et al. 2017a, 433–40. 50 Haour 2003, 224.
51
On foundation legends in Tidikelt and Tuwat, see Lehuraux 1943. For the Medieval emphasis in
general accounts of Saharan oases and trade, see inter alia Austen 2010; Lydon 2009; Thiry
1995.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 20 David J. Mattingly and Martin Sterry

Figure 1.6. The distribution of Proto-East-West Amazigh language (orange) c.500 BC. The
blue and purple colour zones represent earlier expansions of Proto-Amazigh and related
languages (after Fentress 2019).

also an interesting persistence of legends of Jewish or Christian groups

being responsible for the first irrigation works and fortified villages
(qsur).52 The linguistic evidence suggests that between the third
millennium BC and the seventh century AD, many parts of the Sahara
became dominated by Berber speaking groups (Amazigh), though there is
considerable complexity and divergence concerning the links between the
various Berber language branches.53 The best fit interpretation of the data
by Fentress notes a close correlation between major extension of Berber
speaking zones and some of the dating of oasis foundations and expansions
that are discussed in this volume (Fig. 1.6).54 To the north and west of the
Garamantes this relates to the territories traditionally ascribed to the
peoples known collectively as the Gaetuli in the ancient sources. The
Amazigh are often presented as the indigenous population of
Mediterranean North Africa – that is, phenotypically white, but things
appear to have been more complicated in the Saharan Berber-speaking

52
Gautier 1905, 24–28, concerning a strong tradition recording Jewish migration to Tuwat
(Algeria) in the late Roman period; Jacques-Meunié 1982, 173–85, on local tradition in the Draa
(Morocco) of Jewish and Christian groups established there before the arrival of Islam.
53
See in particular the different reconstructions proposed by Blench 2019 and Ehret 2019.
54
Fentress 2019.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 21

zones. The skeletal evidence suggests these were ethnically very mixed
communities of people with Mediterranean, Sub-Saharan and intermedi-
ate ethnic markers.55 There were also areas of the Sahara, where non Berber
populations held sway, most notably the Tubu/Teda of the Tibesti region
on the Libyan/Chad border.56
As this volume will demonstrate, there is now increasing archaeological
evidence of Protohistoric establishment of oasis communities – notably
focused on the Garamantes. The heyday of the Garamantes coincided with
the Roman Empire, but their heartlands in the Libyan Sahara lay far to the
south of the Roman frontier.
Even if the Garamantes are now recognised as an important exception,
Protohistoric development of oases elsewhere in the Sahara remains
underappreciated. One of the reasons for the historiographical blindspot
concerning precocious oasis development relates to the persistent belief in
the modern colonial era that the desert was occupied in Roman times only
by nomads.57 This was sometimes presented as a reason why the Roman
Empire chose not to conquer the desert regions – ignoring the evidence
that significant sectors of the desert were actually incorporated into the
frontier zone. Overall, there has been a general failure to consider either
the possibility that oases already existed in these areas before contact with
the Roman Empire or the implications of this.58 Pastoralists there certainly
were, but they were not alone and, in concert, pastoral and sedentary
groups represented a much more significant scale of Saharan population
than has traditionally been recognised. As we shall see there are plenty of
indicators that oases existed in the Protohistoric period and in fact pre-
Roman origins of many oases are indicated by among other factors the
prominence of date palm iconography in Carthaginian artworks or the
appearance of the date palm on Cyrenaican coinage.59 Pliny’s famous
55
Gatto et al. 2019, especially Chapters 4 and 5.
56
Beltrami 2007; Chapelle 1957; Cline 1950; Nachtigal 1974; Rohlfs 2003.
57
Two paired papers in the very first volume of the Travaux de l’institut de recherché sahariennes
provide a perfect illustration of this, Leschi 1942, writing about Rome’s relationship with
Saharan nomads and the companion paper by Capot-Rey 1942 focusing on nomads in the
French Sahara. This has had important implications for the interpretation of the Roman
frontier, Cagnat 1914; Gsell 1933; Guey 1939. See Shaw 1981; 1983; Trousset 1982; 2012 for
important historiographical discussions of ‘nomads’. See now also the important overview
article of Leveau 2018 on the social implications of the environmental and climatic conditions
of the Roman frontier in the desert margins.
58
See, for example, Toutain 1896.
59
Cherif 2006, 74–75 palms on stelae, on razors 75–76, 76 on coins; Quinn 2018, 86–122. Though
the popularity of palm imagery at Carthage played on the similarity of the Greek terms for the
tree and the ‘Phoenicians’, the allusion only really works in a region where date palms (and
knowledge of them) were well established. Roman iconography of Africa frequently depicts her

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 22 David J. Mattingly and Martin Sterry

account of the oasis of Tacape (Gabes) leaves no doubt about its pre-
Roman origins.60
Certain characteristics of the Garamantian oases, which have been
conclusively dated to Protohistoric periods, are helpful in framing an
agenda for tracing early sedentarism in other parts of the Sahara.61 These
include:
• a concentration of population around water sources or potential hydrau-
lic resources, evidenced by the appearance of new settlement forms such
as hillforts or dense funerary landscapes;
• the adoption of a developed agricultural package, with obvious affinities
with the oases of the Western Egyptian Desert;
• the association with people riding horses and driving chariots and
slightly later also the camel, as evidenced in rock art studies;62
• the movement of ideas and competencies (pyrotechnologies, irrigation
works including the foggara, spinning and weaving, a written Libyan script);
• the construction of distinctive styles of fortified structures (qsur);
• evidence of trade contacts.
In the following chapters, we focus primarily on the physical traces and
chronology of settlements, but in some cases we shall make reference to
wider sources of evidence, including some proxy markers. There are
various strands of evidence that can be called on as indicative proxies of
early origins for the oases. These include a number of sites where there was
a Roman military presence, often supported by epigraphic finds and some
investigation of the forts themselves.63 Even without explicit demonstra-
tion that the adjacent oasis was already developed prior to the arrival of the
army, the existence of an oasis would seem to be a sine qua non for the
support of units posted to remote desert locations. Literary evidence also
provides compelling evidence for Roman or pre-Roman era activity at
a number of centres that can be identified with later oasis sites. Early
European travellers in the Sahara, colonial administrators and mappers
also reported on Protohistoric or Roman ruins and antiquities in many
oasis clusters. Finds of Latin inscriptions, dressed stone blocks (including

personified with an elephant trunk headrest associated with palms, as on a ceramic plate
illustrated by Laporte 2011, 147. For pre-Roman Cyrenaican coinage featuring palms, see
Robinson 1965.
60
Pliny, Natural History 18.188.
61
The work here builds on the Archaeology of Fazzan reports, Mattingly 2003; 2007; 2010; 2013.
62
Now superbly illustrated by Barnett 2019a; 2019b.
63
See Mattingly et al. 2013 for a short summary on the Roman frontier in Africa, taking account
of the new data on oasis development.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 23

architectural elements such as bases, columns and capitals) are strong

indicators at oases in the Roman frontier zone, but these probably only
represent the tip of the iceberg, as mudbrick architecture was the norm in
the oases and stone architecture the exception. Villages built entirely in
mudbrick and which did not gain access to large quantities of imported
Mediterranean goods like pottery will be difficult to place chronologically
without radiocarbon dating. Nonetheless, all finds of Roman material in
the Sahara merit careful consideration as potential markers of
Protohistoric development.64
Pre-Islamic burial monuments are an important category of evidence.
Primarily cairns and drum-shaped tombs, these are typically dated to
c.500 BC–AD 650, but earlier and later examples are known.65 Especially
when concentrated in dense cemeteries around oases, these are a potential
indicator of early origins.66 The organisation of these cairns and drum
shaped tombs into regular and dense cemeteries, as opposed to more
dispersed funerary zones, may be an indication of locations with perma-
nent settlements.
Dating the irrigation technology used in the oases is incredibly difficult,
but one particularly distinctive type of feature can be shown to have
Protohistoric origins and to have played an important part in the early
establishment of oases. This was the foggara, essentially a Saharan variant
of the Persian qanat, which was introduced to Egypt by the Persians.
Foggaras have had a long life in many oases down to the present and
have regularly been assumed to be Medieval in date, but the possibility of
a Protohistoric spread of the technology can no longer be ignored.67
Finally, the abundant rock art of the Sahara includes a number of
depictions of actual oases associated with images of people riding horses,
driving chariots and sometimes also featuring camels.68 There are also
images of what appear to be schematic representations of fortified sites
similar to the qsur of the Garamantes (on which see below). Detailed study
of the Garamantian heartlands has demonstrated that there were dense
concentrations of horse and camel imagery alongside the oasis, indicating

64
For summary accounts of finds of Roman material culture in the Sahara, see Mauny 1956; 1978;
Salama 1981.
65
Gatto et al. 2019.
66
Camps 1961; Gauthier 2015; Grébénart 1985; Paris 1984; 1996; Reygasse 1950.
67
See Wilson et al. Forthcoming.
68
Camps and Gast 1982; Gauthier and Gauthier 2011; Hachid 2000, 136–72; Lhote 1982; Mauny
1978, 277–92 (map 282); Muzzolini 1990. For the depictions of oases and palms, see di Lernia
and Zampetti 2008, 90–97, 127–29; Hachid 2000, 207–14.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 24 David J. Mattingly and Martin Sterry

a close social relationship between the two.69 However, some of the rock art
clusters occur far from oasis centres, showing that it was also produced by
people living in remote locations and more mobile lifestyles.70 In such
cases, it may be that the imagery reflects more on the inter-relationships
between pastoralists and oasis communities in terms of trade, raiding and
other cultural aspects. The extensive distribution (reaching the Western
Sahara) of this sort of historic era rock art (first millennium BC and later),
associated with both pastoral groups and oasis locations, is another possi-
ble marker of more extensive oasis networks that underpinned successive
phases of development. Since oasis networks were highly dependent on
communication across the desert, the spread of horse and camel breeding
was essential to their maintenance.
Libyan inscriptions in early variants of the Tuareg tifinagh script may be
another sign of the emergence of complex societies and is a further beha-
vioural trait that could be linked to early oasis cultivating and trading
societies.71 Whatever the specific meaning of inscriptions and rock art
images, the commonalities between widely separated material within the
Sahara indicates an already connected space in the Protohistoric period.72

Urbanism

Towns are the second theme of this volume and are a defining character-
istic of complex polities,73 though with a multiplicity of definitions having
been put forward. These include those built round checklists of urban
traits,74 those that emphasise the roles of towns within landscapes and
people’s lives,75 those that contrast rural and urban identities76 and those
that dismiss the idea that there are defining features.77
Homogenising models, such as the consumer city that was once
favoured by ancient historians for the Mediterranean in antiquity, have
been increasingly abandoned in favour of comparative or particularist
69
See Barnett 2019a, 230–45 and 258–77, for locational and social analysis of horse/camel period
rock engravings in te Wadi al-Ajal.
70
di Lernia and Zampetti 2008; Hachid 1998; Mori, F. 1998.
71
Barnett 2019a, 161–69; Brogan 1975; Daniels 1975; Hachid 2000, 173–90; Mattingly 2003,
317–24; Rebuffat 1975.
72
Cf. Ennabli 2004, for a fairly recent restatement of the view that the Sahara was impassable
(‘infranchisable’) in the Roman era.
73
Throughout this paper we use towns/cities interchangeably to describe ancient urban
settlements of varied types. On towns in global history, see Clark 2013; Renfrew 2008.
74
See inter alia, Childe 1950; cf. Talbert 2000; Smith, M. 2003; Smith, M.E. 2009.
75
Yoffee 2005; 2009. 76 Cowgill 2004; Reader 2004. 77 Smith, A.T. 2003.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 25

approaches.78 The role of towns as centres or agglomerations remains

important: whether of size and population density;79 of particular func-
tions and craft specialisations;80 of economic networks and development;81
of farming hinterlands or zones of raw material exploitation;82 of social
complexity and authority.83 However, there is increasing interest in the
cost of towns as a form of complexity,84 the relationship between size and
communication,85 their legitimacy and resiliency within different social
strategies86 and spatial forms such as clusters of settlements or low density
settlement.87 Above all, there is awareness that urban societies are not an
inevitability, but one of many variations, with versions distinct to their
ecologies. Cities and systems of cities are multi-scalar, often transformed
by cross-scale interactions and sustained by urban innovation.88 To inves-
tigate these requires substantial exploration and most likely excavation of
both the urban centre and its hinterland.
New additions are still being made to the list of societies identified as
urbanised, especially in regions of the world where archaeological data
have been hitherto limited or under-explored. In such cases, it may be
helpful to look on cities as ‘supernova’ that re-routed and transformed
patterns of everyday life, changing the landscape of country-city
interactions.89 In archaeologically blank parts of the world, the search for
cities should focus on settlements that are very distinct from other settle-
ment types, promoting new forms of social life and changes in interactions
with a hinterland zone.
Urban centres, even of small size, placed increased stress on their
inhabitants’ ability to feed themselves and therefore had to be sustained
or supplemented through other means. Such sites can quickly hit a limit
and become what Fletcher describes as ‘stasis settlements’ – rare, very large
settlements for the region that maintain their size for long periods of
time.90
This model is reminiscent of dendritic settlement systems in which flows
of goods and information branch outward from a single large centre, but
this ‘town’ monopolises economic development at the expense of its
hinterland.91 Junker has recently reformulated the model describing it as
a concentration of political and economic power within primate centres
78
Bowman and Wilson 2011; Marcus and Sabloff 2008; Nijman 2007; Osborne and Cunliffe 2005;
York et al. 2011.
79
Marzano 2011. 80 McIntosh, R. 2005. 81 Wilson 2011. 82 Cowgill 2004.
83
Smith, A.T. 2003; cf. McIntosh, S. 1999. 84 Tainter 2000. 85 Fletcher 1995.
86
Cf. Crumley 2005. 87 McIntosh, R. 2005 . 88 Ernstson et al. 2010.
89
Yoffee 2005, 61–62; contra Morley 2011, 151. 90 Fletcher 1995, 115–17.
91
Johnson 1970; Kelley 1976; Smith, C.A. 1976.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 26 David J. Mattingly and Martin Sterry

that exert weakening authority over a series of linearly radiating

settlements.92 Most often they occurred because of the dominant role
that long-distance trade played in the local economy. Finally, environmen-
tal and/or cultural constraints on transport can result in a linear conver-
gence of trade networks on a single strategically located centre.

Saharan Urbanism and Urban Networks

There is a growing body of work questioning the applicability of European
or Western models of urbanisation to the realities of Africa and to African
societies.93 What we hope to demonstrate in this book is that the Sahara not
only underwent a process of localised sedentarisation in the first
millennia BC/AD, but that some of the key oases that emerged merit
consideration as urban centres. In the same way that oases have tended
to belong to networks, rather than exist in splendid isolation, so it is
plausible that nodal centres within those networks fulfilled roles that may
be considered urban. The extent to which these urban centres were influ-
enced by other urban societies in the Trans-Saharan zone is a more com-
plicated issue. A feature of the debate among contributors to this book
concerns the interplay of exogenous and endogenous factors. Other factors
that are discussed by various contributors include an increase in social
hierarchisation and inequality, issues of security, and the importance of
trade in the emergence of Saharan urbanism. Slave-trading is an important
and controversial aspect of early Saharan trading and urban systems – the
possibility that the Garamantes were an early example of the sort of slaving
state that later dominated the Sub-Saharan zone certainly merits
consideration.94
Trade network models also fit well with oasis centres, like the
Garamantian capital Jarma.95 Jarma was located in an area with accessible
groundwater that could have supplied caravans as well as residents and was
situated on a potential intersection of east-west and north-south commu-
nication routes through Fazzan. Comparative cases in the Hellenistic world
suggest that these centres often underwent rapid population growth
because they were connected to the outside world through trade or as
colonial points of entry.96

92
Junker 2006, 213–14.
93
See MacDonald 2013; Mattingly and MacDonald 2013; McIntosh, R. 2015; McIntosh and
McIntosh 1993; Sinclair 2013.
94
Fentress 2011. 95 Mattingly and Sterry 2013. 96 Morris 2006.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 27

State Formation

It is a logical corollary to the identification of urbanisation in the Sahara

and its neighbouring lands, to ask whether these urban societies merit
recognition as early polities. The third big theme in this book is thus state
formation. Much depends of course on what one means by this and how
exactly we choose to define states.

What Do We Mean by ‘State’?

The debate about what constitutes a state is as heated as that concerning
what defines a town. In the past, the tendency was to link state formation to
an evolutionary vision of human societies moving from simple and scat-
tered bands, to tribes, to chiefdoms, to polities.97 More recently there has
been a reaction against such evolutionary or neo-evolutionary models, at
least when they are broadly drawn and crudely applied.98 Nonetheless, it is
hard to escape entirely from evolutionary judgements in assessing what
marks early states and civilisations out from other ancient societies and
a possible way to mitigate some of the issues with the western bias of
evolutionary theory is to attempt to combine emic and etic perspectives.99
Just as towns often stand out as different from other sorts of settlement, so
early states may also have been differentiated from other contemporary
communities. What is clear from the best recent work is that hierarchy,
agriculture and urbanisation are key social components in defining the
primitive state.100 Much debate still focuses on two categories of state: the
city state and the territorial state. The former relates to the formation of
networks of self-governing polities based on a single main centre, while the
latter concerns the emergence of more hierarchical urban structures and
the coalescence of centralised nodes of power and authority. Territorial
states could evolve out of networks of city states or independently of the
prior existence of such smaller states.101 Most city states tended to remain
relatively small-scale societies (though with some notable exceptions) and
were often parts of cultural koine with wider parameters. Territorial states
97
For classic statements of this kind see Sahlins and Service 1960; Service 1975.
98
See inter alia Trigger 2003, 40–52; Yoffee 2005, 4–21 attack socio-evolutionary theory from
different perspectives, but both ultimately retain the concept of evolution in analysing early
cities and states.
99
Trigger 2003, 62–65.
100
Trigger 2003, highlights kingship, class, urbanism, food production and land ownership, trade
and craft production.
101
Trigger 2003, 92–109.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 28 David J. Mattingly and Martin Sterry

could be very extensive and be responsible for significant cultural changes

within their territories, leading to their recognition as civilisations.102
Some states might thus be identified as civilisations, but not all. So an
important supplementary question to the basic one about whether the
Garamantes constituted a state concerns what sort of state model the
evidence suits best and whether this also equates with recognition of
a distinctive civilisation.
In their influential recent book, Flannery and Marcus have made a good
case for the underpinning significance of the emergence and reinforcement
of social inequality in the creation of kingdoms and empires.103 Drawing
on a wide array of examples from different parts of the world and varied
periods, they demonstrate many commonalities in the formation of early
states, and the human behaviour and physical structures that underpinned
them.104 A key point they make is that once a kingdom existed in a region,
it provided a model for future kingdoms – the social knowledge could not
be uninvented.105 The Garamantes bear comparison with many of the
examples of early kingdoms described by Flannery and Marcus, especially
in the manner in which their society developed out of a base shared in
common with contemporary groupings into something manifestly differ-
ent and distinctive.106

Examples of States
Mario Liverani, who has studied early state formation across a wide range
of contexts from the Near East to North Africa, has been in no doubt about
the recognition of the Garamantes as an early state.107 He suggests that the
concept of ‘mirror state’ may be useful in considering the Garamantian
case.108 This idea, developed specifically in the context of nomadic empires
of the Asian steppe, saw them as examples of societies adapting ideas and
structures of state organisation from neighbouring states/empires – not so
much in terms of simple diffusionist emulation, but as a practical means of
dealing effectively with such neighbouring powers. In organising things
like tribute extraction from agriculture, livestock raising and commerce,
military organisation and levying of troops, labour needs for the construc-
tion of monumental buildings and major irrigation works, religious foci,

102
Trigger 2003, 40–52. 103 Flannery and Marcus 2012, especially 547–64.
104
Flannery and Marcus 2012, 341–471. 105 Flannery and Marcus 2012, 422.
106
For previous claims of Garamantian state formation, see Liverani 2006, 431–44; Mattingly
2003, 346–62; 2004a; 2006; 2011b; 2013, 530–34.
107
Liverani 2006, 431–44; 2007. 108 Liverani 2006, 439–40, following Barfield 1989; 2001.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 29

long-range communications, specialist craft production and markets, the

Garamantes could have drawn on observations of a range of peoples they
had contact with. In a connected Trans-Saharan world, they did not have to
invent everything for themselves.
The Sahara was bordered by a variety of early states by the late first
millennium BC: to east (Egypt),109 south-east (Meroe/Kush),110 south
(Lake Chad area),111 south-west (Niger Bend area)112 and north (kingdoms
of Cyrene; Numidia, Mauretania; empires of Carthage and Rome).113 The
histories, semantics and physical characteristics of these states vary con-
siderably one from another. Given the central position of the Garamantes
within the Trans-Saharan world, we should be careful neither to privilege
nor to exclude any one of these from consideration. The key point we
would make is that the Garamantes developed their society within a Trans-
Saharan world of states and proto-states. Obviously, the lack of detailed
historical documentation on their administrative organisation, legal provi-
sions and so on, makes the assessment of the Garamantian state much
more difficult than say the Roman one. However, we would argue that
there are important markers that identify the Garamantes as certainly
possessing some characteristics of early states and evidence hinting at the
likely presence of other aspects.
A long standing view is the idea that segmented Berber societies have been
antithetical to state formation. In part this has been a consequence of the
modern colonial era’s attitude to indigenous North Africans, denying them an
active role in history in favour of an emphasis on the creative contributions of
incomers and colonists.114 Brett and Fentress typify such views:

There is no evident progression from the relatively anarchic tribal structures to the
Hellenistic state: nor, indeed, is there any reason to expect it. From what we have
seen, the development of the Hellenistic monarchies in North Africa between the
fourth and first centuries BC occurred in emulation of the major polities and was in
no way a spontaneous occurrence.115
109
Flannery and Marcus 2012, 394–421 for a detailed study of the rise of the kingdom of Egypt in
the fourth millennium BC, which had profound regional implications.
110
Edwards 1996; 1998; Chapter 9, this volume; Welsby 1996; 2013.
111
Magnavita, Chapter 14, this volume.
112
McIntosh, R. 2005; McIntosh, S. 1999; Chapter 15, this volume.
113
Cyrene: Chamoux 1953; Numidia: Camps 1960; Kallala and Sanmartí 2011; Sanmartí et al.
2012; Sanmartí et al., Chapter 11, this volume; Mauretania: Bokbot, Chapter 12, this volume;
Carthage: Fantar 1993; Rome: Desanges et al. 2010; Mattingly and Hitchner 1995; Quinn 2009,
Wilson, Chapter 10, this volume.
114
Mattingly 2011a, 43–72. A similar point is made by Monroe 2013, 704–5 regarding modern
colonial myths inhibiting study of early states across Africa.
115
Brett and Fentress 1996, 34, but see also Fantar 1993; Laronde and Golvin 2001.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 30 David J. Mattingly and Martin Sterry

In view of the new evidence of precocious agriculture, technological

advances and proto-urbanism in Numidia, initially independent of
Phoenician influence, it is a moot point whether such dismissive views of
an African component in urbanisation and state formation are still
sustainable.116

The Structure of the Volume

In the next part of the volume, mostly written by the co-editors with
contributions from a number of others, we offer an overall review of the
evidence for early development of Saharan oases, starting in Chapter 2 with
the heartlands of the Garamantes in the Central Sahara (Fig. 1.7). From
there we move back in time and eastwards (Chapters 3–4), to review the
evidence from the oases of the Western Desert of Egypt and eastern Libya.
As well as an overall survey chapter by the editors, we include a more
focused study by Anna Boozer of the important data from the highly
developed Roman-era urban centres of the Western Egyptian oases.
These represent outstanding, but in some ways atypical, examples of
urban development in the Sahara, due to the close links between the
Egyptian oases and the Nile Valley. Chapter 5 explores the northern
oases that formed the Roman frontier from north-western Libya to central
Algeria. Chapter 6 continues further west again to the oases of western
Algeria and southern Morocco. Chapter 7 concludes our survey with
a study of the Southern Sahara to the north of Chad, Niger, Mali and
Mauritania, regions which were at various times dominated by the empires
of Kanim and Songhay as well as the Tubu and Tuareg peoples (including
the sultanate of Agadez).
The first half of this volume thus pulls together a vast dossier of
information and bibliography relating to the main oases groups in the
Sahara. We have also exploited the increasing availability of high resolution
satellite imagery to make assessments of the archaeology, both where sites
have been previously reported and frequently where there has been no
systematic archaeological research of Protohistoric and early Medieval
sites. We believe that these two aspects alone can provide a new starting
point for future work.
One of the major problems with demonstrating Protohistoric origins is
the lack of systematic archaeological investigation and scientific dating for

116
Sanmartí, Chapter 11, this volume; Mattingly 2016.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 Figure 1.7. Places discussed in Chapters 2–8.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 32 David J. Mattingly and Martin Sterry

the vast majority of these oasis groups, the notable exceptions being the
Libyan Central Sahara and the Western Egyptian Desert. The dating of
most sites is very crude and for many locations the best we can say is that
there was activity at some point in the Roman period. As we shall see, the
available evidence suggests that there was a broad chronological shift from
east to west, with the earliest oases in the Western Egyptian Desert dated to
the third and second millennium BC, the earliest indications from the
Central Sahara dated to the end of the second millennium BC and more
extensive development in the first millennium BC. A plausible working
hypothesis might be that the initial development in the Algerian and
Moroccan Sahara occurred in the late first millennium BC and early first
millennium AD, but that remains to be verified.117 It is also possible that
the Western Sahara followed its own distinctive trajectory given the vast
distances involved. For instance, the role of Tichitt culture in the wider
Trans-Saharan sphere is still poorly understood.
Further discussion of the motivations and processes of oasis creation is
the subject of Chapter 8.118 Many uncertainties remain and particularly as
our study progressed further to the west we have had to rely on hints and
suggestions more than hard evidence. The picture is currently hypothetical
in places, but at the same time we believe that the sheer volume of oasis
sites for which Protohistoric development can be demonstrated (or plau-
sibly argued) means that we must view the alternative picture of an under-
developed and barren Saharan world prior to the Medieval period as
equally unproven at present. What is needed is more work on Saharan
oases – however remote a possibility that may seem at the present time –
and especially a concerted approach to radiocarbon dating of sites (see
below).
As well as examining other case studies of urbanisation in the Trans-
Saharan world, the second half of this book explores the theme of state
formation, again with a particular focus on African specificities and con-
texts. A series of case studies focusing on the lands bordering the Sahara
and spanning the Protohistoric and Medieval eras is presented (Fig. 1.8).

117
New radiocarbon datings from Wadi Draa in Morocco on settlements associated with cereal
cultivation and early metallurgy, suggest development in the early centuries AD, but some of
the tombs in the associated cemeteries certainly date back to the first millennium BC. See
Mattingly et al. 2017b and unpublished data.
118
Building on previous discussions of the evolution of social and settlement hierarchy in desert
areas, see inter alia Barker and Gilbertson 2000; Liverani 2006. The relation between oasis
settlement and trade will be further considered there, see Mattingly et al. 2017a; Mitchell 2005;
Wilson 2012.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 Figure 1.8. Places discussed in Chapters 9–18.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 34 David J. Mattingly and Martin Sterry

In looking at the neighbouring zones in Part III, we have chosen to

follow a broadly anticlockwise tour around the Sahara, starting from the
Nile Valley. In Chapter 9, David Edwards presents a case study focused on
the Nubian kingdom of Meroe, asking interesting questions about the
nature of royal power and its manifestation in architecture and settlement
organisation.
The next group of chapters have a focus on the Mediterranean hinter-
land of North Africa, once seen as a zone dependent on external influences
and colonisation for its socio-economic evolution. However, it can now be
argued that both exogenous and endogenous factors contributed to the
emergence of urbanisation and state formation. Andrew Wilson in
Chapter 10 presents an important new survey of the Greek, Phoenician
and especially Roman influences in the development of the Classical cities
of the Maghrib. He shows an awareness of the significance of the important
work of Joan Sanmartí at the site of Althiburos, which is summarised in
Chapter 11. Excavations beneath a small Roman town in western Tunisia
have revealed a deeply stratified and complex settlement dating back to
almost 1000 BC and associated from the outset with sedentary agriculture.
The point at which this settlement might be deemed urban cannot be
settled on current evidence, but the significance of the long endogenous
proto-urban sequence is clear and establishes a very different sort of
analysis. It invites us to explore what an African model of urbanism may
have contributed to the Roman-era cities described by Wilson.119
Althiburos is important not just as a type site for pre-Roman urbanism
in the Maghrib, but also for the window it opens on the nature and early
formation process of the Numidian kingdom that had emerged as
a political force by the third century BC. Similar issues are raised in
Chapter 12 by Youssef Bokbot, discussing the nature of first
millennium BC settlement in Morocco and its relation to the emergence
of social complexity there. The narrative is clearly intertwined with that of
Phoenician entrepôts along the North African coast, but again he argues
persuasively that in the past we have too much emphasised the contribu-
tions of external groups and underestimated the role of indigenous society
in early urbanism. Here too, an early state, the Mauretanian kingdom, had
appeared by the later centuries BC.
The following three chapters all deal with aspects of urbanisation in and
bordering the Southern Sahara. In Chapter 13, Kevin MacDonald reviews
the evidence for the emergence of complex settlements in the Southern

119
See Mattingly 2016, for an extended discussion about this new paradigm.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 35

Sahara and the Niger Bend area, starting with the precocious Tichitt
settlement sequence from the second millennium BC. A complementary
view is offered by Susan McIntosh in Chapter 14, though with a greater
focus on the early Medieval developments in the Niger Bend area and
Senegal Valley. She also highlights the long distance trading contacts of the
early centres as being important in their development. Although neither
MacDonald nor McIntosh can identify precise evidence of direct contacts
between the early West African polities and the Garamantes, in general
they are supportive of the view that there were significant contacts with the
Saharan world in this period that need further evaluation in future work. In
Chapter 15, Carlos Magnavita provides an overview of the emergence of
large fortified sites in the Lake Chad basin in the first millennia BC/AD.
Again, there is a lack of direct evidence of contact with the Garamantes,
though one of the possible explanations of the fortified sites is that they
were a response to slave raiding from the north, whether direct by
Garamantian expeditions or sub-contracted by them to the people of the
Kawar oases. It must be recognised, however, that a key problem in
evaluating the trade contacts between the Sub-Saharan zone and the
Sahara is that so much of Saharan trade is in archaeologically invisible or
vestigial materials (slaves, salt, gold, other metals, textiles, leather goods,
etc.).120
The next pair of chapters has a focus on the Medieval Sahara. We noted
already how interpretation of town formation in Phoenician/Roman Africa
has been skewed hitherto by a dominant colonialist discourse that has
obscured or marginalised indigenous developments. New evidence is
prompting a reappraisal of the role of local actors. Similar themes of
competing narratives of endogenous and exogenous urban foundation
and state formation are explored for the late first millennium AD in
Morocco by Chloé Capel in Chapter 16. The town of Sijilmasa was one
of the key northern portal cities of Medieval Trans-Saharan trade and
attracted alternative foundation myths. There has been a general consensus
that this was an ex novo creation of town and oasis landscape in the mid-
eighth century. While Capel’s analysis offers an important reinterpretation
of different elements of the historiography, she highlights important evi-
dence hinting at Protohistoric activity in the vicinity of Sijilmasa. The
recent work in Wadi Draa, reported in Chapter 6, offers further support
for the view that the early Medieval golden age of Sijilmasa built on an
already established sedentary community in the Tafilalat oasis.

120
Mattingly et al. 2017a.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 36 David J. Mattingly and Martin Sterry

Chapter 17 by Sam Nixon provides a detailed review of the evidence of

the early Medieval/Islamic trading towns in the Southern Sahara and on its
edge, serving as portals into West Africa. He illustrates the distinctive
features of such sites and how they differed from contemporary West
African centres.
While we must be explicit at the outset in saying that we believe the sum
of the evidence supports the idea that the Garamantian kingdom was
a state, we have deliberately sought to consider and to engage with
a diverse array of views in this volume. Judith Scheele makes a strong
case in Chapter 18 for considering such an alternative viewpoint – arguing
that the Sahara has for most of its history been essentially stateless and
dominated by pastoral groups with shifting power structures and attach-
ments, where control of people has tended to be more important than
formal delimitation of territory. Her chapter serves as an important remin-
der that our goal is ultimately not to simply divide between state and non-
state entities and confine activities to one or other of these, but rather to
look at the full range of how power was held and used.
The concluding discussion by the editors in Chapter 19, attempts to pull
together some of the key discussions about the potential linkages between
urbanism and state formation in early historical settings in Africa. This
volume is by no means intended as a comprehensive review of the many
varieties of African urbanism and early polities, but we hope to have
demonstrated the importance of enlarging the sphere of such debates, to
take account of contemporary developments across the Trans-Sahara zone.
We argue strongly that more attention needs to be given to local influences
and models for the institutions that evolved in the Sahara, while also
recognising the inherent connectedness of the Saharan world from the
first millennium BC. The issue of state formation in the Trans-Saharan
world, especially in the Protohistoric period and within the Sahara itself is
controversial. We shall argue that some Protohistoric Saharan societies are
better understood as states or polities, though again this seems to have been
a rare development (Meroe, Siwa and the Garamantes are the outstanding
early cases).
While the authors of the chapters in this book are not in entire agree-
ment about the definitions of state that suit these African polities, there is
a strong consensus that the conventional models of state formation devel-
oped in the context of Europe and Western civilisation are ill-suited to the
African examples. As with urbanisation, we need to be more open to the
importance of ‘African’ ways of defining and enshrining power within
societies. ‘State’ and ‘state formation’ have to be reconceptualised to

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 37

make sense of the Trans-Saharan world. To follow Lekson’s corollary – if

we are arguing over whether the Garamantes or similar groups were or
were not states, then they must be really interesting.121

The Dating of Protohistoric and Historic Sites

in the Saharan Zone

A crucial problem concerning the issues that this volume seeks to address is
the lack of absolute dates from many of the oasis regions of the Sahara,
bearing on their origins and evolution. Most application of radiocarbon
dating in the Sahara has related to prehistory, with relatively little attempt
to date Protohistoric and historic sites. It is not just the early chronology
that is obscure, the Medieval phases of many key sites are equally poorly
understood, with folklore outweighing scientific dating criteria. However,
our investigations have revealed that there is more evidence now available
than is perhaps currently appreciated. Since the late 1990s, our work in the
Central Saharan heartlands of the Garamantes has involved a major pro-
gramme of scientific dating of such sites, with 177 radiocarbon dates in
total.122 The Italian mission in the same broad region has published
a further 85 relevant dates.123 We have now initiated a similar programme
of AMS dating in relation to our survey in the Moroccan Wadi Draa, with
77 dates already available.124 These dating programmes have demonstrated
the potential to differentiate between Protohistoric and Medieval settle-
ments and underpin our new appreciation of the scale and complexity of
Protohistoric Saharan sedentarism.125
In the absence of diagnostic pottery and well dated excavated sequences
in many areas of the Sahara, the close dating of sites and monuments is
fraught with difficulties. Traditional radiocarbon dating, involving sub-
stantial amounts of material for analysis often suffered from the effects of
mixed samples or ‘old wood’ charcoal. However, high precision AMS dates,
often obtained from an individual seed or small amounts of grass temper
and chaff used in the manufacture of mudbrick, offer much greater cer-
tainty that samples are not mixed and relate to annual or short-lived

121
Lekson 2009.
122
Mattingly 2007; 2010; Mattingly et al. 2015; Sterry and Mattingly 2011; 2013; Sterry et al. 2012.
123
Liverani 2006; Mori, L. 2013. 124 Mattingly et al. 2017b.
125
See now Mattingly et al. 2018, for a first overall presentation of the AMS dating programme of
the Trans-SAHARA project.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 38 David J. Mattingly and Martin Sterry

materials.126 While a few anomalous dates have still been delivered, our
experience is that a very high percentage of samples submitted (especially
from temper included in mudbrick and pisé construction) have provided
reliable dates for the construction. There are also plateaux effects in the
radiocarbon calibration curve that extend the range in certain periods, but
it is broadly possible to recognise activity phases and to distinguish
between the early first millennium BC, the later first millennium BC, the
early centuries of the first millennium AD, the later first millennium AD
and the earlier centuries of the second millennium AD. Where sufficient
dates are obtained, and especially where they come from an established
stratigraphic sequence, it is possible to narrow the range through Bayesian
modelling, as we did for phasing activity at the Garamantian capital of Old
Jarma.127
A supplementary, but highly important, contribution of this book is thus
that we have systematically gathered all the available radiocarbon dates for
the Sahara from c.1000 BC onwards and present them together here in
a standardised and consistent manner.128 All dates (both older radiometric
and AMS) have been recalibrated using Oxcal 4.3 and calibration curve
IntCal13 to two standard deviations (95.4 per cent confidence interval).129
Dates are presented in a series of tables related to each region as described
in Chapters 2–7. It is hoped that this will also encourage the spread of
scientific dating and reporting to other Saharan sites. The contextual detail
of some samples is lacking and a few samples yielded modern dates,
a reminder that sites are constantly being reworked and modified down
to the present. We hope that this resource of more than 1,000 absolute
dates will be of value to the scholarly community and regional heritage
organisations and that it will stimulate further attempts to date key sites
and monuments of the historic Sahara. While our main focus in this book
has been with the potential Protohistoric origins of urbanisation and state
formation, it is self-evident from a glance at the tables of dates that the
Medieval history of the Saharan oases can also be brought into much closer
focus through improved dating. If there is an overriding conclusion of this

126
For an explanation of sampling methods, see Sterry et al. 2012, 138–39.
127
For the Bayesian modelling of the Jarma sequence see, Mattingly 2013, 125–34.
128
Two major sources of dates are Vernet and Aumassip 1992 and Manning and Timpson 2014.
We have additionally conducted a review of the journals Radiocarbon and Archaeometry as
well as a systematic oasis by oasis search. Where possible we present the site name, explanation
of site type/context, material dated, Laboratory reference, uncalibrated range and calibrated
date. We shall be grateful to receive additional information on any dates listed as well as
additional dates, so that we can maintain and update this date list in future.
129
Bronk Ramsey 2009; Reimer et al. 2013.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 39

volume, it is that urbanisation and state formation in the Trans-Saharan

zone are themes that need to be studied in a diachronic and spatially
extensive framework.

References

Aillet, C., Cressier, P. and Gilotte, S. (ed.). 2017. Sedrata. Histoire et archéologie
d’un carrefour du Sahara médiéval à la lumière des archives inédites de
Marguerite van Berchem. Madrid: Casa de Velázquez.
Austen, R.A. 2010. Trans-Saharan Africa in World History. Oxford: Oxford
University Press.
Barfield, T. 1989. The Perilous Frontier. Oxford: Blackwell.
Barfield, T. 2001. The shadow empires: Imperial state formation along the
Chinese-nomad frontier. In S. Alcock, T. D’Altroy, K. Morrison and
C. Sinopoli (eds), Empires. Perspectives from Archaeology and History.
Cambridge: Cambridge University Press, 10–41.
Barker, G. and Gilbertson, G. (eds). 2000. The Archaeology of Drylands: Living at
the Margin. London: Routledge.
Barker, G., Gilbertson, D., Jones, B. and Mattingly, D.J. 1996a. Farming the Desert.
The UNESCO Libyan Valleys Archaeological Survey. Volume 1, Synthesis.
(Principal editor, G. Barker). Paris/London: UNESCO, Society for Libyan Studies.
Barker, G., Gilbertson, D., Jones, B. and Mattingly, D.J. 1996b. Farming the Desert.
The UNESCO Libyan Valleys Archaeological Survey. Volume 2, Gazetteer and
Pottery. (Principal editor, D.J. Mattingly). Paris/London: UNESCO, Society for
Libyan Studies.
Barnett, T. 2019a. An Engraved Landscape. Rock Carvings in the Wadi al-Ajal,
Libya. Volume 1: Synthesis. London: Society for Libyan Studies.
Barnett, T. 2019b. An Engraved Landscape. Rock Carvings in the Wadi al-Ajal,
Libya. Volume 2: Gazetteer. London: Society for Libyan Studies.
Battailon, G. 1955. Le Souf. Etude de géographie humaine. Alger: Institut de
Recherches Sahariennes, mémoire n° 2.
Beltrami, V. 2007. Il Sahara Centro-Orientale Dalla Preistoria Ai Tempi Dei
Nomadi Tubu: The Central-Oriental Sahara from Prehistory to the Times of the
Nomadic Tubus. Oxford: BARS.
Blench, R. 2019. The linguistic prehistory of the Sahara. In Gatto et al. 2019,
431–63.
Bowman, A. and Wilson, A. (eds). 2011. Settlement, Urbanization, and Population.
Oxford: Oxford University Press.
Brett, M. and Fentress, E. 1996. The Berbers. Oxford: Blackwell.
Brogan, O. 1975. Inscriptions in the Libyan alphabet from Tripolitania and some
notes on the tribes of the region. In Bynon and Bynon 1975, 267–89.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 40 David J. Mattingly and Martin Sterry

Bronk Ramsey, C. 2009. Bayesian analysis of radiocarbon date. Radiocarbon 51.1:

337–60.
Brooks, N., di Lernia, S., Drake, N., Chiapello, I., Legrand, M., Moulin, C. and
Prospero, J. 2005. The environment-society nexus in the Sahara from prehistoric
times to the present day. The Journal of North African Studies 304: 253–92.
Bynon, J. and Bynon, T. (eds). 1975. Hamito-Semitica. The Hague: Mouston.
Cagnat, R. 1914. La frontière militaire de la Tripolitaine à l’époque romaine.
Mémoires de l’Académie des Inscriptions et Belles Lettres 39: 77–109.
Camps, G. 1960. Massinissa ou les débuts de l’histoire. Libyca 8.1: 1–320.
Camps, G. 1961. Aux origines de la Berbérie. Monuments et rites funéraires proto-
historiques. Paris: Arts et Metiers Graphiques.
Camps, G. and Gast, M. (eds). 1982. Les chars préhistoriques du Sahara. Aix en
Provence: Institut de Recherches Méditerranéennes.
Capot-Rey, R. 1942. Le nomadisme pastoral dans le Sahara français. Travaux de
l’Institut de Recherches Sahariennes 1: 63–86.
Capot-Rey, R. 1953. Le Sahara Français (L’Afrique Blanche Francaise II). Paris:
Presses universitaires.
Chamoux, F. 1953. Cyrène sous la monarchie des Battiades. Paris: BEFAR.
Chapelle, J. 1957. Black Nomads of the Sahara (Translated by F. Schütze). Paris:
Librairie Plon.
Cherif, Z. 2006. Le palmier dattier et son image dans l’iconographie cartaginoise. In
M.H. Fantar (ed.), Le Sahara et l’homme. Un savoir pour un savoir faire. Tunis:
Université de Tunis, 67–84.
Childe, V.G. 1950. The urban revolution. Town Planning Review 21.1: 3–17.
Clark, P. (ed.). 2013. The Oxford Handbook of Cities in World History. Oxford:
Oxford University Press.
Cline, W. 1950. The Teda of Tibesti, Borkou and Kawar. Mwenasha: General Series
in Anthropology.
Cowgill, G.L. 2004. Origins and development of urbanism: Archaeological
perspectives. Annual Review of Anthropology 33: 525–49.
Cremaschi, M. 1998. Late Quaternary geological evidence for environmental
changes in Western Fazzan (Libyan Sahara). In M. Cremaschi and S. di Lernia
(eds), Wadi Teshuinat, Palaeoenvironment and Prehistory in South-Western
Fezzan. Milan: CNR, 13–48.
Cremaschi M. 2003. Steps and timing of the desertification during the Late
Antiquity. The case study of the Tanzzuft oasis (Libyan Sahara). In
M. Liverani (ed.), Arid Lands in Roman Times, AZA 4. Firenze: Edizioni
All’Insegna del Giglio, 1–14.
Cremaschi M. 2006. The Barkat oasis in the changing landscape of Wadi
Tannezzuft during the Holocene. In: Liverani 2006, 13–24.
Cremaschi, M. and Zerboni, A. 2009. Early to Middle Holocene landscape exploi-
tation in a drying environment: Two case studies compared from the Central
Sahara (SW Fezzan, Libya). Comptes Rendus Geoscience 341: 689–702.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 41

Cremaschi, M. and Zerboni, A. 2011. Human communities in a drying landscape.

Holocene climate change and cultural response in the Central Sahara. In
I.P. Martini and W. Chesworth (eds), Landscape and Societies, Selected Cases.
London: Springer, 67–89.
Crumley, C. 2005. Remember how to organize: Heterarchy across disciplines. In C.
S. Beekman and W. Baden (eds), Nonlinear Models for Archaeology and
Anthropology: Continuing the Revolution. Aldershot: Ashgate, 35–50.
Daniels, C.M. 1975. An ancient people of the Libyan Sahara. In Bynon and Bynon
1975, 249–65.
deMenocal, P.B. and Tierney, J.E. 2012. Green Sahara: African humid periods
paced by Earth’s orbital changes. Nature Education Knowledge 3.10: 12.
Desanges, J. 1980. (Pline l’ancien) Histoire naturelle Livre V.1–46 (L’Afrique du
nord). Paris: Collection Budé.
Desanges, J., Duval, N., Lepelley, C. and Daint-Amans, S. 2010. Carte des routes et
des cités de l’est de l’Africa à la fin de l’antiquité. Turnhout (BE): Biblio de
l’Antiquité Tardive, Brepols.
di Lernia, S. 2013. The emergence and spread of herding in northern Africa:
A critical reappraisal. In Mitchell and Lane 2013, 527–40.
di Lernia, S. and Zampetti, D. 2008. Le pitture rupestri dell’Acacus tra passato
e futuro. Firenze: All’Insegna del Giglio.
Dowler, A. and Galvin, E.R. (eds). 2011. Money, Trade and Trade Routes in Pre-
Islamic North Africa. London: British Museum Press.
Drake, N., Wilson, A., Pelling, R., White, K., Mattingly, D.J. and Black, S. 2004.
Water table decline, springline desiccation and the early development of irri-
gated agriculture in the Wadi al-Ajal, Libyan Fazzan. Libyan Studies 35: 95–112.
Drake, N.A., Blench, R.M., Armitage, S.J., Bristow, C.S., White K.H. 2011. Ancient
watercourses and biogeography of the Sahara explain the peopling of the desert.
Proceedings of the National Academy of Sciences of America 108: 458–62.
Duckworth, C., Cuénod, A. and Mattingly, D.J. (eds). Forthcoming. Mobile
Technologies in the Ancient Sahara and Beyond. Trans-Saharan Archaeology
Volume 4. Cambridge: Cambridge University Press and The Society for Libyan
Studies.
Edwards, D.N. 1996. The Archaeology of the Meroitic State: New Perspectives on Its
Social and Political Organisation. Oxford: Archaeopress.
Edwards, D.N. 1998. Meroe and the Sudanic kingdoms. Journal of African History
39.2: 175–93.
Ehret, C. 2019. Berber peoples in the Sahara and North Africa: Linguistic historical
proposals. In Gatto et al. 2019, 464–94.
Eldblom, L. 1968. Structure foncière, organisation et structure sociale. Une étude sur
la vie socio-économique dans les trois oasis libyennes de Ghat, Mourzouk et
particulièrement Ghadamès. Lund: Uniskol.
Ennabli, A. 2004. Entre Afrique du nord antique et Afrique sub-saharienne: un
obstacle infranchisable. In A. Bazzana and H. Bocoum (eds), Du nord au sud du

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 42 David J. Mattingly and Martin Sterry

Sahara. Cinquante ans d’archéologie française en Afrique du ouest et au Maghreb.

Bilan et perspectives. Paris: Editions Sepia, 23–24.
Ernstson, H., van der Leeuw, S.E., Redman, C.L., Meffert, D.J., Davis, G., Alfsen, C.,
and Elmqvist, T. 2010. Urban transitions: On urban resilience and
human-dominated ecosystems. Ambio 39.8: 531–45.
Fantar, M. 1993. Carthage. Approche d’une civilisation, 2 vols. Tunis: Les Éditions
de la Méditerranée.
Fentress, E. 2011. Slavers on chariots. In Dowler and Galvin 2011, 64–71.
Fentress, E. 2019. The archaeological and genetic correlates of Amazigh linguistics.
In Gatto et al. 2019, 495–524.
Flannery, K. and Marcus, J. 2012. The Creation of Inequality. How Our Prehistoric
Ancestors Set the Stage for Monarchy, Slavery and Empire. Cambridge (Mass):
Harvard University Press.
Fletcher, R. 1995 The Limits of Settlement Growth. A Theoretical Outline.
Cambridge: Cambridge University Press.
Gatto, M.C., Mattingly, D.J., Ray, N. and Sterry, M. (eds). 2019. Burials, Migration
and Identity in the Ancient Sahara and Beyond. Trans-Saharan Archaeology
Volume 2. Series editor D.J. Mattingly. Cambridge: Cambridge University Press.
Gautier, E.-F. 1905. Les oasis sahariennes. Algérie: Imp Fontana.
Gautier, E.-F. 1970. Sahara. The Great Desert (translated from the French by D.F.
Mayhew). London: Octagon.
Gauthier, Y. 2015. Pre-Islamic dry-stone monuments of the central and Western
Sahara. In C. Ruggles (ed.), Handbook of Archaeoastronomy and
Ethnoastronomy. New York: Springer Science, 1059–77.
Gauthier, Y. and Gauthier, C. 2011. Des chars et des Tifinagh: étude aréale et
corrélations. Les Cahiers de l’AARS 15: 91–118.
Gauthier, Y. and Gauthier, C. Forthcoming. L’impact du climat sur le peuplement
des zones lacustres sahariennes du Tchad septentrional (Borkou, Ennedi,
Tibesti) à l’Holocène. In Recherches croisées sur les écosystèmes lacustres tcha-
diens (coll. intern., N’Djamena, 25–27 avril 2017).
Goudie, A.S., Atkinson, B.W., Gregory, K.J., Simmons, I.G., Stoddart, D.R. and
Sugden, D. (eds). 1994. Encyclopedic Dictionary of Physical Geography. Second
Edition. Oxford: Blackwell.
Grébénart, D. 1985. Le néolithique final et les débuts de la métallurgie. La région d’In
Gall – Tegidda N Tesemt (Niger). Etudes Nigeriennes no. 49. Niamey: Institut de
Recherches en Sciences Humaines.
Gsell, S. 1972a. Histoire ancienne de l’Afrique du nord. Tome I Les conditions de
développement historique. Les temps primitifs, la colonisation phénicienne et
l’empire de Carthage. Osnabrück: Otto Zeller Verlag.
Gsell, S. 1972b. Histoire ancienne de l’Afrique du nord. Tome V Les royaumes indigènes,
orientation sociale, politique et économique. Osnabrück: Otto Zeller Verlag.
Gsell, S. 1933. La Tripolitaine et le Sahara au III siècle de notre ère. Mémoires de
l’Académie des Inscriptions et Belles Lettres 43.1: 149–66.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 43

Guey, J. 1939. Note sur le ‘limes’ romain de Numidie et le Sahara au IVe siècle.
Mélanges de 1‘Ecole Française de Rome 56: 178–248.
Hachid, M. 1998. Le Tassili des Ajjer. Aux sources de l’Afrique, 50 siècles avant les
pyramides. Paris: Paris-Méditerranée.
Hachid, M. 2000. Les premiertes berbères. Entre Méditerranée, Tassili et Nil. Aix-
en-Provence: Edisud.
Haour, A. 2003. One hundred years of archaeology in Niger. Journal of World
Prehistory 17.2: 181–234.
Jacques-Meunié, D. 1982. Le Maroc saharien des origins à 1670, 2 vols. Paris:
Klincksieck.
Johnson, E.A.J. 1970. The Organization of Space in Developing Countries.
Cambridge (Mass): Harvard University Press.
Junker, L.L. 2006. Population dynamics and urbanism in premodern island south-
east Asia. In G.R. Storey (ed.), Urbanism in the Preindustrial World: Cross-
Cultural Approaches. Tuscaloosa: The University of Alabama Press, 203–30.
Kallala, N. and Sanmartí, J. 2011. Althiburos I. La fouille dans l’aire du capitole et la
nécropole méridionale. Tarragona: Institut Català d’Arqueologia Clàssica.
Kelley, K.B. 1976. Dendritic central-place systems and the regional organization of
Navajo trading posts. In C.A. Smith (ed.), Regional Analysis. Volume 1:
Economic Systems. New York: Academic Press, 219–54.
Kuhlmann, H., Meggers, H., Freudenthal, T. and Wefer, G. 2004. The transition of
the monsoonal and the N Atlantic climate system off NW Africa during the
Holocene. Geophysical Research Letters 31.22.
Kuper, R. and Kröpelin, S. 2006. Climate-controlled Holocene occupation in the
Sahara: motor of Africa’s evolution. Science 313: 803–87.
Laporte, J.-P. 2011. Particularités de la province de Maurétanie Césarienne (Algerie
central et occidentale). In Briand-Ponsart, C. and Modéran, Y. (eds), Provinces et
identités provinciales dans l’Afrique romaine. Caen: Université de Caen, 111–50.
Larrasoaña J.C., Roberts A.P. and Rohling E.J. 2013. Dynamics of Green Sahara
periods and their role in hominin evolution. PLoS ONE 8.10: e76514.
Laronde, A. and Golvin, J.-C. 2001. L’Afrique antique. Histoire et monuments.
Paris: Tallader.
Laureano, P. 1991. Sahara. Jardin méconnu. Paris: Larousse.
Lehuraux, L. 1943. Les origines des oasis du Tidikelt et du Bas-Touat. Travaux de
l’Institut de Recherches Sahariennes 2: 105–20.
Lekson, S.H. 2009. A History of the Ancient Southwest. Santa Fe: School for
American Research Press.
Leschi, L. 1942. Rome et les nomades du Sahara central. Travaux de l’Institut de
Recherches Sahariennes 1: 47–62.
Leveau, P. 2018. Climat, sociétés et environment aux marges sahariennes du
Maghreb: Une approche historiographique. In S. Guédon (ed.), La
frontière méridionale du Maghreb. Approches croisées (Antiquité-Moyen Âge).
Bordeaux: Ausonius, 19–106.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 44 David J. Mattingly and Martin Sterry

Lhote, H. 1982. Les chars rupestres sahariennes des Syrtes au Niger, par le pays des
Garamantes et des Atalantes. Toulouse: Editions des Hespérides.
Lichtenberger, A. 2016. ‘Sea without water’ – conceptualising the Sahara and the
Mediterranean. In M. Dabag, D. Haller, N. Jaspert and A. Lichtenberger (eds),
New Horizons. Mediterranean Research in the 21st Century. Paderborn:
Ferdinand Schoningh, 267–83.
Liverani, M. (ed.). 2006. Aghram Nadharif. The Barkat Oasis (Sha‘abiya of Ghat,
Libyan Sahara) in Garamantian Times. Firenze: All’Insegna del Giglio.
Liverani, M. 2007. La struttura sociale dei Garamanti in base alle recenti scoperte
archeologiche. Rendiconti dell’Accademia Nazionale dei Lincei Ser 9, 18:
155–204.
Lutz, R. and Lutz, G. 1995. The Secret of the Desert. The Rock Art of the Messak
Settafet and Messak Mellet, Libya. Innsbruck: Golf Verlag.
Lydon, G. 2009. On Trans-Saharan Trails: Islamic Law, Trade Networks and Cross-
cultural Exchange in Nineteenth-Century Western Africa. Cambridge:
Cambridge University Press.
MacDonald, K. 2013. Complex societies, urbanism and trade in the western Sahel.
In Mitchell and Lane 2013, 829–44.
McIntosh, R. 2005. Ancient Middle Niger: Urbanism and the Self-Organizing
Landscape. Cambridge: Cambridge University Press.
McIntosh, R. 2015. Different cities: Jenne-jeno and African urbanism. In The
Cambridge World History: Volume 3, Early Cities in Comparative Perspective,
4000 BCE–1200 CE. Cambridge: Cambridge University Press, 364–80.
McIntosh, S.K. (ed.). 1995. Excavations at Jenne-jeno, Hambarketolo and Kaniana
(Inland Niger Delta, Mali): The 1981 Season. University of California
Monographs in Anthropology. Berkeley: University of California Press.
McIntosh, S.K. 1999. Beyond Chiefdoms: Pathways to Complexity in Africa.
Cambridge: Cambridge University Press.
McIntosh, S.K. and McIntosh, R.J. 1993. Cities without citadels: Understanding
urban origins along the Middle Niger. In T. Shaw, P. Sinclair, B. Andah, and
A. Okpoku (eds), The Archaeology of Africa: Food, Metals and Towns. London:
Routledge, 622–41.
Manning, K. and Timpson, A. 2014. The demographic response to Holocene
climate change in the Sahara. Quaternary Science Reviews 101: 28–35.
Marcus, J. and Sabloff, J.A. (eds). 2008. The Ancient City: New Perspectives on
Urbanism in the Old and New World. Santa Fe: School for Advanced Research
Press.
Marzano, A. 2011. Rank-size analysis and the Roman cities of the Iberian Peninsula
and Britain: some considerations. In Bowman and Wilson 2011, 196–228.
Mattingly, D.J. 2003. The Archaeology of Fazzan, Volume 1 Synthesis. London:
Society for Libyan Studies.
Mattingly, D.J. 2004a. Nouveaux aperçus sur les Garamantes: Un état saharien?
Antiquités africaines 37 (2001) [2004]: 45–61.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 45

Mattingly, D.J. 2004b. Surveying the desert: From the Libyan Valleys to Saharan
oases. In M. Iacovou (ed.), Archaeological Field Survey in Cyprus. Past History,
Future Potentials. Proceedings of a Conference Held at the University of Cyprus,
1–2 December 2000. London: BSA monograph 11, 163–76.
Mattingly, D.J. 2006. The Garamantes: The first Libyan state. In D. Mattingly,
S. McLaren, E. Savage, Y. al-Fasatwi and K. Gadgood (eds), The Libyan Desert.
Natural Resources and Cultural Heritage. London: Society for Libyan Studies,
189–204.
Mattingly, D.J. 2007. The Archaeology of Fazzan. Volume 2, Site Gazetteer, Pottery
and other Survey Finds. London: Society for Libyan Studies, Department of
Antiquities.
Mattingly, D.J. 2010. The Archaeology of Fazzan. Volume 3, Excavations Carried
out by C.M. Daniels. London: Society for Libyan Studies.
Mattingly, D.J. 2011a. Imperialism, Power and Identity: Experiencing the Roman
Empire. Princeton: Princeton University Press.
Mattingly, D.J. 2011b. The Garamantes of Fazzan: An early Libyan state with
Trans-Saharan connections. In Dowler and Galvin 2011, 49–60.
Mattingly, D.J. 2013. The Archaeology of Fazzan, Volume 4. Survey and Excavations
at Old Jarma (Ancient Garama) Carried out by C. M. Daniels (1962–69) and the
Fazzan Project (1997–2001). London: Society for Libyan Studies.
Mattingly, D.J. 2016. Who shaped Africa? The origins of agriculture and urbanism
in Maghreb and Sahara. In N. Mugnai, J. Nikolaus and N. Ray (eds), De Africa
Romaque. Merging Cultures across North Africa. London: Society for Libyan
Studies, 11–25.
Mattingly, D.J. and Hitchner, R.B. 1995. Roman Africa: An archaeological review.
Journal of Roman Studies 85: 165–213.
Mattingly, D.J. and MacDonald, K. 2013. Early cities: Africa. In P. Clark (ed.), The
Oxford Handbook of Cities in World History. Oxford: Oxford University Press,
66–82.
Mattingly, D.J. and Sterry, M. 2013. The first towns in the Central Sahara.
Antiquity 87.366: 503–18.
Mattingly, D.J., Rushworth, A., Sterry, M. and Leitch, V. 2013. The African
Frontiers. Frontiers of the Roman Empire series, general editor, D. Breeze.
Edinburgh: Hussar.
Mattingly, D.J., Sterry, M. and Edwards, D. 2015. The origins and development of
Zuwīla, Libyan Sahara: An archaeological and historical overview. Azania 50.1:
27–75.
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017a. Trade in the Ancient Sahara and Beyond. Trans-Saharan
Archaeology Volume 1. Series editor D.J. Mattingly. Cambridge: Cambridge
University Press.
Mattingly, D.J., Bokbot, Y., Sterry, M., Cuénod, A., Fenwick, C., Gatto, M., Ray, N.,
Rayne, L., Janin, K., Lamb, A., Mugnai, N. and Nikolaus, J. 2017b. Long-term

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 46 David J. Mattingly and Martin Sterry

history in a Moroccan oasis zone: The Middle Draa Project 2015. Journal of
African Archaeology, 15: 141–72.
Mattingly, D.J., Sterry, M., al-Haddad, M. and Bokbot, Y. 2018. Beyond the
Garamantes: The early development of Saharan oases. In Purdue et al. 2018a,
205–28.
Mattingly, D.J., Sterry, M. and Fothergill, B.T. Forthcoming. Animal traffic in the
Sahara. In V. Blanc-Bijon (ed.), XIe Colloque international Histoire et
Archéologie de l’Afrique du Nord. Marseille: Presses universitaires de Provence.
Mauny, R. 1956. Monnaies antiques trouvées en Afrique au sud du ‘limes’ romain.
Libyca 4: 249–61.
Mauny, R. 1978. Trans-Saharan contacts and the Iron Age in West Africa. In
J.D. Fage (ed.), The Cambridge History of Africa, Volume 2. Cambridge:
Cambridge University Press, 272–341.
Messier, R. and Miller, J. 2015. The Last Civilized Place. Sijilmasa and Its Saharan
Destiny. Austin: University of Texas Press.
Mitchell, P.J. 2005. African Connections: Archaeological Perspectives on Africa and
the Wider World. Walnut Creek: Altamira Press.
Mitchell, P. and Lane, P. 2013. The Oxford Handbook of African Archaeology.
Oxford: Oxford University Press.
Monroe, J.C. 2013. The archaeology of the precolonial state in Africa. In Mitchell
and Lane 2013, 702–22.
Mori, F. 1998. The Great Civilization of the Ancient Sahara. Rome: L’Erma di
Bretschneider.
Mori, L. (ed.). 2013. Life and Death of a Rural Village in Garamantian Times. The
Archaeological Research in the Oasis of Fewet (Libyan Sahara). Firenze:
All’Insegna del Giglio.
Morley, N. 2011. Cities and economic development in the Roman Empire. In
Bowman and Wilson 2011, 143–60.
Morris, I. 2006. The growth of Greek cities in the first millennium BC. In G.
R. Storey (ed.), Urbanism in the Preindustrial World: Cross-Cultural Approaches.
Tuscaloosa: The University of Alabama Press, 27–51.
Muzzolini, A. 1990. Au sujet de la datation des chars au gallop volant. Sahara 2:
115–18.
Nachtigal G. 1974. Sahara and Sudan, vol I, Tripoli and Fezzan, Tibesti or Tu.
Translated by A.G.B. and H.J. Fisher. London: C. Martin.
Nicholson, S.E. 1979. The methodology of historical climate reconstruction and its
application to Africa. Journal of African History 20.1: 31–49.
Nicolaisen, J. and Nicolaisen, I. 1997. The Pastoral Touareg, Ecology, Culture and
Society, 2 vols. London: Thames and Hudson.
Nijman, J. 2007. Introduction – comparative urbanism. Urban Geography
28.1: 1–6.
Osborne, R. and Cunliffe, B. (eds). 2005. Mediterranean Urbanization 800–600 BC.
Oxford: Oxford University Press.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 47

Paris, F. 1984. La région d’In Gall – Tegidda-n Tesemt (Niger). Programme

archéologique d’urgence 1977–1981. III. Les sépultures du néolithique final à
l’Islam. Niamey: Institut de Recherches en Sciences Humaines.
Paris, F. 1996. Les sépultures du Sahara nigérian, du Néolithique à l’Islamisation, 2
vols. Paris: Bondy.
Pascon, P. 1984. La Maison d’Iligh et l’histoire sociale du Tazerwalt. Rabat: Société
Marocaine des Editeurs Réunis.
Perrin, R.M.S. and Mitchell, C.W. 1967. Glossary of Local Physiographic and
Hydrogeological Terms about World Hot Deserts. Military Engineering
Experimental Establishment, Christchurch. Report No. 1124.
Potchter, O., Goldman, D., Kadish, D. and Iluz, D. 2008. The oasis effect in an
extremely hot and arid climate: The case of southern Israel. Journal of Arid
Environments 72.9: 1721–33.
Purdue, L., Charbonnier, J. and Khalidi, L. (eds). 2018a. Des refuges aux oasis: Vivre
en milieu aride de la Préhistoire à aujourd’hui. XXXVIIIe rencontres internatio-
nales d’archéologie et d’histoire d’Antibes. Antibes: Éditions APDCA.
Purdue, L., Charbonnier, J. and Khalidi, L. 2018b. Introduction. Living in arid
environments from prehistoric times to the present day: Approaches to the
study of refugia and oases. In Purdue et al. 2018a, 9–32.
Quinn, J.C. 2009. North Africa. In A. Erskine (ed.), A Companion to Ancient
History. Oxford: Blackwell, 260–72.
Quinn, J.C. 2018. In Search of the Phoenicians. Princeton: Princeton University
Press.
Rachet, M. 1970. Rome et les Berbères. Un probleme militaire d’Auguste à
Diocletian. Brussels: Collection Latomus.
Reader, J. 2004. Cities. London: William Heinemann.
Rebuffat, R. 1975. Graffiti en libyque de Bu Njem. Libya Antiqua 11–12: 165–87.
Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk Ramsey, C.,
Grootes, P.M., Guilderson, T.P., Haflidason, H., Hajdas, I., Hatté, C., Heaton, T.
J., Hoffmann, D.L., Hogg, A.G., Hughen, K.A., Kaiser, K.F., Kromer, B.,
Manning, S.W., Niu, M., Reimer, R.W., Richards, D.A., Scott, E.M.,
Southon, J.R., Staff, R.A., Turney, C.S.M. and van der Plicht, J. 2013. IntCal13
and Marine13 radiocarbon age calibration curves 0–50,000 years cal BP.
Radiocarbon 55.4: 1869–87.
Renfrew, C. 2008. The city through time and space. In J. Marcus and J.A. Sabloff
(ed.), The Ancient City: New Perspectives on Urbanism in the Old and New
World. Santa Fe: School for advanced Research Press, 29–52.
Reygasse, M. 1950. Monuments funéraires préislamiques de l’Afrique du Nord.
Paris: Arts et Métiers Graphiques.
Robinson, E.S.G. 1965. Catalogue of the Greek Coins of Cyrenaica. Vol. 29. Bologna:
A. Forni.
Rohlfs, G. 2003. Voyages et explorations au Sahara. Tome V, Koufra – les oasis de
Djofra et de Djalo 1878–1879 (translated by J. Debetz). Paris: Karthala.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 48 David J. Mattingly and Martin Sterry

Sahlins, M. and Service, E. (eds). 1960. Evolution and Culture. Ann Arbor:
University of Michigan Press.
Salama, P. 1981. The Sahara in classical antiquity. In G. Mokhtar (ed.), UNESCO
General History of Africa, II Ancient Civilisations of Africa. Paris: UNESCO,
513–32.
Sanmartí, J., Kallala, N., Belarte. M.C., Ramon, J., Telmini, B.M., Jornet, R. and
Miniaoui, S. 2012. Filling gaps in the protohistory of the eastern Maghreb: The
Althiburos Archaeological Project (el Kef, Tunisia). Journal of African
Archaeology 10.1: 21–44.
Scarin, E. 1934. Le oasi del Fezzan. Bologna: Nicola Zanichelli.
Scheele, J. 2010. Traders, saints, and irrigation: Reflections on Saharan
connectivity. The Journal of African History 51.3: 281–300.
Scheele, J. 2012. Smugglers and Saints of the Sahara: Regional Connectivity in the
Twentieth Century. Cambridge: Cambridge University Press.
Sèbe, A. 1989. Moula-Moula. Le Sahara à vol d’oiseau. Vidauban: Collection
Tagoulmoust.
Sèbe, A. and Sèbe, T. 2003. Sahara: The Atlantic to the Nile. London: Hachette.
Service, E. 1975. Origins of the State and Civilization: The process of cultural
evolution. New York: Norton.
Shanahan, T.M., Overpeck, J.T., Wheeler, C.W., Beck, J.W., Pigati, J.S., Talbot, M.
R., Scholz, C.A., Peck, J. and King, J.W. 2006. Paleoclimatic variations in West
Africa from a record of late Pleistocene and Holocene lake level stands of Lake
Bosumtwi, Ghana. Palaeogeography, Palaeoclimatology, Palaeoecology 242.3:
287–302.
Shaw, B.D. 1981. Fear and loathing: The nomad menace in Roman Africa. In C.
M. Wells (ed.), Roman Africa/L’Afrique Romaine. The 1980 Vanier lectures.
Ottowa: Les Presses de L’Université d’Ottawa, 29–50.
Shaw, B.D. 1983. ‘Eaters of flesh, drinkers of milk’: The ancient ideology of the
pastoral nomad. Ancient Society 13–14 (1982–1983): 5–31.
Sinclair, P. 2013. The archaeology of African urbanism. In Mitchell and Lane 2013,
689–702.
Smith, A.T. 2003. The Political Landscape: Constellations of Authority in Early
Complex Polities. Los Angeles: University of California Press.
Smith, C.A. 1976. Exchange systems and the spatial distribution of elites: The
organization of stratification in agrarian societies. In C.A. Smith (ed.), Regional
Analysis. Volume 2: Social Systems. New York: Academic Press, 309–74.
Smith, M. (ed.). 2003. The Social Construction of Ancient Cities. Washington, DC:
Smithsonian Books.
Smith, M.E. 2009. V. Gordon Childe and the urban revolution: an historical
perspective on a revolution in urban studies. Town Planning Review 80: 2–29.
Sterry, M. and Mattingly, D.J. 2011. DMP XIII: Reconnaissance survey of archae-
ological sites in the Murzuq area. Libyan Studies 42: 89–116.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 1 Introduction to Urbanisation and State Formation 49

Sterry, M, and Mattingly, D.J. 2013. Desert Migrations Project XVII: Further AMS
dates for historic settlements from Fazzan, south-west Libya. Libyan Studies 44:
127–40.
Sterry, M, Mattingly, D.J. and Higham, T. 2012. Desert Migrations Project XVI:
Radiocarbon dates from the Murzuq region, southern Libya. Libyan Studies 43:
137–47.
Stone, R.G. 1967. A desert glossary. Earth Science Reviews 3: 211–68.
Tainter, J.A. 2000. Problem solving: complexity, history, sustainability. Population
and Environment 22.1: 3–41.
Talbert, R.J.A. (ed.). 2000. Barrington Atlas of the Greek and Roman World.
Princeton: Princeton University Press.
Thiry, J. 1995. Le Sahara libyen dans l’Afrique du nord médiéval. Leuven: Orientalia
Lovaniensia Analecta, 72.
Toutain, J. 1896. Les Romains dans le Sahara. Mélanges de L’Ecole Française de
Rome 16: 63–77.
Trigger, B. 2003. Understanding Early Civilizations: A Comparative Study.
Cambridge: Cambridge University Press.
Trousset, P. 1982. L’image du nomade saharien dans l’historiographie antique.
Production pastorale et sociéte 10: 97–105.
Trousset, P. 2012. Nomadisme (Saharien en Afrique du nord dans l’antiquité).
Encyclopedie berbère fasc 34. Aix-en-Provence: Edisud, 5578–89.
Valloggia, M. 2004. Les oasis d’Egypte dans l’antiquité. Des origins au deuxième
millénaire avant J.-C. Bischheim: Infolio.
Vernet, R. and Aumassip, G. 1992. Le Sahara et ses marges: Paléoenvironnements et
occupation humaine à l’Holocène: Inventaire des datations 14C. Meudon: CNRS.
Villiers, M. de and Hirtle, S. 2002. Sahara. A Natural History. New York: Walker
and Company.
Welsby, D. 1996. The Kingdom of Kush. London: British Museum.
Welsby, D. 2013. Kerma and Kush and their neighbours. In Mitchell and Lane
2013, 751–64.
Wilson, A.I. 2005. Foggara irrigation, early state formation and Saharan trade: The
Garamantes of Fazzan. Schriftenreihe der Frontinus-Gesellschaft 26: 223–34.
Wilson, A.I. 2011. City sizes and urbanization in the Roman Empire. In Bowman
and Wilson 2011, 161–95.
Wilson, A.I. 2012. Saharan trade: Short-, medium- and long-distance trade net-
works in the Roman period. Azania: Archaeological Research in Africa 47.4:
409–49.
Wilson, A.I. and Mattingly, D.J. 2003. Irrigation technologies: Foggaras, wells and
field systems. In Mattingly 2003, 235–78.
Wilson, A.I., Mattingly, D.J. and Sterry, M. Forthcoming. The diffusion of irriga-
tion technologies in the Sahara in antiquity: Settlement, trade and migration. In
Duckworth et al. Forthcoming.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 50 David J. Mattingly and Martin Sterry

Yoffee, N. 2005. Myths of the Archaic State: Evolution of the Earliest Cities, States
and Civilizations. Cambridge: Cambridge University Press.
Yoffee, N. 2009. Making ancient cities plausible. Reviews in Anthropology 38:
264–89.
York, A.M., Smith, M.E. Stanley, B.W., Stark, B.L., Novic, J., Harlan, S.L.,
Cowgill, G.L. and Boone, C.G. 2011. Ethnic and class clustering through the
ages: A transdisciplinary approach to urban neighbourhood social patterns.
Urban Studies 48.11: 2399–415.

https://doi.org/10.1017/9781108637978.002 Published online by Cambridge University Press

 part ii

Oasis Origins in the Sahara:

A Region-by-Region Survey

Published online by Cambridge University Press

Published online by Cambridge University Press
 2 Garamantian Oasis Settlements in Fazzan
david j. mattingly, stefania merlo, lucia mori
and martin sterry

Introduction

In this chapter we present a case study of early oasis development relating

to the area of south-west Libya known as Fazzan and the ancient people
known as the Garamantes. The Garamantes were first firmly identified
with this area by Duveyrier in the mid-nineteenth century.1 As will be
explained in more detail below, the archaeological rediscovery of the
Garamantes properly commenced with a pioneering Italian mission in
1933.2 There were some large-scale excavations in the 1960s by
Mohammed Ayoub and importantly this included work at the
Garamantian capital of Old Jarma (ancient Garama), but the poor quality
of the work limited the value of the published outputs.3 More reliable
results were achieved by the team led by Charles Daniels between 1959
and 1977, excavating and surveying a number of sites in the area close to
Jarma.4 But, crucially, some of his most important results were not pub-
lished at the time and this allowed misnomers about the Garamantes to
persist well into the 1990s.5 Far from being a predominantly pastoral, tribal
1
Duveyrier 1864, 275–79. Barth 1857 had wanted to identify Jarma with Garama, but was not
convinced by what he saw, seeing the mausoleum as a southern outlier of the Roman
domininion.
2
Pace et al. 1951.
3
Ayoub 1967; 1968. For a review of the issues with his excavations see Mattingly 2010, 214–21;
2013, 67–71.
4
For some of his key summary publications, see Daniels 1969; 1970; 1971a; 1989.
5
To give a few examples, Connah 2004, a book on ‘Forgotten Africa’, does not mention the
Garamantes or the early emergence of oasis agriculture; Ehret 2002, 221–22 discusses the Berber
migration in the Sahara of the last two millennia BC in terms of scattered groups, without the
rise of any polity level units, emphasising trade and warfare as key factors and ignoring the
potential importance of oasis agriculture; Lassère 2015, 61, while acknowledging the possible
importance of the Garamantes, specifically denies that other Saharan peoples (Gaetules and
Aethiopes) were oasis cultivators. 53

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 54 David J. Mattingly et al.

society, the Garamantian kingdom needs reassessing as being founded on

multiple oasis communities (some of urban character) and arguably being
identifiable as an early state.6
Several distinct projects involving the authors of this chapter over the
last 20 years have contributed to this dramatic new assessment of the
Garamantes. With the exception of the Western Desert of Egypt, this is
archaeologically the most intensively investigated area of the Sahara. While
the proximity of the Western Desert to the Nile Valley can account for the
precocious development there, the implications of the work on the
Garamantes are paradigm changing precisely because Fazzan is a remote
Central Saharan location, at least 1,000 km from the nearest contemporary
urban polities.
Fazzan comprises three main east-west bands of oases, the Wadi ash-
Shati to the north, the Wadi al-Ajal in the centre and the Murzuq/Hufra
depression (ending in the ash-Sharqiyat area) to the south.7 There are two
outlying clusters of oases in north-south aligned depressions to the south-
west (Wadi Tanzzuft) and south-east (Wadi Hikma) that have also been
associated with the Garamantes.
The Garamantes first appear in history in Herodotus’ account of the
Libyan desert, but can be traced archaeologically to populations who were
already sedentary and practising agriculture by the early first
millennium BC.8 The earliest, and overall by far the most characteristic
archaeological evidence for the Garamantes comes from the Wadi al-Ajal,
close to Jarma. While there are cultural links with the other areas of Fazzan,
there are also differences. It is thus likely that the term ‘Garamantes’
evolved over time to accommodate the spread of the power of the group
based in the Wadi al-Ajal over neighbouring oases and Saharan commu-
nities. The development of oasis communities did not follow exactly the
same path or same chronology in Fazzan and, below, we review the
principal evidence for each of the zones defined above.
In the Neolithic period, much of Fazzan comprised a savannah land-
scape in between more verdant sand sea landforms, punctuated by many
small lakes, and with some notable upland features (mountains and

6
For developing argument about the Garamantes as a state, see Mattingly 2003, 346–62; 2004;
2006; 2013, 530–34. Hachid 2000, 92–194, despite being unaware of the latest archaeological
evidence, makes a good attempt to contextualise the Garamantes in Saharan and Berber long-
term history and to bring out their distinctiveness.
7
Key studies on the oases of Fazzan include Despois 1946; Gigliarelli 1932; Lethielleux 1948;
Sahara Italiano 1937; Scarin 1934. For a useful summary of Medieval sources on these sites, see
Thiry 1995.
8
Mattingly 2003, 76–90, for the literary sources,

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 55

plateaux) supplied by springs. Following major climatic change c.5000

years ago, hyper-arid desert conditions became established and over time
the lakes and springs diminished and dried up.9 Population groups that
had at one time been fairly mobile became tied to the remaining water
sources and congregated more and more in the lowest lying areas of the
landscape where water was still available at (or easily reachable from) the
surface. The later phases of rock art in Fazzan feature horses, chariots,
images of palm trees and finally camels, providing strong proxy evidence
for the emergence of oasis communities.10
Hydraulic conditions varied considerably across Fazzan. Artesian springs
are relatively uncommon apart from in the Wadi ash-Shati and the Ghat/
Tanzzuft area. However, the difficulty of enhancing the flow of the natural
springs in ash-Shati and Ghat with dug artesian wells may have limited the
scale of oasis development in these artesian areas for a considerable time. The
parts of Fazzan that show the most significant evidence of early oasis devel-
opment were those that have depended to a greater extent on an elevated water
table below the floor of the depression, which could be tapped by means of
shallow balance wells, or more elaborately by underground irrigation channels
(foggaras). Deeper animal powered wells (dalw) appear to be a development of
the Medieval period, by which time the level of the groundwater appears to
have fallen in some areas to a depth beyond the reach of the balance wells.

History of Research
A number of early European travellers passed through Fazzan in the
nineteenth century describing various features of historical interest and
remaining a valuable source, especially for the early modern period. The
first formal archaeological investigations in Fazzan took place in the winter
of 1933–1934 during the Italian colonial period as part of scientific mis-
sions into all aspects of life in Fazzan.11 The majority of the archaeological
work was in the Wadi al-Ajal where c.100 tombs were excavated in the
cemeteries close to Jarma.12 The mission also conducted a lightning tour of
the major centres to the south – Murzuq, Traghan and Zuwila,13 and
concluded their work with a short expedition to Ghat where they excavated
four tombs on the Quqaman hill overlooking the old caravan town

9
Cremaschi 2003.
10
Barnett 2019a; 2019b; Barnett and Mattingly 2003, 283–88, 301–17; di Lernia and Zampetti
2008.
11
Pace et al. 1951; for a summary see Mattingly 2003, 16–18. 12 Pace et al. 1951, 211–386.
13
Pace et al. 1951, 416–19.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 56 David J. Mattingly et al.

(medina).14 They also identified the first evidence of stone-footed and

mudbrick buildings at Jarma, Zinkekra and Qasr Larku (though were not
able to demonstrate conclusively the Garamantian date). A French expedi-
tion was carried out in the area of Ghat from 1944 to 1945, and Leschi
undertook a systematic excavation of the cemeteries at Barkat (related to
the fortified citadel of Aghram Nadharif) and Tin Alkum (now in Algeria),
whose results were mostly unpublished.15 The French geographers Despois
and Letheilleux surveyed some of the ruined villages and fortifications in
Fazzan.16 These include the first serious reports from the Wadi ash-Shati,
where Despois described Qasr Bin Aghenneb, north of Winzrik, compar-
ing it with the Tripolitanian qsur of Roman date.17 Despois suggested
a connection between a group of large rectangular qsur with projecting
angle towers and the foggara irrigation systems: ‘il est intéressant de
constater que presque partout, les vestiges de foggara voisinent avec les
ruines des enceintes fortifiées ou des forts’.18 Another French expedition
led by Pauphillet in 1949 excavated tombs in the eastern Wadi al-Ajal and
at Tajirhi in the Wadi Hikma.19
In 1961, Mohammed S. Ayoub, the controller of Antiquities in Fazzan,
started a series of excavations at cemeteries and settlements in the Wadi al-
Ajal, most notably Jarma itself between 1962 and 1967. Research into the
Garamantes was also conducted by Charles Daniels from 1958 to 1977.
Drawing on aerial photographs of the region, Daniels conducted the first
systematic surveys of the Wadi al-Ajal and Murzuq-Hufra depression and
also excavated at Zinkekra, Jarma, Saniat Jibril and a number of escarp-
ment settlements.20 Daniels also made a short visit to DabDab in the ash-
Shati where he recorded a cemetery, a foggara and a settlement. Daniels’
research established that, from the early first millennium BC, there were
permanent settlements whose inhabitants were practising oasis agriculture.
Between the 1960s and 1990s, Helmut Ziegert carried out a number of
excavations at settlements in Fazzan including Budrinna, Jarma, Zuwila
and at a suspected Garamantian settlement at Idri in the Wadi ash-Shati.21

14
Pace et al. 1951, 386–91; De Agostini 1934.
15
Leschi 1945. For an overview of the history of research on the Garamantes in the Wadi
Tanzzuft, see Gatto 2006; Mori 2013.
16
Despois 1946; Lethielleux 1948. 17 Despois 1946, 223.
18
Despois 1946, 57–61, quote at p. 57.
19
Bellair et al. 1953, 71–98; Bellair and Pauphillet 1959.
20
Daniels 1968; 1969; 1970; 1971a; 1971b; 1973; 1975; 1977; 1989; the final publication of Daniels
work took place after his death and can be found in Mattingly 2007; 2010; 2013.
21
Most of this research remains unpublished: Mattingly 2003, 20; Ziegert 1969. Note also the
summary account of the Garamantes in German, Ruprechtsberger 1997.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 57

The most recent phase of archaeological research into historic period

Fazzan began in 1997 with the renewal of the Italian-Libyan Archaeological
Mission in Messak and Akakus of Sapienza University of Rome under
Mario Liverani and the British-Libyan Fazzan Project under David
Mattingly. The Italian-Libyan Archaeological Mission, based in Ghat,
excavated two major sites at Aghram Nadharif and Fewet as well as several
smaller explorations of outlying sites in the Wadi Tanzzuft.22 The Fazzan
project excavated at Jarma and conducted systematic surveys in the Wadi
al-Ajal and in the western part of the Murzuq/Hufra depression.23
From 2007 to 2011, the Desert Migrations Project (DMP) followed up
the British-Libyan work, with a particular emphasis on burials, but also
with some exploratory survey in other areas of the main oasis sectors in
Fazzan. In 2007–2008, Merlo conducted a first systematic survey of the
Wadi ash-Shati as part of the DMP, confirming a long-term chronology of
occupation and a substantial Garamantian presence.24 In 2011, just prior to
the Libyan Civil War, Mattingly and Sterry led a short survey of the central
Murzuq/Hufra depression, again with important results.25
In the synthesis that follows, we present data from each of the main oases
zones of Fazzan in turn and draw attention both to similarities and con-
trasts in the sort of oasis landscapes and settlement histories that have been
recorded (Fig. 2.1). Cumulatively, the evidence from Fazzan makes an
incontestable case for large-scale pre-Islamic development of sedentary
oasis societies alongside Saharan pastoral groupings. Coupled with some
explicit historical source information, there is a wide range of archaeolo-
gical evidence (settlements, burials, irrigation technologies, material cul-
ture, rock art imagery, botanical remains) and above all scientific dating.
A total of >270 radiocarbon (mostly AMS) dates are now available from
Protohistoric and historic sites in several areas of Fazzan and these demon-
strate the scale and spread of oasis settlement in this region, as well as
providing a level of chronological detail that is unparalleled for this period
elsewhere in the Sahara (Table 2.1 summarises the data).26 Although, this
chapter will focus primarily on the Garamantian period, the list also
includes some relevant Medieval and early modern dates, as well as a few
from the late Pastoral phase.

22
Liverani 2006; Mori 2013. 23 Mattingly 2003; 2007; 2010; 2013.
24
Mattingly et al. 2007, Merlo et al. 2008, Merlo et al. 2013.
25
Sterry and Mattingly 2011; 2013; Sterry et al. 2012.
26
For previous overview of the dates, see Mattingly et al. 2018 and for subsets of the dates, see
Liverani 2006, 363–74; Mattingly 2007, 294–302; 2010, 78–82; 2013, 43–63, 125–34, 525–29;
Mori 2013, 66–69; Sterry and Mattingly 2013; Sterry et al. 2012.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 58 David J. Mattingly et al.

Figure 2.1. Fazzan showing main regions and sites discussed in the text. The brown dots
are certain or probable Garamantian oasis settlements.

The evidence reviewed here highlights four major stages of settlement

development in Fazzan that correlate with the Early Garamantian period
(1000–500 BC), the Proto-Garamantian period (500–1 BC), the Classic
Garamantian period (AD 1–300) and the Late Garamantian period (AD
300–700).

Wadi al-Ajal and Jarma (Ancient Garama)

The Wadi al-Ajal comprised the heartlands of the Garamantes (Fig. 2.2). It
is not a natural water course, but a 150-km-long, narrow depression,
mostly only a few kilometres wide, between the Ubari sand sea to the
north and a steep cliff-like sandstone escarpment of Massak Sattafat to the
south. There are no artesian spring sources in Wadi al-Ajal, but a high
water table, perhaps adventitiously still accessible at some points in the
form of phreatic springs in the first millennium BC, will have supported
palms and encouraged the development of additional irrigation measures.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 Table 2.1 Protohistoric and historic radiocarbon dates from the Central Sahara (Fazzan)

Area/Site Context Material Reference Date Calibrated

Wadi al-Ajal
TIN001 Hillfort, Area A2, building by gate Sorghum Beta-194236 2180±40 BP 371–113 calBC
TIN001 Area A2, building by gate Pearl millet Beta-194237 2130±40 BP 355–46 calBC
TIN001 Area A2, building by gate Barley OxA-9573 2117±37 BP 351–43 calBC
TIN001 Area A2, building by gate Barley OxA-9574 2074±40 BP 198 calBC–calAD 5
EDS010 Within amphora at qasr Grape OxA-X-2632–15 1536±25 BP calAD 427–586
EDS027 Qasr Wood OxA-X-2632–23 1260±25 BP calAD 670–860
TAG001 Cemetery, Area 3 Date stone OxA-X-2632–20 1592±26 BP calAD 408–539
TAG001 Cemetery, Area 3 Date stone OxA-29179 1106±23 BP calAD 890–989
TAG001 Cemetery, Area 3 Date stone OxA-29244 364±25 BP calAD 1451–1633
TAG001 Cemetery, Area 3 Date stone OxA-X-2632–21 309±25 BP calAD 1492–1648

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

TAG001 Cemetery, Area 5 Date wood? OxA-X-2555–52 1830±80 BP calAD 24–385
TAG001 Cemetery, Area 5 Date stone OxA-29180 1144±25 BP calAD 777–975
TAG006 Cemetery Wood OxA-X-2554–7 2366±27 BP 533–389 calBC
TAG006 Cemetery Date stone OxA-29245 2228±29 BP 382–204 calBC
TAG006 Cemetery Date stone OxA-29246 138±25 BP calAD 1671–1943
TAG006 Cemetery Date stone OxA-29247 141±28 BP calAD 1669–1945
TAG011 Qasr Wood OxA-9935 2055±70 BP 352 calBC–calAD 83
TAG021 Cemetery Date wood OxA-32219 1527±26 BP calAD 428–600
TAG063 Cemetery Human tissue OxA-32321 2283±33 BP 405–210 calBC
TAG063 Cemetery Date stone OxA-29248 2153±28 BP 356–94 calBC
ELH013 Mudbrick lining of foggara Unknown OxA-26746 2243±26 BP 391–206 calBC
ZIN003.5 Hillfort Unknown OxA-X-2632–27 2281±27 BP 402–231 calBC
ZIN013 Hillfort, context 10 Date stone OxA-3072 2620±70 BP 927–540 calBC
 Table 2.1 (cont.)

Area/Site Context Material Reference Date Calibrated

ZIN013 Hillfort, context 155 hearth Charcoal I-6323 2695±100 BP 1127–543 calBC
ZIN013 Hillfort, context 33 Date stone OxA-3073 2530±70 BP 807–430 calBC
ZIN013 Hillfort, context 51 hearth Charcoal I-6321 2595±90 BP 918–430 calBC
ZIN013 Hillfort, apsed room Charcoal OxA-6322 2560±110 BP 905–405 calBC
ZIN062 Hillfort, context 1 within building Wheat OxA-3070 2560±70 BP 835–430 calBC
ZIN071 Hillfort, accumulation within building Wheat OxA-3071 2670±70 BP 1010–591 calBC
ZIN071 Hillfort, mixed debris Charcoal, I-6341 2560±110 BP 905–405 calBC
ZIN105 Below hillfort, predates mudbrick Charcoal I-6324 2410±120 BP 803–209 calBC
building
ZIN105 W area in mudbrick building Barley OxA-3075 2490±70 BP 790–416 calBC
ZIN218 Occupation levels Date stone OxA-X-2632–19 2515±27 BP 792–542 calBC
ZIN218 Occupation levels Date stone OxA-29242 2510±30 BP 791–540 calBC
ZIN218 Occupation levels Date stone OxA-X-2633–16 2328±28 BP 474–262 calBC

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

ZIN220 Cemetery plant remains OxA-X-2635–54 2567±27 BP 806–570 calBC
ZIN220 Cemetery Textile OxA-X-2557–9 2067±26 BP 171–1 calBC
ZIN337 Hearth in building Barley OxA-3074 2620±70 BP 927–540 calBC
ZIN700 Beneath large cairn Date stone OxA-29243 2295±28 BP 405–234 calBC
UAT004 Burial cairn Seeds celtis sp. OxA-29241 2173±30 BP 361–119 calBC
UAT008 Cemetery Human tissue OxA-32324 2242±30 BP 391–206 calBC
UAT008 Cemetery Date stone OxA-X-2632–24 1948±27 BP 20 calBC–calAD 126
UAT008 Cemetery Human tissue OxA-32223 1917±26 BP calAD 23–133
UAT008 Cemetery Textile OxA-X-2635–52 1913±27 BP calAD 20–206
UAT008 Cemetery Unknown OxA-32322 1886±30 BP calAD 60–218
UAT008 Cemetery Human tissue OxA-32222 1867±28 BP calAD 76–227
UAT008 Cemetery other unknown OxA-32323 1852±29 BP calAD 85–234
UAT050 Cemetery Date stone OxA-32214 1719±27 BP calAD 250–390
 UAT050 Cemetery Date stone OxA-32213 1716±28 BP calAD 251–392
UAT050 Cemetery Charcoal OxA-32215 1279±26 BP calAD 671–771
UAT051 Cemetery Charcoal OxA-32293 1425±26 BP calAD 585–658
UAT056 Cemetery Charcoal OxA-32294 1006±25 BP calAD 985–1147
UAT056 Cemetery Charcoal OxA-32295 966±27 BP calAD 1018–1155
GER001 Old Jarma, Garamantian capital Charcoal Hv-4805 2700±100 BP
GER001 G1 Date wood OxA-32292 2112±29 BP 204–48 calBC
GER001 G1 (542) 5-late Barley OxA-26708 1837±28 BP calAD 86–244
GER001 G1 (469) 4-mid Barley OxA-26749 1774±25 BP calAD 142–338
GER001 G1 (468) 4-early Barley OxA-27254 1685±24 BP calAD 260–415
GER001 G1 (391) 4-late Grape OxA-26748 1576±24 BP calAD 420–543
GER001 G1 (389) 3-mid Barley OxA-27005 1542±25 BP calAD 427–575
GER001 G1 (283) 3-early Barley OxA-26747 1274±25 BP calAD 673–772
GER001 G1 (1009) 10-early Date stone OxA-12707 2247±31 BP 394–206 calBC
GER001 G1 (358) 3-late Date stone OxA-10340 122±35 BP calAD 1677–1941
GER001 G1 (375) 2-early Cucurbitaceae OxA-10341 812±35 BP calAD 1165–1271

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

GER001 G1 (376) 4-late Date stone OxA-10342 204±34 BP calAD 1643–. . .
GER001 G1 (391) 4-late Barley OxA-10343 1657±39 BP calAD 257–535
GER001 G1 (402) 3-mid Date stone OxA-12701 1584±28 BP calAD 411–543
GER001 G1 (4031) 8-late Date stone OxA-12705 2224±28 BP 379–204 calBC
GER001 G1 (468) 4-early Barley OxA-10344 1730±37 BP calAD 231–398
GER001 G1 (535) 5-late Cotton Beta-194241 1600±40 BP calAD 383–557
GER001 G1 (542) 5-late Barley grain OxA-10347 1776±37 BP calAD 133–345
GER001 G1 (572) 5-late Date stone OxA-10345 1661±35 BP calAD 257–532
GER001 G1 (586) 3-early Wheat OxA-10346 1554±36 BP calAD 417–582
GER001 G1 (607) 5-mid Pearl millet OxA-10290 1930±80 BP 160 calBC–calAD 314
GER001 G1 (61) 1-early Pumpkin or OxA-10288 220±55 BP calAD 1518–. . .
squash
GER001 G1 (611) 5-early Date stone OxA-10352 1690±38 BP calAD 252–420
 Table 2.1 (cont.)

Area/Site Context Material Reference Date Calibrated

GER001 G1 (65) 1 to 2 Maize OxA-10339 95±37 BP calAD 1681–1938

GER001 G1 (652) 5-late Date stone OxA-10349 1269±35 BP calAD 663–865
GER001 G1 (655) 5-late Date stone OxA-10350 1285±36 BP calAD 656–857
GER001 G1 (661) 6-late Cotton Beta-194240 1770±40 BP calAD 135–379
GER001 G1 (661) 6-late Barley OxA-10351 1727±36 BP calAD 236–396
GER001 G1 (665) 6-early Barley OxA-10289 1670±55 BP calAD 243–535
GER001 G1 (678) 6-mid Date stone OxA-10353 1764±37 BP calAD 141–381
GER001 G1 (686) 6-late Date stone OxA-10354 1657±37 BP calAD 257–534
GER001 G1 (753) 4-late Charcoal OxA-12709 950±26 BP calAD 1025–1154
GER001 G1 (82) 1 to 2 Maize OxA-10338 161±35 BP calAD 1663–. . .
GER001 G1 (824) 6-early Date stone OxA-12702 1817±30 BP calAD 125–322
GER001 G1 (864) 8-late Pearl millet Beta-194239 1930±40 BP 40 calBC–calAD 209
GER001 G1 (925) 7-late Date stone OxA-12703 1856±29 BP calAD 83–232

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

GER001 G1 (951) 9-early Date stone OxA-12706 2219±29 BP 373–203 calBC
GER001 G1 (951) 9-early Pearl millet Beta-194242 2160±40 BP 361–92 calBC
GER001 G1 (955) 10-late Date stone OxA-12713 2246±27 BP 391–207 calBC
GER001 G1 (962) 9-early Date stone OxA-12712 2133±28 BP 350–54 calBC
GER001 G1 (963) 7-late Date stone OxA-12704 1960±30 BP 40 calBC–calAD 121
GER001 G1 (978) 8-early Charcoal OxA-12708 2123±28 BP 346–52 calBC
GER001 G2 section (cf. 2057) 5-late Unknown unknown 1285±95 BP calAD 598–970
GER001 G2 section (2057) 5-late Date stone OxA-12710 1303±26 BP calAD 660–769
GER001 G2 section (2057) 5-late Grape OxA-12711 1322±26 BP calAD 653–766
GER001 G3 (3014) 1 to 2 Pearl millet OxA-10348 170±34 BP calAD 1656–. . .
GER001.4 (219) Date stone OxA-9575 2260±34 BP 399–207 calBC
GER001.5 Below wall Charcoal OxA-9576 1928±34 BP 37 calBC–calAD 138
GER001.5 Below wall Charcoal OxA-9820 1901±38 BP calAD 24–219
 GER001.6 In mosque foundations Charcoal OxA-9632 921±36 BP calAD 1026–1203
GER001.9 City wall plant remains OxA-26494 972±29 BP calAD 1016–1155
tamarisk?
GER001.9 City wall Charcoal OxA-9933 1135±45 BP calAD 774–990
GER001.13 D-shaped tower Charcoal OxA-9934 270±40 BP calAD 1486–. . .
GER001.15 Below pier of mosque Date stone OxA-9577 1062±34 BP calAD 895–1025
GER001.15 Below pier of mosque Date stone OxA-9578 716±35 BP calAD 1225–1386
GER001.34 NW gate Date stone OxA-26739 194±23 BP calAD 1657–. . .
GER001.65 Qasaba tower Twigs OxA-9769 492±34 BP calAD 1330–1454
GER001.65 Qasaba wall Date stone OxA-9579 532±32 BP calAD 1318–1440
GSC030 Royal Cemetery, tomb 8 (original burial) Date stone OxA-29181 1555±29 BP calAD 422–567
GSC030 Royal Cemetery, tomb 8 (possible Date stone OxA-29249 206±27 BP calAD 1648–. . .
robbing)
GSC031 Royal Cemetery, XX Date stone OxA-29250 1747±27 BP calAD 235–381
Tuwash area Burial from wadi collected by Ulrich Unknown Bonn-145 1950±60 BP 92 calBC–calAD 220

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

Hallier, university Bonn
Wadi al-Ajal Ziegert – unknown context Unknown Hv-5479 2115±130 BP 415 calBC–calAD 214
Wadi al-Ajal Ziegert – unknown context Unknown Hv-6980 1965±420 BP 1018 calBC–calAD 889
Wadi al-Ajal Ziegert – unknown context Unknown Hv-5487 1830±85 BP
Wadi al-Ajal Ziegert – unknown context Unknown Hv-5486 660±165 BP calAD 1020–1632
FUG022 Qasr Date stone OxA-26737 399±23 BP calAD 1440–1619
FUG022 Qasr Date stone OxA-26738 374±23 BP calAD 1448–1630
CHA031 Infill of foggara channel Charcoal OxA-9581 modern± BP calAD 1896–1904
TEK010 Qasr (medieval type) Charcoal OxA-10100 1110±40 BP calAD 778–1018
FJJ013 Qasr Unknown OxA-26493 1645±31 BP calAD 267–535
FJJ056 Qasr Date stone OxA-26736 1687±25 BP calAD 258–415
GRE015 Qasr Date stone OxA-26751 1581±25 BP calAD 417–541
GRE015 Qasr Date stone OxA-26750 1542±25 BP calAD 427–575
 Table 2.1 (cont.)

Area/Site Context Material Reference Date Calibrated

LEK018 Qasr (medieval type) Date stone OxA-9821 404±32 BP calAD 1434–1625
LEK021 Qasr Charcoal OxA-9823 1632±35 BP calAD 342–536
Budrinna Child (mummified)? Human tissue KI-394 1470±110 BP calAD 340–770
Budrinna Child (mummified)? Human tissue KI-396 1330±40 BP calAD 643–770
Budrinna Child (mummified) Human tissue KI-392 1360±60 BP calAD 563–774
ZOU015 Fortified stone village Twigs OxA-9822 279±32 BP calAD 1496–1796
GBD001 Qasr Charcoal OxA-9853 1560±45 BP calAD 405–595
GBD001 Enceinte Date stone OxA-9580 1614±35 BP calAD 356–543
GBD002 Qasr Charcoal OxA-9941 1654±35 BP calAD 259–534
GBD007 Qasr Charcoal OxA-9825 1684±35 BP calAD 255–423
BNH005 Qasr (medieval) preserved timber OxA-9824 582±35 BP calAD 1298–1419
Massak Sattafat Unknown context Charcoal UGAMS-5854 2980±25 BP 1277–1121 calBC

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

Massak Sattafat Unknown context Charcoal UGAMS-5857 1790±25 BP calAD 136–326
Wadi Barjuj/Wadi
Utba
MAR001 Qasr Tamarisk OxA-9940 1600±40 BP calAD 383–557
SCH020 Qasr ah-Sharraba, qasr B Date stone OxA-26740 1687±24 BP calAD 259–413
SCH020 Qasr ah-Sharraba, qasr A Date stone OxA-26742 1509±24 BP calAD 432–615
SCH020 Qasr ah-Sharraba, NW tower of fort Plant remains OxA-26741 795±23 BP calAD 1210–1274
SCH020 Qasr ah-Sharraba, build-up within Charcoal OxA-9826 918±32 BP calAD 1029–1187
qasr A
SCH020 Qasr ah-Sharraba, qasr A Charcoal OxA-9631 1721±39 BP calAD 235–405
SCH020 Qasr ah-Sharraba, qasr A Charcoal OxA-9942 1431±33 BP calAD 569–659
 Murzuq Basin
MZQ001 Old Murzuq, S side town wall Cereal OxA-26492 415±31 BP calAD 1427–1620
MZQ001 Old Murzuq, house in abandoned S area Unknown OxA-26735 380±23 BP calAD 1446–1626
MZQ001 Old Murzuq, SW city wall Date stone OxA-26733 1.04707±0.00289 BP calAD 1896–1904
MZQ001 Old Murzuq, SW city wall Date stone OxA-25826 1.06715±0.00272 BP calAD 1896–1904
MZQ001 Old Murzuq, D-shaped tower, S town Unknown OxA-26734 71±22 BP calAD 1695–1919
wall
MZQ001 Old Murzuq, D-shaped tower, town wall Date stone OxA-25827 98±21 BP calAD 1691–1925
MZQ001 Old Murzuq, late buttress, qasaba Date stone OxA-25828 143±22 BP calAD 1669–1945
MZQ007 Qasr Wood OxA-26725 1621±23 BP calAD 386–535
MZQ007 Qasr (medieval phase) Acacia OxA-25825 572±24 BP calAD 1308–1418
MZQ021 Qasaba in fortified village Unknown OxA-25796 477±26 BP calAD 1411–1450
MZQ021 Enceinte of fortified village Rodent bone OxA-X-2475–37 97±26 BP calAD 1687–1927
MZQ031 Dispersed village Tamarisk OxA-25797 270±28 BP calAD 1520–1798
HHG001 Gate of central qasr Unknown OxA-26726 1578±24 BP calAD 420–541
HHG001 SW corner of central qasr Date stone OxA-25821 1577±23 BP calAD 421–541
HHG001 SW corner outer enceinte Unknown OxA-26487 1581±30 BP calAD 410–545

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

HHG006 Qasr Unknown OxA-26490 1840±38 BP calAD 76–314
HHG007 Qasr Date stone OxA-25829 1772±26 BP calAD 142–340
HHG008 Qasr Unknown OxA-26728 1614±24 BP calAD 392–536
HHG012 Qasaba of fortified village Date stone OxA-25831 543±21 BP calAD 1321–1431
HHG012 Mosque in fortified village Date wood OxA-25830 530±22 BP calAD 1327–1436
HHG013 Qasr Cereal OxA-26488 1714±31 BP calAD 249–395
HHG014 Qasr Date wood OxA-26727 1630±24 BP calAD 348–535
HHG014 Qasr Date wood OxA-25832 1585±23 BP calAD 416–540
HHG014 Late disturbance/fire on qasr Unknown OxA-26489 1.36896±0.00437 BP calAD 1896–1904
GAT001 Qasr Unknown OxA-25833 1582±22 BP calAD 419–539
GAT007 Qasr Date stone OxA-25834 117±21 BP calAD 1682–1936
GAT010 Qasr, buttressing NE tower Date stone OxA-25823 1580±24 BP calAD 420–540
GAT010 Qasr, E wall Olive? OxA-25822 1577±22 BP calAD 421–541
 Table 2.1 (cont.)

Area/Site Context Material Reference Date Calibrated

GAT012 Qasr Cereal OxA-26491 1783±29 BP calAD 136–333

ZZW013 SE tower secondary qasr Date stone OxA-25824 1644±26 BP calAD 337–532
ZZW013 SE tower secondary qasr Cereal OxA-26729 1589±23 BP calAD 414–539
ZZW014 Qasr Unknown OxA-26731 1587±24 BP calAD 414–540
ZZW016 Qasr Unknown OxA-25869 1662±25 BP calAD 266–426
ZZW018 SE tower qasr Date stone OxA-25870 1730±25 BP calAD 247–383
ZZW018 SE corner of qasr Unknown OxA-26732 1552±24 BP calAD 426–563
ZZW101 Qasr Unknown OxA-26730 1507±25 BP calAD 432–619
Zuwila
ZUL001 Zuwila, pisé wall of citadel Date stone OxA-26743 1065±23 BP calAD 900–1020
ZUL002 Zuwila, ‘White mosque’ Date stone OxA-26744 1260±24 BP calAD 670–857
ZUL003 Zuwila, tombs of Banu Khattab Unknown OxA-26745 1029±24 BP calAD 976–1031
ZUL003 Zuwila, tombs of Banu Khattab Unknown OxA-26495 1038±27 BP calAD 904–1032

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

Wadi Tanzzuft
Wadi Tanzzuft Burial 00/195 Gx-28482 3530±40 BP 1971–1745 calBC
Wadi Tanzzuft Burial 00/195bis Gx-28483 3030±40 BP 1407–1131 calBC
Wadi Tanzzuft Burial 00/98 T4 Gx-28440 2310±40 BP 482–209 calBC
Wadi Tanzzuft Burial 96/129 S2 Gx-28441 3370±40 BP 1752–1533 calBC
Wadi Tanzzuft Burial 96/129 T1 Gx-28476 3850±40 BP 2461–2204 calBC
Wadi Tanzzuft Burial 96/129 T1 Gx-26255 3080±40 BP 1431–1231 calBC
Wadi Tanzzuft Burial 96/129 T10 Gx-28479 2720±190 BP 1386–406 calBC
Wadi Tanzzuft Burial 96/129 T2 Gx-26256 2910±40 BP 1224–980 calBC
Wadi Tanzzuft Burial 96/129 T3 Gx-27386 3760±40 BP 2293–2036 calBC
Wadi Tanzzuft Burial 96/129 T3bis Gx-28477 2930±40 BP 1258–1011 calBC
Wadi Tanzzuft Burial 96/129 T4 Gx-28478 2550±30 BP 801–551 calBC
Wadi Tanzzuft Burial 97/5 RT Date stone Gx-27385 1700±40 BP calAD 246–416
 Wadi Tanzzuft Burial 97/5 SF 25 Charcoal Gx-27383 1850±80 BP 18 calBC–calAD 380
Wadi Tanzzuft Burial 97/5 SF 7 Charcoal Gx-27384 1740±40 BP calAD 174–400
Wadi Tanzzuft Burial 97/5 T1 Gx-28450 4280±100 BP 3326–2580 calBC
Wadi Tanzzuft Burial T43 Gx-28481 3310±40 BP 1687–1503 calBC
Wadi Tanzzuft Burial T6 Gx-28480 3330±40 BP 1731–1511 calBC
Wadi Tanzzuft THA4 Charcoal GX-20011-AMS 3890±60 BP 2564–2155 calBC
Fewet Settlement Fw1 Gx-28452 2950±70 BP 1390–943 calBC
Fewet Settlement Fws1b3 Gx-30340-AMS 2230±50 BP 396–186 calBC
Fewet Settlement Fw23 Gx-30339 2260±70 BP 507–114 calBC
Fewet Settlement FwS1b2 Gx-30335-AMS 2070±40 BP 195 calBC–calAD 16
Fewet Settlement Fw5 UGAMS-13136 2020±20 BP 88 calBC–calAD 50
Fewet Settlement Fw6 UGAMS-13137 1990±25 BP 44 calBC–calAD 63
Fewet Settlement Fw4 Gx-30341-ext 2060±70 BP 353 calBC–calAD 79
Fewet Settlement Fw13 Gx-30337-AMS 2000±40 BP 111 calBC–calAD 83
Fewet Settlement Fw16 Gx-30343-AMS 1980±40 BP 88 calBC–calAD 124

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

Fewet Settlement Fw9 Gx-30342-AMS 1970±40 BP 50 calBC–calAD 125
Fewet Settlement Fw14 Gx-30338-AMS 1940±40 BP 45 calBC–calAD 136
Fewet Settlement Fw28 Gx-31078-AMS 1950±70 BP 151 calBC–calAD 233
Fewet Settlement FW11 Gx-30336-AMS 1890±60 BP 38 calBC–calAD 252
Fewet Settlement kiln Gx-31081 710±60 BP calAD 1210–1400
Fewet necropolis T1287 H4 UGAMS-02205 3590±40 BP 2117–1779 calBC
Fewet necropolis T669 H1 UGAMS-02203 2520±40 BP 798–521 calBC
Fewet necropolis T41 H1 UGAMS-02204 2390±40 BP 747–389 calBC
Fewet necropolis T1210 H1 UGAMS-08704a 2320±25 BP 411–361 calBC
Fewet necropolis T719 H1 UGAMS-02200 2360±70 BP 760–231 calBC
Fewet necropolis T1226 H1 UGAMS-02209 2290±40 BP 408–208 calBC
Fewet necropolis T1197 H1 UGAMS-02201 2230±40 BP 388–202 calBC
Fewet necropolis T1191 UGAMS-02210 2220±40 BP 387–197 calBC
 Table 2.1 (cont.)

Area/Site Context Material Reference Date Calibrated

Fewet necropolis T669 H1 LTL-13707A 2220±45 BP 391–184 calBC

Fewet necropolis T1261 H1 UGAMS-02206 2160±40 BP 361–92 calBC
Fewet necropolis T1223 H1 UGAMS-8705a 2120±25 BP 334–53 calBC
Fewet necropolis T1223 H1 UGAMS-8705 2060±25 BP 166 calBC–calAD 1
Fewet necropolis T715 H1 LTL-13441A 1962±40 BP 45 calBC–calAD 125
Fewet necropolis T976 H1 UGAMS-02208 1960±40 BP 43 calBC–calAD 125
Fewet necropolis T914 H1 UGAMS-08703 1940±25 BP calAD 7–126
Fewet necropolis T1210 H1 UGAMS-08704 1900±25 BP calAD 32–210
Fewet necropolis T954 H1 UGAMS-02202 1920±40 BP 19 calBC–calAD 214
Fewet necropolis T716 H1 UGAMS-08702a 1840±25 BP calAD 88–240
Fewet necropolis T914 H1 UGAMS-08703a 1810±25 BP calAD 130–317
Fewet necropolis T716 H1 UGAMS-08702 1740±25 BP calAD 240–381
Aghram Nadharif Citadel Gx-27380 2380±50 BP 751–377 calBC

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

Aghram Nadharif Citadel Gx-28443-AMS 2250±40 BP 397–204 calBC
Aghram Nadharif Citadel Gx-27379 2290±90 BP 750–109 calBC
Aghram Nadharif Citadel GX-2090-AMS 2090±30 BP 195–42 calBC
Aghram Nadharif Citadel Gx-28453 1980±60 BP 166 calBC–calAD 134
Aghram Nadharif Citadel Gx-27382 1990±70 BP 194 calBC–calAD 208
Aghram Nadharif Citadel Gx-28444-AMS 1930±40 BP 40 calBC–calAD 209
Aghram Nadharif Citadel Gx-28445-AMS 1920±40 BP 19 calBC–calAD 214
Aghram Nadharif Citadel Gx-24267 1880±75 BP 42 calBC–calAD 329
Aghram Nadharif Citadel Gx-24268 1760±70 BP calAD 87–416
Aghram Nadharif Citadel Gx-28442 1590±40 BP calAD 391–560
Aghram Nadharif Citadel Gx-27920-AMS 1580±40 BP calAD 397–565
Aghram Nadharif Citadel Gx-27029 1610±90 BP calAD 241–630
Aghram Nadharif Citadel Gx-31080 770±70 BP calAD 1046–1390
 Aghram Nadharif Citadel Gx-27381 670±60 BP calAD 1251–1411
el-Barkat I Burial Sa-92 1330±120 BP calAD 431–975
el-Barkat II Burial Sa-93 1680±150 BP calAD 29–649
Tadrart Akakus
Adad Qasr Unknown Gx-30334-AMS 2090±40 BP 333 calBC–calAD 2
Adad Qasr Unknown Gx-30329-AMS 2040±40 BP 167 calBC–calAD 51
Imessarajen Qasr Unknown Gx-30331-AMS 2060±30 BP 170 calBC–calAD 4
Imessarajen Qasr Unknown Gx-30334-AMS 2050±40 BP 174 calBC–calAD 49
Imessarajen Qasr Unknown Gx-30332-AMS 1950±40 BP 41 calBC–calAD 129
Imessarajen Qasr Unknown Gx-30328-ext 2020±100 BP 356 calBC–calAD 220
Imessarajen Qasr Unknown Gx-30330-AMS 1550±40 BP calAD 418–594
Erg Uan Kasa 94/106 Wood Gx-20710 2375±110 BP 791–206 calBC
Uan Amil Rock shelter Dung Bo-341 1260±60 BP calAD 655–891

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

Uan Muhuggiag Rock shelter deposit B/1 Date stone OxA-4389 2130±70 BP
Uan Muhuggiag Rock shelter deposit B/1 citrullus Gd-4290 2220±220 BP 807 calBC–calAD225
colocynthis
Uan Muhuggiag Rock shelter deposit B/1 Coprolite Gd-4288 2770±80 BP 1127–798 calBC
Uan Muhuggiag Rock shelter wall Dung Enea 3195±70 BP 1627–1292 calBC
Uan Muhuggiag Rock shelter Tr 1A Level 2 Animal bone OxA-17909 2925±27 BP 1214–1030 calBC
Uan Muhuggiag Rock shelter Tr 1A Level 2 Animal bone OxA-17960 2915±28 BP 1209–1017 calBC
Wadi Hikma
Tajirhi Burial Leather Sa-78 1190±120 BP calAD 615–1118
 70 David J. Mattingly et al.

Figure 2.2. Garamantian settlement (and probable Garamantian settlement) in the

Wadi al-Ajal.

Jarma itself sits adjacent to a spring which in Medieval times still served to
fill the moat around its walls. However, mechanical exploitation of ground-
water has always been the predominant irrigation strategy here. In the
nineteenth century water was still to be found in places at a depth of only
a few metres, easily reachable with simple balance wells (the shaduf). The
most remarkable aspect of Garamantian irrigation in the Wadi al-Ajal (and
to a lesser extent in other areas) was its strong reliance on more advanced
technology, the foggara irrigation systems. These are underground water
channels similar to the Near Eastern qanat and the technology was almost
certainly transferred from Persia to Egypt in the sixth–fifth centuries BC
and from there along the oasis chain that links the Western Desert with
Fazzan.27 Examples in the Western Desert oases have been dated to the
fifth century BC and, as we shall see, one example in Fazzan has now been
dated scientifically to the fourth–third century BC.28
Along the southern escarpment of the Wadi al-Ajal there are notable
concentrations of pre-Islamic burial monuments – variously estimated at
60,000–120,000 or even 250,000.29 Although the Garamantian dates of the

27
Wilson 2006; 2009; Wilson and Mattingly 2003.
28
See Wilson et al. Forthcoming, for full presentation of the date.
29
These comprise a mixture of circular shafts, cairns, drum tombs and mudbrick constructions,
Pace et al. 1951, 210–12; Daniels 1989, 49. Recent surveys in some of these cemeteries suggest
that the older estimates of numbers were significantly undercounted, Mattingly et al. 2008;
2009; 2010a; 2010b; 2011. For a summary account, Mattingly et al. 2019.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 71

settlements at Zinkekra, Jarma and Saniat Jibril were first demonstrated in

the 1960s–1970s, until recently almost no other habitation sites had been
conclusively assigned to the Garamantian era. Based largely on Daniels’
work, the Barrington Atlas sheet compiled by Mattingly in the late 1990s
showed only 10 settlements and 32 major cemeteries in the Wadi al-Ajal.30
Subsequent fieldwalking in the oasis has identified the existence of many
sites, though surface traces are not always distinct enough to clarify whether
we are dealing with settlements or cemeteries. However, many sites have
some upstanding remains, in particular of fortified square buildings (qsur),
and surface pottery and AMS dates have revealed a surprising number of
these to predate the Medieval period. We can now recognise c.100
Garamantian villages and hamlets in the Wadi al-Ajal, with very high
densities in two areas where our survey work has been more intensive (the
Taqallit to al-Hatiya sector of the wadi, c.30 km west of Jarma, and the
immediate environs of Jarma).31 There was a complex chronological
sequence of development here, from a relatively small number of sites that
appear to have originated in the early first millennium BC, to a major phase
of expansion in the latter centuries BC, consolidation in the early
centuries AD and apparently some contraction and concentration on for-
tified oasis villages in late antiquity.
There was an initial (Early Garamantian) phase of hillfort activity at
suitable sites along the escarpment edge (Fig. 2.3).32 The type site is
Zinkekra,33 where a settlement was established by the early first
millennium BC, associated from the beginning with an agricultural pack-
age including wheat, barley and the grape vine alongside the date palm.34
The early buildings at the site were oval and sub-rectangular huts with one
to three rooms, but by the later centuries BC, the site took on a more proto-
urban character with mudbrick construction of complex multi-roomed
rectilinear buildings and surrounding walls and embankments.35

30
Mattingly 2000 for the Barrington Atlas map. The total number of cemetery sites is actually
much larger than portrayed there (Mattingly 2007, includes >300 cemeteries of Garamantian,
or probable Garamantian, date from the Wadi al-Ajal alone) and this always made it probable
that significant numbers of contemporary settlements would be found when searched for. The
Fazzan Project work (1997–2001) and subsequent work as part of the Desert Migrations Project
and the Trans-SAHARA Project has dramatically added to this total, see Mattingly 2013,
525–34.
31
Mattingly 2007, 64–167; 2013, 525–29; Mattingly et al. 2010b, 117–31 (Taqallit).
32
About 20 such sites are now known, though not all can be closely dated, Mattingly 2003,
136–42; 2010, 19–119.
33
Daniels 1968; Hawthorne et al. 2010, 19–84.
34
Van der Veen 1992; Van der Veen and Westley 2010.
35
Mattingly 2010, 22–26, on the walls.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 72 David J. Mattingly et al.

Figure 2.3. Comparative plans of Garamantian hillfort sites in Wadi al-Ajal (see
Mattingly 2007; 2010 for details of these sites).

The later phases at Zinkekra overlapped with the earliest evidence for the
construction of new mudbrick settlements in the centre of the depression,
a kilometre or more away from the escarpment and now more closely
connected with the probable location of the irrigated gardens. The occupa-
tion sequence at Zinkekra is well-dated by a sequence of AMS and
Radiometric radiocarbon dates to the first millennium BC, with an earlier
group and a later group indicating prolonged habitation (see Table 2.1).36
The long-term capital of the Garamantes was one of these new oasis
settlements, established at Old Jarma (ancient Garama) from c.400 BC.37
As the Garamantian capital (described as a metropolis by the ancient
sources), it is perhaps unsurprising to find some unusual features, as yet
unparalleled at other Garamantian sites.38 This site was c.9 ha in area and
from the first century BC featured buildings utilising dressed stone
masonry alongside regular mudbrick construction (Fig. 2.7a). In the first
and second centuries AD, some buildings were erected in ashlar quality
36
Mattingly 2007, 294–95; 2010, 78–82.
37
Mattingly 2013; 2016; cf. also the publications of Ayoub 1967; 1968.
38
Pliny, Natural History 5.35–38, Ptolemy, Geography 4.6.30, on Jarma; cf. Mattingly 2013, 9–10,
530–34, for discussion.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 73

masonry and utilising Mediterranean influenced architectural features

(columns, capitals and architrave, etc.).39 One of these was almost certainly
a Garamantian temple.40
Botanical remains attest to the oasis agricultural economy, which seems to
have become more intensive by the Classic Garamantian period with the
introduction of summer crops like sorghum and pearl millet, as well as
cotton.41 An extensive series of AMS dates allow us to trace occupation at
the site from c.400 BC until its final abandonment in 1935.42 Garama merits
recognition as a true urban settlement and it may have had defences from an
early date, though evidence is inconclusive. In late antiquity it appears to have
had a central fortified compound (qasr), a feature now widely recognised at
other Late Garamantian sites.43
Close by was a c.3.5 ha undefended village settlement, Saniat Jibril
(Fig. 2.5 – GER002).44 The site is dated by imported ceramics to the first
to fourth centuries AD, though there are some hints of earlier and later
activity. Of particular interest is the recognition of extensive manufactur-
ing activity (metallurgy, bead making, weaving and glass working).45
Similar evidence was also noted at Jarma and indicates that Garamantian
society was not wholly dependent on oasis agriculture.46
A particularly instructive example of the pattern of development of oasis
centre settlement relates to the area around the Taqallit headland, c.30 km west
of Jarma. Here there are very limited traces of Early Garamantian activity, but
there was a dramatic transformation of the landscape in the Proto-Urban
phase, with the creation of at least 11 main settlements linked to major
cemeteries of drum and corbelled cairns along the escarpment by the lines of
shafts of foggara irrigation systems. Dating evidence links the initial develop-
ment of all of these elements in the oasis landscape to the latter centuries BC,
although the scale and complexity of the archaeology certainly represents
a great deal of subsequent consolidation. The lack of evidence for
a substantial localised population prior to this point, supports the identifica-
tion of this as pioneer farming development, with settlers expanding into this
area from parts of the wadi where oasis farming was already established
(Fig. 2.4). The firm dating of new crops and of the construction of a foggara
to this phase suggests that the expansion was also enabled by technological
developments. The settlement pattern in this area suggests a very dense

39
Mattingly 2013, 67–115. 40 Mattingly 2013, 93–100, 203–25, 291–92.
41
Pelling 2005; 2008; 2013a; 2013b. 42 Mattingly 2007, 294–302; 2013, 125–34.
43
Mattingly 2013, 90–92, 513, 532; Mattingly and Sterry 2013.
44
Mattingly et al. 2010c, 123–204. 45 Mattingly 2007, 448–62; 2010, 126–30.
46
Mattingly 2013, 511–12, 515–17.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 74 David J. Mattingly et al.

Figure 2.4. Hypothetical development sequence in the Taqallit landscape (WorldView-

2 image, 14 January 2013, copyright DigitalGlobe)

network of oasis villages, mostly of a few hectares in size, lying at the delivery
end of foggaras that originated at the escarpment, where substantial drum
cairns and corbel cairns close to the mother-wells made a statement about
communal rights to the precious water that was mined at great cost by
construction of the foggaras. That implies a link between the village

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 75

Figure 2.5. Examples of Garamantian village settlements from the Wadi al-Ajal:
a) ELH003; b) GBD001; c) GER002; d) FJJ056.

communities and the construction and enjoyment of the advantages of foggara

systems, with each settlement having control of at least one or a small group of
foggaras. It is notable that in this detailed case study, fortified structures (qsur)
were eventually constructed at the 11 main settlement locations, though only
two sites show strong indications of continuity into the Medieval period.
An AMS dating programme, focused on qsur in the eastern Wadi al-
Ajal, has established a Late Garamantian date for many fortified sites in the
wadi, a conclusion supported by surface finds of Roman pottery at many
sites when visited (Fig. 2.5). This trend towards construction of fortified
buildings within previously open settlements or to construct fortified
village-size compounds appears to be a general characteristic of Late
Garamantian settlement. The numbers, and close proximity of many of
these sites, strongly suggest that most of these sites were not military
outposts, but part of a changing settlement hierarchy.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 76 David J. Mattingly et al.

The revised picture of settlement in the Wadi al-Ajal thus reveals an

initial phase of oasis cultivation associated with hillfort sites, beginning in
the early first millennium BC (or perhaps a little earlier).47 The initial
settlement appears to have been fairly sparse, but by the latter centuries BC
there was evidently a significant increase in the numbers of villages and
these now spread onto low-lying locations immediately adjacent to the
cultivated area in the wadi.48 This was accompanied by the construction of
major irrigation works (foggaras). The foggaras and the new low-lying
villages evidently appeared in the Wadi al-Ajal landscape at the same time
as a number of new crops from Sub-Saharan Africa (pearl millet, sorghum
and cotton) and this can hardly be coincidental. The Proto-Urban phase
was a time of dramatically increasing population and expanding pioneer
cultivation within the valley.
Many of the villages of the Proto-Urban and Classic Garamantian
periods appear to have been open undefended settlements, but in the
Late Garamantian era there was an increasing emphasis on fortified set-
tlements, contained within walls with projecting towers, or fortified build-
ings (qsur) at the hearts of settlements.
Barnett’s recently published survey of the rock art of the Wadi al-Ajal
has revealed a wealth of material that is contemporary with or later than the
Garamantes, including significant clusters of images of horses, horsemen,
chariots and ancient Libyan inscriptions at key entry points into the valley
and at prominent promontory locations.49 The imagery is normally asso-
ciated with natural features, rather than Garamantian settlements and
burials.
A final point of distinction to be made about the Wadi al-Ajal is that the
funerary archaeology of this oasis belt was significantly more complex and
varied than that of many of the other areas identified as ‘Garamantian’.50 In
addition to standard Saharan styles of cairn burial, drum tombs and shaft
burials, the population of the Garamantian heartlands close to Jarma also
experimented with a wide range of novel tomb types, especially rectangular
and stepped monuments, but also pyramids and even Roman style

47
di Lernia and Manzi 2002, 179, for a date stone with an AMS date of 1200 calBC from a late
Pastoral burial in the Wadi Tanzzuft, c.400 km south-west of Jarma. This possibly indicates the
existence of some pioneer cultivation of the date palm in the central Sahara by that date.
48
For possible traces of Garamantian gardens in Wadi al-Ajal, see Mattingly 2007, 157–58,
196–98.
49
Barnett 2019a, 113 (59 horse representations and 157 camels), 149 (nine chariot
representations), 267–76 (for clustering of such imagery along main routeways along and out of
Wadi al-Ajal).
50
See now Gatto et al. 2019a, for the Trans-Saharan Archaeology volume focused on burial rites.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 77

Figure 2.6. Distribution of Garamantian oasis sites in the Murzuq depressions.

mausolea.51 The heartland cemeteries and settlements also reveal much

more extensive evidence of trade contacts between the Garamantes and the
Mediterranean world.52 The fact that there was significant divergence in
trade contacts, funerary and settlement archaeology between the
Garamantian heartlands and the other oasis clusters within Fazzan may
indicate some social distinctions between the peoples of these discrete
oases. At the same time, there is also sufficient shared in common between
different oasis groups to suggest that there were overarching factors that
bound these disparate oases together with the Garamantes.

The Murzuq/Hufra Basin and ash-Sharqiyat

The southern oasis band in Fazzan, comprising a series of contiguous

depressions (the Wadi Barjuj, Wadi Utba, the Murzuq depression, the
Hufra and ash-Sharqiyat), runs for over 200 km east to west along the

51
See the complementary overview study of Garamantian burials, Mattingly et al. 2019; also
Mattingly and Edwards 2003.
52
Leitch et al. 2017, 323–33.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 78 David J. Mattingly et al.

northern edge of the Murzuq sand sea. Until recently no certain

Garamantian era sites were known in this area, though some reconnais-
sance survey work by Daniels in 1968 had hinted at it.53 There are numer-
ous abandoned mudbrick qsur and settlements in this zone, though the
general assumption previously has been that they were of Medieval or early
modern construction.54 Although the Fazzan Project work initially high-
lighted the possibility of an early date for some of these sites,55 the more
recent availability of high resolution satellite imagery has transformed the
picture. Satellite remote-sensing has identified a large dossier of aban-
doned settlement sites and, with the benefit of some limited ground visits,
we can divide these into three broad categories:
1. Mudbrick rectilinear arrangements around qsur, often with fortified
enceintes and in close association with rectilinear field systems and
cairn cemeteries.
2. Mudlump nucleated sites with irregular shapes and arrangements.
3. Stone, mudlump and zariba (palm front huts) dispersed settlements
with a mixture of rectilinear and curvilinear structures.
Field survey of 79 sites in 2011, from which ceramics and AMS dating
samples were obtained, have demonstrated that the first group date to c.AD
100–700 with the more heavily fortified examples in the latter half of this
range. Conversely, the second group were of late Medieval date and the
final group were of more recent date implying that there was a radical
change in settlement forms in the Medieval period.56 We have demon-
strated the Garamantian nature of many of the fortified settlements iden-
tified on the satellite imagery and there are clear links between these
settlements and irrigation works and pre-Islamic cemetery types.57
Figure 2.6 shows dense settlement clusters of suspected or proven
Garamantian date.

Qasr ash-Sharraba
A particularly important site is Qasr ash-Sharraba, located towards the
western limits of oasis cultivation and which we have identified as an
53
Daniels 1968; 1989; cf. also Boxhall 1968.
54
Despois 1946, 59–60; Lethielleux 1948, 13, 48–50; Mattingly 2003, 146–54; Sterry et al. 2012.
55
Edwards 2001; Mattingly 2007, 254–88.
56
Mattingly and Sterry 2013; Sterry and Mattingly 2011; 2013; Sterry et al. 2012.
57
The dating of a substantial number of qsur to the Late Garamantian era is particularly
significant as previously these mudbrick castles had been uniformly assumed to be Medieval in
date, see Sterry and Mattingly 2013; Sterry et al. 2012.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 79

Figure 2.7. Possible Garamantian urban centres: a) Old Jarma; b) Qasr ash-Sharraba.

urban centre (Fig. 2.7b).58 This is very similar in form to the settlements
described above, but on a much larger scale. The site comprises an unwalled
settlement of c.15–18 ha, at the heart of which lay a large fort-like compound
(0.7 ha) with external towers. In the north-east corner of the fort was
a tower-like qasr and a second qasr stood independently in the settlement
to the west of the fort. A very extensive field system (556 ha) is preserved
around the town which contains two further qsur and some isolated build-
ings. On the surrounding hills, especially to the south, there were at least 24
drum cairn and shaft cemeteries containing thousands of tombs. The earliest
surface pottery from the cemeteries dates to the second century AD, perhaps
indicating the initial growth of the site, but six AMS dates from the fortified
citadel and qsur suggest that construction and occupation of these continued
through Late Garamantian times into the early Islamic era. The nature of the
twelfth-century activity is currently unclear. To the west of Qasr ash-
Sharraba, the Wadi Barjuj is uncultivated today, but there are some addi-
tional traces of Garamantian cemeteries, settlements and hydraulic systems
(wells and a variant on the foggara), associated with Roman pottery.59

The Murzuq Basin

The area east of the early modern capital Murzuq has been investigated in
detail, following the release of high resolution satellite imagery. Because
modern oasis cultivation has retreated from parts of this area, ancient sites

58
Mattingly 2003, 149–50; 2007, 261–65; Mattingly and MacDonald 2013; Mattingly and Sterry
2013, 506–10.
59
Mattingly et al. 2008, 250–51.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 80 David J. Mattingly et al.

Figure 2.8. Detailed mapping of the Garamantian fortified settlements and their
associated gardens in the Zizaw area (WorldView-2 image, 23 November 2002,
copyright DigitalGlobe).

and abandoned gardens are particularly well preserved. Four main areas of
sites were visited on the ground in 2011: Murzuq (MZQ), Gawat (GWT),
Hij Hijayl (HHG) and Zizaw (ZZW), including a total of 33 qsur from
many of which Roman-era pottery was recovered, subsequently confirmed
by AMS dates from 15 sites. A particularly striking cluster of sites was
recorded in a 6 km2 area at Zizaw. Here no less than 11 fortified settlements
can be identified, each located at the centre of a set of gardens and with
each village only a few hundred metres from the next one (Fig. 2.8).
Garamantian style burials are identifiable close to several of the sites
(comprising shaft burials and drum cairns). The irrigation here was by
means of shallow shaduf wells dug in the centre of many of the gardens.
A variant form of shallow foggara has been identified in one part of this
area (HHG), but never seems to have supplanted the use of shallow wells.
The architecture of these fortified sites is strikingly similar to that of late
Roman military outposts, which may have provided remote inspiration,
though here the interpretation of the complex as a group of fortified oasis
villages seems inescapable, given the density of these sites and their clear
agricultural associations.60 The Garamantian origin of this type of rectan-
gular fortified site, with external bastion towers, is now firmly established

60
Mattingly et al. 2013b, for a contextual study of the Garamantian qsur in relation to similar
fortified sites in the Roman frontier zones of North Africa. Cf. also Mattingly et al. 2013a, for
Roman military architecture.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 81

Figure 2.9. Comparative plans of fortified sites (qsur) in the Murzuq depression (imagery: Google,
DigitalGlobe).

and a number of representative plans of examples (for many of which we

have secure AMS dates in the range calAD 300–600) are shown in Figure 2.9.
Our investigation of the settlement pattern has been led by the visibility of
substantial fortified structures on the satellite imagery and, as stated, the date
of the qsur is predominantly Late Garamantian (late antique). The ultimate
origins of oasis cultivation in the Murzuq depression thus remain uncertain, as
the apparent lack of first millennium BC settlement here could be due to the
greatly reduced susceptibility of undefended settlements to identification in
satellite remote sensing. Nonetheless, in the Murzuq area the fortified build-
ings and fortified settlements sit at the centre of clusters of gardens in a way
that suggests they were laid out together. Although there were some foggara
channels here, the more common irrigation method seems to have involved
centrally placed shaduf wells within fields/gardens. Therefore, while there may
have been some earlier sites, our preliminary reading of the evidence is that the
extensive development of the southern oasis zone occurred at a later date than
the Wadi al-Ajal – specifically in the Classic and Late Garamantian period.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 82 David J. Mattingly et al.

Zuwila
Another important centre in eastern Fazzan was Zuwila, long recognised as
the pre-eminent site in the Early Islamic period.61 Survey has identified the
existence of a significant Garamantian settlement at this location. The early
Medieval site was embellished with a 4.5 ha fortified enclosure to the north of
a 20 ha open settlement. The larger settlement area appears to have originated
in the Garamantian period and seems to have related to both gardens with well
irrigation and a zone of foggara-fed irrigation (Fig. 2.10). There are also several
cemeteries that have yielded Roman material. Towards the centre of the
unenclosed settlement there was a (now destroyed) large qasr with projecting
towers (60 × 70 m, ZUL004 on Fig. 2.10). Roman imports have been found
close to this structure, which we believe to be Late Garamantian in date. The

Figure 2.10. Detail of possible Garamantian centre at Zuwila (Worldview-2 image, 30

October 2011, copyright DigitalGlobe).

61
Mattingly et al. 2015; Mattingly and Sterry 2018 on Zuwila; see also Ziegert 1969, on
Garamantian rock-cut burials.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 83

earliest origins of Zuwila are unknown, though Daniels reported seeing an

eye-bead there, which is a hint of trade connections by the latter centuries BC.
It is possible that Zuwila had already become an urban-scale centre in the Late
Garamantian period (compare the size of the possible settlement area around
the central qasr on Fig. 2.10 with Jarma and ash-Sharraba, as shown in Fig. 2.7)
and that this might have played a part in its supplanting Jarma as the pre-
eminent early Islamic centre in Fazzan.

Umm al-Aranib, al-Bdayir, Humera, Misqwin, Tmissa

At Umm al-Aranib there is a group of c.10 foggaras that run south-north into
a small depression (Fig. 2.11). At least 15 foggaras run north-south at al-Bdayir
and feed into an extensive area of field-systems on the edge of a playa.62 There
are several thin scatters of cairns and possible hut clearings on the plateaux to
the north. South of al-Bdayir another foggara group (about five in number)
run south to north into the same depression. At the southern end of these there
is a badly preserved mudbrick settlement of c.0.5 ha that may contain a for-
tification. It looks similar to Garamantian settlements in the Murzuq region
and 500 m to the north-west of the settlement there is a nucleated shaft
cemetery (c.6 ha) that is almost certainly of Classic Garamantian date.
Another foggara group of c.20 channels runs into the oasis of Misqwin. Two
square qsur with corner towers are visible (c.15 × 15 m and c.28 × 28 m), both
very similar morphologically to Late Garamantian examples.
To the east of Zuwila, on a secondary route towards Zala (and onwards
towards Egypt), Tmissa is a large multi-phase settlement on the edge of

Figure 2.11. Garamantian sites in eastern Fazzan.

62
An unusual 0.8 ha sub-circular fortification with ten evenly spaced towers may be dated to the
period of Kanimi dominance in the thirteenth–fifteenth centuries.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 84 David J. Mattingly et al.

a large playa. At the centre of the settlement there appears to have been
a square qasaba, which was later heavily modified and incorporated into
the fabric of the town’s housing. There is also a sub-oval enclosure c.5.5 ha
with the qasaba built into the south-east side.63 Around 1.8 km to the west
are the remains of a large field system with numerous wells on the western
side of which is a possible Late Garamantian qasr.
Some 30 km to the south of Zuwila, across an intervening area of sand
dunes, around Tirbu there are a few further sites with suspected Garamantian
origins, again associated with qsur and foggaras.64

Wadi ash-Shati

The Wadi ash-Shati is a topographic depression that runs east-west,

between the southern flanks of the al-Qarqaf arch to the north and the
Ramlat Zellaf to the south (c.160 km long and 40–60 km wide). Unlike
the majority of oases in Fazzan, those of ash-Shati are not positioned in the
lowest portion of the depression. Instead, they are located in the northern
part of the depression, on a sandstone piedmont. The sandstone stratum
appears as a sequence of small sterile mesas and stepped profiles, resulting
in a landscape of mounds and spurs, which stand out from the slight
depressions covered by alluvial sands.
Wadi ash-Shati is unusual in Fazzan in deriving its water from both
a confined aquifer, located under a roof of compact crystalline sandstone,
and from phreatic surface water.65 As a consequence, humans exploited both
the numerous artesian springs and the shallow ground water which was easy to
reach by means of dug wells. In the 1930s Scarin documented 277 natural
springs and 612 wells in the wadi. Some of the artesian springs were natural (for
example, the six small springs aligned along a natural fissure in Maharuqa), but
the majority were artificial.66 This aspect has made ash-Shati the most agricul-
turally developed and populated region of Fazzan in the Medieval and early
modern periods.67 The region is also known for its iron ore deposits.68

63
This was in disrepair by the time the site was visited by Hornemann 1802, 53–55, though he
mentions that there were reputedly inscriptions among the ruins. See also Despois 1946, 95, for
a plan and description of these features.
64
Mattingly 2007, 282, Tirbu. 65 Despois 1946; Dubay 1980.
66
Scarin 1937a; 1937b; cf. also Despois 1946.
67
The colonial Italian census of 1936 (Scarin 1937b) reported 13,769 inhabitants as opposed to
6,398 in the Wadi al-Ajal and 5,504 in the Southern oases, see also Mattingly 2013, 537.
68
These were described by Desio 1937 and then further studied by Goudarzi 1970; 1971 and Turk
et al. 1980.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 85

Figure 2.12. Garamantian sites recorded in the Wadi ash-Shati.

Despite being the most populated part of Fazzan in the early twentieth
century, the Wadi ash-Shati has been little explored archaeologically.69
Barth recorded the existence at Idri, of rock-cut chambers, possibly origin-
ally intended for burial.70 Better explored for its rock art panels,71 the wadi
was visited by Daniels, recording several sites in the area of Dabdab,
including a cemetery, a settlement and a foggara.72 A first systematic
survey from 2007 to 2008,73 and ongoing intensive remote sensing based
mapping, has now confirmed a long-term chronology of occupation and
a substantial Garamantian presence (Fig. 2.12).
The particularly favourable hydrological conditions made the wadi one
of the prime targets of government-funded agricultural schemes in the
early 1970s. Land reclamation at that time coupled with massive infra-
structural development,74 has had a significant effect on the preservation of
the archaeological record, with sites much altered or destroyed.
So far 12 qsur of probable Garamantian origins have been identified, and
three fortified villages of the same period (Table 2.2). This does not exclude
possible Garamantian origins of other settlements, which are characterised
by the presence of inner walls and fortified structures and which were
occupied until the colonial periods (for example Quttah). Both these

69
Barich and Baistrocchi 1987; Biagetti and di Lernia 2008; Cremaschi and di Lernia 1998; 2001;
di Lernia and Manzi 2002; di Lernia et al. 2001; Liverani 2006.
70
Barth 1857, 154. 71 Graziosi 1942; Le Quellec 1987.
72
The sites with associated photographs have been published in Mattingly 2007.
73
Merlo et al. 2008; 2013. 74 Pliez 2004.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 86 David J. Mattingly et al.

Table 2.2 List of qsur and fortified settlements surveyed in the Wadi ash-Shati. The site of
Bir al-Qasr was reported by the local informant Mustafa Habib

Locality (local Long. Lat. Altitude

Code name of structure) E (deg/min) N (deg/min) (m asl) Size (m)

Qsur
ADW001 Adwesa 12° 35.455’ 27° 44.489’ 409 50 × 20
(Qasr Bin Maherik)
IDR002 Idri 13° 3.327’ 27° 26.579’ 383 Not ascertainable
IDR023 Idri 13° 2.352’ 27° 26.634’ 373 30 × 20
IDR025 Ramlat Zellaf 13° 11.144’ 27° 19.888’ 402 20 × 16
IDR030 (Bir al-Qasr)* 13° 0.987’ 27° 30.924’ 11 × 10
ABY001 (Qasr Abyad) 13° 30.919’ 27° 32.396’ 383 22 × 18
ABY006 13° 30.614’ 27° 31.905’ 367 20 × 14
ABY007 13° 30.022’ 27° 31.925’ 378 20 × 15
WIN001 Hatiya Winzrik 13° 14.061’ 27° 29.680’ 373 30 × 25
Qasr Ain Omar
WIN002 Hatiya Winzrik 13° 14.721’ 27° 30.918’ 390 50 × 35
Qasr Bin Aghenneb
DBD002 Dabdab 14° 22.534’ 27° 35.289’ 390 56 × 40
BRK010 Tamzawa 14° 13.379’ 27° 34.022’ 380 25 × 26 (remains
of two walls
only)
Other Fortified Sites
IDR001 Idri 13° 03.165’ 27° 26.707’ 406 100 × 90
TIM001 Tmisan 13° 07.030’ 27° 28.849’ 394 70 × 20
BRG001 Birgin 13° 36.676’ 27° 33.388’ 372 190 × 23/70

settlement forms are characterised by a building technique that was based

on stone. It was used either for the entirety of the structure or at least for its
base, on which a mudbrick superstructure was built. Two isolated and very
poorly preserved qsur – IDR002 and ABD006 – form the only exceptions to
this, since they consist of a small mound of decaying mudbricks. The
arrangement of the stone blocks, although varied both in type of shape,
material and pattern, tends to be very regular. This is a distinctive character
that differentiates the qsur and fortified settlements from other stone-built
structures in the wadi (isolated houses, dispersed and nucleated settle-
ments), which are rather irregular in arrangement and varied in the type of
construction material. Further, it suggests a broad chronological
commonality.
Surface pottery (including Roman imports) collected from around
a number of the isolated qsur can be ascribed to the Classic and Late
Garamantian periods. Amongst the isolated qsur, only one was located in

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 87

the southern part of the depression, Qasr Zellaf (IDR025) a well-defined

stone and mudbrick structure with an internal partition. The others lay near
the northern edge of the depression and were all located above 360 masl.
The qsur were located on eroded relicts of Pliocene sedimentary
sequences, which serve as upstanding topographical features in the sur-
rounding landscape. No ditches were noted surrounding the qsur and in
general, their plans were irregular since they followed the shape of these
natural spurs/hills and they lack evidence of towers. However, Qasr Abyad
(ABY001), although on a natural spur, was rectangular in plan and pos-
sessed two towers. Frequently the qsur were either in proximity to an
extended cemetery or had a limited number of cairns (generally quite
large in size) in their vicinity. In some cases additional structures were
noted in the vicinity of the qsur, generally a rectangular building with at
least one or two internal partitions. Qasr Bin Maherik (ADW001) and Qasr
Zellaf (IDR025) were more isolated, lacking surrounding traces of ceme-
teries or field systems (either ancient or modern). These qsur were located
respectively to the north-west and south of the depression, on the line of
previously hypothesised routes from Ghadamis to Idri and from Idri to
Jarma. Most qsur in the wadi occurred in groups of two or three within
a radius of less than 5 km rather than being isolated.
Three fortified settlements of a different type were identified, at Idri,
Tmisan and to the west of Birgin. The first two occupy the flat top of
a substantial isolated hill (and have evident similarities with the
Zinkekra-type hillforts of the Wadi al-Ajal), whilst the Birgin site is at
an elevation of only 6 m above the present day surrounding sand dune
surface. The three sites had outer walls that followed the profile of the
natural relief on which they are located. This determined their shape,
which is irregular in nature and size. Idri and Birigin were built with
a mix of stone and mudbrick both for the external wall and internal
partitions; Tmisan is characterised by the exclusive use of cut stone for
the entire settlement. Surface material culture points to a Garamantian
origin for these sites. Whilst Birgin was eventually abandoned and the
settlement moved c.1 km east, Idri and Tmisan experienced a different
pattern during later times. The nucleated village settlements were both
moved down to the foot of the elevated area. The non-contemporaneity
of the top and bottom settlements in both instances is demonstrated by
the substantially different building techniques used, with the introduc-
tion of mudbrick and irregular stone built structures.
Substantial nucleated cemeteries characterised the area surrounding
both the qsur (with the exception of the isolated examples) and fortified

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 88 David J. Mattingly et al.

sites.75 Their morphology and spatial distribution is clearly connected with

the settlements, confirming the probable Classic Garamantian age of the
overall system. The organisation of settlements analysed thus far presents
a concentric configuration, with the qasr or fortified village in the middle,
a first circle of sparse graves or small nucleated cemeteries in close proxi-
mity to the settlement, a circle where modern (and presumably ancient)
wells and field systems are located, and finally a number of longitudinal
and/or nucleated cemeteries that radiated in all directions from the culti-
vated areas. It therefore seems that the particular topography of the north-
ern part of the depression was used to create a distinctive spatial pattern,
probably reflecting the chronology, but possibly also the social stratifica-
tion of occupation in the Garamantian period.
There is limited evidence as yet for a pre-Garamantian funerary land-
scape and, with caution, it is therefore suggested that the early oasis
occupation of the wadi is to be linked with the expansion of the
Garamantian kingdom, most probably during the Classic Garamantian
period. On the other hand, there is some rock art imagery of chariots in
the ash-Shati and two of the places named in Pliny’s account of the
campaign of Balbus in 20 BC chime with Wadi ash-Shati toponyms:
Dedris with modern Idri and Baracum with modern Brak. Idri as noted
was a hillfort site, which again might suggest a first millennium BC
origin. Pliny also mentions a location called Thelgae or Matelge with
a hot spring, which would suit the Wadi ash-Shati hydrology.76 This
might suggest that some initial development had begun around the
main natural springs in the first millennium, but that settlement expan-
sion occurred later. The continuity of exploitation of the northern oases
during the Late Garamantian period is also highly probable, with
a possible abandonment of some of the isolated outposts but
a continuous occupation of the majority of the settlements. As in the
Murzuq area we need to note that research has been focused on the more
upstanding fortified sites, which were mainly of Late Garamantian date.
We cannot exclude that there were undefended settlements that have so
far escaped detection on the satellite images. However, the overall num-
ber of Garamantian burials identified in the ash-Shati is consistent with

75
So far over 12,000 single cairn and drum graves have been plotted in the wadi thanks to the
increasing availability of high resolution satellite imagery through Google Earth. It is expected
that the total number will be in the region of 20–30,000, once the systematic study of the
imagery is completed.
76
On the rock art, see Le Quellec 1987. Pliny, Natural History 5.35–37; Desanges 1980, for
commentary.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 89

the relatively less dense settlement distribution observed here when

compared with the Wadi al-Ajal.

Some Outlying Oases in Eastern and Northern Fazzan

Ghuddwa
Located mid-way between the Wadi al-Ajal, Murzuq/Hufra basin and ash-
Sharqiyat, Ghuddwa is a small, but strategic oasis in Fazzan (Fig. 2.1).
Several Garamantian cemeteries with very characteristic Garamantian
funerary furniture and imported Roman pottery have been identified
here, as well as the remains of five or six qsur, two of which have yielded
some imported Roman pottery. No AMS dates have been obtained here,
but the morphology of some of the qsur is also consistent with a Late
Garamantian date.77 There is a string of Garamantian cemeteries and
possible settlements along the line of the Wadi an-Nashwa that connects
Ghuddwa with the Wadi Utba and Qasr ash-Sharraba area to the south-
west, a few of the former were visited by Daniels in 1968.78

The Ubari Sand Sea Lake Villages

A number of salty lakes are located in the Ubari sand sea between the Wadi
al-Ajal and Wadi ash-Shati. A brief survey of the village of Mandara,79 on
the shores of the homonymous salty lake in the Ubari sand sea has
identified a mudbrick qasr (30 × 20 m), which, according to local infor-
mants, has been in ruins in living memory. The site yielded Roman pottery,
brought to light during the excavation of deep wells. This is testimony of
early oasis occupation in the Ubari sand sea of at least one of its oasis
villages by the well-known lakes of that area.80 Survey connected with oil
prospection in this area also revealed a number of concentrations of
Roman pottery, possibly linked to exploitation of mineral salt deposits in
and around these lakes.81

77
Mattingly 2007, 267–70. 78 Mattingly 2007, 271–72.
79
Part of a survey of the recently abandoned lake villages of Gabroun, Trouna and Mandara
conducted in 2007 and 2008 (Mattingly et al. 2007 and Merlo et al. 2008).
80
On the lakes of the Ubari sand sea and the Dawada people in general, see Bellair 1951; Bruce-
Lockhart and Wright 2000.
81
Unpublished work involving Mattingly, see Lahr et al. 2009.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 90 David J. Mattingly et al.

Sabha, Samnu and al-Abid

There is no certain archaeological evidence of ancient origins of Sabha
(Fig. 2.1). However, there has not been any archaeological investigation of
the several villages that housed the Early Modern population of the oasis
and it is possible that these overlay earlier settlements. The name Sabai
appears linked with Garama in Ptolemy and is a suggestive hint that there
was a settlement of some sort in the Sabha oasis by the second
century AD.82 Some 3 km to the west of modern Sabha there is
a recently bulldozed mudbrick mound that was likely a qasr, if so this
would suggest some form of Garamantian settlement.
A number of hills in the Sabha area have evidence of scattered cairns.
A range of hills 45 km to the east of Sabha features a group of hilltop
settlements with small enclosures similar to those found in the Wadi al-
Ajal and are likely Proto-Urban Garamantian or earlier in date. At Samnu
there are the remains of two or perhaps three small qsur isolated amongst
the modern gardens. On the edge of the cultivated area there is a low hill on
the south-east spur of which is a small settlement (45 × 15 m). Also on this
hill are c.200 shaft and drum tombs clustered into five groups and
a building aligned east-west 20 × 10 m that is similar to the buildings
sometimes found on the east side of qsur and the burial structure
UAT003.83
An extensive network of foggaras is known in relation to the al-Abid
oasis north-east of Sabha and although conclusive evidence is lacking this
could indicate an early development on the routes between Fazzan and
al-Jufra oases to the north and Zala in the east (see Chapter 3).84 The main
foggara group led into a system of gardens of c.200 ha extent. Further
foggaras fed gardens to the north-east and north-west. Settlement remains
are few, but include at least two qsur and a large hilltop enclosure (with
only a few internal features). The largest area of settlement appears to
underlie a colonial era fort and the remains are much disturbed. Several of
the hills on either side have rings of drum cairns encircling them, though
there are no visible shaft cemeteries.

Wadi Tanzzuft and the Tadrart Akakus

Wadi Tanzzuft is a 200 km long, north-south oriented valley with an

ephemeral stream, that runs along a cuesta, whose western side is formed
82 83 84
Ptolemy, Geography 4.6.12. Mattingly 2007, 107–09. Nachtigal 1974, 61.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 91

Figure 2.13. Garamantian sites recorded in the Wadi Tanzzuft and Wadi Awis.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 92 David J. Mattingly et al.

by the edge of the Tassili range, while the eastern side in marked by the
steep cliff of the Tadrart Akakus mountain. Garamantian-era activity has
been traced in the main valley and in the mountain and desert landscapes
adjoining it.

Wadi Tanzzuft
The source of the wadi is located in the Takarkori area (southern Tadrart
Akakus), and its northern reach has been identified in a wide endorheic
depression at the western fringe of the Edeyen of Ubari.85 The main course
of the wadi is surrounded by lowlands with inselberg/pediment type relief,
alluvial fans, dried lakes, and small sand seas. Radiocarbon dates on archae-
ological sites buried inside the alluvial plain indicate that the northern part of
Wadi Tanzzuft was still fed by the river in the second and first millennia BC,
while a prominent contraction of the oasis followed a dramatic drop in
precipitation and subsequent unsteady climatic phase in the first half of the
first millennium AD. The presence of artesian springs in several parts of the
valley is an important aspect of the hydrology. This resulted in the formation of
three separate oases (Ghat, al-Barkat and Fewet) in the southern part of the
wadi (Fig. 2.13).86 The existence of another natural artesian spring at al-
Uwaynat (Sardalas) at the northern end of the Tanzzuft valley may have
supported an oasis settlement there too, but no conclusive evidence has been
found beneath the modern small town.
An incipient sedentism has been hypothesised for the Ghat oasis from the
beginning of the first millennium BC by the finding of an open-air site with
evidence of irrigation devices and a hybrid ceramic production, combining
decorative patterns typical of the late Pastoral culture (simple impressions and
rocker impression) with the Garamantian ones (twisted cord simple or roulette
impressions).87 But settlements dated to the Garamantian period have not been
identified at Ghat, probably because they are deeply buried under the Medieval
and modern town.88 There are major Garamantian cemeteries adjacent to Ghat
that were excavated by Pace et al. – these surely attest to a major settlement
close by.89 The cemeteries recorded in the al-Barkat and Fewet oases, by their
scale and locations, also strongly suggest the presence of additional undiscov-
ered Garamantian settlement sites in the other Tanzzuft oasis areas.90 In
addition, a number of fortified qsur have been recorded, some on the edge of
the oasis zone, others along the approach routes, like the Wadi Awis.
85
Cremaschi and Zerboni 2009; 2013. 86 Cremaschi and Zerboni 2011; 2013.
87
Mori et al. 2013. 88 Liverani 2000; Scarin 1937c. 89 Pace et al. 1951; Gatto et al. 2019a.
90
For the funerary remains in Wadi Tanzzuft, see the overview study by Gatto et al. 2019b.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 93

Fewet
The emergence of domestic architecture in the form of small compounds
made of mudbricks is dated to the third century BC, as illustrated by
research on the oasis of Fewet. However, radiocarbon dates from a large
cemetery in the same area suggest an earlier occupation of the oasis, from
at least the sixth century BC.91 At the edge of the modern village of Tan
Afella, the remains of a small village covering an area of c.850 m2 were
brought to light, though only its south-east corner was well preserved
(Fig. 2.14b). The site remained in use until the beginning of the first
century AD. It was planned and built as a single structure, for the dwelling
of a small clan with an egalitarian socio-economic base.92 The settlement
had a defensive character, with a single gate and a communal well, which
ensured water availability also in the event of an external danger. A large
central courtyard facilitated interior circulation and most probably pro-
vided a protected space for working activities, such as threshing, grinding
of cereals, pottery, basketry and leather production. The dwelling rooms
were built against the inner face of the perimeter, and were generally
arranged in units composed of two rooms: a larger square room flanked
by a smaller rectangular one. The surface of the five preserved dwelling
units ranged from 10 to 28 m2, with an average of c.18.5 m2.93 Building
techniques and materials were rather homogeneous: roughly dressed sand-
stone slabs were used almost exclusively for the mudbrick wall foundations
and for the footing of the village wall. Mudbrick was the main building
material and bricks were highly standardised both in size (50 × 35 × 8 cm,
implying the use of moulds of the same shape) and in composition. The
Fewet compound was the earliest structure excavated in the Tanzzuft
region built in mudbricks, but the craft and standardisation of the mud-
brick work implies an introduction and experimentation prior to the third
century BC.94 Local hand-made pottery, grinding stones, polishers, bone
tools, vesicular basalt lamps and iron tools were found inside the rooms,
together with faunal and botanical remains showing a community with
a fully developed irrigated agriculture, based on Near-Eastern crops,95 and
91
Mori and Ricci 2013.
92
A total number of 13 residential units for the excavated compound has been supposed, with
about 40 individuals. See Castelli et al. 2005 especially section by Mori, 73–84; Liverani 2006;
Mori 2010; 2013.
93
The two-roomed unit is an architectural pattern which is attested in other settlements of the
Fazzan, in Garamantian times. In the Wadi al-Ajal it was found at Zinkekra from the Early
Garamantian phase (Daniels 1968; Mattingly 2010, 83), and was still in use up to the second
century AD in the first phase of Saniat Jibril (Daniels 1971b, 6–7; Mattingly 2010, 155).
94
Mori 2013, 63–66. 95 Mercuri et al. 2013.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 94 David J. Mattingly et al.

Figure 2.14. Comparative plans of Garamantian sites in the Tanzzuft area.

the herding of livestock together with the hunting of wild animals like
Barbary sheep, hares and gazelles.96
A cemetery composed of 1,329 stone tumuli, already in use during the
late Pastoral period, developed as the burial ground of the entire commu-
nity of the oasis from the sixth century BC to the third century AD.97

Aghram Nadharif
The oases of Garamantian Fazzan reached their peak in the first
centuries AD.98 In Wadi Tanzzuft this meant the building of larger fortified
citadels protecting the oases villages and the construction of a series of forts/
castles at strategic points in the open desert in order to control the main
passageways of the caravan routes.99 On the eastern fringe of the Barkat
oasis, the stone citadel of Aghram Nadharif, ‘the city of alum’, was built on
a low sandstone terrace, overlooking a long stretch of the Wadi Tanzzuft.
The settlement was built in the first half of the first century BC and was used
till the fourth century AD.100 The fortified citadel has an irregular oblong
shape, following the morphology of the rock spur on which it was built, with
an inner surface area, excluding the city-wall of approximately 0.6 ha (Fig.
2.14a).101 The outer enceinte was a 2-m-thick stone wall, reaching c.2.5 m in
height. The city wall was provided with two towers on the east side, still
standing to a maximum height of 4.5 m, possibly duplicated by two more

96 97
Alhaique 2013. Liverani et al. 2013; Mori et al. 2013.
98
Liverani 2006, 2007a/b; Mattingly 2013, 530–34.
99
Liverani 2006, 363–74; Biagetti and di Lernia 2008.
100
For the sequence of the 14C datings upon which this reconstruction was proposed see Liverani
2006, 363–74 and in particular tables 30.I and II.
101
Putzolu 2006.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 95

whose remains are too poorly preserved to be interpreted with certainty.

Four gates gave access to the citadel. Three areas of the interior were
excavated, mainly uncovering residential houses with storage facilities –
pits dug in the bedrock and bins built above the floor, with thin partition
walls – and small working areas, together with lanes and open courtyards.
Once again, the houses had a two-roomed standard plan, similar to the ones
in the Fewet compound, but here entirely built in stone. The layout of
dwelling modules was much more articulated and the layout more complex
than that of the Fewet compound and built to host a larger part of the
sedentary oasis community.102 Clusters of fireplaces in the area surrounding
the citadel testify to the contemporary presence of ephemeral settlements;
a scattered population living in pastoral encampments has also been identi-
fied throughout the southern Tanzzuft valley.103 The settlement seems
entirely built up of domestic units and no ‘public’ building has been identi-
fied, a situation which seems again to reflect the presence of an ‘egalitarian’
community, based on the local kin structure. This social structure is also
reflected in the nearby cemetery, located just on the opposite bank of the
wadi, on a rocky terrace. Approximately 590 tumuli were identified on the
basis of a satellite Ikonos image,104 and from this preliminary observation
they show rather uniform typologies, with conical and drum-shaped tumuli.
The greater richness and volume of Mediterranean imports in the burials
close to Ghat hints that the region’s political elite, which probably promoted
the building of similar citadels like Aghram Nadharif, was located there.105
Two of the burials excavated by Leschi have dates that are broadly contem-
porary with Aghram Nadharif.106
In Wadi Tanzzuft this settlement system was apparently strongly related
to the development of the caravan trade and it seems to have collapsed by
the Late Garamantian phase, as suggested by the abandonment of the stone
citadels and forts. As far as settlements are concerned, from the fourth
century AD this region underwent a shift from an oasis-based territorial
society to the probable hegemony of pastoral tribes.107

102
See Liverani 2006, 395–409; 2007b: 165–68 for spatial analysis and demographic calculations.
According to his reconstruction the site could include a maximum of 65 houses hosting an
estimated population of approximately 260 inhabitants, at its peak.
103
Cremaschi 2006, 18–19. 104 Castelli and Liverani 2006, Table 4.1.
105
Liverani 2006, 416–17.
106
Leschi 1945. The dates are reported in Fagan 1965, 115: Sa-92, 1330±120, calAD 431–975; Sa-
93, 1680±150, calAD 29–649.
107
Liverani 2006, 461–62.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 96 David J. Mattingly et al.

Tadrart Akakus
The deeply incised valleys on the east side of the Akakus range and the
main passes across the mountains have also yielded traces of Garamantian
era activity. Rock art in many of the rock shelters of the Akakus wadis
extends to horse period imagery, assumed to be contemporary with the
Garamantes and featuring depictions of chariots, biconical figures and
palm trees. Radiocarbon dates from some of the rock shelters confirm
the intensity of occupation at this time (Table 2.1). Such sites provide
tantalising glimpses of the relationships between pastoral groups and the
oasis dwellers. The building of the Aghram Nadharif citadel adjacent to the
oasis of Barkat was undertaken contemporary to the construction of
a network of fortified sites in the Akakus region. Three isolated qsur have
been identified and preliminary investigation showed a foundation date
which ranges from the end of the first century BC to the first century AD
(Figs 2.13–2.14).108 The best preserved structures were found along the
eastern fringes of the Akakus mountain, facing the Wadi Awiss where two
forts/castles have been identified that strongly resemble Aghram Nadharif
both in building techniques and in dating. The forts/castles of Imassarajen
and Adad were located far from the oasis of Wadi Tanzzuft, but lie along
one of the caravan routes connecting that valley with Wadi Barjuj and the
Murzuq region through the aqba (pass) of Aghelachem, which crosses the
Akakus.109 A unique funerary monument, the so-called ‘Royal tumulus’ of
In-Aghelachem, was built in Wadi Tanzzuft, close to this aqba. It was
composed of a huge stepped tomb enriched by a considerable number of
small features, heaps of stones and U-shaped structures which do not have
parallels in our area.110 It was probably the burial of a chieftain, controlling
the important mountain pass, and testifies to the presence of autonomous
pastoral groups contemporary to the occupation phase of Aghram
Nadharif and the desert qsur.
The castles/forts of Adad and Imassarajen were protected by a massive
stone wall and were surrounded by complex open-air sites, formed by
a number of stone structures of different shape and nature and fire places.
Small test trenches showed the use of mudbricks with painted plaster for
the inner partition walls. Both structures, did not survive the collapse of the
Garamantian kingdom, just like Aghram Nadharif.
No detail of the inner planning was readable from the extant structures but
some information on demography and population came from the cemeteries
related to each site, which were located a few hundred metres from the living
108 109 110
Mori 2012. Biagetti and di Lernia 2008. di Lernia and Manzi 2002, 102–16.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 97

areas. A total of 325 tumuli were mapped for Imassarajen, and 156 for Adad;
the most recurrent type was the drum-shaped tomb followed by the conical
cairn.111 The itineraries of caravan routes on the western side of the Akakus
were conditioned by water availability and by the possibility of crossing the
mountain through the aqbas. Two main mountain passes in the northernmost
part of the Akakus mountain were probably in use in Garamantian times,
demonstrated by the presence of rock inscriptions and engravings attributable
to that period. At the aqba Irlarlaren, the longest inscription in Old Libyan
characters known in Fazzan up to now has been recorded.112 The concentra-
tions of the rock art sites featuring Horse and Camel styles, traditionally dated
to the Protohistorical and historical periods, seems to partially replicate the
distribution of Tifinagh scripts.113

Wadi Hikma

Al-Qatrun and Tajirhi are the largest of a number of minor oasis settle-
ments along the Wadi Hikma, a linear depression running from north to
south parallel to the eastern edge of the Murzuq sand sea. A series of qsur
have been reported down this corridor that links Fazzan with Tibesti and
Lake Chad, though hitherto these have generally been assumed to be
Medieval in date (Fig. 2.1).114 Many of these are large and sit within
extensive garden systems that are far from any of the known early modern
and Medieval villages (Fig. 2.15). At Tajirhi, a foggara has been identified
and a burial in the pre-Islamic tradition has been radiocarbon dated to
the second half of the first millennium AD.115 Given the close similarity in
the architecture and layout of the Wadi Hikma qsur and those of the
Murzuq depression to the north, early origins for some elements of this
line of oases seem highly probable.

Comparing Settlement Trajectories

and Concluding Discussion

Fazzan now provides ample evidence for the origins of oasis cultivation in
the first millennium BC in the central Sahara. The data available indicate
111
Biagetti and di Lernia 2008. 112 Liverani 2006, 437–39; Ait Kaci 2007.
113
Biagetti et al. 2012. 114 Despois 1946, 57–61; cf. also Bruce-Lockhart and Wright 2000.
115
Bellair and Pauphillet 1959; Bellair et al. 1953; Fagan 1965, 115, Sa-78 1190±120, calAD
615–1118.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 98 David J. Mattingly et al.

Figure 2.15. Detail of abandoned oasis settlement and gardens in the Wadi Hikma area,
c.5 km south-east of Tajirhi (WorldView-2 image, 29 April 2012, copyright
DigitalGlobe).

that all the main oasis areas of Fazzan were developed to a greater or lesser
extent during the Garamantian period and that the total number of oasis
settlements numbered in the hundreds, including many substantial villages
and a few sites of urban scale. This seems to have begun in different sub-
regions at different moments in time and with varying intensity of devel-
opment. Currently, the earliest evidence relates to the Wadi al-Ajal, with
precocious agriculture in the first half of the first millennium BC and more
intensive expansion in the latter centuries BC. There is also strong evidence
for late first millennium BC development in the Ghat area of the Wadi
Tanzzuft. On the other hand, the evidence from the Murzuq depression
and the Wadi ash-Shati seems at present to point to somewhat later
development, mainly focused on the first half of the first millennium AD,
though as noted Pliny’s account of the campaign of Balbus suggests that
some of the natural springs of the Wadi ash-Shati were already exploited by
the late first millennium BC.
The Early Garamantian period (1000–500 BC) is quite distinct from the
later phases and is firmly associated with promontories and high ground.
These hilltop and escarpment sites are predominantly found in the Wadi
al-Ajal and around Sabha, but with examples known from the Wadi

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 99

ash-Shati, Wadi Tanzzuft and Wadi Hikma. They are extremely hetero-
geneous in size, form and layout, but there are some commonalities in
features. The occupied area on top of hilltops was generally very small and
only in one example did it exceed 1 ha, although where there were addi-
tional enclosing walls at the scarp base much larger areas were enclosed.
A majority of examples have occupation on the slopes of the hills although
again most of these were fairly small in size and some lacked significant
fortifications. A subset of these sites had large complexes of walls on and at
the base of slopes and enclosing large areas (these reached 10 ha and more,
for example: Zinkekra, Tinda, al-Khara’iq and Ikhlif 2) in some cases this
also includes some evidence of terracing.116 These encircling embank-
ments are thought to be a development of the Proto-Urban Garamantian
phase (500–1 BC).
The hilltops were generally difficult to access and, while ideal as refuges and
for storage, they were somewhat impractical for daily use by farming com-
munities. The slope settlements were better suited for occupation by these
farmers, while still remaining relatively secure. The larger complexes com-
bined elements of refuge with occupation on the low slopes. The archaeobo-
tanical remains leave no doubt that the inhabitants practised oasis agriculture
and Van der Veen’s analysis of the botanical samples from the site has
indicated that much of the basic processing work, especially of grain was
carried out on top of the plateau, with more fruits and less processing waste
found in the terraced housing on the scarps of the hillfort.117 The development
of the large complexes seems to have been a relatively late phenomenon,
perhaps as late as the third to first centuries BC and linked to a widespread
expansion of agriculture and population in Fazzan.
It was also in the Proto-Urban period that we have the earliest evidence of
settlement within oases, notably the walled Tin Afella compound at Fewet in
the Ghat region, but also at Jarma and at multiple sites close to the Taqallit
peninsular in the Wadi al-Ajal.118 At this date we can also demonstrate the
introduction of foggara irrigation technology, the development of dense cairn
cemeteries on the slopes of the Hamada in the Wadi al-Ajal and around the
Ghat oasis and the first indications of trade with the Mediterranean in the
form of imported ceramics.
In these early periods we can question to what extent we have evidence
of either state formation or urbanisation. During the period 1000–500 BC

116
Mattingly 2003, 139–41. 117 Van der Veen and Westley 2010, 517–18.
118
Note AMS dates in Table 2.1. Also Mattingly et al. 2010b, 117–25 for the settlements near
Taqallit and notes on possible Punic and Garamantian Proto Urban pottery.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 100 David J. Mattingly et al.

the number of permanent settlements and their scale appears small, quite
unlike those found in parts of the south-Western Sahara at this time.119 In
the period 500–1 BC, it has been argued that the one- and two-roomed
buildings of Fewet (up to 28 m2 in size) provide evidence of an egalitarian
society.120 This basic form of structure is found amongst all the hilltop sites
but in the more densely occupied sites (such as Tinda) they are commonly
arranged into compounds (not dissimilar to the Fewet compound itself).
While the cairn cemeteries contemporary to these sites do show signs of
diverse identities, the lack of excavated settlements apart from Fewet
means that we are still some way off from being able to distinguish
evidence of social differentiation, especially between compounds, from
the material culture. While they vary in size and complexity (no doubt in
part from their varied histories), the only evidence of more substantial
constructions are the enclosing walls and terraces. The wider arrangement
would suggest clumps of oasis cultivation that developed around natural
springs or in places where there were other accessible water sources.
Settlement in the Proto-Urban phase appears to have been based around
numerous smaller and quite egalitarian settlements, with not much sense
of an emerging social or settlement hierarchy at this date.
At the start of the Classic Garamantian phase, around the start of the first
century AD, there are the earliest indications of more substantial buildings
and more complex architecture focused on sites round Jarma. At Jarma itself
a stone-footed temple was constructed in the centre of the settlement, on the
slopes of Zinkekra a series of multiple-roomed rectilinear structures were
built with ashlar footings, at Uatuat (UAT003) a tomb was elaborated with
a large mudbrick enclosure. Elsewhere, the compound architecture contin-
ued at Aghram Nadharif, Ghat and at a number of sites in Wadi ash-Shati
(such as ADW001) in which small low hills were enclosed and filled with
structures on a rectilinear arrangement (these are smaller and denser than
the hilltop examples described above, but they represent a persistent settle-
ment form that continued on into the Medieval period). The most important
development in the settlement record in the early first millennium AD
concerns large nucleated sites with evidence of monumental buildings,
craft specialisation and in particular control of Trans-Saharan trade. The
two prime examples are Jarma in the Wadi al Ajal and Qasr ash-Sharraba in
the Murzuq/Hufra depression. Another possible urban site, Zuwila, in the
ash-Sharqiyat region, was potentially larger than both, but the lack of

119
Those found in the Dhars of Mauritania, see MacDonald, Chapter 13, this volume.
120
Mori 2013, 63–66; Mori et al. 2013.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 101

structured survey on the now destroyed site makes it impossible to distin-

guish between the Medieval and Garamantian remains (Fig. 2.10). In the
Wadi al-Ajal and Murzuq/Hufra basin there were large numbers of smaller
sites consisting of rectilinear mudbrick buildings, square qsur or both.
Outside these regions there were also qsur in Wadi ash-Shati and Hikma,
but only a few with associated settlement and a few of somewhat irregular
shape in the Tanzzuft. The Saniat Jibril type of open site seems largely
confined to the area around Jarma and Taqallit in the Wadi al-Ajal, but
this may be due in part to the lack of intensive survey in other regions.121 The
largest number of settlements can be attributed to this period, especially in
the Wadi al-Ajal and Wadi Tanzzuft although this is partly due to the
widespread distribution of distinctive Mediterranean imports that make
identifying settlements easier than in other periods.
In the Late Garamantian period (AD 300–700) there was an increasing
emphasis on fortification throughout the Wadi al-Ajal and the Murzuq/
Hufra basin with all known settlements from this period featuring central
qsur and many further fortified with enceintes and protruding towers.
There are signs of a general shift in settlement towards central and eastern
Fazzan with no known Late Garamantian settlements in the Wadi
Tanzzuft. The many fortified sites in the Murzuq-Hufra basin and Wadi
Hikma (paralleled by some indications of a shift in wealth and importance
of villages from west to east within the Wadi al-Ajal)122 foreshadow the
eclipse of Jarma and the further development of Zuwila as the chief centre
in Fazzan in the early Medieval period.
Table 2.3 summarises the distribution of different types of settlement
sites within the various regions of Fazzan. These differences in part relate
to topographical variability, but do suggest that there were some under-
lying regional preferences, chronological factors and localised trends at
play. Perhaps this is one of the most interesting conclusions to emerge
from this overview. The label ‘Garamantian’ covers a wide range of settle-
ment forms within Fazzan, yet it is only through intensive survey work and
direct scientific dating that many of these sites have been finally recognised
as Garamantian. Similar work is needed in other Saharan regions where the
same sort of architectural forms occur to see whether some of those
settlements may also be of pre-Islamic date.

121
Only the Tanzzuft area has seen any substantial pedestrian survey, but this seems to have
focused mainly on burials.
122
This is evident in terms of the numbers of Garamantian qsur, the distribution of late Roman
pottery and more elite burial monuments/cemeteries. Nonetheless, Jarma seems to have
remained the Garamantian capital at this date.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 Table 2.3 The distribution of site types between the different regions of Fazzan.

Mudbrick Fortified low

Hilltop and compound/ hilltop Fortified
Hilltop Slope slope settlement (rectilinear Village with settlement
Region Refuge Settlement complex (w/o qasr) arrangement) Urban site Isolated Qasr qsur with qsur

Wadi ash-Shati X? X X? X X

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

Wadi al-Ajal X X X X X X X X
Murzuq/Hufra X? X X X X X
Basin
Wadi Tanzzuft X X X X
Wadi Hikma X? X X X
 2 Garamantian Oasis Settlements in Fazzan 103

Tracing early oasis settlements in the Sahara is not straightforward as

they are often buried and masked beneath later settlements and gardens.
The Garamantian sites are to some extent unusual for their visibility due to
the intensity of trade with the Roman empire, but it is salutary to observe
that it is only in the last 20 years that fieldwork and AMS dating has allowed
us to demonstrate in Fazzan the occupation of the hundreds of early first
millennium AD oasis settlements. Even in the absence of ground truthing
and direct dating at many other locations, defensive features such as the
qsur and dense cairn and shaft cemeteries help us highlight areas for
consideration. Other characteristics of the Garamantian oases which may
be helpful for tracing early sedentarism in other parts of the Sahara include:
• the adoption of a developed agricultural package, with obvious affinities
with the oases of the Western Egyptian Desert;
• the association with rock art images of people riding horses, driving
chariots and the first images of camels;
• the movement of ideas and competencies (pyrotechnologies, irrigation
works including the foggara, spinning and weaving, written script);
• the evidence of trade contacts.123

A final observation of the Garamantian settlement pattern is that it did not

follow the same distribution found in later periods. In the 1930s the Wadi
ash-Shati accounted for more than a third of the c.33,500 inhabitants of
Fazzan, whilst the Wadi al-Ajal had only c.6,500. In the Wadi al-Ajal there
are 125 identified Garamantian settlements compared to just 29 villages or
hamlets still occupied in 1931, whereas in the Wadi ash-Shati there are just
15 known Garamantian settlements – less than half the 1931 total. While
varying levels of archaeological preservation and visibility are factors, the
consecutive capitals of Fazzan: Jarma, Zuwila, Traghan, Murzuq and Sabha
should make clear that oases are not static environments and that where we
find settlement today is not where we should necessarily expect it to have
been located in the past.

References

Ait Kaci, A. 2007. Recherche sur l’ancêtre des alphabets libyco-berbères. Libyan
Studies 38: 13–37.
Alhaique, F. 2013. The faunal remains. In Mori 2013, 191–98.

123
Mattingly 2017, for an overview of Garamantian and Saharan trade.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 104 David J. Mattingly et al.

Ayoub, M.S. 1967. Excavations in Germa between 1962 and 1966. Tripoli: Ministry
of Education.
Ayoub, M.S. 1968. Excavations in Germa (Fezzan). Cemetery of Saniat Ben-
Howidy. Tripoli: Ministry of Education.
Barich, B.E. and Baistrocchi, M. 1987. Archaeology and Environment in the Libyan
Sahara: The Excavations in the Tadrart Acacus, 1978–1983 (Cambridge
Monographs in African Archaeology 23/ BAR International Series 369).
Oxford: British Archaeological Reports.
Barnett, T. 2019a. An Engraved Landscape. Rock Carvings in the Wadi al-Ajal,
Libya. Volume 1: Synthesis. London: Society for Libyan Studies.
Barnett, T. 2019b. An Engraved Landscape. Rock Carvings in the Wadi al-Ajal,
Libya. Volume 2: Gazetteer. London: Society for Libyan Studies.
Barnett, T. and Mattingly, D.J. 2003. The engraved heritage: Rock art and inscrip-
tions. In Mattingly 2003, 279–326.
Barth, H. 1857. Travels and Discoveries in North and Central Africa (reprint 1965).
London: Frank Cass.
Bellair, P. 1951. La ramla des Daouada (Fezzan). Travaux de l’Institut de Recherches
Sahariennes 7: 69–86.
Bellair, P. and Pauphillet, D. 1959. L’âge des tombes préislamiques de Tejerhi
(Fezzan). Travaux de l’Institut de Recherches Sahariennes 18: 183–85.
Bellair, P., Gobert, E.-G., Jodot, P. and Pauphilet, D. 1953. Mission au Fezzan.
Tunis: Institut Hautes Etat Tunis.
Biagetti, S. and di Lernia, S. 2008. Combining intensive field survey and digital
technologies: New data on the Garamantian castles of Wadi Awiss, Acacus Mts,
Libyan Sahara. Journal of African Archaeology 6.1: 57–85.
Biagetti, S., Ait Kaci, A., Mori, L. and di Lernia, S. 2012. Writing the desert. The
‘Tifinagh’ rock inscriptions of the Tadrart Acacus (south-west Libya). Azania 47:
153–74.
Boxhall, P. 1968. Near East Land Forces Expedition to Murzuk, Libya. Expedition
Report. Nicosia: Nicolaou and Sons Ltd.
Bruce-Lockhart, J. and Wright, J. 2000. Difficult and Dangerous Roads. Hugh
Clapperton’s Travels in Sahara and Fezzan. London: Sickle Moon.
Castelli, R. and Liverani, M. 2006. Cemeteries around the Barkat oasis. In Liverani
2006, 25–28.
Castelli, R., Cremaschi, M., Gatto, M.C., Liverani, M. and Mori, L. 2005.
A preliminary report of excavations in Fewet, Libyan Sahara. Journal of
African Archaeology 3.1: 69–102.
Connah, G. 2004. Forgotten Africa. An Introduction to its Archaeology. London:
Routledge.
Cremaschi M. 2003. Steps and timing of the desertification during the Late
Antiquity. The case study of the Tanzzuft oasis (Libyan Sahara). In
M. Liverani (ed.), Arid Lands in Roman Times, AZA 4. Firenze: Edizioni
All’Insegna del Giglio, 1–14.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 105

Cremaschi M. 2006. The Barkat oasis in the changing landscape of Wadi

Tannezzuft during the Holocene. In Liverani 2006, 13–24.
Cremaschi, M. and di Lernia, S. (eds). 1998. Wadi Teshuinat. Palaeoenvironment
and Prehistory in South-Western Fezzan (Libyan Sahara). Survey and
Excavations in the Tadrart Acacus, Erg Uan Kasa, Messak Sattafet and Edeyen
of Murzuq, 1990–1995. Milan: All’Insegna del Giglio.
Cremaschi, M. and di Lernia, S. 2001. Environment and settlements in the
Mid-Holocene palaeo-oasis of Wadi Tanzzuft (Libyan Sahara). Antiquity 75:
815–25.
Cremaschi M. and Zerboni A. 2009. Early to middle Holocene landscape exploita-
tion in a drying environment: two case studies compared from the Central
Sahara (SW Fezzan, Libya). Comptes Rendus Géoscience 341: 689–702.
Cremaschi M. and Zerboni A. 2011. Human communities in a drying landscape.
Holocene climate change and cultural response in the Central Sahara. In I.
P. Martini and W. Chesworth (eds), Landscape and Societies, Selected cases.
New York: Springer Science, 67–89.
Cremaschi M. and Zerboni A. 2013. Fewet: an oasis at the margin of Wadi
Tanezzuft. In Mori 2013, 7–15.
Daniels, C.M. 1968. Garamantian excavations: Zinchecra 1965–1967. Libya
Antiqua 5: 113–94.
Daniels, C.M. 1969. The Garamantes. In W.H. Kanes (ed.), Geology, Archaeology
and Prehistory of the Fezzan, Libya. Tripoli: Petroleum Exploration Society of
Libya, Eleventh Annual Field Conference, 31–52.
Daniels, C.M. 1970. The Garamantes of Southern Libya. London: Oleander.
Daniels, C.M. 1971a. The Garamantes of Fezzan. In F.F. Gadallah (ed.), Libya in
History. Proceedings of a Conference Held at the Faculty of Arts, University of
Libya 1968. Benghazi: University of Benghazi, 261–85.
Daniels C.M. 1971b. Excavations at Saniat Gebril, Wadi el-Agial, Fezzan. Libyan
Studies 2: 6–7.
Daniels, C.M. 1973. The Garamantes of Fezzan – an interim report of research,
1965–1973. Libyan Studies 4: 35–40.
Daniels, C.M. 1975. An ancient people of the Libyan Sahara. In J. Bynon and
T. Bynon (eds), Hamito-Semitrica. The Hague: Mouston, 249–65.
Daniels, C.M. 1977. Garamantian excavations (Germa) 1977. Libyan Studies 8: 5–7.
Daniels, C.M. 1989. Excavation and fieldwork amongst the Garamantes. Libyan
Studies 20: 45–61.
De Agostini E. 1934. La conca di Gat. In Governo della Tripolitania e della
Cirenaica, Servizio di studi, Bollettino Geografico 5–6: 14–27.
Desanges, J. 1980. (Pline l’ancien) Histoire naturelle Livre V.1–46 (L’Afrique du
nord). Paris: Collection Budé.
Desio, A. 1937. Geologia e morfologia. In Sahara Italiano 1937, 39–94.
Despois, J. 1946. Mission scientifique du Fezzan (1944–1945) III Géographie
humaine. Algiers: Université d’Alger.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 106 David J. Mattingly et al.

di Lernia, S. and Manzi, G. (eds). 2002. Sand, Stones and Bones. The Archaeology of
Death in the Wadi Tanzzuft Valley (5000–2000 BP). Firenze: All’Insegna dell
Giglio.
di Lernia, S. and Zampetti, D. (eds). 2008. La memoria dell’arte. Le pitture rupestri
dell’Acacus tra passato e futuro. Firenze: All’Insegna dell Giglio.
di Lernia, S., Bertolani, G.B., Merighi, F., Ricci, F.R., Manzi, G. and Cremaschi M.
2001. Megalithic architecture and funerary practices in the late prehistory of
Wadi Tenezzuft (Libyan Sahara). Libyan Studies 32: 29–48.
Dubay, L. 1980. Ground water in Wadi Ash Shati, Fazzan – a case history of
resource development. In M.J. Salem and M.T. Busrewil (eds), The Geology of
Libya II. London: Academic Press, 611–26.
Duveyrier, H. 1864. Les Touareg du nord. Exploration du Sahara. Paris: Challamel
Ainé.
Edwards, D. 2001. Archaeology in the southern Fazzan and prospects for future
research. Libyan Studies 32: 49–66.
Ehret, C. 2002. The Civilizations of Africa. A History to 1800. Oxford: James Currey.
Fagan, B.M. 1965. Radiocarbon dates for Sub-Saharan Africa (from c.1000 B.C.).
Journal of African History 6.1: 107–16.
Gatto, M. 2006. The Garamantes in the Ghat-Barkat area: previous research. In
Liverani 2006, 21–24.
Gatto, M., Mattingly, D.J., Ray, N. and Sterry, M. (eds). 2019a. Burials, Migration
and Identity in the Ancient Sahara and Beyond. Trans-Saharan Archaeology
Volume 2. Series editor D.J. Mattingly. Cambridge: Cambridge University Press
and The Society for Libyan Studies.
Gatto, M.C. Mori, L. and Zerboni, A. 2019b. Identity markers in south-western
Fazzan. Were the people of the Wadi Tanzzuft/Tadrart Akakus region
Garamantes? In Gatto et al. 2019a, 108–33.
Gigliarelli, U. 1932. Il Fezzan. Tripoli: Governo della Tripolitania.
Goudarzi, G.H. 1970. Geology and Mineral Resources of Libya – a Reconnaissance.
Washington, DC: US Geological Survey Prof Paper 660.
Goudarzi, G.H. 1971. Geology of the Shati valley area iron deposit, Fezzan, Libyan
Arab Republic. In C. Gray (ed.), First Symposium on the Geology of Libya.
Tripoli: Faculty of Science, University of Libya, 489–500.
Graziosi, P. 1942. L’arte rupestre della Libia. Napoli: Edizioni della Mostra
d’Oltremare.
Hachid, M. 2000. Les premiers Berbères. Entre Méditerannée, Tassili et Nil. Aix-en-
Provence: Edisud.
Hawthorne J, Mattingly D.J. and Daniels C.M. 2010. Zinkekra: An Early
Garamantian escarpment settlement and associated sites. In Mattingly 2010:
19–84.
Hornemann, F. 1802. The Journal of Frederick Hornemann’s Travels from Cairo to
Mourzouk the Capital of the Kingdom of Fezzan in Africa in the Years 1797–9.
London: G. and W. Nichol.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 107

Lahr, M.M., Foley, R.F., Mattingly, D.J. and Le Quesne, C. 2009. Area 131, Jarma,
Fazzan: Archaeological Mitigation of Seismic Acquisition 2006–2008 – Final
Report. Unpublished Consultancy report for Occidental Libya Oil and Gas BV
and Libyan Department of Antiquities.
Lassère, J.-M. 2015. Africa, quasi Roma (256 av. J.-C. – 711 apr. J.-C.). Paris: CNRS
editions.
Leitch, V., Duckworth, C., Cuénod, A., Mattingly, D.J., Sterry, M. and Cole, F.
2017. Early Saharan trade: The inorganic evidence. In Mattingly et al. 2017,
287–340.
Leschi, L. 1945. La mission scientifique au Fezzan. Archéologie. Travaux de
l’Institut de Recherches Sahariennes 3: 183–86.
Le Quellec, J.-L. 1987. L’art rupestre du Fezzan septentrional (Libye): Widyan Zreda
et Tarut (Wâdi esh-Shâti) (BAR International Series 365). Oxford: British
Archaeological Reports.
Lethielleux, J. 1948. Le Fezzan. Ses jardins, ses palmiers. Notes d’ethnologie et
d’histoire. Tunis: Imp. Bascone et Muscat.
Liverani, M. 2000. The Libyan caravan road in Herodotus IV.181–184. Journal of
the Economic and Social History of the Orient 43.4: 496–520.
Liverani, M. (ed.). 2006. Aghram Nadharif. The Barkat Oasis (Sha‘abiya of Ghat,
Libyan Sahara) in Garamantian Times. Firenze: All’Insegna del Giglio.
Liverani, M. 2007a. Cronologia e periodizzazione dei Garamanti. Acquisizioni e
prospettive. Athenaeum 95: 633–62.
Liverani, M. 2007b. La struttura sociale dei Garamanti in base alle recenti scoperte
archaelogiche. Rendiconti dell’Accademia Nazionale dei Lincei Ser 9.18: 155–204.
Liverani, M., Barbato, L., Cancellieri, E., Castelli, R. and Putzolu, C. 2013. The
survey of the Fewet necropolis. In Mori 2013, 199–252.
Mattingly, D.J. 2000. Garama. In R. Talbert (ed.), Barrington Atlas of the Greek and
Roman World. Princeton: Princeton University Press, map 36 and Map-by-Map
Directory pp. 545–51.
Mattingly, D.J. 2003. The Archaeology of Fazzan. Volume 1, Synthesis. London:
Society for Libyan Studies, Department of Antiquities.
Mattingly, D.J. 2004. Nouveaux aperçus sur les Garamantes: Un état saharien?
Antiquités africaines 37 (2001) [2004]: 45–61.
Mattingly, D.J. 2006. The Garamantes: The first Libyan state. In Mattingly et al.
2006, 189–204.
Mattingly, D.J. 2007. The Archaeology of Fazzan. Volume 2, Site Gazetteer, Pottery
and other Survey Finds. London: Society for Libyan Studies, Department of
Antiquities.
Mattingly, D.J. 2010. The Archaeology of Fazzan. Volume 3, Excavations Carried
out by C.M. Daniels. London: Society for Libyan Studies, Department of
Antiquities.
Mattingly, D.J. 2013. The Archaeology of Fazzan. Volume 4, Survey and Excavations
at Old Jarma (Ancient Garama) Carried out by C. M. Daniels (1962–69) and the

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 108 David J. Mattingly et al.

Fazzan Project (1997–2001). London: Society for Libyan Studies, Department of

Antiquities.
Mattingly, D.J. 2016. The Garamantes and after: The biography of a Central
Saharan oasis 400 BC – AD 1900. In G. Bartoloni and M.G. Biga (eds), Not
Only History: Proceedings of the Conference in Honor of Mario Liverani Held in
Sapienza–Università di Roma, Dipartimento di Scienza dell’Antichità,
20–21 April 2009. Winina Lake, Indiana: Eisenbrauns, 147–69.
Mattingly, D.J. 2017. The Garamantes and the origins of Saharan trade: State of the
field and future agendas. In Mattingly et al. 2017, 1–52.
Mattingly D.J. and Edwards D. 2003. Religious and funerary structures. In
Mattingly 2003, 177–234.
Mattingly, D.J. and MacDonald, K.C. 2013. Africa. In P. Clark (ed.), Oxford
Handbook of Cities in World History. Oxford: Oxford University Press, 66–82.
Mattingly, D.J. and Sterry, M. 2013. The first towns in the Central Sahara.
Antiquity 87.366: 503–18.
Mattingly, D.J. and Sterry, M. 2018. Zuwila and Fazzan in the 7th–10th
centuries AD. In G. Anderson, C. Fenwick and M. Rosser-Owen (eds), The
Aghlabids and Their Neighbours: An Interdisclipinary Workshop on Art &
Material Culture in Ninth-Century North Africa. Leuven: Brill, 551–72.
Mattingly, D.J., McLaren, S., Savage, E. al-Fasatwi, Y. and Gadgood, K. (eds). 2006.
The Libyan Desert. Natural Resources and Cultural Heritage. London: Society for
Libyan Studies.
Mattingly, D.J., Lahr, M., Armitage, S., Barton, H., Dore, J., Drake, N., Foley, R.,
Merlo, S., Salem, M., Stock, J. and White, K. 2007. Desert Migrations: People,
environment and culture in the Libyan Sahara. Libyan Studies 38: 115–56.
Mattingly, D.J., Dore, J. and Lahr, M. (with contributions by others). 2008. DMP II:
2008 fieldwork on burials and identity in the Wadi al-Ajal. Libyan Studies 39:
223–62.
Mattingly, D.J., Lahr, M. and Wilson, A. 2009. DMP V: Investigations in 2009 of
cemeteries and related sites on the west side of the Taqallit promontory. Libyan
Studies 40: 95–131.
Mattingly, D.J., Abduli, H., Aburgheba, H., Ahmed, M., Ali Ahmed Esmaia, M.,
Baker, S., Cole, F., Fenwick, C., Gonzalez Rodriguez, M., Hobson, M., Khalaf, N.,
Lahr, M., Leitch, V., Moussa, F., Nikita, E., Parker, D., Radini, A., Ray, N., Savage,
T., Sterry, M. and Schörle, K. 2010a. DMP IX: Summary report on the fourth
season of excavations of the Burials and Identity team. Libyan Studies 41: 89–104.
Mattingly, D.J., al-Aghab, S., Ahmed, A., Moussa, F., Sterry, M and Wilson, A.I.
2010b. DMP X: Survey and landscape conservation issues around the Taqallit
headland. Libyan Studies 41: 105–32.
Mattingly D.J., Hawthorne J. and Daniels C.M. 2010c. Excavations at the Classic
Garamantian settlement of Saniat Jibril. In Mattingly 2010, 123–204.
Mattingly, D.J., Abduli, H., Ahmed, M., Cole, F., Fenwick, C., Fothergill, B.T.,
Gonzalez Rodriguez, M., Hobson, M., Khalaf, N., Lahr, M., Moussa, F.,

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 109

Nikita, E., Nikolaus, J., Radini, A., Ray, N., Savage, T., Sterry, M. and Wilson, A.I.
2011. DMP XII: Excavations and survey of the so-called Garamantian Royal
Cemetery (GSC030-031). Libyan Studies 42: 89–102.
Mattingly, D.J., Rushworth, A., Sterry, M. and Leitch, V. 2013a. The African
Frontiers. Frontiers of the Roman Empire series, general editor, D. Breeze.
Edinburgh: Hussar.
Mattingly, D.J., Sterry, M. and Leitch, V. 2013b, Fortified farms and defended
villages of Late Roman and Late Antique Africa. Antiquité Tardive 21: 167–88.
Mattingly, D.J., Sterry, M. and Edwards, D. 2015. The origins and development of
Zuwila, Libyan Sahara: An archaeological and historical overview of an ancient
oasis town and caravan centre. Azania 50.1: 27–75. Available at: http://dx
.doi.org/10.1080/0067270X.2014.980126 [last accessed 4 September 2019].
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017. Trade in the Ancient Sahara and Beyond. Trans-Saharan
Archaeology Volume 1. Series editor D.J. Mattingly. Cambridge: Cambridge
University Press and The Society for Libyan Studies.
Mattingly, D.J., Sterry, M., al-Haddad, M. and Bokbot, Y. 2018. Beyond the
Garamantes: The early development of Saharan oases. In Purdue et al. 2018a,
205–28.
Mattingly, D.J., Gatto, M., Ray, N. and Sterry, M. 2019. Dying to be Garamantian:
Burial and migration in Fazzan. In Gatto et al. 2019a, 51–107.
Mercuri, A.M., Bosi, G. and Buldini, F. 2013. Seeds, fruit and charcoal from the
Fewet compound. In Mori 2013, 177–90.
Merlo, S., Hakenbeck, S.E. and Balbo, A. 2008. DMP IV: 2008 fieldwork on historic
settlement in the Wadi ash-Shati and the Dawada lake villages. Libyan Studies
39: 295–98.
Merlo, S., Hakenbeck, S. and Balbo, A.L. 2013. Desert Migrations Project XVIII:
The archaeology of the northern Fazzan: A preliminary report. Libyan Studies
44: 141–61.
Mori, L. 2010. Between the Sahara and the Mediterranean coast: The archaeological
research in the oasis of Fewet (Fazzan, Libyan Sahara) and the rediscovery of the
Garamantes. In M. Dalla Riva (ed.), Meetings between Cultures in the Ancient
Mediterranean, XVII Congresso Internazionale di Archeologia Classica, Roma
22–26 settembre 2008. Rome: Bollettino di Archeologia, volume speciale, 17–29.
Mori, L. 2012. Fortified citadels and castles in southern Fazzan (Libyan Sahara) in
Garamantia times. In F. Jesse and C. Vogel (eds), The Power of Walls.
Fortifications in Ancient Northeastern Africa. Cologne: Heinrich-Barth-Institut
of Cologne, 195–216.
Mori, L. (ed.). 2013. Life and Death of a Rural Village in Garamantian Times. The
Archaeological Research in the Oasis of Fewet (Libyan Sahara), AZA
Monographs 6. Firenze: All’Insegna del Giglio.
Mori, L. and Ricci, F. 2013. The excavation of the Fewet necropolis. In Mori 2013,
253–317.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 110 David J. Mattingly et al.

Mori, L., Gatto M.C., Zerboni A. and Ricci F. 2013. Life and death at Fewet. In Mori
2013, 375–87.
Nachtigal G. 1974. Sahara and Sudan, vol I, Tripoli and Fezzan, Tibesti or Tu.
Translated from the German by A.G.B. and H.J. Fisher. London: C. Hurst
and Co.
Pace, B., Sergi, S. and Caputo, G. 1951. Scavi sahariani. Monumenti Antichi 41:
150–549.
Pelling, R. 2005. Garamantian agriculture and its significance in a wider North
African context: The evidence of the plant remains from the Fazzan Project. The
Journal of North African Studies 10.3–4: 397–411.
Pelling, R. 2008. Garamantian agriculture: The plant remains from Jarma, Fazzan.
Libyan Studies 39: 41–71.
Pelling, R. 2013a. The archaeobotanical remains. In Mattingly 2013, 473–94.
Pelling, R. 2013b. Botanical data appendices. In Mattingly 2013, 841–52.
Pliez, O. 2004. La fin de l’état demiurge? Les nouvelles facettes de l’urbain dans le
Sahara libyen. Autrepart 31: 59–74.
Putzolu, C. 2006. The topography of the site. In Liverani 2006, 29–40.
Ruprechtsberger, E.M. 1997. Die Garamanten, Geschichte und Kultur eines
Libyschen Volkes in der Sahara. Mainz: Verlag Philipp von Zabern.
Sahara Italiano 1937 = Il Sahara Italiano, I. Fezzan e oasi di Gat. Rome: Società
Italiana arto grafiche.
Scarin, E. 1934. Le oasi del Fezzan, 2 vols. Bologna: Nicola Zanichelli.
Scarin, E. 1937a. Insediamenti e tipi di dimore. In Sahara Italiano 1937, 515–60.
Scarin, E. 1937b. Descrizione delle oasi e gruppi di oasi. In Sahara Italiano 1937,
603–44.
Scarin, E. 1937c. L’insediamento umano della zona Fezzanese di Gat. Firenze:
Centro di studi coloniali.
Sterry, M. and Mattingly, D.J. 2011. DMP XIII: Reconnaissance survey of archae-
ological sites in the Murzuq area. Libyan Studies 42: 103–16.
Sterry, M, and Mattingly, D.J. 2013. Desert Migrations Project XVII: Further AMS
dates for historic settlements from Fazzan, south-west Libya. Libyan Studies 44:
127–40.
Sterry, M, Mattingly, D.J. and Higham, T. 2012. Desert Migrations Project XVI:
Radiocarbon dates from the Murzuq region, southern Libya. Libyan Studies 43:
137–47.
Thiry, J. 1995. Le Sahara Libyen dans l’Afrique du Nord Medievale. Leuven:
Orientalia Lovaniensia Analecta, 72.
Turk, T.M., Doughri, A.K. and Banerjee, S. 1980. A review of the recent investiga-
tion on the Wadi ash-Shati iron ore deposits, Northern Fazzan, Libya. In
M.J. Salem and M.T. Busrewil (eds), The Geology of Libya III. London:
Academic Press, 1019–43.
Van der Veen, M. 1992. Garamantian agriculture: The plant remains from
Zinchecra, Fezzan. Libyan Studies 23: 7–39.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 2 Garamantian Oasis Settlements in Fazzan 111

Van der Veen, M. and Westley, B. 2010. Palaeoeconomic studies. In Mattingly

2010, 488–522.
Wilson, A.I. 2006. The spread of foggara-based irrigation in the ancient Sahara. In
Mattingly et al. 2006, 205–16.
Wilson, A.I. 2009. Foggaras in ancient North Africa or how to marry a Berber
princess. In Contrôle et distribution de l’eau dans le Maghreb antique et medieval.
Rome: CEFR, 19–39.
Wilson, A.I. and Mattingly, D.J. 2003. Irrigation technologies: Foggaras, wells and
field systems. In Mattingly 2003, 234–78.
Wilson, A.I., Mattingly, D.J. and Sterry, M.S. Forthcoming. The diffusion of
irrigation technologies in the Sahara in antiquity: Settlement, trade and migra-
tion. In C. Duckworth, A. Cuénod and D.J. Mattingly (eds), Mobile Technologies
in the Sahara and Beyond. Cambridge: Cambridge University Press.
Ziegert, H. 1969. Überblick zur jüngeren Besiedlungsgeschichte des Fezzan.
Berliner Geographischen Abhandlungen 8: 49–58.

https://doi.org/10.1017/9781108637978.003 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern
Sahara
david j. mattingly, martin sterry, louise rayne
and muftah al-haddad

Introduction

This chapter will review the evidence of early oasis development in

Western Egypt and Eastern Libya, broadly following the course of the
‘route of the oases’, running west from the Nile to Siwa, then onwards
to Awjila and al-Jufra in Libya, where it met the major north-south
route from the Mediterranean to Garamantian Fazzan and beyond to
Chad.1 The evidence presented for pre-Islamic oasis development is
particularly strong in this part of the Sahara; indeed the origins of
agriculture at some of the Egyptian oases went back to the third
millennium BC and the route as a whole seems to have been well-
developed by the fifth century BC.
We suggest that the ultimate origins of oasis agriculture in the Western
Desert are to be sought in the Nile Valley and the Fayum, with a package of
plants and irrigation techniques first developed there, then adopted in the
oasis depressions of the Western Desert – notably Kharga, Dakhla, Farfara,
Bahariya and Siwa (Fig. 3.1). From Siwa the line of oases was extended
along the northern edge of the great Libyan sand sea, to al-Jiarabub and the
Awjila group to the south of Cyrenaica (Fig. 3.2). Beyond Awjila, the Syrtic
desert crossing was facilitated by the existence of a number of small oases,
Marada, Zala, Tagrifet. The al-Jufra group at the south-western boundary
of Syrtica has been little explored hitherto, but new satellite image analysis
and some ground visits allow a new appreciation of the importance of this
cluster.2

1
On the oasis route, see Herodotus 4.181–184; Liverani 2000; 2006, 445–56; Rebuffat 1970b.
2
The account below summarises past publication and new work carried out as part of the Trans-
112 SAHARA project.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 113

Figure 3.1. Western Desert showing main regions and sites discussed in the text.

Western Desert of Egypt

The next chapter by Boozer presents a more detailed picture of the Roman
era settlement in the Western Desert oases and in consequence we only
highlight some of the main bibliography and provide a concise outline for
that area here. The quantity of archaeological research conducted in the
Western Desert is disproportionately large in comparison to the rest of the
Sahara, with a bibliography now running to hundreds of items, reflecting
the work of a number of large international and Egyptian teams working in
the oases over many years.3 However, the orientation of research is to
a large extent Egyptological, with the gaze primarily directed east to the
Nile, rather than further west along the Saharan trails.4 To the extent that

3
See for instance, the excellent online bibliography of the Dakhla oasis project, available at: www
.amheida.org/inc/pdf/amheida_bibliography.pdf [last accessed 4 September 2019].
4
See Kuhlmann 2013 for a rare exception.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 114 David J. Mattingly et al.

Figure 3.2. Eastern Libya showing main regions and sites discussed in the text.

the Libyan population of the desert are considered, it is mainly as pastoral

and warlike people who needed to be pacified and controlled by the ruling
power of Egypt.5 This is, of course, a reflection of the influence and impact
of Pharaonic and later Hellenistic and Roman overlordship of these oases,
backed up by some profound shared cultural preferences (temple archi-
tecture, hieroglyphs, mummies, etc.). In Classical writings, the perspective
of the nature of desert peoples became even more based around crude
stereotypes.6 Approaches that stress the external interventions of the

5
Bates 1914; Leahy 1990; Midant-Reynes and Tristant 2008; O’Connor and Reid 2003. Cf.
Bowman 1989, 12: ‘To the east and west . . . lay inhospitable or mountainous desert. The western
desert was punctuated by a series of oases, supporting a small population . . . their secure
occupation was an important factor in controlling incursion or potential disruption by bands of
desert nomads.’
6
Colin 2000 provides a particularly uncritical perspective on the Libyan peoples between
Cyrenaica and Egypt in the Classical sources, taking at face value the emphasis on pastoralism
(40–86), with only limited mention of oasis cultivation (115–19) and a complete denial of the
existence of long-range trade in the Sahara (45–57). Cf. Mattingly 2003, 76–90.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 115

Pharaonic state of Egypt or its Hellenistic and Roman successors in the

development of the oases miss an important element of the story, not just
in the opening chapters of oasis formation, but in relation to the Sahara
overall. The Sahara was not empty space when oases were established and
some of the pre-existing populations were important agents in the process.
The later prehistory of the Eastern Sahara has revealed the existence
there of wide-ranging pastoral peoples, whose lives were transformed
following climatic change around 5,000 years ago.7 Some of these Libyan
peoples no doubt sought refuge within the Nile Valley and played a part in
the transformation of Nilotic society at that time, but others seem to have
remained in the desert regions, now focused increasingly on the springs
and relict lakes of the depressions, where oases were later to appear. The
communications between the Nilotic peoples and the Libyans of the
Western Desert were important and these groups had symbiotic reasons
as well as issues of neighbourly competition and warfare. Though the
Egyptian sources emphasise the pastoral and barbaric character of the
Libyan peoples, the representations of them show them when in battle
having access to the high technology of warfare – metal swords, chariots –
and being able to deliver significant levels of tribute. Although many
accounts of the Libyan peoples (variously called Meshwesh, Libu and
Tehenu) still emphasise that these were nomadic peoples, it seems more
plausible to consider that by the first millennium BC (and perhaps con-
siderably earlier) these were societies comprising both pastoral and oasis
cultivating elements. In addition, trade between the Nile and the desert
regions seems to have been initiated early in the Old Kingdom.8
The peculiar hydrology of the River Nile, with its annual floods, created
what was in effect an immense linear oasis of cultivated land along its
course. From there, irrigation techniques and a package of cultivated crops
spread into surrounding desert areas. The earliest oases in the Sahara are
plausibly to be sought in the Western Egyptian Desert, where there are
certain traces of Old Kingdom (third millennium BC) activity at several
locations (see below). Although the settlement trajectory was not always
consistent, by the Roman era there was very extensive settlement within
these oasis zones, which fall into several main groups.9 An important

7
Jennerstrasse 8 2002; cf. Barich and Hassan 1990; McDonald 2003; Wendorf and Schild 1980;
Wendorf et al. 1984.
8
See the important article by Kuhlman 2002, arguing for its existence by the III or IV Dynasty
(mid-third millennium BC). Also now, Morkot 2016.
9
See Mattingly 2000a, map 73; Müller-Wollermann 2000, map 75; Wagner 2000, map 79, for the
Oasis Magna, Oasis Parva and Siwa.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 116 David J. Mattingly et al.

recent overview of Ptolemaic activity in the oases has shown that pre-
Roman activity was much more extensive and well developed than some-
times imagined.10 The oases of the Western Desert as a group were well
connected with the Nile Valley from early times and were frequently
mentioned in historical sources both of the Pharaonic periods and in
later Classical times.11 Nonetheless, the oases were distinctively different
hydrologically compared with the Nile Valley in that water, though a scarce
resource, was available all year round. The water also emerged naturally
from artesian thermal springs in these depressions, obviating the need to
search for it.12 Both these aspects favoured early development of oasis
agriculture as a complement to the lands irrigated annually by the Nile
flood. Table 3.1 presents the relevant radiocarbon dates for sites discussed
in this chapter.

The Fayyum
Arguably, the earliest oases of the Western Desert were those of the
Fayyum.13 The Fayyum is a triangular depression, c.100 × 90 km, with
a 40 km long lake along its northern edge that was sacred to the crocodile
god.14 Sedentary settlement and agriculture in the depression around the
Fayyum lake began during the Old Kingdom (2686–2181 BC). However,
the major expansion of agriculture and population in Fayyum began
during the Middle Kingdom (2055–1650 BC) when Nile floodwaters
were diverted into the depression to enhance irrigation there. There were
114 villages in Fayyum in Hellenistic times and this intensity of activity
continued into Roman times, with well-developed towns such as Boubastis,
Dionysias, Karanis, Philadelphia, Theadelphia and many villages.15
Detailed plans of a number of the Fayyum cities have been established
and their overall location is very interesting within the depression, in that
many of them were established around the margins of the lakes and the
fringes of the cultivated zone.16 Although the Fayyum is highly atypical of
the other Saharan oases, because of the augmentation of the irrigation

10
Gill 2016.
11
Abboudy Ibrahim 1992; Abd el-Ghany 1992; Bagnall 1997; Ball 1942; Giddy 1987; Reddé 1989;
Wagner 1987.
12
Gautier 1970, 139–59, for a general overview of hydrology of the Egyptian oases.
13
Vivian 2000, 211–17. 14 Müller-Wollermann 2000, 1125–26.
15
Many toponyms mentioned in papyri cannot be located precisely, as is evident from Müller-
Wollermann 2000; Rathbone 1996. It is clear, however, that the Fayyum was very densely
inhabited by the Roman period and a highly productive agricultural region.
16
Davoli 1998; 2012. See also Carpentiero 2016 for discussion of town planning in Fayyum.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 Table 3.1 Protohistoric and historic radiocarbon dates from the Eastern Sahara

Area/Site Context Material Reference Date (BP) Calibrated

Kharga
Abu Ziyar Cooking remains Charcoal Not given 3605±48 2134–1781 calBC
Kysis Mudbrick (temple) Charcoal Gif? 2260±60 415–166 calBC
Al-Deir Irrigation canal Charcoal CEDAD-LTL-13096A 2138±40 356–49 calBC
Al-Deir Irrigation canal Charcoal UCIAM-76668/ULA-1623 1770±15 calAD 226–331
Al-Deir Palaeosoil Charcoal UGAMS-1124/ULA-3179 2810±25 1025–901 calBC
Al-Deir Palaeosoil Charcoal UGAMS-11244/ULA-3180 2210±25 364–202 calBC
Al-Deir Palaeosoil Charcoal UCIAM-76668/ULA-1625 2130±15 204–96 calBC
Al-Deir Palaeosoil Charcoal Poz-56029 1850±15 calAD 91–231
Dakhla
Yardang AM66 Palaeosoil Charcoal LTL-8373 2805±34 1049–848 calBC

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

Yardang GS001 Palaeosoil Charcoal LTL-8364 2710±55 980–797 calBC
Yardang GS001 Palaeosoil Charcoal LTL-8369 2599±36 835–590 calBC
Yardang GS001 Palaeosoil Charcoal LTL-8366 2527±40 799–540 calBC
Spring GS022 Palaeosoil Charcoal LTL-8368 2315±42 510–210 calBC
Yardang AM72 Palaeosoil Charcoal LTL-8375 2220±33 379–202 calBC
Yardang GS001 Palaeosoil Charcoal LTL-8365 2186±40 377–118 calBC
Spring GS022 Palaeosoil Charcoal LTL-8367 2101±35 342–40 calBC
Yardang AM70 Palaeosoil Charcoal LTL-8374 2062±35 176 calBC–calAD 17
Yardang GS011 Palaeosoil Charcoal LTL-8372 1935±33 21 calBC–calAD 133
Yardang GS011 Palaeosoil Charcoal LTL-8370 1933±33 21 calBC–calAD 134
Ein Tirghi cemetery Burial E31 Sk25 Human remains TO-4476 2745±70 1071–796 calBC
Ein Tirghi cemetery Burial E31 Sk36 Human remains TO-4476 2750±60 1030–803 calBC
Kellis 1 cemetery Mummy 1 Human tissue GX-19942 2130±75 374 calBC–calAD 4
 Table 3.1 (cont.)

Area/Site Context Material Reference Date (BP) Calibrated

Kellis 1 cemetery Mummy 1 Human tissue Beta-119805 1940±60 88 calBC–calAD 230

Kellis 1 cemetery Mummy 1 Textile Beta-119803 1960±40 43 calBC–calAD 125
Kellis 1 cemetery Mummy 2 Textile Beta-119811 1860±60 calAD 23–326
Kellis 1 cemetery Mummy 3 Human tissue GX-20367 2075±60 351 calBC–calAD 58
Kellis 1 cemetery Mummy 4 Human tissue Beta-119806 2210±50 395–164 calBC
Kellis 1 cemetery Mummy 4 Human tissue Beta-170565 2010±40 156 calBC–calAD 75
Kellis 1 cemetery Mummy 4 Human tissue Gx-19943 2280±75 730–116 calBC
Kellis 1 cemetery Mummy 4 Human tissue Beta-119810 2280±70 703–164 calBC
Kellis 1 cemetery Mummy 5 Human tissue Gx-20366 1830±60 calAD 57–341
Kellis 1 cemetery Mummy 5 Human tissue Beta-119809 3070±70 1496–1126 calBC
Kellis 1 cemetery Mummy 5 Resin Gx-19941 2225±105 728 calBC–calAD 4
Kellis 1 cemetery Mummy 5 Resin Beta-119802 2320±40 511–214 calBC
Kellis 1 cemetery Mummy 6 Human tissue Gx-19944 2020±75 347calBC–calAD 133

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

Kellis 1 cemetery Mummy 7 Human tissue Gx-19945 2335±155 800–55 calBC
Kellis 1 cemetery Mummy 7 Human tissue Gx-19946 2005±105 356 calBC–calAD 238
Kellis 1 cemetery Mummy 7 Resin Beta-120426 2580±50 836–541 calBC
Kellis 1 cemetery Mummy 7 Resin Beta-170566 2660±40 901–792 calBC
Kellis 1 cemetery Mummy 7 Resin Beta-172212 2660±40 901–792 calBC
Kellis 1 cemetery Mummy 7 Textile Gx-19947 2695±75 1051–760 calBC
Kellis 1 cemetery Mummy 7 Textile Beta-119804 1950±40 41 calBC–calAD 129
Kellis 1 cemetery Mummy 8 Human tissue Beta-143632 1890±40 calAD 28–230
Kellis 1 cemetery Mummy 8 Human tissue Beta-143633 1870±30 calAD 73–226
Kellis 1 cemetery Mummy 8 Human tissue Beta-119807 1800±60 calAD 82–380
Kellis 1 cemetery Mummy 9 Human tissue GX-20368 2245±60 406–167 calBC
Kellis 1 cemetery Mummy 9 Human tissue Beta-170567 2240±40 393–204 calBC
Kellis 1 cemetery Mummy 9 Human tissue Beta-172213 2160±40 361–92 calBC
 Kellis 1 cemetery Mummy 10 Human tissue Beta-119807 1800±60 calAD 82–380
Kellis 1 cemetery Mummy 15 Human tissue Gx-19938 1880±95 91 calBC–calAD 383
Kellis 1 cemetery Mummy 15 Human tissue Gx-19937 2205±90 416 calBC–calAD 3
Kellis 1 cemetery Mummy 15 Textile Gx-19939 2515±375 1607 BC–calAD 233
Kellis 1 cemetery Mummy 15 Wood Gx-19940 2005±115 358 calBC–calAD 240
Kellis 2 Cemetery Burial Human bone TO-5475 1840±60 calAD 52–355
Kellis 2 Cemetery Burial K2-B6 Human bone TO-5476 1670±50 calAD 244–535
Kellis 2 Cemetery Burial K2-B6 Human bone TO-6104 1600±50 calAD 344–569
Kellis 2 Cemetery Burial Human bone TO-6256 1720±50 calAD 145–423
Kellis 2 Cemetery Burial Human bone TO-6257 1920±50 38 calBC–calAD 219
Kellis 2 Cemetery Burial Human bone TO-6013 1590±50 calAD 352–580
Kellis 2 Cemetery Burial Human bone TO-10274 1800±50 calAD 85–345
Kellis 2 Cemetery Burial Human bone TO-10275 1840±50 calAD 65–326
Kellis 2 Cemetery Burial Human bone TO-10276 1850±50 calAD 56–322
Kellis 2 Cemetery Burial Human bone TO-10277 1780±40 calAD 131–377

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

Kellis 2 Cemetery Burial Human bone TO-10278 1880±40 calAD 53–236
Bahariya
Qasr Allam Mudbrick structure Not given Not given 2402±50 753–394 calBC
Qasr Allam Mudbrick structure Not given Not given 2410±53 756–396 calBC
Qasr Allam Level above child inhumations Date stone Not given 1585±45 calAD 386–575
Abu Ballas trail
Staging post From underneath stashed pots Basket UtC-8868 3226±88 1732–1290 calBC
Abu Ballas 85/55–2 With cups and bowls Charcoal KIA-20684 3785±25 2289–2140 calBC
Abu Ballas 85/55–2 Hearth Charcoal KIA-20682 3675±35 2192–1952 calBC
Abu Ballas 85/55–2 Donkey droppings Barley KIA-23062 3095±30 1429–1280 calBC
Jaqub 00/20 Storage Jar Barley Poz-23221 3520±35 1939–1749 calBC
Jaqub 99/30–1 Hearth Charcoal Erl-2876 3232±53 1627–1412 calBC
Jaqub 99/30–1 Hearth Charcoal Erl-2877 2963±52 1377–1015 calBC
 Table 3.1 (cont.)

Area/Site Context Material Reference Date (BP) Calibrated

Jaqub 99/30–1 Hearth Charcoal KN-5359 2895±40 1211–946 calBC

Jaqub 99/30–1 Hearth Date stone KIA-12423 2880±35 1194–936 calBC
Jaqub 99/31–3 With cups and bowls Charcoal KIA-20683 3755±30 2284–2041 calBC
Jaqub 99/31–3 Charcoal KIA-21008 3240±20 1607–1446 calBC
Al-Jiarabub
Site not clear Funerary bed within tomb Wood GX-30372 1980±70 173 calBC–calAD 210
Awjila
Al-Darb al-Kabir Rock-cut burial Textile OxA-33721 2177±31 362–164 calBC
Al-Jufra

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

Delbak (WDN001) From mudbrick wall of settlement Charcoal OxA-33869 1281±32 calAD 659–776
Delbak (WDN001) From mudbrick wall of settlement Seeds OxA-33964 7590±40* Date rejected
HUN006 From mudbrick wall of settlement Charcoal OxA-33870 996±27 calAD 987–1151
HUN006 From mudbrick wall of settlement Plant remains OxA-33871 1257±25 calAD 672–862
HUN003 From mudbrick wall of settlement Charred seeds OxA-33872 97±27 calAD 1685–1928
HUN003 From mudbrick wall of settlement Plant remains OxA-33873 79±25 calAD 1692–1920
South-east Libya
No data No data Tamarisk Hv-3761 2300±145 780–51 calBC
No data No data Wood Hv-3762 1625±145 calAD 85–659
 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 121

system by canalised Nile floodwaters, the region was probably of crucial

importance for the evolution of irrigation and cultivation technologies in
a fundamentally desert landscape.

Kharga
Kharga (c.200 km west of the Nile to the south of Fayyum) was known with
neighbouring Dakhla in Roman times as Oasis Magna (Great Oasis). The
depression, bordered by escarpments to the east and north, is 220 km long
north-south, but only 15–40 km wide east-west. There are two discrete
zones of oasis settlement within the basin, towards its northern and south-
ern ends, each dominated by a major town – Hibis in the north and Kysis
(Dush) in the south.17 The archaeology has been examined by a number of
projects in recent decades.18 There is textual and archaeological evidence of
Pharaonic activity here and, though not proven archaeologically as yet, this
must extend back here (as at Dakhla further west, see below) to the Old
Kingdom. There were three major sites in the Roman period,19 but numer-
ous other villages (some fortified) and traces of ancient activity, including
important temple complexes at both major centres and some smaller
settlements. The Christian cemetery at Bagawat close by Hibis is notable
for its domed and painted mausolea.20
An important aspect of recent work has been the thorough investigation
of extensive oasis garden areas irrigated by foggaras. Some of the foggaras
have been dated to the fifth and fourth centuries BC and seem to link also
to documentary records post-dating the Persian take-over of Egypt, which
probably marked the introduction of this technology from the Persian
heartlands.21 The documentary evidence relating to the irrigation regime
in the Kharga oasis during the Persian period provides confirmation of the
importance and sophistication of oasis agriculture by this date.22
The many fortified sites in the oasis have traditionally been interpreted
as Roman military garrison points. While some undoubtedly were Roman
military bases, others were potentially of more local origin, especially

17
Jackson 2002, 163–97; Reddé et al. 2004; Vivian 2000, 52–105; Wagner 2000.
18
See inter alia, Bravard et al. 2016b; Dunand et al. 2010; 2012; 2013; Fakhry 1942; Gill 2016,
130–34; Ibrahim 2013; Tallet et al. 2013.
19
In addition to Hibis and Kysis, there was another extensive settlement in the northern area
known today as Ain Umm Dabadib, Jackson 2002, 190–91; Rossi 2000.
20
Jackson 2002, 180–83.
21
Agut-Labordère 2018; Bousquet 1996; Gonon 2018; Wuttman et al. 2000; cf. Wilson 2006.
22
Chaveau 2001; 2005; Newton et al. 2013.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 122 David J. Mattingly et al.

bearing in mind the parallels of late antique fortified settlements in Fazzan

(described in the previous chapter).23
An important aspect of Kharga is that it was the point of departure of
important overland routes to Nubia and to the Chad/Darfur region, the so-
called 40-day route, along which a number of minor oases developed (such
as Selima).24 The route that followed a desert detour to the Nubian centres
on the Upper Nile was in many ways more practicable than following the
course of the river south. We need more information on the oases of
northern Sudan, but it is a reasonable assumption that their origins were
also early.

Dakhla
The Dakhla oasis group (c.150 km west of Kharga) has also been the focus
of intensive archaeological research for many years.25 The oasis depression
measures c.70 km east-west and 30 km north-south, with steep escarp-
ments marking the north and east limits.26 Roman era settlement and oasis
cultivation fell into three main areas: an eastern area with a main site near
Tineida, a central area with two main towns, Kellis and Mothis, and
a western area with a main urban site called Trimithis.27 There is now
good excavated evidence of Old Kingdom activity, notably at Ain Asil
(Balat), evidently the Old Kingdom capital.28 The fortified town of Ain
Asil was the base for the pharaoh’s governors in the oasis, as several burial
monuments and important epigraphic finds attest. Some accounts are
cautious regarding the development of the oasis at this early date, seeing
Ain Asil as a control fort, from which trading parties could be sent towards
Nubia.29 However, the accumulating archaeological evidence of other Old
Kingdom sites calls this into question. There was another 5-ha mudbrick
settlement called Ain al-Gazzareen, and around 30 other locations have
produced evidence of Pharaonic activity.30 The documentation from Ain
23
Jackson 2002, 166–67, 176–77, 183–89; Reddé 1999; Rossi 2013; Vivian 2000, 65–66; Wagner
1987.
24
Kirkwan 1971; Förster and Riemer 2013; Morkot 1996; 2016; Welsby 1996.
25
See inter alia, the Dakhla oasis Project monograph series (15 volumes), notably Bagnall et al.
2013; Bowen and Hope 2003; Hope and Bowen 2002; Hope and Mills 1999; Wiseman 2008.
26
Jackson 2002, 197–227; Vivian 2000, 106–42.
27
Davoli 2012; 2013; Warner 2013, for a good summary study of the towns.
28
Valloggia 2004, for an overview account of excavations at the Old Kingdom site of Balat/Ain
Aseel/Ayn Asil.
29
Boozer 2015b, 7–10.
30
Mills 2002; 2013; Mills and Kaper 2003; Pettman 2013; Smekalova et al. 2003. See also Hope and
Pettman 2013; Mills 1999a for a summary of evidence of Pharaonic activity in Dakhla.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 123

Asil indicates the collection of agricultural produce among other things,

presumably from oasis farmers within the depression. There are also
references to goods that appear to be traded (including commodities like
salt, carnelian, ochre) and possible stockpiling of goods for carriage on
Saharan trails.31 From the earliest times, therefore, Dakhla oasis appears to
have had a double focus on agriculture and trade.
Ptolemaic activity in the oasis is now known to have been far more
extensive than at one time believed.32 By Hellenistic and Roman times the
main centres were at Ismant al-Kharab (Kellis)33 and Amheida (Trimithis)
and another substantial site existed at Mut al-Kharab.34 The extensive
excavations at Kellis have revealed a complex Roman town, where substan-
tial temple complexes, some decorated with elaborate wall paintings, later
gave way to Christian churches, and where bath-houses and wealthy houses,
lay alongside extensive zones of manufacturing activity (furnaces and
kilns).35 Trimithis was a particularly large urban site at c.100 ha, though
excavation there has mainly focused on domestic housing of Roman date,
with important results.36 Geoarchaeological survey around Trimithis has
identified the formation of irrigation soils from the Late Pharaonic period
and degradation of the soils in the third and fourth centuries AD.37
Cemeteries have been identified close to many of the main settlements.38
A notable feature of the Dakhla oasis is the relative lack of fortified settle-
ments there (only two compared with ten or more at Kharga), perhaps
related to the fact that west of Dakhla extends a great sand sea and security
may have been less of an issue than in Kharga which could be accessed by
a multiplicity of routes.39

Abu Ballas
At 200 km from Dakhla on the trail towards Gilf Kebir lies Abu Ballas
(literally ‘pottery hill’), one of a string of archaeological sites.40 Here there
is a scatter of late Old Kingdom or early First Intermediate Period ceramics
(c.2200–2100 BC) making it the earliest evidence of incipient Trans-
Saharan networks. This and the other sites have no suitable water sources
31
Valloggia 2004, 99–102, 144–49. 32 Gill 2016, especially 97–127 for the sites.
33
Hope and Bowen 2002 (168–69 for the overall plan); Knudstad and Frey 1999.
34
Boozer 2013a; 2013b; Jackson 2002, 213–16. For Mut al-Kharab, see Gill 2016, 19–41.
35
See in particular, various papers in Hope and Bowen 2002. Also Mills 1999b on pottery
production in the oasis.
36
Boozer 2010; 2011; 2012; 2015a. 37 Bravard et al. 2016a.
38
Aufderheide et al. 2004; Bashendi 2013; Stewart et al. 2003. 39 Kucera 2013.
40
Förster 2013; 2015.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 124 David J. Mattingly et al.

nearby and all food and water appears to have been carried in by donkey to
create a supply depot. The trail was in use through the second
millennium BC, but there is very little evidence of use in more recent
periods. Of particular significance are plant remains of dates and barley
from 1194–936 calBC to 1429–1280 calBC, respectively.41 This is the ear-
liest evidence of these important crops west of the Egyptian oases.

Farfara
Some 310 km north of Dakhla and 175 km south-west of Bahariya is the
small depression of Farfara (or Farafra).42 Only a couple of relatively small
settlements and three cemeteries are known in the oasis, all evidently of
Roman date, though textual sources seem to indicate Pharaonic origins
here.43 There are indications in recent work that the ancient activity in the
oasis was in fact more extensive.44 Libyan inscriptions are known from the
Farfara basin, but have not been transcribed.45

Bahariya
Bahariya is a substantial oasis depression 350 km west of Cairo (94 km
north-south and 42 km east-west maximum). Together with the outlying
oasis depression of Farfara to the south, Barhariya was known as Oasis
Parva (Lesser Oasis) in antiquity. The Barhariya oasis group lies c.500 km
north of the Great Oasis and 240 km west of Fayyum. Pharaonic (Middle
Kingdom) and Roman activity have long been known here.46 Bahariya
consists of two separate areas of oasis, separated by c.40 km of desert:
a northern area around al-Qasr/Bawiti (ancient Psobthis) and a southern
area sometimes referred to as al-Hayz oasis. Around Bawiti there is
a standing Roman triumphal arch and Fakhry uncovered a temple dedi-
cated to Alexander the Great,47 while there has been an important dis-
covery of unrobbed Roman era mummified burials at Bawiti in the 1990s.48
More recent work has also focused on sites in the southern parts of the

41
Förster 2015.
42
Barich and Hassan 1990; Beadnell 1901, Gallinaro 2018, for topographical setting and
prehistoric evidence.
43
Fakhry 1974; Jackson 2002, 230–32; Mattingly 2000a; Vivian 2000, 143–73.
44
Gill 2016, 134–35. 45 Jackson 2002, 297, n. 5.
46
Ball and Beadnell 1903; Fakhry 1942; 1942/1950; 1974; Gill 2016, 135–37; Gosline 1990; Jackson
2002, 233–39; Vivian 2000, 174–212. For the latest dating evidence of Pharaonic activity, see
Dospěl and Suková 2013, 185–67.
47
Fakhry 1942; 1942/1950. 48 Hawas 2000.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 125

basin.49 Here again irrigation seems to have been improved at some point
by the introduction of foggara technology.50 In the al-Hayz cluster 11
settlements of Pharaonic or Hellenistic or Roman/Byzantine date have
been identified and there were at least nine settlements in the northern
oasis.51 Some of the largest settlements like al-Ris are urban in scale.52
There were at least two major fortresses in the Bahariya depression, at al-
Hayz in the south and Qasr Muharib in the north.

Siwa
Since the first reports of lost Persian armies and the pilgrimage of Alexander
the Great, the oasis of Ammon (Siwa) has always been the most celebrated spot
in the Eastern Sahara. There was a hot spring here that reputedly (though
inaccurately) varied in temperature between night and day, but its greater fame
was due to its status as the great oracular centre of the desert god Ammon.53
The relative difficulty of access has merely deepened Siwa’s mystique.54 The
Siwa depression is c.80 km long east-west with a series of spring-fed lakes
between the north side of the great Libyan Sand Sea and south-west of the
Qattara depression. There are a few indications of outlying settlements and
cemeteries in the direction of Barhariya to the south-east and al-Jiarabub
c.50 km to the north-west is arguably an associated secondary depression of
the Siwa group.55 There are at least 14 known ancient cemeteries spaced out
along the depression, implying at least an equivalent number of settlements, of
which eight can be demonstrated archaeologically and several of which were
ornamented by temples using ashlar masonry.56 The main temple of the oracle
can be identified with a substantial structure of the sixth century BC at
Aghurmi, 4 km from modern Siwa at the heart of the oasis. Many of the rock-
cut tombs of the ancient cemeteries contained mummified bodies.

Al-Jiarabub
As noted, the oasis cluster of al-Jiarabub is really a north-western outlier of the
Siwa group, lying only 50 km from the north-western edge of the Siwa
depression. However, today it is located mainly on the Libyan side of the
49
Dospěl and Suková 2013. 50 Dospěl and Suková 2013, 271–85.
51
Dospěl and Suková 2013, 6–7. 52 Dospěl and Suková 2013, 32.
53
Ball 1942 for the sources. See also Fakhry 1944; 1950; 1973.
54
Belgrave 1923; Hornemann 1802, 14–29; Prorok 2001; Rohlfs 2002a/b.
55
Herschend 2009, 298–321 for an unexpectedly detailed study of Siwa in a book on the Swedish
Iron Age; Mattingly 2000a.
56
Gill 2016, 137–40; Jackson 2002, 240–60; Kuhlmann 1998; 2013.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 126 David J. Mattingly et al.

Egyptian border. There are still a few lakes within the depression (the depres-
sion is generally below sea level), reminiscent of the Siwa lakes. The water
resources are rather saline and there are few palms here today.57 Yet despite this
unpropitious appearance, the oasis has had strategic importance as a cross-road
location between the main east to west route and routes running south to Kufra
and north towards Cyrenaica. The four main archaeological sites lie at spring
sites close to the edge of the basin, controlling the main approach routes from
the east, west, north and south. Most of the evidence recorded to date relates to
several cemeteries of rock-cut tombs (Ain Melfa – with hundreds of burials,
Abbiar Zergum, Ain Abu Zaid, ad-Fregia – the last two with smaller numbers
of tombs), from which mummified bodies and a range of Roman artefacts have
been retrieved.58 The tombs are rock cut chamber tombs, generally square and
with multiple loculi or at any rate used for multiple burials. The dead appear in
general to have been properly mummified in the Egyptian fashion, with linen
or cotton wrappings and removal of vital organs.59 The bodies were buried in
extended supine position on rock-cut loculi shelves or on wooden beds. One
mummy of second-century AD date has been analysed in Italy and is now in
the Tripoli museum.60 Further mummies, retrieved during work in the 1990s
by the Libyan Department of Antiquities, are stored in the museum stores at
Cyrene and have been partly studied by an Italian team.61 The tombs have been
heavily pillaged over the years, but from the better preserved examples that
have been excavated, it is clear that they were often equipped with wooden
doors and accompanied by a wide range of grave goods (coloured textiles,
wheel-made pottery, glass, combs, metal jewellery, glass and stone beads,
spindle whorls etc.). A Roman date is indicated by the finds and radiocarbon
dates have confirmed this, including a date on a wooden bed of 170 calBC to
calAD 180.62 The burials have many similarities with those of the Siwa oasis to
the south-east and the burial rite especially emphasises the connections with
the Egyptian oases. Oases to the west of al-Jiarabub have not produced evidence
of deliberately mummified burials and in those western locations the body was
normally laid on the side in a flexed position, more typical of the Saharan
tradition.63 Al-Jiarabub, thus seems to mark a point of division in funerary

57
Lloyd-Owen 2009, 42: ‘The area of Giarabub at the best of times is not very attractive and its
water supply was always brackish and unpleasant.’
58
D’Ercole and Martellone 2006; 2009; Saraullo 2009; Wright 1997.
59
D’Ercole and Martellone 2009, 57–59. 60 Kenrick 2013, 329–30.
61
D’Anastasio 2009; d’Anastasio et al. 2009.
62
D’Ercole and Martellone 2009, 66 (GX-30372, 1980±70 BP).
63
Two mummified bodies found in the Garamantian heartlands appear to be the result of natural
desiccation, rather than deliberate treatment of the body and both were covered with simple
shrouds, rather than being systematically wrapped in cloth bandages, Mattingly et al. 2008, 233–39.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 127

practices at Saharan oasis.64 On the other hand, Libyan inscriptions in

a primitive script have been reported from the Ain Melfa cemetery, demon-
strating some cultural connection of the people buried here with the Libyan
Saharan peoples further west.65
The lack of settlement evidence reported is almost certainly a factor of the
state of modern research here, which has focused almost exclusively on the
visible (and threatened) tombs. In fact, the most recent Italian work has
identified at least two settlement sites. The first is a hilltop site immediately
above the Ain Melfa tombs, with a second settlement site, including a tower
built of well-dressed stone blocks reported by the smaller cemetery of Sand Hill
(ad-Fregia).66 The modern town of al-Jiarabub, at the extreme western edge of
the depression, also looks to have traces of earlier settlement below the modern
one. Though it is 20–30 km away from the three main cemeteries, which
strongly suggests that there are additional settlements to be discovered, there
are reported burials from just north of al-Jiarabub.67 Ain Melfa and Abbiar
Zergum are located close to the extant lake in the far north-eastern part of the
basin and could plausibly relate to a single settlement. Ad-Fregia lies in
a subsidiary basin, to the south-west of the main depression and about 15 km
south of al-Jiarabub.68

Awjila and Syrtica

There are a number of important oases along the main Saharan route
running to the north of the great Libyan sand sea and south of Cyrenaica
and Syrtica.69 In this section we discuss the Awjila group, Marada, Tagrifet
and Zala, while the next section will deal with the al-Jufra oases, which
marked the western terminus of this route. The oasis of Kufra, which lies to
the south of the main east-west route is discussed in Chapter 7, in relation
to the Tibesti and the southern Libyan desert.70

Awjila
One of the most celebrated Libyan oases in the ancient sources after
Ammon (Siwa) and Garama (Jarma), this was evidently another oracular

64
Gatto et al. 2019, 528–29. 65 Saraullo 2009, 55. 66 D’Ercole and Martellone 2009, 59–60.
67
D’Ercole and Martellone 2009. 68 Mattingly 2000b; Mohamed 1998; 2007; Wright 1997.
69
For cartography see Goodchild 1954a; 1954b; Mattingly 2000b; 2000c. On the desert route from
Egypt into Libya, see Guédon 2010; Liverani 2000; 2006; Rebuffat 1970a; 1970b; 2004.
70
Sterry and Mattingly, Chapter 7, this volume.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 128 David J. Mattingly et al.

centre of Ammon71 and the principal centre of the Libyan people known
as the Nasamones. The encounter of the Nasamones with Rome followed
a similar path to that of the Garamantes. There were early hostile
encounters, with Roman campaigns launched against them in both the
Augustan and Flavian periods.72 There has been comparatively little
archaeological investigation of Awjila; the massive scale of modern oil
exploration in the desert to west of here has led to much redevelopment
over recent decades and the near total obliteration of the traditional oasis
settlements, though a few distinctive Medieval mosques survive at Awjila
itself.73 There are two mosques in the oasis attributed to the seventh
century, one associated with the reputed tomb of Sidi Abdallah ibn Sa’ad
ibn Abi-Sarh (died AD 656).
The Awjila group actually comprises three distinct clusters of oasis,
arranged in a triangular pattern and separated one from the other by
c.25–30 km. Awjila is the most westerly of these and surely coincides
with the ancient site of Augila, mentioned by Herodotus and others.74
A Libyan colleague, Fuaad Bentaher, sent us news of the discovery of three
Classical period rock-cut tombs at al-Darb al-Kabir, Awjila, to the north of
the historic core. A sample of textiles from the burial has now been radio-
carbon dated to 362–164 calBC confirming the first millennium BC activity
at this oasis.75 The early Arab military conquest of Awjila clearly attests to
its importance as a population centre at that time. To the north-east
(Gicherra/Jakharrad) and south-east (Jialo) are two additional foci, at
both of which Pacho noted traces of ruins and ancient habitations in the
1820s.76 While it cannot be demonstrated that these go back to pre-Islamic
times, on the balance of probability it must be likely that all three of the
Awjila oases were developed in pre-Islamic times.

Marada
Marada is a small oasis between Awjila and the al-Jufra group, marking an
important watering point along the east to west route running south of
Syrtica, along with Zelden and Zala. There were evidently abandoned
structures seen by Pacho, but no archaeological research has ever been

71
Mattingly 1995, 39, 168. 72 Mattingly 1995, 70–75. 73 Kenrick 2013, 25–28.
74
Pacho 1827; Scarin 1937, 13–44. 75 OxA 33721, 2177±31 BP.
76
Pacho 1827, map (inset detail). Cf. also Rohlfs 2002a, 156–68; 2003, 152–67 for two dismal stays
in Awjila and Djalo; Hornemann 1802, 37–39 also gives a brief account, but makes no specific
reference to ancient remains.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 129

carried out here and the area lies at the heart of the modern oil fields. Its
antiquity is thus unproven, but likely.77

Zala
Zala is a small oasis at a key location south of the Syrtic coast and just east
of the forbidding basalt plateau lands of the Jabal bin Ghanima. This was an
important stage between Awjila and Fazzan, from where trails diverge
towards Sirte, al-Jufra and Zuwila. The early accounts of Rohlfs and
Scarin highlight an important water source in the 12 × 5 km depression
supporting 100,000 palms.78 The old village at Zala was originally
a fortified site on top of a flat-topped hill (rather reminiscent of Waddan,
see below), but this was entirely demolished to make way for an Italian
colonial era fort, with the settlement then displaced to around the foot of
the hill.79 There was certainly pre-Islamic activity in the oasis, as Rebuffat
recorded two additional stone fortified structures (28 × 23 m and
18 × 18 m), both evidently associated with some Roman pottery, which
he interpreted as Roman outposts.80 These are to be found in the south-
western part of the modern oasis, close to the modern fort. There is no
evidence in fact to suggest that either were Roman forts rather than
examples of indigenous fortifications (qsur) found throughout Fazzan
and Tripolitania. Ward and Scarin both mention a foggara in this oasis,
although this has not been relocated.81 At c.10 km to the north of the main
oasis is a small abandoned oasis (Tirsa) that is bypassed by the modern
road over the Jabal. There are two small clusters of buildings of unknown
date on the east side of the palms and a few scattered cairns on the
surrounding hills.82

Tagrifet
This small oasis lies east of Waddan and directly north of Zala towards the
Syrtic coast, with a route running up to the Sirte/Madina Sultan area. Di
Vita recorded a fortified site 200 m in circumference in a dominant

77
Mattingly 2000c; Rebuffat 1970a.
78
Scarin 1938, 75–83; Rohlfs 2003, 136–39 for general description.
79
Scarin 1938, TAV XIX for air-photograph and ground shot.
80
Rebuffat 1970a, 17–18; 1970b. For the Roman pottery, see Rebuffat et al. 1970, 64, 80, 83 (sherds
C102, A90, A120).
81
Scarin 1938, 81–82; Ward 1968, 35.
82
Nachtigal 1974, 167 indicates that this Tirsa was still occupied in the nineteenth century.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 130 David J. Mattingly et al.

position overlooking an area of ancient cultivation.83 There is also an

intriguing report of Roman coins found here in 1964.84 Inspection of
available satellite imagery shows that there are at least three large ruined
settlements in this oasis. The first is a fortification on a hilltop spur with
two enclosures (almost certainly di Vita’s fortified site). It has signs of
internal stone buildings, but these are in a poor state of repair. On a slight
rise 1.8 km to the north is a 3.5 ha settlement within an enceinte. There are
at least two major phases to this settlement visible from its plan and a large
rectangular building at its core. This site has some morphological simila-
rities with the al-Jufra settlements with associated Roman pottery (see
below).85 An abandoned field system 1 km to the west is probably asso-
ciated with this settlement. Further areas of gardens are visible around the
same playa, but with no obvious settlement remains. The third settlement
is 10 km to the west and consists of a large rectilinear enclosure within
a small area of abandoned gardens. To the north, close to the standing
remains of a colonial era fort are two lines of mounds that may be foggaras,
although there are no other visible traces that can be linked with them.
There is very little funerary evidence in this oasis, but a small cluster of
mounds, at least some of which are cairns, are located to the south.

Al-Fuqha
A tiny oasis and also one of the most isolated, al-Fuqha was mentioned by
al-Bakri as on the subsidiary route between Zala and Zuwila, via Tmissa.86
The oasis consists of a small mudbrick village and a few scattered gardens.
Notable archaeological remains include a hilltop settlement, probably
walled, that sits on a small plateau and a handful of small foggaras that
feed the gardens just to the south of the hilltop.

The Oases of al-Jufra

The oases of al-Jufra are an important cluster lying between Fazzan and the
Syrtic coast and the evidence from Fazzan creates a strong presumption

83
See also di Vita 1964, 94; Rebuffat 1970b, 181–83.
84
See also di Vita 1964, 94 n. 158 on the unpublished find of second-century bronze coins
reported in Sunday Ghibli 9/2/1964. See also Rebuffat 1970b, 183.
85
This is also possibly the site of Tagrifet mentioned by al-Bakri, de Slane 1859, 30–31.
86
Thiry 1995, 435 see also Scarin 1938, 54–57, and TAV IX. al-Bakri, de Slane 1859, 31.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 Figure 3.3. Sites in al-Jufra oasis.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 132 David J. Mattingly et al.

that this oasis zone was also active in pre-Islamic times.87 The three main
oasis settlements of Waddan, Hun and Sukna are 16–20 km apart in an
east-west line (Fig. 3.3). The early modern fortified towns were described
in detail by nineteenth-century travellers88 and there is a good overall study
of the oases from the 1930s.89 However, there has been little archaeological
work in the al-Jufra area. The oases were still Berber speaking in the
nineteenth century.
Our remote-sensing investigation of the overall al-Jufra area has
revealed the presence of a sequence of urban-scale settlements and old
irrigation systems that clearly predate the modern period. Subsequently,
a team of Libyan archaeologists led by Muftah al-Haddad conducted
a short survey of the region and obtained samples for radiocarbon dating
(see below). An interesting aspect of al-Jufra is its intermediate position
between Mediterranean and Fazzan, where as Nachtigal put it, ‘stone is still
a commoner building material than it is farther south, but poorly con-
structed bricks of sun-dried clay, used instead of stone, appear more
frequently than they do near the north coast’.90 This hybridity in construc-
tion materials is also evident in the archaeology recorded.91

Waddan
Waddan was evidently the dominant oasis in the al-Jufra group at the time
of the Arab conquest in the AD 640s, though in more recent times it seems
that Sukna has been the regional centre.92 The early Arab sources identify
Waddan as the capital of a kingdom and a key site that the Arab authorities
established treaty relations with.93 There is some archaeological evidence
from the immediate environs of the early modern town. Rebuffat evidently
found a Roman amphora fragment 6 km to the north (about twice as far
from Waddan as the site of Busi, WDN002/4/5, see below) and also
recovered third-century Roman pottery from the castle at the centre of

87
Mattingly 1995, 7, 48.
88
Rohlfs 2003, 93–131 for a general account of the al-Jufra. See also, Denham and Clapperton
1826, xxv–xxvii; Lyon 1821, 70–80; Nachtigal 1974, 50–55.
89
Scarin 1938. 90 Nachtigal 1974, 52.
91
Following initial reconnaissance work by Martin Sterry, Libyan colleagues led by Muftah al-
Haddad visited a number of the sites on the ground in early 2014, taking photographs,
recording structural detail and obtaining material for dating. Further mapping has been carried
out in 2015 by Louise Rayne and Martin Sterry as a joint undertaking of the Trans-SAHARA
and EAMENA projects.
92
Nachtigal 1974, 50–53 on the predominance of early modern Sukna.
93
De Slane 1859, 30–31; Levtzion and Hopkins 1981, 12–13 and 63; Thiry 1995, 76–109.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 133

Figure 3.4. The settlement and associated field-system of Busi (WDN002, 004–005, 007).

the modern town.94 This castle comprised an irregular fortified enceinte on

an elevated rocky outcrop and has long been recognised as of likely pre-
Islamic date.95
Our analysis of satellite imagery has now identified several additional
population centres in this oasis. To the north-west of the modern town are
a group of three settlements collectively known as Busi (WDN002,
WDN004 and WDN005), covering at least 8.5 ha on the edge of a large
(116 ha) area of abandoned gardens (WDN007) (Fig. 3.4).96 Just 2.5 km to
the south-west (west of modern Waddan) are two further settlements of
8.5 ha (Delbak, WDN001) and 1.6 ha (WDN006) (Fig. 3.5). Material from
the northern group is consistent with Rebuffat’s amphora fragment and of
likely Roman date. Finds from the south-west group include glass and
ceramics that appear to be of Roman date, as well as some probable Islamic

94
Rebuffat 1969, 181, 187; 1970b, 187; Rebuffat et al. 1970, 121. See also Pace 1951, 425–29 for
a gold treasure found there of unusual representational style (Saharan?) and of possible pre-
Islamic type.
95
Scarin 1938, 50–52 (with plan); Tav. X for air-photographs.
96
Substantial bulldozing of these settlements hides their true extent.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 134 David J. Mattingly et al.

Figure 3.5. Comparative plans of larger settlements in al-Jufra oasis.

material. A sample obtained from the exposed mudbrick wall of an other-

wise buried structure at WDN001 returned a date of calAD 659–776.97
All five of these settlements were probably occupied into the Medieval
period and their names recall the descriptions by al-Bakri and al-Idrisi of
two centres called Tum (possibly abandoned at this time) and Dilbak.
However, the nearby hills also have cemeteries of several hundred clustered
cairns, suggesting that these centres were occupied at least in late antiquity
if not earlier. Elsewhere, between the modern gardens to the south-east of
Waddan, there is a rectangular fortification (60 × 36 m) that sits within

97
The site numbering in this section and in Table 3.1 is revised from the initial publication,
Mattingly et al. 2018. Future reference should be to the dates and site numbers given here.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 Figure 3.6. Foggaras and settlements in al-Jufra oasis.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 136 David J. Mattingly et al.

a small area of heavily eroded settlement. Patches of abandoned gardens

are visible to the east and a group of three heavily truncated foggaras
(visible in two areas 2 km apart) may have irrigated these. Moving further
to the south-west there are more cairns on small hills, but the most
remarkable archaeological remains are found along a 13-km stretch of
escarpment edge. Here there are 23 foggaras leading to at least 14 small
settlements and gardens clusters (Fig. 3.6). Only one of these was surveyed
on the ground and it was not possible to obtain suitable material for
radiocarbon dating, but ceramics from the site were from fabric groups
that are indicative of a Roman date. Both the settlements and the irrigation
works are partly truncated and at imminent risk of destruction from
modern agricultural developments.

Hun
Although truncated by the modern town, Hun is the only one of the three oases
to still retain a sizeable portion of its early modern buildings. Clapperton
described this as ‘recently built’ when he visited in the 1820s and commented
that the community had moved from a nearby settlement c.4 km to the north.98
In fact, there are two abandoned settlements in that direction (see Fig. 3.5). The
first (HUN003) is a walled settlement, much enveloped by sand dunes, but still
visible just under 3 km to the north. This was certainly occupied in the post-
Medieval period as demonstrated by two dated samples from the enclosing wall
of calAD 1685–1928 and calAD 1692–1920.99 Yellow and dark green glazed
ceramics along with a mosque, marabout and cemetery further demonstrate
the Medieval and later date of this site. The second (HUN006), a further 2 km to
the north, is another much larger settlement, badly damaged by modern
bulldozing. It is undoubtedly of greater antiquity, the ceramics appear to
include Roman coarsewares, amphorae and a pale cream-coloured pottery
that is possibly early Islamic in date. Two radiocarbon samples have been
dated to calAD 672–862 and calAD 987–1151.100 However, despite its large
size (15.5 ha), the only possible mention of the site is al-Bakri who describes
a town near Waddan called Holl as having a large population and many date
palms and springs. A fourth town-sized unwalled settlement (HUN004) has
been located 4 km to the north-west of the modern town, beyond the sand
dunes that cover HUN003. This lies on the edge of a large playa, the remains of
irrigation works and gardens are found across its entirety. Garden walls can also
be found in the area of dunes between the large settlements. There is very little

98 99 100
Scarin 1938, 47. OxA-33872 and OxA-33873. OxA-33870 and OxA-33871.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 137

archaeology visible to the south of the town, but a group of foggaras (now
largely destroyed by modern developments, but detected on 1970s Corona
imagery) fed a depression 4 km to the south of Hun, the date of these is
unknown but most likely of similar age as those to the south-east of Waddan.

Sukna
The Sukna oasis has seen the most modern development, leaving little
visible or known archaeologically about its deeper history.101 Lyon thought
about 2,000 people still dwelt within the early modern walled town, noting
its seven gates (only one being wide enough to admit a loaded camel), its
projecting towers, the narrow streets and two storey houses constructed in
stone and mud.102 According to the informants of Nachtigal, this walled
town was only 300 years old and in full decline by the 1860s:
The town forms an elongated heptagon . . . there are seven gates and thirty-two bastions
in the surrounding walls, which are made from limestone and cement, but failed to
make any impression of security . . . the wooden supports of the gates . . . seemed to be
the most solid element in the enclosing wall . . . The gigantic castle, completely ruined
and no longer serving any useful purpose, towers above everything else.103

Walls, qasbah and domestic quarters of the Medieval/early modern town

have all been swept away in modern redevelopment across the last 20
years.104 However, 6 km to the south we have identified another large
unwalled settlement of 5 ha with distinctive long rectangular houses and
several hectares of abandoned gardens. Another notable site is a hilltop
settlement, almost certainly of the first millennium BC, indicating
a possible early occupation of this oasis group.

Conclusion

The archaeological evidence demonstrates quite clearly the early develop-

ment of oasis settlements and trading contacts in the Egyptian Western
Desert by the mid-third millennium BC. Although we only have termini ante
quem for the oases extending further west into the Eastern Sahara, Siwa and
Awjila were certainly in existence by the mid first millennium BC (with
archaeological evidence and the textual attestation of Herodotus for both).
For the oases of Syrtica and al-Jufra at present we can only affirm the
101 102
Scarin 1938, 41–50, TAV VII–VIII for air-photos. Lyon 1821, 72.
103
Nachtigal 1974, 52. 104 Nachtigal 1974, 150–52.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 138 David J. Mattingly et al.

evidence of Roman period activity. In all probability, however, we can use

the evidence for very early first millennium BC agriculture in Fazzan to
cross-date the initial evolution of the oases between Bahariya and the
Garamantian heartlands to the late second or early first millennia BC.
Only detailed fieldwork at some of these intermediate points will allow
that to be demonstrated explicitly, but this should be the default hypothesis.
It is no surprise to encounter urban-scale oasis centres in the Western
Desert, where the journey from Fayum and Nile was only a few days’ travel.
But it is less expected to see urbanisation in less accessible early oases.
A second really important observation of this chapter concerns the scale
and sophisticated plans of a number of the sites identified among the
Libyan oases, in particular in al-Jufra. As noted, several of these appear
to be pre-Islamic in origin and match the evidence already presented from
the Garamantian heartlands for possible urban sites there.
The third, key conclusion is that many of these Eastern Saharan oases
share elements in common with each other: an agricultural package (invol-
ving dates, cereals, figs, grapes, even olives), access to Roman pottery and
other material culture, knowledge of the foggara irrigation system. There
are also some important differences. Among the Egyptian oases, the
influence of Nilotic civilisation is strong in all periods from the Old
Kingdom to the Byzantine era. The burial rite involving mummified supine
bodies, often laid in rock-cut chamber tombs, can be traced as far as al-
Jirabub, but no further. The other Libyan oases seem to adhere most closely
to a Saharan tradition of crouched burials under circular cairns (though the
Garamantes also had a significant engagement with rectangular burial
monuments, mastaba-style stepped tombs and even small pyramids. In
the Egyptian oases there is much evidence of external writing systems
(hieroglyphs, demotic Egyptian, Greek), with occasional but rare examples
of Libyan scripts. From al-Jiarabub westwards, the primary written lan-
guage appears to have been Libyan, with virtually no evidence of Latin or
Greek beyond the frontiers of the Mediterranean states.

References

Abboudy Ibrahim, M. 1992. The western desert of Egypt in the classical writings. In
Carratelli 1992, 209–17.
Abd el-Ghany, M. 1992. The oases in Roman Egypt in the light of papyri. In
Carratelli 1992, 3–12.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 139

Agut-Labordère, D. 2018. The agricultural landscape of Ayn Manawir (Kharga

oasis, Egypt) through the Persian period ostraca (Vth–IVth century BC). In
Purdue et al. 2018, 359–77.
Aufderheide, A.C., Nissenbaum, A. and Cartmell, L. 2004. Radiocarbon date
recovery from bitumen-containing Egyptian embalming resins. Journal of the
Society for the Study of Egyptian Antiquities 31: 87–96.
Bagnall, R.S. 1997. The Kellis Agricultural Account Book (P. Kell. IV Gr. 96) Dakhleh
Oasis Project: Monograph 7. Oxford: Oxbow Books (Oxbow Monograph 92).
Bagnall, R.S., Davoli, P. and Hope, C.A. (eds). 2013. The Oasis Papers 6: Proceedings
of the Sixth International Conference of the Dakhleh Oasis Project. Oxford:
Oxbow Books.
Ball, J. 1942. Egypt in the Classical Geographers. Cairo: Government Press.
Ball, J. and Beadnell, H.J.L. 1903. Baharia Oasis: Its topography and geology. Cairo:
Survey Department, Public Works Ministry.
Barich, B.E. and Hassan, F.A. 1990. Il Sahara e le oasi: Farafra nel deserto occi-
dentale egiziano. Sahara 3: 53–62.
Bashendi, M. 2013. Cemeteries in Dakhleh. In Bagnall et al. 2013, 249–62.
Bates, O. 1914. The Eastern Libyans. London: MacMillan (reprint 1970).
Beadnell, H.J.L. 1901. Farafra Oasis: Its Topography and Geology. Cairo: National
Printing Department.
Belgrave, C.D. 1923. Siwa: The Oasis of Jupiter Ammon. London: John Lane, The
Bodley Head Ltd.
Boozer, A. 2010. Memory and microhistory of an Empire: domestic contexts in
Roman Amheida, Egypt. In D. Borić (ed.), Archaeology and Memory. Oxford:
Oxbow Books, 138–57.
Boozer, A. 2011. Forgetting to remember in the Dakhleh Oasis, Egypt. In
M. Bommas (ed.), Cultural Memory and Identity in Ancient Societies. London
and New York: Continuum Publishers, 109–26.
Boozer, A. 2012. Globalizing Mediterranean identities: The overlapping spheres of
Egyptian, Greek, and Roman worlds at Trimithis. Journal of Mediterranean
Archaeology 25.2: 93–116.
Boozer, A. 2013a. Archaeology on Egypt’s edge: Archaeological research in the
Dakhleh Oasis, 1819–1977. Ancient West and East 12: 117–56.
Boozer, A. 2013b. Frontiers and borderlands in imperial perspectives: Exploring
Rome’s Egyptian frontier. American Journal of Archaeology 117.2: 275–92.
Boozer, A. 2015a. A Late Roman-Egyptian House in the Dakhla Oasis: Amheida
House B2. New York: Institute for the Study of the Ancient World, New York
University Press.
Boozer, A. 2015b. The social impact of trade and migration: The Western Desert in
Pharaonic and post-Pharaonic times. In C. Riggs (ed.), Oxford Handbooks
Online in Archaeology. Oxford: Oxford University Press. DOI:10.1093/
oxfordhb/9780199935413.013.37.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 140 David J. Mattingly et al.

Bousquet, B. 1996. Tell-Douch et sa région. Géographie d’une limite de milieu à une

frontière d’empire. Cairo: IFAO.
Bowen, G.E. and Hope, C.A. 2003. The Oasis Papers 3: Proceedings of the Third
International Conference of the Dakhleh Oasis Project. Oxford: Oxbow Books.
Bowman, A. 1989. Egypt after the Pharaohs. Berkeley: California University Press.
Bravard, J.-P., Mostafa, A., Davoli, P., Adelsberger, K.A., Ballet, P.,
Garcier, R., Calcagnile, L. and Quarta, G. 2016a. Construction and deflation
of irrigation soils from the Pharaonic to the Roman period at Amheida
(Trimithis), Dakhla Depression, Egyptian Western Desert. Géomorphologie
22.3: 305–24.
Bravard, J.-P., Mostafa, A., Garcier, R., Tallet, G., Ballet, P, Chevalier, Y. and
Tronchère H. 2016b. Rise and fall of an Egyptian oasis: artesian flow, irrigation
soils, and historical agricultural development in El-Deir, Kharga depression,
Western Desert of Egypt. Geoarchaeolgy 31.6: 467–86.
Carpentiero, G. 2016. Continuity and change in Hellenistic town planning in
Fayum (Egypt). Between tradition and innovation. In Mugnai et al. 2016,
73–100.
Carratelli, G. (ed.). 1992. Roma et l’Egitto nell’antichità classica. Cairo 6–9 Febbraio
1989. Rome: Istituto poligrafico e Zecca dello Stato.
Chaveau, M. 2001. Les qanâts dans les ostraca de Manawir. In P. Briant (ed.),
Irrigation et drainage dans l’antiquité, qanâts et canalisations souterraines en
Iran, Egypte et en Grèce. Paris: Thotm editions, 137–42.
Chaveau, M. 2005. Irrigation et exploitation de la terre dans l’oasis de Kharga à
l’époque perse. Les cahiers de recherches de l’institut de papyrologie et égyptologie
de Lille 25: 157–63.
Churcher, C.S. and Mills, A.J. 1999. Reports from the Survey of Dakhleh Oasis
1977–87. Dakhleh Oasis Project Monograph 2. Oxford: Oxbow Books.
Colin, F. 2000. Les peuples Libyens de la Cyrénaique à l’Egypte d’après les sources de
l’antiquité classique. Louvain: Académie Royale de Belgique.
D’Anastasio, R. 2009. Proposal for a project on the anthropological and paleo-
pathological study and preservation of the human mummies from Jarabub
(Lybia). In Menozzi 2009, 51–52.
D’Anastasio, R., Vitullo, G. and Urso, M. 2009. Studio preliminare antropometrico
e paleopatologico delle mummie e dei reperti ossei da Giarabub e ora al
magazzino di Cirene. In Menozzi 2009, 45–51.
D’Ercole, V. and Martellone, A. 2006. Ricognizione preliminare nell’oasi di
Giarabub. In Fabricotti and Menozzi 2006, 457–66.
D’Ercole, V. and Martellone, A. 2009. Ricognizione preliminare nell’oasi di
Giarabub. In Menozzi 2009, 57–67.
De Slane, M.G. 1859. Description de l’Afrique septentrionale par el-Bekri. Paris:
Imprimerie impériale.
Davoli, P. 1998: L’archeologia urbana nel Fayyum di età ellenistica e romana.
Napoli: G. Procaccini.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 141

Davoli, P. 2012. Reflections on urbanism in Graeco-Roman Egypt: A historical and

regional perspective. In E. Subías, P. Azara, J. Carruesco, I. Fiz and R. Cuesta
(eds), The Space of the City in Graeco-Roman Egypt: Image and Reality.
Tarragona: Institut Català d’Arqueologia Clàssica, 69–92.
Davoli, P. 2013. Amheida 2007–2009: New results from the excavations. In Bagnall
et al. 2013, 263–78.
Denham, D. and Clapperton, H. 1826. Narration of Travels and Discoveries in
Northern and Central Africa in the Years 1822–1824. London: John Murray.
Di Vita, A. 1964. Il ‘limes’ Romano di Tripolitania nella sua concretezza archae-
ologica e nella sua realtà storica. Libya Antiqua 1: 65–98.
Dospěl, M. and Suková, L. 2013. Bahriya Oasis. Recent Research into the Past of an
Egyptian Oasis. Prague: Charles University.
Dunand, F., Heim, J.-L. and Lichtenberg, R. 2010. El-Deir. Nécropoles I. La
nécropole Sud. Paris: Cybèle.
Dunand, F., Heim, J.-L. and Lichtenberg, R. 2012. El-Deir. Nécropoles II. Les
nécropoles Nord et Nord-Est. Paris: Cybèle.
Dunand, F., Heim, J.-L. and Lichtenberg, R. 2013. Les nécropoles d’el-Deir (Oasis
de Kharga). In Bagnall et al. 2013, 279–96.
Fabricotti, E. and Menozzi, O. 2006. Cirenaica: Studi, scavi e scoperte. Atti del
X Convegno di Archeologia Cirenaica Chieti 24–26 Novembre 2003 Nuovi dati da
città e territorio. Oxford: BAR S1488.
Fakhry, A. 1942. Recent Explorations in the Oases of the Western Desert. Cairo:
French Institute of Oriental Archaeology.
Fakhry, A. 1942/1950. Bahria Oasis, 2 vols. Cairo: Service des antiquités de
l’Egypte.
Fakhry, A. 1944. Siwa Oasis. Its History and Antiquities. Cairo: Government Press.
Fakhry, A. 1950. The Oasis of Siwa. Its Customs, History and Monuments. Cairo:
Wadi El-Nil Press.
Fakhry, A. 1973. The Oases of Egypt. I, Siwa. Cairo: American University in Cairo Press.
Fakhry, A. 1974. The Oases of Egypt. II, Bahariyah and Farfara Oases. Cairo:
American University in Cairo Press.
Förster, F. 2013. Beyond Dakhla: The Abu Ballas trail in the Libyan Desert (SW
Egypt). In Förster and Reimer 2013, 297–337.
Förster, F. 2015. Der Abu Ballas Weg. Eine pharanische Karawanenroute durch die
Libysche Wüste. Africa Praehistorica 28. Cologne: Heinrich-Barth Institut.
Förster, F. and Reimer, H. 2013. Desert Road Archaeology in Ancient Egypt and
Beyond. Cologne: Heinrich-Barth Institut.
Gallinaro, M. 2018. Mobility and Pastoralism in the Egyptian Western Desert.
Steinplätze in the Holocene Regional Settlement Patterns. Firenze: All’Insegna
del Giglio.
Gatto, M.C., Mattingly, D.J., Ray, N. and Sterry, M. (eds). 2019. Burials, Migration
and Identity in the Ancient Sahara and Beyond. Trans-Saharan Archaeology
Volume 2. Series editor D.J. Mattingly. Cambridge: Cambridge University Press.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 142 David J. Mattingly et al.

Gautier, E.-F. 1970. Sahara. The Great Desert (translated from the French by D.F.
Mayhew). London: Octagon.
Giddy, L.L. 1987. Egyptian Oases. Bahariya, Dakhla, Farafra and Kharga during
Pharonic Times. Warminster: Aris & Phillips Ltd.
Gill, J.C.R. 2016. Dakhleh Oasis and the Western Desert of Egypt under the
Ptolemies. Oxford: Oxbow Books.
Gonon, T. 2018. La gestion de l’eau dans le désert oriental égyptien durant les
temps historiques, de l’époque perse à nos jours: Le site de ‘Ayn Manâwîr et la
prospection du bassin sud de l’oasis de Kharga. In Purdue et al. 2018, 269–90.
Goodchild, R.G. 1954a. Tabula Imperii Romani: Lepcis Magna (Sheets H.33 1.33).
Oxford: Society of Antiquaries of London.
Goodchild, R.G. 1954b. Tabula Imperii Romani: Cyrene (Sheets H.34 I.34). Oxford:
Society of Antiquaries of London.
Gosline, S.L. 1990. Bahariya Oasis Expedition. Season Report for 1988 Part I Survey
of Qarat Hilwah. San Antonio: Van Siclen Books.
Guédon, S. 2010. Le voyage dans l’Afrique romaine. Bordeaux: Ausonius.
Hawas, Z. 2000. Valley of the Golden Mummies. Cairo: AUC Press.
Herschend, F. 2009. The Early Iron Age in South Scandinavia. Social Order in
Settlement and Landscape. Uppsala: Uppsala University Press.
Hope, C.A. and Bowen, G.E. 2002. Dakhleh Oasis Project: Preliminary Reports on
the 1994–1995 to 1998–1999 Field Seasons. Oxford: Oxbow Books.
Hope, C.A. and Mills, A.J. 1999. Dakhleh Oasis Project: Preliminary Reports on the
1992–1993 and 1993–1994 Field Seasons. Oxford: Oxbow Books.
Hope, C.A. and Pettman, A.J. 2013. Egyptian connections with Dakhleh oasis in
the Early Dynastic period to Dynasty IV: New data from Mut al-Kharab. In
Bagnall et al. 2013, 147–66.
Hornemann, F. 1802. The Journal of Frederick Hornemann’s Travels from Cairo to
Mourzouk the Capital of the Kingdom of Fezzan in Africa in the Years 1797–9.
London: G. and W. Nichol.
Ibrahim, A.I. 2013. Major archaeological sites in Kharga Oasis and some recent
discoveries by the Supreme Council of Antiquities. In Bagnall et al. 2013, 1–8.
Jackson, R.B. 2002. At Empire’s Edge. Exploring Rome’s Egyptian Frontier. New
Haven: Yale University Press.
Jennerstrasse 8. 2002. Tides of the Desert. Contribution to the Archaeology and
Environmental History of Africa in Honour of Rudolph Kuper. Cologne: Heinrich
Barth Institut.
Kenrick, P. 2013. Archaeological Guides: Cyrenaica. London: Society for Libyan
Studies.
Kirkwan, L.P. 1971. Roman expeditions to the Upper Nile and the Chad-Darfur region.
In F.F. Gadallah (ed.), Libya in History. Proceedings of a Conference Held at the
Faculty of Arts, University of Libya 1968. Benghazi: University of Benghazi, 253–61.
Knudstad, J.E. and Frey, R.A. 1999. Kellis, the architectural survey of the Romano-
Byzantine town at Ismant el-Kharab. In Churcher and Mills 1999, 189–214.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 143

Kucera, P. 2013. Al-Qasr: the Roman castrum of Dakhleh Oasis. In Bagnall et al.
2013, 205–16.
Kuhlmann, K.P. 1998. Roman and Byzantine Siwa: Developing a latent picture. In
O.E. Kaper (ed.), Life on the Fringe: Living in the Southern Egyptian Deserts
during the Roman and Early Byzantine Periods. Leiden: Research School CNWS,
159–80.
Kuhlmann, K.P. 2002. The ‘oasis bypath’ or the issue of desert trade in Pharaonic
times. In Jennerstrasse 8 2002, 125–70.
Kuhlmann, K.P. 2013. The realm of two deserts: Siwa oasis between east and west.
In Förster and Reimer 2013, 133–66.
Leahy, A. 1990. Libya and Egypt c1300–750 BC. London: Society for Libyan Studies.
Levtzion, N. and Hopkins, J.F.P. (eds). 1981. Corpus of Early Arabic Sources for
West African History. Cambridge: Cambridge University Press.
Liverani, M. 2000. The Libyan caravan road in Herodotus IV.181–184. Journal of
the Economic and Social History of the Orient 43.4: 496–520.
Liverani, M. (ed.). 2006. Aghram Nadharif. The Barkat Oasis (Sha‘abiya of Ghat,
Libyan Sahara) in Garamantian Times. Firenze: All’Insegna del Giglio.
Lloyd-Owen, D. 2009. The Long Range Desert Group 1940–1945: Providence Their
Guide. Barnsley: Pen and Sword.
Lyon, G.F. 1821. A Narrative of Travels in Northern Africa in the Years 1818–1819
and 1820. London: John Murray.
McDonald, M.M.A. 2003. The early Holocene Masara A and Masara C cultural
sub-units of Dakhleh Oasis, within a wider cultural setting. In Bowen and Hope
2003, 43–70.
Mattingly, D.J. 1995. Tripolitania. London: Batsford.
Mattingly, D.J. 2000a. Ammon. In R. Talbert (ed.), Barrington Atlas of the Greek
and Roman World. Princeton: map 73 and Map-by-Map Directory pp. 1108–16.
Mattingly, D.J. 2000b. Cyrene. In R. Talbert (ed.), Barrington Atlas of the Greek and
Roman World. Princeton: map 38 and Map-by-Map Directory pp. 558–69.
Mattingly, D.J. 2000c. Syrtica. In R. Talbert (ed.), Barrington Atlas of the Greek and
Roman World. Princeton: map 37 and Map-by-Map Directory pp. 552–57.
Mattingly, D.J. 2003. The Archaeology of Fazzan. Volume 1, Synthesis. London:
Society for Libyan Studies, Department of Antiquities.
Mattingly, D.J., Dore, J. and Lahr, M. (with contributions by others). 2008. DMP II:
2008 fieldwork on burials and identity in the Wadi al-Ajal. Libyan Studies 39:
223–62.
Mattingly, D.J., Sterry, M., al-Haddad, M. and Bokbot, Y. 2018. Beyond the
Garamantes: The early development of Saharan oases. In L. Purdue,
J. Charbonnier and L. Khalidi (eds), Des refuges aux oasis: Vivre en milieu
aride de la Préhistoire à aujourd’hui. XXXVIIIe rencontres internationales
d’archéologie et d’histoire d’Antibes. Antibes: Éditions APDCA, 205–28.
Menozzi, O. (ed.). 2009. Chieti University Cyrene Season June 2009. Preliminary
Report. Privately circulated report.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 144 David J. Mattingly et al.

Midant-Reynes, B. and Tristant, Y. (eds). 2008. Egypt at Its Origins 2. Dudley, MA:
Uitgeverij Peeters en Departement Oosterse Studies.
Mills, A.J. 1999a. Pharaonic Egyptians in the Dakhleh oasis. In Churcher and Mills
1999, 171–78.
Mills, A.J. 1999b. Pottery manufacture in the Dakhleh oasis. In Churcher and Mills
1999, 215–43.
Mills, A.J. 2002. Deir el-Hagar, Ain Birbiyeh, Ain el-Gazzareen and el-
Muzawwaqa. In Hope and Bowen 2002, 25–30.
Mills, A.J. 2013. An Old Kingdom trading post at ‘Ain el-Gazzareen, Dakhleh
Oasis. In Bagnall et al. 2013, 177–80.
Mills, A.J. and Kaper, O.E. 2003. ‘Ain el-Gazzareen: Developments in the Old
Kingdom settlement. In Bowen and Hope 2003, 123–29.
Mohamed, F.A. 1998. El Jaghbub. In E. Catani and S. Maria Marengo (eds), La
Cirenaica in età antica. Macerata: Istituti editoriali e poligrafici internazionali,
263–73.
Mohamed, F.A. 2007. Trial excavations at el-Jaghbub. In L. Gasperini and S.
M. Marengo (eds), Cirene e la Cirenaica nell’antichità: atti del convegno internazio-
nale di studi: Roma-Frascati, 18–21 dicembre 1996. Tivoli: Edizioni Tored, 1–16.
Morkot, R. 1996. The Darb el-Arbain, the Kharga oasis and its forts and other
desert routes. In D.M. Bailey (ed.), Archaeological Research in Ancient Egypt.
Ann Arbor: JRA supplementary volume, 82–94.
Morkot, R. 2016. Before Greeks and Romans: Eastern Libya and the oases, a brief
review of interconnections in the Eastern Sahara. In Mugnai et al. 2016, 27–38.
Mugnai, N., Nikolaus, J. and Ray, N. (eds). 2016. De Africa Romaque. Merging
Cultures Across North Africa. London: Society for Libyan Studies, 27–38.
Müller-Wollermann, R. 2000. Memphis-Oxyrhynchus. In R. Talbert (ed.),
Barrington Atlas of the Greek and Roman World. Princeton: map 75 and Map-
by-Map Directory pp. 1125–39.
Nachtigal G. 1974. Sahara and Sudan, Vol I, Tripoli and Fezzan, Tibesti or Tu.
Translated from the German by A.G.B. and H.J. Fisher. London: C. Hurst.
Newton, C., Whitbread, T., Agut-Labordere, D. and Wuttman, M. 2013.
L’agriculture oasienne à l’époque perse dans le sud de l’oasis de Kharga
(Egypte, Ve–IVe s. AEC). Revue d’ethnoécologie 4: 2–18.
O’Connor, D. and Reid, A. (eds). 2003. Ancient Egypt in Africa. London: UCL
Press.
Pace, B. 1951. Parte I. In B. Pace, S. Sergi and G. Caputo, Scavi sahariani.
Monumenti Antichi 41: 151–200.
Pacho, J.R. 1827. Relation d’un voyage dans la Marmarique, la Cyrénaique et les
oasis d’Audjelah et de Maradeh. Paris: Lib. Firmin Didot.
Pettman, A. 2013. The date of the occupation of ‘Ain el-Gazzareen based on
ceramic evidence. In Bagnall et al. 2013: 181–208.
Prorok, B.K. de. 2001. In Quest of Lost Worlds: Five Archaeological Expeditions
1925–1934. Santa Barbara: Narrative Press (reprint of 1935 book).

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 3 Pre-Islamic Oasis Settlements in the Eastern Sahara 145

Purdue, L., Charbonnier, J. and Khalidi, L. (eds). 2018a. Des refuges aux oasis: Vivre
en milieu aride de la Préhistoire à aujourd’hui. XXXVIIIe rencontres internatio-
nales d’archéologie et d’histoire d’Antibes. Antibes: Éditions APDCA.
Rathbone, D. 1996. Towards a historical topography of the Fayyum. In D.M. Bailey
(ed.), Archaeological Research in Roman Egypt. Ann Arbor: JRA Supplement 19,
50–56.
Rebuffat, R. 1969. Deux ans de recherches dans le sud de la Tripolitaine. Comptes
rendus de l’académie des inscriptions et belles lettres 1969: 189–212.
Rebuffat, R. 1970a. Routes d’Egypte de la Libya Interieure. Studi Magrebini 3: 1–20.
Rebuffat, R. 1970b. Zella et les routes d’Egypte. Libya Antiqua 6–7: 181–87.
Rebuffat, R. 2004. Les romains et les routes caravanieres africaines. In M. Fantar (ed.), Le
Sahara. Lien entre les peuples et les cultures. Tunis: Université de Tunis, 221–60.
Rebuffat, R., Gassend, J.M., Guery, R. and Hallier, G. 1970. Bu Njem 1968. Libya
Antiqua 6–7: 9–105.
Reddé, M. 1989. Les oasis d’Egypte. Journal of Roman Archaeology 2: 281–90.
Reddé, M. 1999. Sites militaires romains de l’oasis de Kharga. Bulletin de l’Institut
Française d’Archéologie Orientale 99: 377–96.
Reddé, M., Ballet, P., Barbet, A. and Bonnet, C. 2004. Kysis: fouilles de l’Ifao à
Douch, oasis de Kharga, 1985–1990. Cairo: Institut Français d’Archéologie
Orientale.
Rohlfs, G. 2002a. Voyages et explorations au Sahara. Tome III, Tripolitaine –
Cyrenaique – Siwah 1868–1869 (translated by J. Debetz). Paris: Karthala.
Rohlfs, G. 2002b. Voyages et explorations au Sahara. Tome IV, Désert Libyque.
Siwah et les oasis d’Egypte 1873–1874 (translated by J. Debetz). Paris: Karthala.
Rohlfs, G. 2003. Voyages et explorations au Sahara. Tome V, Koufra – les oasis de
Djofra et de Djalo 1878–1879 (translated by J. Debetz). Paris: Karthala.
Rossi, C. 2000. Umm el-Dabadib, Roman settlement in the Kharga oasis:
Description of the visible remains, with a note on Ayn Amur. Mitteilungen des
Deutschen Archäologischen Instituts Abteilung Kairo 56: 235–52.
Rossi, C. 2013. Controlling the borders of the Empire: The distribution of Late-
Roman ‘forts’ in the Kharga Oasis. In Bagnall et al. 2013, 331–36.
Saraullo, L. 2009. L’oasi di el-Jaghbub. In Menozzi 2009, 52–57.
Scarin, E. 1937. Le oasi cirenaiche del 29˚ parallelo. Firenze: Sansoni.
Scarin, E. 1938. La Giofra e Zella (le oasi del 29 parallelo della Libia occidentale).
Firenze: Sansoni.
Smekalova, T.N., Mills, A.J. and Herbich, T. 2003. Magnetic survey at ‘Ain el-
Gazzareen. In Bowen and Hope 2003, 131–35.
Stewart, J.D., Molto, J.E. and Reimer, P. 2003. The chronology of Kellis 2: The
interpretive significance of radiocarbon dating of human remains. In Bowen and
Hope 2003, 345–64.
Tallet, G., Bravard, J.-P., Garcier, R., Guédon, S. and Mostapha, A. 2013. The
survey project at el-Deir, Kharga Oasis: First results, new hypotheses. In Bagnall
et al. 2013, 349–61.

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 146 David J. Mattingly et al.

Thiry, J. 1995. Le Sahara Libyen dans l’Afrique du nord medievale. Leuven:

Orientalia Lovaniensia Analecta, 72.
Valloggia, M. 2004. Les oasis d’Egypte dans l’Antiquité. Des origins au deuxiène
millénaire avant J.-C. Bischhelm: Infolio.
Vivian, C. 2000. The Western Desert of Egypt. An Explorer’s Handbook. Cairo:
American University in Cairo Press.
Wagner, G. 1987. Les Oasis d’Egypte. Cairo: Institut Français d’Archéologie
Orientale du Caire.
Wagner, G. 2000. Oasis Magna. In R. Talbert (ed.), Barrington Atlas of the Greek
and Roman World. Princeton: map 79 and Map-by-Map Directory pp. 1164–69.
Ward, P. 1968. Touring Libya. The Southern Provinces. London: Faber.
Warner, N. 2013. Amheida: Architectural conservation and site development,
2004–2009. In Bagnall et al. 2013, 363–79.
Welsby, D. 1996. The Kingdom of Kush. London: British Museum.
Wendorf, F. and Schild, R. 1980. Prehistory of the Eastern Sahara. New York:
Academic.
Wendorf, F., Schild, R. and Close, A.E. (eds). 1984. Cattle-Keepers of the Eastern
Sahara: The Neolithic of Bir Kiseiba. Dallas: Southern Methodist University.
Wilson, A.I. 2006. The spread of foggara-based irrigation in the ancient Sahara. In
D.J. Mattingly, S. McLaren, E. Savage, Y. al-Fasatwi and K.Gadgood (eds), The
Libyan Desert. Natural Resources and Cultural Heritage. London: Society for
Libyan Studies, 205–16.
Wiseman, M.F. 2008. The Oasis Papers 2: Proceedings of the Second International
Conference of the Dakhleh Oasis Project. Oxford: Oxbow Books.
Wright, G.R.H. 1997. Tombs at the oasis of Jeghbub: An exploration in 1955.
Libyan Studies 28: 29–41.
Wuttmann, M., Gonon, T. and Thiers, C. 2000. The qanats of ‘Ayn-Manâwîr
(Kharga Oasis, Egypt). Journal of Achaemenid Studies and Researches 1: 1–11.
www.achemenet.com

https://doi.org/10.1017/9781108637978.004 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert
under Roman Rule
anna lucille boozer

Introduction

The oases of Egypt’s Western Desert had long been a focus of Egyptian,
Persian, and Ptolemaic exploitation and investment by the time they were
incorporated into the Roman Empire.1 Despite this long history of inter-
vention, it is clear that the oases of antiquity reached their greatest occupa-
tional density under Roman rule. Along with this population increase came
substantial investment in cities. Several of these centres acquired formal
urban status within the Empire, signalling their importance in the Roman
political, social, and economic organisation of their Empire.
As systematic research in the Western Desert began in the 1970s, we can
make only tentative suggestions about diachronic change, the morphology
of urban sites, and the causes of these developments.2 Despite the many
lacunae in what follows, it is important to explore the urbanisation of the
Western Desert now in order to guide future research in this remarkable
region. In particular, this chapter examines the relationship between urba-
nisation and phases of colonialism and imperialism, which brought these
oases into new social, economic and political relationships.
My starting point for these queries focuses upon diachronic change in
the Western Desert, with the caveat that we still have much to learn about
the development of this region. One of the hindrances to learning more
about the changes in this region is that settlement appears to have been
continuous in many areas, resulting in tells.3 Roman ruins are typically the

1
As discussed more fully by Mattingly et al., Chapter 3, this volume.
2
For a historiography of archaeological research in the Dakhla oasis, see Boozer 2013a.
3
Contra Giddy 1987, who argued that the oases had single-phase settlements. This problem of
later settlements masking earlier ones is common among early oasis settlements across the
Sahara, see discussion by Mattingly and Sterry, Chapter 1, this volume. 147

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 148 Anna Lucille Boozer

only visible remains on these sites, which prevents us from describing

accurately what pre-Roman settlement forms looked like.4 Despite this
paucity of data, keyhole excavations, documentary sources, and the occa-
sional accessible remains of earlier phases can shape hypotheses about the
developmental trajectory of urbanism in this region. A sub-set of this first
query is an in-depth analysis of Roman urbanism in this region. While
under Roman rule, the Western Desert reached its greatest occupational
density from antiquity until the modern era.5 There are only two well-
mapped and well-preserved settlement sites to discuss: Ismant al-Kharab/
Kellis and Amheida/Trimithis. Mut el Kharab/Mothis, Kysis and Kharga
City were also important sites, but their urban form is less clear due to
modern occupation. Amheida (known as Trimithis in Roman times) was
occupied over thousands of years and ultimately achieved city status. It is
relatively unencumbered with later settlement, modern architecture and
agricultural intrusions. Trimithis will be discussed in more detail here as
a key case study of urbanism in the Western Desert.
The second line of enquiry draws together evidence from the pre-
vious section to explore the potential economic links between the
Western Desert and the Nile Valley, as well as Nubia and other areas
of the Sahara. These economic links probably hold the answer as to
how the region was developed and why it eventually collapsed. The
question of urban collapse in the Western Desert will be the final
theme explored below. It appears that the Western Desert experienced
a significant population reduction in the late fourth century AD. Major
sites such as Amheida, Kellis, and Kysis were abandoned at this time.
Habitation seems to have re-focused on fortresses, some of which were
continuously occupied from approximately the fifth century AD
through to the recent past (with some in continuing occupation).6 In
the absence of destructive layers or any documentary ‘explanation’ for
this significant change, we must explore comparisons and logical
hypotheses. Despite all of this uncertainty, it is essential to explore
these three questions relating to the urbanisation of the Western Desert
now, rather than at a later phase of research, so we can plan our
research strategies.

4
The Roman development of the Western Desert finds comparisons in other areas of the Sahara,
see Mattingly et al., Chapter 5, this volume.
5
Although, see Gill 2015, who argues that the Romans built upon Ptolemaic developments. The
imprint of Ptolemaic activity is less visible than Roman activity due to tells.
6
On the Roman castrum, al-Qasr, see Kucera 2013.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 149

Geography of the Western Desert

From the Pharaonic to Roman Periods, the basic geographic compass
constituting Egypt was the Nile Valley, which terminates in the Delta,
and the Fayum (Fig. 4.1). Beyond this core area lay: the Eastern Desert;
the oases of the Western Desert; further out to the west the Libyan Desert
beyond the oases; to the south, Nubia. Egypt proper incorporated these
marginal zones during times of prosperity, but often lost control of them
during periods of insecurity. The Western Desert comprises two-thirds of
the land within the current boundaries of Egypt. The primary ecological
niches within an otherwise harsh environment are the five oases – Siwa,
Bahariya, Farafra, Dakhla, and Kharga. Bahariya and Farafra together were
known as the Oasis Parva (Lesser Oasis) in antiquity. The Dakhla oasis and

Figure 4.1. Map of Egypt showing oases of the Western Desert (Margaret Mathews).

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 150 Anna Lucille Boozer

Figure 4.2. The Great Oasis, map of Kharga and Dakhla (Margaret Mathews).

the Kharga oasis were grouped together as the Oasis Major (the Great
Oasis). In Arabic, Dakhla means the Inner Oasis and Kharga means the
Outer Oasis, which is interesting, as distance is expressed with respect to
the desert, rather than the Nile (Fig. 4.2).7
Climatic conditions in the Western Desert are unforgiving. The south-
ern half of this desert is one of the driest regions on earth, with almost no
rain. The presence of oases in the Western Desert makes sedentary life
possible in this arid region, with the exception of the Qattara Depression
where the subterranean water is too salty. The water that forms the oases
derives from the artesian sources in the sandstone underlying the entire
Western Desert.8
The sand-laden winds from the north were as much an obstacle in
antiquity as they are today. The wind flattens crops, fills in houses, and
deposits dunes over paved roads. Today the winds also serve as an obstacle
to fieldwork, which is only possible in the winter months (October–
March), and even then formidable sandstorms are not uncommon. The
extreme temperatures also make life and fieldwork difficult. They range

7
For the Oasis Magna, Oasis Parva, and Siwa, see Wagner 2000, map 79.
8
Schild and Wendorf 1977, 10.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 151

from 0–2° C just before sunrise to 20–25° C by midday; in summer

maximums reach at least 40° C for extended periods.9
Egyptians considered ‘oasis country’ to be the desert region west of the
Nile, extending until ‘nearly’ [haud procul] the Fayyum (T3 wḥ 3tjw;
Arabic: Bilad al-Wahat).10 Regionally, locals were called wḥ 3tw or oasites,
but the term also applied to oases west of Egyptian-held territory, accord-
ing to a list from Edfu.11 Egypt’s rulers incorporated these peripheral areas
into Egypt proper during times of prosperity, but these regions often
slipped into self-rule during periods of greater uncertainty. Although
peoples of the Eastern and Western Deserts of Egypt were connected to
Egypt proper throughout Egyptian history they remained peripheral to
mainstream Egyptian culture and world-views.

Diachronic Change: Pre-Roman Settlement Patterns

and Urbanism
The starting point for this chapter is an examination of diachronic change
in the Western Desert. The oases were occupied from prehistoric to Roman
times, but our ability to accurately locate sites and interpret the site
morphology is difficult before the Roman Period.12 There appears to
have been continuous occupation at many sites (for example, Mut,
Amheida, Balat), which created tells of material. Archaeological material
deriving from desert roads peripheral to the oases themselves fills in some
of the knowledge gaps created by these tells. From these roads, we see clear
evidence of prehistoric and Protohistoric long-distance travel involving the
oases and there are signs of old connections between Sudan and the
oases.13
These desert roads mostly appeared during a time when the Saharan
environment had begun to shift into a fully arid landscape due to climactic
drying, beginning in about 5000 BC.14 A consequence of this aridification of
the Western Desert was the movement into the oases and the Nile Valley of
hunter-gatherers and pastoral nomads. They established more sedentary
communities clustered around these more fertile regions.15 These sedentary
communities were then compelled to find the most favourable networks of
transportation between the oases and areas further afield to reduce the

9
Giddy 1987, 3. 10 Kamal 1935, vol. 1, 395. 11 Aufrère 2000.
12
See Mattingly and Sterry, Chapters 2–3, 5–8, this volume, who argue that pre-Islamic sedentary
oasis settlements can be found across the Sahara, contrary to past expectations.
13
Gatto 2002; Kröpelin and Kuper 2007; Riemer 2004; 2009; Riemer and Kuper 2000.
14
Kuper and Kröpelin 2006. 15 Kuper 2002.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 152 Anna Lucille Boozer

impact of the harshness of the desiccated environment.16 Under the Fourth

Dynasty (c.2613–2494 BC), the Egyptian Pharaoh, Khufu (2589–2566 BC)
sent expeditions to the oases to seek a mysterious item called ‘mefat’.17
Egyptian control over the ‘inner’ oases was established shortly thereafter.
Settlement data within the oases also first appears during the Old
Kingdom (c.2649–2150 BC). At the juncture of two major roads in the
east of Dakhla, the Darb el-Tawil and another road that leads to the Darb
el-Arbain, there is a large multi-phase settlement at Ain Asyl/Balat and
a cemetery at Qila el-Dabba. The dates of these sites range from the Old
Kingdom (late Fifth/early Sixth Dynasties, c.2400 BC) to the Second
Intermediate Period (c.1650–1550 BC), with a late reoccupation in the
Roman Period. Ain Asyl/Balat seems to have been the site of the Old
Kingdom capital in Dakhla. The settlement was once surrounded by
a large mud-brick enclosure wall, which subsided towards the east in
antiquity and was later built on top of. The interior of this urban complex
was well-built and planned. The funerary chapels of three governors of the
oasis were located at this site. A stele was found in one of these buildings,
which contains a copy of a royal decree of Pepi II that mentions establish-
ing a ‘dwelling of vital strength’.18 Initially, Ain Asyl may have been an
outpost in a ‘foreign’ territory, but it seems that the oasites and those
manning the outpost integrated over time. Epigraphic material excavated
at Balat/Ain Asyl strongly suggests that Egyptian expeditions to the far
west, as well as to other desert regions, operated from the Dakhla oasis.19 In
recent years, evidence has shown that Pharaonic Egypt attempted to
explore the desert west of the oases, from at least the Old Kingdom.20
Since the Late Old Kingdom or the First Intermediate Period (c.2200/
2100 BC), Egyptians were able to traverse the nearly 400 km stretch from
Dakhla to the Gilf Kebir, and potentially even further.21 Dakhla’s gover-
nors were called ‘admirals’ and subsequent Egyptian expedition leaders
carried nautical titles underscoring the explorative nature of their roles.22
There may have been raids into Upper Nubia at this time, which were part
of the same Egyptian expansion of exploration, trade, and extraction.23
Additionally, more than 20 potential Old Kingdom ‘satellite sites’ in the
Dakhla oasis attest to the increasing importance of the oasis from the Old
Kingdom onward.24 Excavations at Mut el-Kharab, a significant locale

16
Riemer and Förster 2013, 33. 17 Kuhlmann 2005. 18 Pantalacci 1985.
19
Pantalacci 2013. 20 Kuhlmann 2002; Kröpelin and Kuper 2007.
21
Riemer and Förster 2013, 48. 22 Helk 1975, 127–32; Valloggia 1985. 23 Török 2009, 56.
24
Giddy and Grimal 1979; Osing 1982, 1; Soukiassian et al. 1990. On pharaonic Dakhla, see also
Hope and Pettman 2013; Mills 1999b; 2002; 2013; Mills and Kaper 2004.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 153

from at least as early as the Old Kingdom until today, have revealed
ceramics and architectural features dating to the Old Kingdom.25 The
western portion of Dakhla, near the village of al-Qasr, has revealed
a larger cluster of Old Kingdom settlements. None of these sites is fully
excavated and they are only partially published.26
There is only sparse archaeological data relating to the Middle Kingdom
(c.2055–1650 BC), although evidence of outposts along desert roads sug-
gest that there were once Middle Kingdom settlements, as well as trade
links, in Dakhla and Kharga.27 Similarly, little is known about New
Kingdom (c.1550–1070 BC) occupation. This lacuna is largely because
oasis sites were continuously occupied and the older remains are covered
by subsequent occupation. A few stone blocks from Kharga’s Hibis Temple
indicate that there was once a New Kingdom temple in the same location,
but nothing is known of its form.28 There is some ceramic and documen-
tary evidence of the Third Intermediate Period (c.1069–664 BC) from Mut
el-Kharab, suggesting continued occupation from the Old Kingdom.29
Under the Persians (525–402; 343–332 BC), the oasis region seems to
have become an important strategic area since there are growing signs of an
imperial presence in the form of temples and waterworks. Settlements, if
not cities, must have accompanied this investment. It is unclear why the
Persians devoted such time and energy to this region. Agricultural goods
from the area may have been attractive as may have been the trade links
that the oases provided to the south. Kuhlmann suggests that gold may
have been the primary incentive. Gold was the chief Nubian commodity
and the oases and Nubia were closely interconnected. Over the course of
the fifth century BC, signs of activity along the roads connecting the Great
Oasis to the Nile Valley increased.30 New technologies for water manage-
ment, such as qanats, also appear to have been developed during the
25
Some Sheikh Muftah handmade wares, taken as the work of indigenous Dakhla inhabitants,
were found alongside a smaller number of Nile Valley wares, including some identifiable
carinated Meidum bowl fragments, Hope 2005, 50–53. These Meidum bowl fragments were
made of Nile Valley marl fabric and were certainly imported. Hope suggests that these ceramics
may date to Dynasties IV and V and therefore to the initial phases of Egyptian annexation and
colonisation of Dakhla, Hope et al. 2009; see also Kuhlmann 2002.
26
Kilns for ceramics production were found at one site, Mills 1999a; 1999b.
27
Castel and Tallet 2001; Darnell 2002a, 172–73; 2007b, 38, figure 3; 2013b, 22, 25, 257.
28
Cruz-Uribe 1999.
29
Hope 2004; 2005. Attempts to make general statements about the Third Intermediate Period in
Dakhla have been forced to grasp at evidence dating to many hundreds of years after the Third
Intermediate Period, Hubschmann 2012.
30
Kuhlmann 2013, 159–60. This is attested by graffiti naming Darius I and by the presence of
sigha-pots (a variety of flask), datable to the fifth century BC, which have been found along
these same roads, Darnell 2000; Di Cerbo and Jasnow 1996).

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 154 Anna Lucille Boozer

Persian Period.31 This technology almost certainly passed further west

across the Sahara along the chain of oases that links the Western Desert
with Fazzan.32
Although the Kharga oasis has some evidence for Pharaonic activity,
significantly more material survives from the Late Period onward. The
Persians were active in the Kharga area and the sites around Kharga rose in
importance at this time.33 At least four temples were constructed in
Kharga, along with an increased number of structures.
The temple of Amun at Hibis, located just north of the modern town,
Kharga city, is the best known of Kharga’s Persian temples.34 Qasr el-
Ghuieta, 17 km south of Hibis, also has a temple (ancient Perwesekh).35
Ain Manawir, in the southern reaches of the oasis, has an agricultural
settlement on a small hill, which includes a mud-brick temple dedicated to
Osiris, as well as two groups of houses.36 At least 22 qanats fed the
agricultural fields below this site.37 Several hundred ostraka, written in
Demotic (a script form of writing derived from a northern form of Hieratic
in a late phase of the Egyptian language), attest to the Persian period
occupation of this site.38 French archaeologists working at Dush/Kysis
found a temple in the southernmost edge of the oasis and dating to the
Persian period.39 A settlement dating from the Persian period has revealed
a temple, some important documents written in Demotic script and traces
of the irrigation system dating to about 500 BC.40 This building activity and

31
Qanat (plural qanatha, qanawat; Anglicised as ‘qanats’; also known as foggaras) is an Arabic
term for an underground gallery connecting a water source with a cistern some distance away
from it. On qanats, see Newton 2013. For qanats in Kharga, see Bousquet 1996; Chauveau 2005;
Schacht 2003; Wuttman 2001; Youssef 2012. See also the fuller discussion of foggaras by
Mattingly et al., Chapters 2 and 3, this volume.
32
Mattingly et al., Chapter 2, this volume; Wilson 2002; 2006; 2009; Wilson and Mattingly 2003.
33
Cruz-Uribe 1999.
34
Temple construction began in the Saite Period, likely during the reign of Psametik II, with
additions by Darius I, Hakor, Nectanebo I and II, and possibly Ptolemy IV, Davies 1953;
Winlock 1941; for an alternative dating interpretation, see Cruz-Uribe 1988.
35
Approximately 19 × 10 m with some early portions that appear to date to the 25th or 26th
Dynasties. It contains additional contributions by Darius I, and Ptolemies III, IV, and X. The
Theban triad is depicted in relief decoration, suggesting connections with both Hibis and
Thebes. This temple is surrounded by a fortress considered to be of Roman date, but this dating
conjecture has not been verified, Cruz-Uribe 1999; Darnell 2007a; 2013a.
36
Grimal 1997, 340–42; Wuttman 1996, 393–40.
37
Wuttmann et al. 1998; 2000; Wuttman 2001.
38
These ostraka also provide details concerning water rights and land management, Chauveau
1996; 2001; 2005; 2008.
39
Conceptually, the south of Kharga seems to have been considered to be ‘Kush’. Dush<Greek:
Kysis<Old Persian: Kusiya, Mitchell 1992.
40
Bousquet 1996, 195–202.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 155

the Kharga qanats attest to a growing population, but urban sites are
unknown from the Kharga oasis during the Late Period.
The Dakhla oasis provides a patchy, but growing, corpus of data on
Persian Period occupation. The epigraphic material from Amheida’s tem-
ple has provided an important collection of cartouches with Late Period
names that indicate governmental investment in the oases.41 Mut el-
Kharab has revealed substantial quantities of Late Period ceramics in the
vicinity of the Seth Temple, which require further study, but once again
suggest the continuing significance of Mut.42 At some time before this
period, the centre of power in Dakhla seems to have shifted from further
east in Balat/Ain Asil to Mut, where it still remains.43 Qanats have also
been identified in Dakhla, although they have not been thoroughly
researched yet.
Until recently, evidence for Ptolemaic (305–30 BC) activity was thought
to be limited in comparison to the Roman activity in the Western Desert.
Recent research on Dakhla ceramics revises this assumption and suggests
that substantial settlement and agricultural activities occurred in Dakhla
during this period.44 These results suggest that ceramics that were often
lumped into Roman temporal brackets ought to be re-examined.
Architectural remains are fragmentary during this period, based upon
our current understanding, but may be hidden beneath Roman ruins. It
is equally possible that some ruins categorised as Roman during initial
surveys may be re-attributed as Ptolemaic upon further examination.45
One of the largest ancient towns of Dakhla lay in the eastern part of the
oasis at the site of Ain Birbiya, although it is the temple that has been the
focus of the excavations.46 Modern-day Mut (ancient Mothis) seems to
have been the capital of Dakhla during the Ptolemaic and Roman Periods.
The ancient town is only preserved in small islands of tell formations
within the modern city. Among these disturbed urban remnants is
a large Ptolemaic and Roman era temple dedicated to Seth, the patron
deity of the oasis.47
Once again, the Kharga oasis provides obscure evidence of settlement
data. Qasr Zaiyan (ancient Tchonemuris), 27 km south of Hibis, appears to
date from the Ptolemaic era, although the only dated inscription is Roman
41
Kaper 2009; Kaper and Davoli 2006; Kaper and Demarée 2006. 42 Hope et al. 2009.
43
Sterry and Mattingly, Chapter 8, this volume, argue that Saharan towns frequently experience
power shifts and relocations over time, rather than remaining fixed on single sites.
44
Gill 2012a; 2012b. 45 For a summary of the initial survey, see Churcher and Mills 1995.
46
Mills 1999b. It was called ‘Imret in Egyptian and perhaps Mesobe in Greek (P Kellis IV 74–75).
On the temple, see Kaper 2010.
47
Hope et al. 2009. Much of Gill’s ceramics data derives from Mut, Gill 2012a; 2012b.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 156 Anna Lucille Boozer

and dates from the reign of Antoninus Pius (AD 140). Egyptian archae-
ologists working at the site cleared much of the temple precinct, exposing
a substantial granary and associated administrative quarters.48 Other areas
of the site remain unexcavated and the finds are unpublished.

Roman Urbanisation of the Western Desert

The Western Desert experienced a large population increase under Roman

rule, which appears to have developed and expanded upon pre-existing
settlement patterns. This expansion can be characterised as urbanisation in
the case of some oases, namely Kharga and Dakhla, but potentially others
as well. This pattern of urbanisation complements the broad trend of
urbanisation found across Roman North Africa, which became one of
the most densely urbanised regions of the Empire.49

Kharga
Under Roman rule (30 BC–AD 641), Kharga experienced its greatest
prosperity and took shape with new fortresses.50 The location of these
forts suggests that Roman administrators were concerned with internal
security as well as trade routes, but it is uncertain if they represent
a response to changing circumstances or were part of an overarching
strategy, because dating remains uncertain.51
Hibis was the ancient capital of the Kharga oasis during the Roman
Period. Little is known of this settlement’s form beyond the Metropolitan
Museum of Art excavations 1909–1910, which uncovered a few houses
with vaulted ceilings and frescoed paintings. It appears to have been
garrisoned and the Temple to Amun-Re stood at the centre of town.52
The towns of Kysis, Ain Manawir and Shams el-Din/Mounesis, appear to

48
Cruz-Uribe 1999. 49 See Wilson, Chapter 10, this volume.
50
Reddé 1999. The character of the fortresses themselves signifies that they were primarily
residential since the walls were not nearly wide enough to sustain an earnest attack, Ikram and
Rossi 2004; Rossi 2000.
51
Morkot 1996, 87.
52
Winlock 1941. The Metropolitan Museum of Art (New York, NY, USA) had an exhibition that
focused on objects from Kharga in 2013, see: www.metmuseum.org/exhibitions/listings/2012/
kharga-oasis [last accessed 4 September 2019]. They have also digitised many of Herbert
E. Winlock’s notes and journals from his excavations: http://libmma.contentdm.oclc.org/cdm/
landingpage/collection/p16028coll10 [last accessed 4 September 2019].

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 157

have had populations engaged in trade and cultivation, both important

practices on the trade route that the fortress guarded.53
In southern Kharga, the Institut Français d’Archéologie Orientale have
been excavating at Qasr Dush (ancient Kysis) from 1976 onwards.54 It is the
best-documented site in this oasis and, although the Roman occupation is
best-known, it may have had occupation, at least temporarily, since the Fourth
Dynasty. In the Roman Period, Dush was a garrisoned fortress with
a settlement and temple to Sarapis (originally Osiris). Cemeteries lie to the
north and west of the town itself, with the Roman cemeteries being the largest
and located to the south-east.55 Inscriptions dating to the reigns of Roman
emperors Domitian (AD 51–96), Hadrian (AD 76–138), and Antoninus Pius
(AD 86–161) have been exposed.56

Siwa
The Siwa oasis was far removed from the rest of the oases in the Western
Desert. There is little evidence of prehistoric occupation at Siwa.57 No
ancient Egyptian sources concerning the oases mention Siwa.58 Egypt
seems to have found few appealing luxury goods or specialty items in
Siwa and so the oasis connected more to eastern Libya for most of its
history. The development of Siwa seems to have taken off particularly once
the Greeks began to immigrate into Cyrenaica from the seventh
century BC onwards.59 These Greeks may have been interested in Siwa as
a connection point to Nubian gold.60 There is no evidence of Carthaginians
exchanging goods with or engaging with Siwa or areas east of Siwa.61 The
Siwa oasis remains a mystery in the Roman Period, although markedly less
so than in earlier periods. Siwa lost its political sovereignty and became
a province of the Roman Empire (nomus hamoniacus) in c.74 BC when the
Pentapolis came under Roman Rule (c.96 BC). The oasis only became part
of Egypt after Muhammad ‘Ali forcibly incorporated it into Egypt proper
(AD 1819–1820).62

53
Wagner 1987, 182–83.
54
Dunand et al. 1992; 2005; Harrison 2007; Reddé et al. 2004. On the ceramics from Douch/Kysis,
see Ballet 1990; 2000; 2004.
55
Dunand et al. 1992; Reddé et al. 2004. 56 Reddé 1990.
57
On the available evidence of prehistoric habitation, see Hassan 1976; 1978.
58
Kuhlmann 2013, 144. The only possible reference is from the Edfu topographical list, dating to
Hellenistic times, Aufrère 2000.
59
Kuhlmann 2013, 147. 60 Kuhlmann 2013, 160. 61 Kuhlmann 1988, 97–101.
62
Kuhlmann 2013, 146.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 158 Anna Lucille Boozer

At Siwa the Supreme Council of Antiquities has carried out excavations

at Bilad el-Rum, a cemetery with elaborate rock-cut tombs replete with
entrance doorways and exterior chambers constructed in limestone. The
results of the excavation have not been published yet. Tombs in Siwa
suggest a moderate resident population in the Ptolemaic and Roman
Periods. Although these tombs have been scavenged over the years, it is
clear that they follow mortuary patterns common to the Nile Valley.63
Siwan settlement sites, such as Ain el-Qurayshat and Abu Shuruf, have
been under exploration only recently. Ain el-Qurayshat has many olive oil
presses concentrated together, an adjacent residential zone, and a temple.
The houses were mud-brick and built around large courtyards, reflecting
a more Romanised plan than traditional Egyptian practices.64 These houses
somewhat resemble those from Kellis in the Dakhla oasis. Industrial oil
production can also be found in other areas in eastern Siwa, such as Abu
Shuruf, Timeira, and al-Ma’asir.65 Under the reign of Trajan (AD 98–117),
it seems that wide-scale olive cultivation was introduced. At Siwa, the olive
became the oasis’ foremost crop, above even the date palm.66 Oil from
Siwa, known as the ‘oil from Ammon’, fetched a higher price than ‘oasis
oil’.67

Bahariya
Bahariya oasis shows signs of occupation throughout antiquity and parti-
cularly in the Roman Period.68 The oasis is divided into two regions: the
north clustered around al-Qasr/Bawiti (ancient Psobthis) and the southern
area, known as al-Hayz. There appears to have been heavy occupation in
the northern part of the oasis, but the ruins (east of Bawiti) have not been
excavated. Ahmed Fakhry uncovered a temple dedicated to Alexander the
Great at Bawiti and there is a Roman triumphal arch nearby.69 Qusur
Muharib consists of a Roman Period village with buildings of two storeys
and some very large houses as well as a mud brick fortress.70 Little is known

63
Fakhry 1990, 173–74.
64
Kuhlmann 1998. Egyptian houses typically had a courtyard off to one side or behind houses.
65
Kuhlmann 1988, 167–73.
66
Kuhlmann 2013, 140. Siwa was also a source for fossilised ammonites, which resembled the ram
horns, and were considered to be a sacred gemstone, Rohlfs 1875, 169–70, Kuhlmann 2013, 154.
See also Pliny, Natural History 37.60.167.
67
Kuhlmann 2013, 150.
68
Ball and Beadnell 1903; Fakhry 1974; Gosline 1990; Thissen 2013. For popular accounts of
Bahariya, see Jackson 2002, 233–39; Vivian 2002, 174–212.
69
Fakhry 1974. 70 Fakhry 1974, 106, 108.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 159

about the individual houses or their contents. Among the most spectacular
finds of the Roman Period are the remains of ‘golden mummies’ at
Bawiti.71 The settlement to which this cemetery belongs has not been
identified yet, but they derive from a necropolis associated with the capital
of Bahariya and give some indication as to the wealth and importance of
this oasis during the Roman Period. The gold used to make these masks
indicates links with Nubia, which was the closest source for this precious
metal. The only known mine of iron ore in the Western Desert can be
found on the edge of the escarpment north of Bahariya, which seems to
have been exploited in antiquity.72 To the south of this oasis, in the area of
el-Haiz, scant evidence of occupation can be found that dates to prehis-
toric, Old Kingdom, Second Intermediate Period and the New Kingdom.73
By contrast, Roman occupation can be found at 10 out of the 13 surveyed
sites.74 Two small churches and a late Roman cemetery also have been
found in this area.75 Recently, excavation has begun at a late Roman
settlement, known as Bir el-Showish.76 Military architecture as well as oil
and wine presses have been found elsewhere at el-Haiz, again attesting to
the introduction of crops prized by the Roman Empire as well as a resident
population about which little is known.77 Qanat irrigation technologies
were also introduced in Bahariya to enable this agricultural exploitation.78

Farafra
Most of the antiquities found in Farafra oasis date to Roman times, but
these are relatively insignificant. The visible Roman ruins consist of Ain
Della (north of Farafra) and Ain Besay to the south. Ain Besay consists of
an ancient cemetery, two mud brick buildings, rock-cut tombs, and a
chapel.79 Qanats have also been identified in Farafra.80

Dakhla
The Dakhla oasis was profoundly impacted by Roman rule, as shown by
the Dakhla Oasis Project (DOP) archaeological survey of the oasis between
1977 and 1983, which revealed over 200 Roman Period sites in the oasis.81
This survey was the first broad scale research conducted in the oases. Since

71
Hawass 2000. 72 Kuhlmann 2013, 147. 73 Dospěl and Suková 2013, 6–7.
74
Dospěl and Suková 2013, 7. 75 Mullins 2011. 76 Verner and Benesovska 2008, 54.
77
Hawass 2000, 148–67. 78 de Angeli 2013, 276–79.
79
Barich and Hassan 1990; Fakhry 1974. See also, Jackson 2002, 230–32; Vivian 2002, 143–73.
80
de Angeli 2013, 279–80. 81 Churcher and Mills 1995.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 160 Anna Lucille Boozer

1986, Colin A. Hope of Monash University has conducted excavations at

Ismant al-Kharab/Kellis, which appears to have been occupied during the
first to fourth centuries AD. It had a monumental temple to the local god,
Tutu.82 Two monumental, classical style tombs dating to the third
century AD were found in the north of Kellis. According to Hope, these
tombs find parallels only in Libya; they also resemble depictions of build-
ings on coins from North Africa (first century BC) and Alexandria (second
century AD), signalling potential connections to these regions.83 Small
houses, wealthy houses, cemeteries, and other tombs have also been exca-
vated. Of particular interest are economic and literary texts that place Kellis
at the centre of a regional economy during the fourth century AD.84
Importantly, these excavations have revealed architectural and decorative
motifs, as well as portable objects, that suggest Roman Mediterranean
influences and interaction.85 The settlement form shows clusters of struc-
ture types and organisation but little formal planning. Excavations at Mut
el-Kharab, like those from Trimithis, point towards a wealthy local popu-
lation. The Temple of Seth and its associated structures, which originated
(as noted earlier) in the Late Period demonstrate significant activity
throughout the Roman period.86

Trimithis: A Roman Period City in the Western Desert

Trimithis offers an ideal case study for Roman urbanism in Egypt’s
Western Desert since it had city status and sufficient visible architec-
ture for mapping its urban layout. The lack of other demonstrable
urban sites within the oases, despite their many villages, follows the
pattern Mattingly and Sterry describe for pre-Islamic Sahara oasis
settlements.87
Excavations at Amheida/Trimithis since 2004, led by Roger S. Bagnall,
have contributed valuable information on Roman Dakhla urbanism.
However, despite active research at Amheida, there are, as elsewhere in
Dakhla, hindrances to exploring diachronic change. Most of the visible
architectural remains at Amheida relate to Trimithis, the Roman city and it
is the best-understood phase of the site. Earlier phases of occupation have
been recovered, particularly in the area of the temple mound, where
excavation has revealed a great depth to the tell formation that underlies

82
On Tutu, see Kaper 2003. 83 Hope 1999. 84 Bagnall 1997; 2008.
85
Hope and Whitehouse 2004; 2006; Kaper 2012. 86 Hope et al. 2009.
87
Sterry and Mattingly, Chapter 8, this volume.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 161

the many phases to the destroyed temple.88 Despite these keyholes into
earlier phases of occupation at Amheida, a secure timeline of activity is not
possible at this time.
Topographical survey of visible remains continues to define the extent
and form of Trimithis’s urban fabric, although preliminary conjectures
regarding the layout can be made at this time (Fig. 4.3).89 Trimithis is
a large urban site that may be as large as 100 ha, although the core
occupied area was much smaller. Defining the perimeter of the site is
difficult because it is surrounded by ancient farmhouses and villages, as
well as modern agricultural fields, roads, and Bedouin villages that overlie
parts of the site.90 Trimithis manifests a diversity of structure types within
its urban core, as can be expected for a major regional centre. This
diversity attests to the multi-purpose function of the city for such likely
roles as administration, tax collection, law, governance, craft production,
and trade. This range in functions relates to political and cultural refer-
ence points of Western Desert cities being focused upon the Nile Valley
and Egypt.
The Trimithis houses seem to be single- or, in some cases, possibly two-
storey mudbrick structures with mainly barrel-vaulted roofs and some
palm reed and mud flat roofs. Industrial areas can be found distributed
among the domestic structures, but primarily along what we currently
understand to be the edges of the city. A temple mound on the west side,
around which the Roman settlement curves, and mortuary structures along
the southern side are also clearly evident among the surface remains.
Agricultural fields occupied the low-lying surrounding landscape.
Governmental, administrative, or so-called ‘public buildings’ have not yet
been identified securely.
There are two major streets currently identifiable at Trimithis. First,
a broad east-west oriented road (S1) provided access into the city from the
east. It leads from the industrial and domestic area on the north-eastern
portion of the site towards an area north of the temple mound, but it turns
88
Hypotheses regarding diachronic change and urban development can be found in Boozer 2014.
For a more thorough discussion of changes in the temple area, and particularly from Egyptian
religion to Christianity, see Bagnall and Cribiore 2012. Temples, as spiritual centres, could offer
an additional reason for urbanism in a particular locale. See Scheele, Chapter 18, this volume.
89
See Amheida.org. On diachronic change within Trimithis’ urban form, see Boozer 2014.
Geophysics was employed for several seasons at Trimithis, but the results did not yield
significant new data, except upon the temple mound, Smekalova and Smekalov 2005; 2006.
Instead, most of the observations are based upon mapping completed using a standard Total
Station.
90
On the so-called columbarium farmhouses of Dakhla, which cluster around Trimithis, see Mills
1993; also Boozer 2019.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 162 Anna Lucille Boozer

Figure 4.3. Plan of Trimithis (Amheida) (Courtesy of the Amheida Project).

sharply off axis and then stops abruptly before reaching the temple mound
itself. At 6.82 m across, it is the widest identified road at Trimithis. It
appears that this road and the structures along its east side may have been
built during a single phase of construction. Area 1, which clusters around
this street, appears to have been occupied from the early-mid third to early
fourth century AD (on the basis of surface ceramics) and seems to have
been constructed at the same time as the street itself.
The only modifications to the Area 1 street are small features con-
structed just outside doorways that may have served as windbreaks, refuse
breaks, or bollards to protect house fronts from traffic (or all three). This
situation can be found just west of House B2, and comparanda to this

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 163

situation can be found at Kellis, Karanis and Dimé/Soknopaiou Nesos.91

The maintenance of a road without blockages suggests that there may have
been institutional involvement in maintaining the road.92 By contrast,
blockages often signify lack of civic control, as can be seen elsewhere at
Trimithis (for example, streets S2 and S3, see below).
The second major street is a north-south oriented road that extends
from the mortuary area in the south-eastern portion of the city (Area 3) to
the north (Area 1). The major north-south oriented road is approximately
5.73 m wide, at its greatest extent, although the width varies considerably
depending upon the location. This road is the only clear north-south artery
across the city. The road shifts and winds along its passage and this road
appears to be less planned than the Area 1 road. This road appears to have
had structures built into it over time, which made it less effective as a major
conduit across the city. Nonetheless, it was probably an important road as
it linked up the northern and southern zones of Trimithis. It meets up with
the main road in Area 1 after a sharp turn to the north-east. Smaller east-
west throughways and alleys intersect with the road. This street might well
have served a performative function, since it is likely that it was used for
public funeral – among other – processions down to the tomb area (Area 3)
from the rest of the city. Such processions of honour and acclaim were
common throughout Roman Egypt as well as the Roman Empire more
broadly.93
In addition to these major roads there are some less-fully traced out
streets and alleys. One street runs roughly parallel to the major north-south
road and is located between this road and the temple mound. Some
additional east-west aligned streets also occur. Curiously, the project has
not yet identified any streets connecting the eastern settlement area with
the temple mound, nor have we found any trace of the temple dromos
(entrance passage).94
The buildings at Trimithis are preserved up to 3 m in height above the
site surface, although most of the visible architecture peeks up above the
sand by only 0.5 m or so. From these visible remains, some structures
appear to be laid out in an orderly plan of large blocks (insulae), such as in
Area 1, the immediate vicinity around House B1, and south of Area 4. The
north-eastern part of the site (Area 1) appears to have been laid out in
a single phase and the structures in this area share similar plans. The Area 1
layout – along with structures to the south-west – is oriented north/north-

91
Hope et al. 1989; 1992, 41–42. For a summary, see Boozer 2015. 92 Keith 2003, 63.
93
Haas 1993; MacCormack 1981. 94 Congedo and De Santis 2006, 20.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 164 Anna Lucille Boozer

west, south/south-east. The centre of the Roman site (Area 2) also contains
several houses that, from the surface, appear to share similar layouts. Area
2 structures have a general north-south orientation. This area contains
mostly fourth-century ceramics on the surface. Likewise, in an area just
south of the temple (Area 4), a series of roads delineate the eastern
boundary of a housing block where at least one significant, decorated
house is identifiable on the basis of surface architecture and pigmented
plaster.95 South of Area 4 the layout follows a similar north-south orienta-
tion. It is unclear when Trimithis took on this urban form, although it is
evident that some form of urban planning led to its present shape.
Other areas of the site seem to have grown up much more organically,
perhaps filling in spaces between planned areas. These areas can be found
north and west of House B1 and west of House B2. Area 4 is architecturally
complex and highly deteriorated so its overall plan remains unclear. It is likely
that the temple of Area 4 resembled other Roman Period Egyptian temples,96
and the temples of Roman Dakhla in particular.97 Like the temple itself, the
area surrounding the temple is complex and largely destroyed so it will require
substantial research before the layout and phasing are understood.

Trimithis Comparanda
Documentary research has shown that Trimithis continued to grow in size
into the fourth century and it officially achieved city status by AD 304, when
it was described as ‘Trimithis polis’ in a papyrus.98 Although population
estimates for Trimithis cannot be given accurately at this time, we do have
several cities we can use for comparison. Bagnall estimates the population of
Hermopolis to be between 25,000 and 50,000, on the basis of the list of oikia
houses (SPV V. 101) and the average Egyptian metropolis to be at 25,000
(although we may consider Trimithis to be on the smaller side).99 With these
estimates in mind, it would not be unreasonable to suggest a population of
approximately 10,000–15,000 or so in late Roman Trimithis, but such con-
jectures probably will change as we gain more data.100

95
Congedo and De Santis 2006, 20.
96
Kaper and Davoli 2006. On late phases of the temple mound, see also Bagnall and Cribiore
2012.
97
Kaper 1997, 1998b.
98
P.Kell. I G.49.1–2; Bagnall and Ruffini 2004, 143–44, for the latest discussion of the urban
status.
99
Bagnall 1993, 53; Bagnall and Frier 1994, 55.
100
This number corrects an earlier estimate (Boozer 2014), thanks to critiques provided by
A. Wilson that a density of 250 people per ha would be more likely. For a discussion of the size

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 165

The urban status of Trimithis gives us some sense of civic governance at

this oasis city. The political constitutions of cities required the elite to serve
as local councillors (decuriones) and to hold a sequence of magistracies and
priesthoods that required them to exercise their political, judicial, and
religious authority on a regular basis. The size of each local council varied
according to the city’s population. It is likely that individual towns had
fixed sizes for their town councils and that council size was subject to
official regulation.101
By the early fourth century, the top echelons of the provincial aristocracy
made their careers in state service and as councillors. Typically the heads of
wealthier families took collective responsibility for city finances.102 Such
magistrates and councillors dominated the public life of late Roman
cities.103 At Trimithis, we excavated the house (B1) of a city councillor
named Serenos, who was identified from a substantial corpus of ostraka
recovered from this house.104 The structure (B1) that housed Serenos was
emblematic of his prestigious civic role and the wall paintings in one of the
rooms underscored Trimithis’s city status.105
Promotions to city status often led to significant expansion and devel-
opment of the urban form. Upon acquisition of city status, the foundation
of a city might involve building operations, particularly in the erection of
public buildings.106 Scholars have identified the development of monu-
mental civic structures as a key component of the Roman redesign of cities
such as Carthage/Colonia Concordia Iulia (Tunisia), Lepcis Magna
(Libya), Italica (Spain), Colchester (Britain) and other cities across the
empire.107 Sometimes these urban redesigns involved the substantial clear-
ance of private houses for the emplacement of new civic buildings that were
emblematic of the empire. This clearance, in turn, created the opportunity
to design Roman-style houses in place of local styles.108
On the other hand, some cities experienced difficulties following upon
the acquisition of city status, as at Oxyrhynchus (Egypt). By AD 235 many
houses in the city were abandoned (P.Oslo 3.111). At some time
between AD 232 and 247 the Oxyrhynchites lamented their own ruin,
although the city continued to be occupied well beyond this period.109

and population of other North African Roman urban sites, see Wilson, Chapter 10, this
volume.
101
Tacoma 2006, 126. 102 Mitchell 2007, 180–81. 103 Edmondson 2006.
104
Bagnall and Ruffini 2004; 2012. 105 Boozer 2010; 2012. 106 Jones 1964, 720.
107
Edmondson 2006, especially 251, 268–69.
108
For example, Julio-Claudian Mérida (Lusitania), Alba Calzado 2004; Edmondson 2006, 561.
109
On the inhabitants of Oxyrhynchus lamenting their ruin, see Parsons 2007, 58–59.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 166 Anna Lucille Boozer

Comparisons across Roman North Africa suggest that urban expansion

took place primarily between the late first through the early third
centuries AD. There was a notable decline in public construction by the
late third century, but individuals maintained a high standard of living
until at least the later fourth or early fifth century AD.110 In most cases,
excavation and survey have shown that the area of urban settlement in the
early Imperial period remained fully occupied under the Late Empire. In
some cities (for example, Cuicul/Djemila, Sitifis/Setif and Thamugadi/
Timgad), new built-up areas were constructed in the fourth century,
attesting to population increases.111
There has been remarkably little research on diachronic urban devel-
opment in Roman North Africa, although there have been some recent
developments on this front for Late Antiquity.112 For example, Wilson
suggests that a typical feature of revitalised Roman North African cities
may have been densely packed urban areas of mixed usage, containing
houses and workshops.113 In particular, Wilson cites Sabratha (Libya)
as an example of a city that experienced this form of development in
the late second or early third century AD. These dates are slightly
earlier (c.50 years) than we find in Area 1 at Trimithis, but the type
of urban space that Wilson describes is reminiscent of the Dakhla case
study.
It is crucial to question the origins of town planning traditions in the
Western Desert.114 The impact of Roman rule on cities has received much
attention by scholars.115 Often, scholars have argued that a Roman influ-
ence can be seen clearly through formal urban planning. In particular,
orthogonal or Hippodamean planning has been considered a signature
attribute of Roman urban planning. Yet, cities with regular layouts occur in
many different regions and historic periods for pragmatic reasons. Rather
than simply betraying cultural roots, regularity may simply indicate the
new foundation of a site or neighbourhood as well as a centralised author-
ity guiding development. We should remain mindful that Greek models of
urbanism often overshadow the equally legitimate urbanisms found in the

110
Lepelley 1992; Mattingly and Hitchner 1995, 185.
111
Duval 1982a; 1982b; Février 1964; Lassus 1969.
112
For critiques, see Mattingly 1994, 11; Mattingly and Hitchner 1995, 186. On Late Antique
developments, see Leone 2007.
113
Wilson 2002, 258–59.
114
For similar queries into the cultural roots of urban design, see Wilson, Chapter 10, Bokbot,
Chapter 12, and Nixon, Chapter 17, this volume.
115
For example, Alston 2002; Fentress 2000; Goodman 2007; Laurence et al. 2011; Liebeschuetz
2000; Lomas 1989.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 167

Near East and Africa. ‘Grid-plans’ and urban planning more broadly were
neither exclusive to the Greek and Roman worlds, nor universal within
them.
With this disclaimer in mind, caution must be exercised when attribut-
ing elements of planning to any cultural group at Trimithis given that
Egyptians, Greeks, and Romans had strong traditions of urban planning.
The town plans of Ptolemaic and Roman settlements in Egypt are often
more regularly laid out than Trimithis, following north-south/east-west
orientations across the entire site or, sometimes, other orientations related
to the placement of a central temple or the geographical terrain. This
regularity may be due, in part, to the more rigid grid-planning that
occurred during Ptolemaic Egypt for the foundation of new cities, as can
be expected for new design projects. The potential Ptolemaic impact upon
Trimithis’s form is not visible at this time, but must be considered to be
possible.
Pragmatism seems to have been the major factor influencing the orches-
tration of city plans and it is not necessary to interpret planning as the
victory of Egyptian, Greek, or Roman factors in the final choice of design.
A settlement’s layout (regular versus irregular) is often an index of the
circumstances of urban growth, and particularly the presence of
a centralised authority overseeing urban development.116
Parallels for portions of Trimithis can be found in the array of Roman
Period settlement types in Egypt, rather than elsewhere in the Sahara since
the Western Desert was oriented primarily towards the Nile Valley.
Geography served as a determining factor for a range of both planned
and organic cities of Roman Egypt. For example, parts of Alexandria follow
the Hippodamean system – a rectangular grid divided into regular
blocks.117 Even so, the natural topography was exploited in the construc-
tion of public architecture; a great theatre and the Serapeum were built on
the limestone ridge that overlooked the entire city. Streets and canals
leading from the Great Harbour to Lake Mareotis emphasised the north-
south axis of Alexandria. Alexandria contained typical Greek elements
such as an agora,118 a theatre, a bouleuterion (council hall),119 law courts,
a gymnasium, a hippodrome, temples to Greek gods,120 and an
armoury.121 On the domestic level, the housing complex at Kom el-
Dikka in Alexandria resembles a hybrid between the large insulae of late
antique Ostia and Rome and the multi-storey houses from Egyptian towns

116 117 118

Joshi 2000, 34. Empereur 1998, 56. Arrian, Anabasis. 3.1.5.
119 120 121
Delia 1991, 15. Strabo 17.1.9–10. Philo, In Flaccus 92.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 168 Anna Lucille Boozer

such as Karanis and Bacchias.122 Excavations at Kom el-Dikka by Polish

archaeologists have provided critical information about both public and
private facilities in Alexandria for the late Roman and Byzantine
centuries.123
Likewise, geography seems to have determined planning at smaller sites,
such as Berenike, on the Red Sea.124 Berenike covered approximately 7 ha,
and was a major Red Sea port with extensive commercial contacts locally,
regionally, and in the greater world context.125 It was occupied from
c.300 BC to early AD 600 and its peak periods were in the early (first–
second centuries AD) and late (fourth–fifth centuries AD) Roman Periods.
Berenike also seems to have been laid out in a rough pattern of blocks and
streets and did not have an enclosure wall. Over time the town shifted
slowly to the east.126 As with these sites, the layout of Trimithis exploited
the natural topography for promoting religious spaces and organising the
location of domestic structures. Geography also determined the fertile
areas around Trimithis, necessitating that elevated areas be devoted to
construction and lower areas to agriculture.
Other settlements show a sense of spatial organisation rather than
concerted planning, with houses arranged in single rows or built back-to-
back in double rows that roughly formed rectangular blocks.127 For exam-
ple, Karanis, in the Fayum, was built following two major north-south
thoroughfares, with no clear east-west through streets. The life span of
Karanis extended from c.300 BC to AD 600. It was an enormous settle-
ment, covering 60 ha including areas of occupational debris. The town had
an irregular street layout and appears to have expanded organically from its
Ptolemaic centre, first to the east and west, and then to the north. As
population levels receded, the occupied area shrank back from the north
and returned to its central core.128 It is evident from the plans that there
was a rough system of grouping houses into insulae, particularly during the
later periods.129
Likewise, the Ptolemaic through early Islamic structures at Syene/Aswan
are oriented on narrow streets running north-south, intersected by smaller
lanes.130 Various teams excavated a portion of Syene/Aswan, which the
Swiss Institute for Archaeological Research in Ancient Egypt (Cairo,
Egypt) eventually surveyed and published.131 The exposed area is quite
modest (40 × 100 m) and includes the Ptolemaic Temple of Isis and
122
Haas 1997, 195. 123 Rodziewicz 1984; 1991. 124 Sidebotham and Wendrich 1998, 87.
125
Whitcomb and Johnson 1979, 5. 126 Wendrich 1998, 246. 127 Badawy 1978, 33.
128
Boak and Peterson 1931, 3–9. 129 Boak and Peterson 1931, 40.
130
Jaritz and Rodziewicz 1994, 119. 131 Jaritz and Rodziewicz 1994; 1996.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 169

a residential area to the north that consists of approximately 23 houses. An

enclosure wall, possibly constructed in the third century BC, seems to have
ordered the orientation of the streets and structures.132 Although Karanis
and Syene do not exhibit clear planning, they show a preference, like
Trimithis, for major north-south streets. Likewise, the situation of consis-
tently oriented housing can be found in Area 1 at Trimithis and in discrete
patches across the site. Indeed, I suggest that there are several formally
planned insulae at Trimithis.

Economic Links with the Western Desert

The economic links between the Western Desert and the Nile Valley,
Nubia, and Libya may provide some explanation for Roman urbanisation
in the Western Desert. In particular, trade and resource extraction/exploi-
tation can be linked to imperial exploitation of the region.133
Three themes emerge from analysis of Roman Period economic links:
(1) the agricultural goods that were suitable to oasite conditions and that
seem to have spread west across the Sahara, (2) the technologies used to
provide water that were essential to managing these goods and which also
spread west, and (3) the evidence of trade contacts found through road
networks and material indices, although the latter are more difficult to
trace.
First, the Roman Empire exploited the Western Desert oases for their
agricultural potential, as it did with other parts of Roman North Africa.134
There was change over time and summer crops, such as sorghum, pearl
millet, and cotton appear to have become more common during the
Roman Period. Such crops were particularly suited to the oases because
of the perennial water supply, unlike the Nile River.135 It is not clear exactly
how these crops reached Egypt, as currently they are known at earlier dates
in the Central Sahara.136 The particular environmental conditions for
cultivating these agricultural goods created a clear motivation for develop-
ing the Western Desert. Unfortunately the ultimate destinations of these
agricultural goods are difficult to trace, even in the desiccated conditions of
132
Jaritz and Rodziewicz 1994, 11; 1996.
133
Evidence of development within desert oases under the Garamantian Kingdom and the
Roman Empire can be found elsewhere in the Sahara, see Mattingly et al., Chapter 3, this
volume. The Roman exploitation of Trans-Saharan caravan routes has been demonstrated in
the Sahara, Liverani 2006; Mattingly 2013, 530–34; see now, Mattingly et al. 2017.
134
See Wilson, Chapter 10, this volume. 135 Bagnall 1997; 2008.
136
Mattingly et al., Chapter 2, this volume.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 170 Anna Lucille Boozer

the Sahara, because the botanical remains do not survive well and/or
archaeologists did not sample for them.
Second, irrigation technologies including qanats/foggaras were essential
to the development of urban and smaller-scale sites in the Western Desert
from the Persian Period onward. These technologies seem to have moved
westward across the Sahara, also enabling population growth and
urbanisation.137 In the Western Desert, it is clear that this technology
allowed oasite farmers to move into new and more difficult hydrological
zones to grow agricultural products that are particularly suited to oasis
environments.
Third, the oases had well-developed trade contacts from the earliest
stages of their sedentary occupational history. Trade contacts with the
Nile Valley are visible from at least as early as the Old Kingdom, but it is
not clear that the Nile Valley inhabitants only sought oasite products.
Exotic goods west and south of the oases may have been their primary
interest, along with the control of trade routes that led south. Although it
would be tempting to argue for Trans-Saharan trade, evidence currently
supports only local-level trade within a network that consisted of the
Western Desert oases, the Nile Valley, and regions to the south.138
Archaeological research on the roads connecting the oases to the Nile
Valley is still in its infancy.139 Even so, we can recognise signs of trade through
the spread of objects and materials, technical knowledge, and other cultural
traits.140 There is evidence of trade centres, caravanserais, trading depots, and
checkpoints along the evident major road systems. Giddy compiled data on
old roads and tracks to produce an outline of Western Desert oasis routes that
is currently being investigated archaeologically.141

137
Mattingly and Sterry, Chapter 8, this volume.
138
There is much debate about the possibility of Trans-Saharan trade with the inner oases (namely
Dakhla and Kharga). On the ‘myth’ of Trans-Saharan trade, see Swanson 1975. Those who support
such trade rely upon new evidence from the ‘Abu Ballas Trail’, which has ceramic assemblages
dating from the late Old Kingdom/early First Intermediate Period and a hieroglyphic graffito,
Clayton et al. 2008; Kröpelin and Kuper 2007. Those who find such claims of a trade network
dubious include Bagnall 2012; Kuhlmann 2013; O’Connor and Reid 2003.
139
For an overview of the paucity of research on roads, with particular reference to Ancient Egypt and
the Sahara, see Riemer and Förster 2013. On Saharan networks in Roman times, see Wilson 2012.
For a road map of the historical Trans-Saharan trade network, see Ross 2011, XV.
140
See above discussion for imports to the oases. It is difficult to identify oasis goods in the Nile
Valley (and elsewhere). For a discussion of the spread of objects, ideas, and language across the
Sahara, see Masonen 2012.
141
Giddy 1987, 5–27. For a review of Giddy, with a comment on the road networks in Dakhla, see
Mills 1991. See also the Theban Desert Road Survey, directed by John and Deborah Darnell
since 1993 (for example, Darnell 2002b). For the impact of the Theban Desert Road Survey on
perceptions of the desert, see Friedman 2002b.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 171

The North Kharga oasis survey project has demonstrated that there was
a major expansion of road traffic during late Roman times.142 Oil, wine,
and cotton were the major commodities shipped from Kharga to the Nile
Valley during this period. Kuhlmann argues that gold and, to a lesser
extent, salt, may have played a major role in trade in pre-Islamic times.143
An important aspect of Kharga is that it was the point of departure of
a number of overland trade routes to Nubia and the Chad/Darfur region,
the so-called Forty Days Road, along which a number of minor oases
developed (such as Selima).144
Until recently, material evidence of trade connections and goods has not
been easy to trace because of both preservation and recognition issues.
Ceramics and other goods from the oasis have only been recognised and
seriated in the past few decades.145 As a result, it is likely that there are
additional oasite goods that have been overlooked in the analyses of
material beyond the oases. Growing comparisons of the pottery fabrics of
New Kingdom amphorae from both the Nile Valley and the Egyptian oases
already are beginning to shed light on over-land connections in Egypt’s
Western Desert.146
Various types of salt (alum; rock- or table salt; also vitriol) exist in the
oases themselves and were mined there throughout history.147 Dates were
another significant commodity. These two products, dates and salt, were
attractive beyond the fringes of the oases. Accounts from Herodotus
(4.172), as well as much later writers describe travellers, marauders, and
raiders journeying great distances to loot date plantations.148
Another debatable, and often invisible, commodity is a human one:
slaves. We know little about the ancient slave trade in Africa, but modern
comparisons can be informative. Slaves were sold and traded at Siwa from
early Muslim times until fairly recently.149 Kuhlman suggests that the
history of slavery in Siwa may extend back even further.150 Kulhman
adds that the introduction of the camel might have aided the importation
of slaves from greater distances, including locales in Sudan.151 There is

142
Rossi and Ikram 2013. 143 Kuhlmann 2013.
144
Kirkwan 1971; Rossi and Ikram 2013; Welsby 1996.
145
Research by Pascale Ballet, Deborah Darnell, Colin Hope, and James C.R. Gill, among others,
have made crucial advances in Western Desert Ceramics. See bibliography for citations
elsewhere to their works.
146
Hope 2002. 147 Kuhlmann 2013, 140.
148
On the Awjila from Kufra, see Hamilton 1856, 181, 191, 197. On marauders from Kufra/Tubu
and regions south of Dakhla (Bidayat from Ennedi or Mergah) who raided Siwa and Dakhla,
see Nachtigal 1974, vol. 2, 36; Stanley 1912, 32; Rohlfs 1875, 250.
149
Belgrave 1923, 96, 150. 150 Kuhlmann 2013, 156. 151 Kuhlmann 2013, 157.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 172 Anna Lucille Boozer

little published about the slave trade in Egypt, Fazzan, and elsewhere for
the pre-Islamic era, but the potential for the movement of peoples to
provide labour for new agricultural investments created by reclaimed
land through qanat technologies should not be overlooked.152

The Collapse of Urbanism in the Western Desert

By the fifth century there is evidence of a significant reduction in occupa-

tion at major sites such as Dush/Kysis in Kharga and Ismant al-Kharab/
Kellis in Dakhla.153 Amheida/Trimithis in Dakhla, with its lack of clear
evidence of any fifth century activity, likely shared the same destiny. This
collapse mirrors a pattern of shrinking urban sites and populations across
Roman North Africa.154 This regional collapse is reminiscent of a broader
Roman imperial collapse that is theorised, but rightly queried, for this time
period.155
It is unclear what factors catalysed urban collapse in the Western Desert.
Saharan comparanda suggest that the disruption of trade routes detrimen-
tally impacted cities, which stagnated and experienced massive population
declines.156 Within the local economy consisting of the Western Desert,
the Nile Valley and Upper Nubia, the often-blamed Blemmyes and other
nomadic groups of the area may have had a hand in disrupting trade. Raids
upon caravans may have made the already-expensive overland trade less
desirable. It is also possible that the oasite agricultural package (wine, dates,
cotton, and other summer crops) could be accessed more easily elsewhere
after the fourth century, thereby rendering Western Desert trade routes
less attractive. Equally, it is possible that the protection of trade routes in
the Western Desert was de-prioritised. All of these possibilities may have
worked in concert to disrupt trade and, in turn, urbanism in the Western
Desert.157

152
However, see Fentress 2011; Wilson 2012. There are more published accounts beginning with
the nineteenth-century slave trade, during a period when it was in decline and European
abolitionist governments took an active interest in its cessation, Lane and MacDonald 2011;
Savage 1992; Wright 1988.
153
On Dush, see Wagner 1987. On Ismant al-Kharab, see Marlow and Mills 2001.
154
Wilson, Chapter 10, this volume. 155 Jones 1964, 1025–68; Liebeschuetz 2000, 29–32.
156
Erdblom 1968, 305–14; Sterry and Mattingly, Chapter 8, this volume.
157
See Fenwick, forthcoming for a discussion of ‘boom-and-bust’ cycles in pre-Islamic and
Islamic North Africa.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 173

Discussion

I return to the questions posed at the beginning of this chapter regarding

(1) diachronic change, and, particularly, the character of the Romano-
Egyptian urbanisation, (2) economic links, and (3) the causes of abandon-
ment in order to draw together these queries to see if we can sketch out
testable hypotheses for understanding the development and collapse of
urbanism in Egypt’s Western Desert.
(1) Diachronic change: it appears that the Old Kingdom was a significant
phase of Nile Valley Egyptian colonisation. Following this interven-
tion, it appears that the Persian Period involved the next most visible
imprint upon the landscape of the Western Desert with the introduc-
tion of qanat technology. The Roman Period was the most heavily
occupied phase for the oases until the modern era.
Given the tell formation of sites, it is difficult to deduce the impact of
Persian and subsequent Ptolemaic urban development on the later Roman
urbanisation of the Western Desert. Even so, it seems that there are some
Roman North African patterns of urban development evident at least at
Amheida, if not other sites in the oasis. Large-scale urbanisation of the
Western Desert occurred along with Roman rule. This urbanisation may be
different in scale but not necessarily in kind (at least in all respects) in
comparison to earlier phases, about which we know frustratingly little.
This Roman investment may have been due to the economic potential of
the oases themselves.158 There also were imperial benefits to creating
a buffer zone around the Nile Valley and in securing trade routes.159 In
sum, it seems that Roman investments in the region built upon and
intensified settlement at important pre-existing locales in the oases, but
we cannot yet define the attributes of these pre-existing settlements.160
(2) Economy: it is clear that goods intrinsic to the oases were of interest to
various large state systems, but the links to further exotica may have
been even more attractive. The long-term, if episodic, use of these
trade routes may be the reason for the development of urbanism in
particular locales within the oases as well as their eventual decline.

158
For example, Mills 1993 argues that the Romans created a ‘New Valley’ scheme, much like the
modern Egyptian government did, to increase agricultural land and outputs.
159
Boozer 2013b.
160
Compare to Nixon, Chapter 17, this volume, who detects a pre-existing, if poorly understood,
settlement at Essouk, which later became an early Islamic entrepôt.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 174 Anna Lucille Boozer

(3) Abandonment: The oases continued to be occupied after Roman rule,

but the population density declined substantially. Decreases in trade
may be the cause. As Kuhlmann observes, ‘trade is like water; it will
flow either along the easiest or the only possible way, and it abides by
its course unless extraordinary circumstances force a change’.161 When
trade became too difficult in the Western Desert, or when other
options became viable, it took another course. From the seventh
century AD, it seems that Siwa was the starting point for journeys
leading into the Bilad al-Sudan (Takrur, Ghana, Mali, Gao, Bilma
[Kawar] or Chad) from Fazzan or Tibesti.162 Dakhla appears to have
been forgotten as a connection to the south, at least from what is
known from Classical Arab Sources.163
Marauders are another possibility that could have made Western Desert
trade less attractive and economically viable. The introduction of the camel
would have made the oases more accessible to nomads and other groups
that would have found them out of range for foot travel. These marauders
may have targeted caravans or simply crops themselves, such as dates (as
described above), which made oasis goods too costly and troublesome to
continue to import.
Despite the abandonment of major cities in the Western Desert, the
subsequent shape of settlement life in Dakhla rested on many old founda-
tions. The refocus of settlement to smaller villages and fortresses shows
continuity in location, if not form.164 Moreover, some cities, such as Mut
in Dakhla and Kharga City in Kharga, appear to have had continual occupa-
tion through to the present day. A narrative for the next phase of settlement
life in the Western Desert remains to be told. Perhaps the patterns we might
find are not nearly as different as it appears from our current vantage.

References

Alba Calzado, M. 2004. Arquitectura doméstica. In X. Dupré Raventós (ed.),

Mérida. Colonia Augusta Emerita (Las capitales provinciales de Hispania 2).
Rome: L’Erma di Bretschneider, 67–83.

161
Kuhlmann 2013, 133. 162 Kuhlmann 2013, 134.
163
Kuhlmann 2013, 135. Under Arab rule, it was Siwa’s extreme western position among the
oases along with its role as a gateway to Egypt for Maghribinian and African pilgrims that
made this oasis, rather than Dakhla, the starting point for Trans-Saharan camel caravans to the
south and south-west, Kuhlmann 2013, 161.
164
Kucera 2013.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 175

Alston, R. 2002. The City in Roman and Byzantine Egypt. London and New York:
Routledge.
Aufrère, S. 2000. La liste des sept oasis d’Edfou. Bulletin de l’Institut Français
d’Archeologie Orientale 100: 79–127.
Badawy, A. 1978. Coptic Art and Archaeology: The Art of the Christian Egyptians
from the Late Antique to the Middle Ages. Cambridge, MA: MIT Press.
Bagnall, R.S. 1993. Egypt in Late Antiquity. Princeton: Princeton University Press.
Bagnall, R.S. (ed.). 1997. The Kellis Agricultural Account Book (P.Kell. IV Gr. 96).
Oxford: Oxbow Monograph 92.
Bagnall, R.S. 2008. Economy and ecology in the Kellis agricultural account book. In
M.F. Wiseman (ed.), The Oasis Papers 2: Proceedings of the Second International
Conference of the Dakhleh Oasis Project. Oxford: Oxbow Books, 115–18.
Bagnall, R.S. 2012. Dakhla and the West in late antiquity: framing the problem. In
S. Guédon (ed.), Entre Afrique et Égypte: relations et échanges entre les espaces au
sud de la Méditerranée à l’époque romaine. Paris: Editions Ausonius, 39–44.
Bagnall, R.S. and Cribiore, R. 2012. Christianity on Thoth’s Hill. In Bagnall et al.
2012, 409–15.
Bagnall, R.S. and Frier, B.W. 1994. The Demography of Roman Egypt. Cambridge
and New York: Cambridge University Press.
Bagnall, R.S. and Ruffini, G.R. 2004. Civic life in fourth-century Trimithis: two
ostraka from the 2004 excavations. Zeitschrift fur Papyrologie und Epigraphik
149: 143–52.
Bagnall, R.S. and Ruffini, G.R. 2012. Ostraka from Trimithis Volume 1: Texts from
the 2004–2007 Seasons. New York: Institute for the Study of the Ancient World
and New York University Press.
Bagnall, R.S., Davoli, P. and Hope, C.A. (eds). 2012. The Oasis Papers 6: Proceedings
of the Sixth International Conference of the Dakhleh Oasis Project. Oxford:
Oxbow Books.
Ball, J. and Beadnell, H.J.L. 1903. Bahariya Oasis: Its Topography and Geology.
Cairo: Survey Department, Public Works Ministry.
Ballet, P. 1990. Annexe II. La céramique du site urbain de Douch/Kysis. Bulletin de
l’Institut Français d’Archeologie Orientale 90: 298–301.
Ballet, P. 2000. La céramique. Notes et remarques. In A. Hussein (ed.), Le sanc-
tuaire rupestre de Piyris à Ayn al-Labakha. Cairo: Institut Français
d’Archéologie Orientale, 91–105.
Ballet, P. 2004. Jalons pour une histoire de la céramique romaine au sud de Kharga,
Douch 1985–1990. In Reddé 2004, 209–240.
Bard, K.A. (ed.). 1999. Encyclopaedia of the Archaeology of Ancient Egypt. London:
Routledge.
Barich, B.E. and Hassan, F.A. 1990. Il Sahara e le oasi: Farafra nel deserto occi-
dentale egiziano. Sahara 3: 53–62.
Belgrave, C.D. 1923. Siwa: The Oasis of Jupiter Amman. London: William Brendon
and Son, Ltd.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 176 Anna Lucille Boozer

Boak, A.E.R. and Peterson, E.E. 1931. Karanis: Topographical and Architectural
Report of Excavations during the Seasons 1924–28. Ann Arbor: University of
Michigan.
Boozer, A.L. 2010. Memory and microhistory of an empire: domestic contexts in
Roman Amheida, Egypt. In D. Borić (ed.), Archaeology and Memory. Oxford:
Oxbow Books, 138–57.
Boozer, A.L. 2012. Globalizing Mediterranean identities: the overlapping spheres
of Egyptian, Greek and Roman worlds at Trimithis. Journal of Mediterranean
Archaeology 25.2: 93–116.
Boozer, A.L. 2013a. Archaeology on Egypt’s edge: archaeological research in the
Dakhleh oasis, 1819–1977. Ancient West & East 12: 117–56.
Boozer, A.L. 2013b. Frontiers and borderlands in imperial perspectives: exploring
Rome’s Egyptian frontier. American Journal of Archaeology 117.2: 275–92.
Boozer, A.L. 2014. Adapting urban space in late Roman Trimithis (Dakhleh oasis,
Egypt). In E. O’Connell (ed.), Egypt in the First Millennium AD: Perspectives
from New Fieldwork. Leuven: Peeters, Orientalia Lovaniensia Analecta, 23–42,
plates 1–3.
Boozer, A.L. 2015. Inside and out: Romano-Egyptian housing from the Fayyum
and the Dakhleh Oasis. In C. Hope and A. Di Castro (eds), Housing and Habitat
in the Mediterranean World: Responses to Different Climates. Leuven: Peeters,
187–200.
Boozer, A.L. 2019. Looking for singles in the archaeological record of Roman
Egypt. In S.R. Huebner and C. Laes (eds), The Single Life in the Roman and
Later Roman Worlds. Cambridge: Cambridge University Press.
Bousquet, B. 1996. Tell-Douch et sa région : géographie d’une limite de milieu à une
frontière d’Empire. Cairo: Institut français d’archéologie orientale.
Bowen, G.E. and Hope, C. (eds). 2004. The Oasis Papers 3, Proceeding of the Third
International Conference of the Dakhleh Oasis Project, Dakhleh Oasis Project:
Monograph 14. Oxford: Oxbow Books.
Castel, G. and Tallet, P. 2001. Les inscriptions d’El-Harra, oasis de Bahareya.
Bulletin de l’Institut Français d’Archeologie Orientale 101: 99–136.
Chauveau, M. 1996. Les archives d’un temple des oasis au temps des Perses.
Bulletin de l’Institut Français d’Archeologie Orientale 137: 32–47.
Chauveau, M. 2001. Les qanâts dans les ostraca de Manâwir. In P. Briant (ed.),
Irrigation et drainage dans l’Antiquité, qanâts et canalisations souterraines en
Iran, en Égypte et en Grèce. Paris: Thotm Éditions, 137–42.
Chauveau, M. 2005. Irrigation et exploitation de la terre dans l’oasis de Kharga à
l’époque perse. Les Cahiers de Recherches de l’Institut de Papyrologie et
d’Égyptologie de Lille 25: 157–63.
Chauveau, M. 2008. Les archives démotiques d’époque perse: À propos des
archives démotiques d’Ayn-Manawîr. In P. Briant, W.F.M. Henkelman and
M.W. Stolper (eds), Les archives des fortifications de Persépolis dans le monde
acheménide. Paris: De Boccard, 517–24.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 177

Churcher, C.S. and Mills, A.J. 1995. Reports from the Survey of the Dakhleh Oasis:
1977–1987. Oxford: Oxbow Books.
Clayton, J., de Trafford, A. and Borda, M. 2008. A hieroglyphic inscription found at
Jebel Uweinat mentioning Yam and Tekhebet. Sahara 19: 129–34.
Congedo, F. and De Santis, V. 2006. Topographical Survey. New York: Columbia
University.
Cruz-Uribe, E. 1988. Hibis Temple Project, Vol. 1: Translations, Commentary,
Discussions and Sign List. San Antonio: Van Siclen Books.
Cruz-Uribe, E. 1999. Kharga oasis, Late Period and Graeco-Roman sites. In Bard
1999, 406–08.
Darnell, D. 2000. Oasis ware flasks and kegs from the Theban desert. Cahier de la
céramique égyptienne 6: 227–33.
Darnell, J.C. 2002a. The narrow doors of the desert: discoveries of the Theban
desert road survey. In B. David and M. Wilson (eds), Inscribed Landscapes.
Honolulu: University of Hawaii Press, 104–21.
Darnell, J.C. 2002b. Opening the narrow doors of the desert: discoveries of the
Theban desert road survey. In Friedman 2002a, 132–55.
Darnell, J.C. 2007a. The antiquity of the Ghueita temple. Göttinger Miszellen 212:
29–40.
Darnell, J.C. 2007b. The deserts. In T. Wilkinson (ed.), The Egyptian World.
London and New York: Routledge, 29–48.
Darnell, J.C. 2013a. A bureaucratic challenge? Archaeology and administration in
a desert environment (second millennium B.C.E.). In J.C.M. Garcia (ed.),
Ancient Egyptian Administration. Leiden: Brill, 785–830.
Darnell, J.C. 2013b. The Girga road: Abu Ziyâr, Tundaba, and the integration of
the southern oases into the Pharaonic state. In Förster and Riemer 2013, 221–63.
Davies, N.D.G. 1953. The Temple of Hibis in El Khargeh Oasis III: The Decorations.
New York: Metropolitan Museum of Art.
de Angeli, S. 2013. Qanat landscapes in the oases of the Western Desert of Egypt:
the cases of the Bahariya and Farafra oases. In Dospěl and Suková 2013, 271–85.
Delia, D. 1991. Alexandrian Citizenship during the Roman Principate. Atlanta:
Scholars Press.
Di Cerbo, C., and Jasnow, R. 1996. Five Persian period demotic and hieroglyphic
graffiti from the site of Apa Tyrannos at Armant. Enchoria 23: 32–38.
Dospěl, M. and Suková, L. 2013. Explorations of the El-Hayz oasis: issues,
approaches, challenges. In Dospěl and Suková 2013, 3–18.
Dunand, F. Heim, J.-L. and Henein, N. 1992. Douch I: La nécropole. Exploration
archéologique, Monographie des tombes 1 à 72: structures sociales, économiques,
religieuses de l’Égypte Romaine. Cairo: Institut Français d’Archéologie Orientale
du Caire.
Dunand, F., Heim, J.-L. Henein, N. and Lichtenberg, R. 2005. Douch V. La
nécropole de Douch. Exploration archéologique vol. 2. Monographie des tombes
73 à 92. Cairo: Institut Français d’Archéologie Orientale du Caire.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 178 Anna Lucille Boozer

Duval, N. 1982a. L’urbanisme de Sufetula (Sbeitla). Aufstieg und Niedergang der

römischen Welt 2 (10.2): 596–632.
Duval, N. 1982b. Topographie et urbanisme d’Ammaedara. Aufstieg und
Niedergang der römischen Welt 2 (10.2): 633–71.
Edmondson, J. 2006. Cities and urban life in the western provinces of the Roman
Empire, 30 BCE–250 CE. In D. Potter (ed.), Companion to the Roman Empire.
Oxford: Blackwell, 250–80.
Empereur, J.-Y. 1998. Alexandria Rediscovered. New York: G. Braziller.
Erdblom, L. 1968. Structure foncière: organisation et structure sociale. Lund:
Uniskol.
Fakhry, A. 1974. Bahriyah and Farafra (2003 reprint). Cairo and New York:
American University in Cairo Press.
Fakhry, A. 1990. Siwa Oasis. Cairo: American University in Cairo Press.
Fenwick, C. Forthcoming. Archaeology, empire and the Arab conquest of North
Africa. Past and Present.
Fentress, E. (ed.). 2000. Romanization and the City: Creation, Transformations, and
Failures. Portsmouth: Journal of Roman Archaeology Supplement.
Fentress, E. 2011. Slavers on chariots. In A. Dowler and E.R. Galvin (eds), Money,
Trade and Trade Routes in Pre-Islamic North Africa. London: British Museum
Press, 65–71.
Février, P.A. 1964. Notes sur le développement urbain en Afrique du Nord: les
examples comparés de Djémila et de Sétif. Cahiers Archeologiques 14: 1–26.
Förster, F. and Riemer, H. (eds). 2013. Desert Road Archaeology in Ancient Egypt
and Beyond. Köln: Heinrich-Barth-Institut.
Friedman, R. 2002a. Egypt and Nubia: Gifts of the Desert. London: British Museum
Press.
Friedman, R. 2002b. Foreword. In Friedman 2002a, xiii–xv.
Gatto, M.C. 2002. Two Predynastic pottery caches at Bir Sahara (Egyptian Western
Desert). Sahara 13: 51–60.
Giddy, L.L. 1987. Egyptian Oases: Baḥ ariya, Dakhla, Farafra, and Kharga during
Pharaonic Times. Warminster: Aris & Phillips Ltd.
Giddy, L.L., and Grimal, N. 1979. Balat: sondage sur le site de Ain Aseel, rapport
préliminaire. Bulletin de l’Institut Français d’Archeologie Orientale 79: 21–30.
Gill, J.C.R. 2012a. Ptolemaic activity in Dakhleh oasis. In The 7th International
Conference of the Dakhleh Oasis Project: New Developments in the Archaeology of
the Egyptian Western Desert and Its Oases. Leiden: Leiden University.
Gill, J.C.R. 2012b. Ptolemaic Period pottery from Mut al-Kharab, Dakhleh oasis. In
Bagnall et al. 2012, 231–41.
Gill, J.C.R. 2015. Dakhleh Oasis and the Western Desert in Egypt under the
Ptolemies. Oxford: Oxbow Books.
Goodman, P. 2007. The Roman City and Its Periphery from Rome to Gaul.
New York: Routledge.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 179

Gosline, S.L. 1990. Bahariya Oasis Expedition Season Report for 1988: Part 1, Survey
of Qarat Hilwah. San Antonio, Texas: Van Siclen Books.
Grimal, N. 1997. Travaux de l’institut français d’archéologie orientale en
1996–1997. Bulletin de l’Institut Français d’Archeologie Orientale 97: 313–429.
Haas, C. 1993. Alexandria’s Via Canopica: political expression and urban topo-
graphy from Augustus to cAmr Ibn al-cAs. In N. Swelim (ed.), Alexandrian
Studies in Memoriam Daoud Abdu Daoud. Alexandria: Société Archéologique
d’Alexandrie, 123–38.
Haas, C. 1997. Alexandria in Late Antiquity: Topography and Social Conflict.
Baltimore: The Johns Hopkins University Press.
Hamilton, J. 1856. Wanderings in North Africa (1985 reprint). London: John
Murray.
Harrison, R. 2007. Materiality, ambiguity and the unfamiliar in the archaeology of
inter-societal confrontations: a case study from northwest Australia. In
P. Cornell and F. Fahlander (eds), Encounters, Materialities, Confrontations:
Archaeologies of Social Space and Interaction. Newcastle: Cambridge Scholars
Publishing.
Hassan, F. 1976. Prehistoric studies of the Siwa oasis region, northwestern Egypt
(1975 field season). Nyame Akuma 9: 18–34.
Hassan, F. 1978. Archaeological exploration of the Siwa region. Current
Anthropology 19: 146–48.
Hawass, Z. 2000. Valley of the Golden Mummies. New York: Harry N. Abrams.
Helk, W. 1975. Wirtschaftsgeschichte des Alten Ägypten im 3. und 2. Jahrtausend
vor Chr. Leiden/Cologne: Brill.
Hope, C.A. 1999. Dakhla oasis, Ismant el-Kharab. In Bard 1999, 222–26.
Hope, C.A. 2002. Oases amphorae of the New Kingdom. In Friedman 2002a,
95–131.
Hope, C.A. 2004. The excavations at Ismant el-Kharab and Mut el-Kharab in 2004.
Bulletin of the Australian Centre for Egyptology 15: 19–49.
Hope, C.A. 2005. Report on the excavations at Ismant el-Kharab and Mut
el-Kharab in 2005. Bulletin of the Australian Centre for Egyptology 16: 35–83.
Hope, C.A. and Pettman, A.J. 2013. Egyptian connections with Dakhleh oasis in
the Early Dynastic period to Dynasty IV: new data from Mut el-Kharab. In
Bagnall et al. 2012, 147–66.
Hope, C.A. and Whitehouse, H. 2004. The gladiator jug from Ismant el-Kharab. In
Bowen and Hope 2004, 291–310.
Hope, C.A. and Whitehouse, H. 2006. A painted residence at Ismant el-Kharab
(Kellis) in the Dakhleh oasis. Journal of Roman Archaeology 19: 312–28, colour
plates 1–10.
Hope, C.A., Kaper, O.E., Bowen, G.E. and Patten, S.F. 1989. Dakhleh Oasis Project:
Ismant el-Kharab 1991–92. Journal of the Society for the Study of Egyptian
Antiquities 19: 1–26.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 180 Anna Lucille Boozer

Hope, C.A., Kaper, O.E. and Bowen, G.E. 1992. Excavations at Ismant el-Kharab –
1992. The Bulletin of the Australian Centre for Egyptology 3: 41–49.
Hope, C.A., Bowen, G.E., Cox, J., Dolling, W., Milner, J. and Pettmen, A. 2009.
Report on the 2009 season of excavations at Mut el Kharab, Dakheh oasis.
Bulletin of the Australian Centre for Egyptology 20: 47–86.
Hubschmann, C. 2012. Searching for an oasis identity: Dakhleh oasis in the Third
Intermediate period. In C.M. Knoblauch and J.C Gill (eds), Egyptology in
Australia and New Zealand 2009. Oxford: Archaeopress, 61–70.
Ikram, S. and Rossi, C. 2004. North Kharga Oasis Survey 2001–2002 preliminary
report: Ain Gib and Qasr el-Sumayra. Mitteilungen des Deutschen
Archäologischen Institutes, Abteilung Kairo 60: 69–92, Tafel 8–9.
Jackson, R.B. 2002. At Empire’s Edge: Exploring Rome’s Egyptian Frontier. New
Haven: Yale University Press.
Jaritz, H. and Rodziewicz, M. 1994. Syene – review of the urban remains and its
pottery. Mitteilungen des Deutschen Archäologischen Institutes Abteilung Kairo
50: 115–41.
Jaritz, H. and Rodziewicz, M. 1996. Syene – investigation of the urban remains in
the vicinity of the temple of Isis (II). Mitteilungen des Deutschen Archäologischen
Institutes Abteilung Kairo 52: 233–49.
Jones, A.H.M. 1964. The Later Roman Empire, 284–602: A Social, Economic and
Administrative survey III (1986 reprint). Baltimore: Johns Hopkins.
Joshi, M.C. 2000. Application of geometry in the planning of an ancient settlement:
Sisupalgarh, a case study. In J.M. Malville and L.M. Gujral (eds), Ancient Cities,
Sacred Skies: Cosmic Geometries and City Planning in Ancient India. New Delhi:
Aryan Books International, 34–39.
Kamal, Y. 1935. Monumenta Cartographica Africae et Aegypti [6 vols] (1987
reprint). Cairo: Le Caire.
Kaper, O.E. 1997. Temples and Gods in Roman Dakhleh: Studies in the Indigenous
Cults of an Egyptian Oasis. PhD, Rijksuniversiteit te Groningen.
Kaper, O.E. (ed.). 1998a. Life on the Fringe: Living in the Southern Egyptian Deserts
during the Roman and Early-Byzantine Periods. Leiden: Research School CNWS.
Kaper, O.E. 1998b. Temple building in the Egyptian deserts during the Roman
period. In Kaper 1998a, 139–58.
Kaper, O.E. 2003. The Egyptian God Tutu: A Study of the Sphinx-God and Master of
Demons with a Corpus of Monuments. Leuven: Peeters.
Kaper, O.E. 2009. Epigraphic evidence from the Dakhleh oasis in the Libyan
period. In G.P.F. Broekman, R.J. Demarée and O.E. Kaper (eds), The Libyan
Period in Egypt, Historical and Cultural Studies into the 21st–24th Dynasties:
Proceedings of a Conference at Leiden University, 25–27 October 2007. Leiden:
Netherlands Institute for the Near East, 149–59.
Kaper, O.E. 2010. Galba’s cartouches at Ain Birbiyeh. In K. Lembke, M. Minas-
Nerpel and S. Pfeiffer (eds), Transition and Transformation: Egypt under Roman
Rule. Leiden: Brill, 181–201.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 181

Kaper, O.E. 2012. The western oases. In C. Riggs (ed.), The Oxford Handbook of
Roman Egypt. Oxford: Oxford University Press, 717–35.
Kaper, O.E. and Davoli, P. 2006. A new temple for Thoth in the Dakhleh oasis.
Egyptian Archaeology 28: 12–14.
Kaper, O.E. and Demarée, R.J. 2006. A donation stela in the name of Takeloth III
from Amheida, Dakhleh oasis. Jaarbericht Ex Oriente Lux 39: 19–37.
Keith, K. 2003. The spatial patterns of everyday life in Old Babylonian neighbor-
hoods. In M.L. Smith (ed.), The Social Construction of Ancient Cities.
Washington and London: Smithsonian Books, 56–80.
Kirkwan, L.P. 1971. Roman expeditions to the Upper Nile and the Chad-Darfur region.
In F. Gadallah (ed.), Libya in History. Proceedings of a Conference Held at the Faculty
of Arts, University of Libya 1968. Benghazi: University of Libya, 253–61.
Kröpelin, S. and Kuper, R. 2007. More corridors to Africa. Cahiers de recherches de
l’Institut de Papyrologie et d’Égyptologie de Lille 26: 219–29.
Kucera, P. 2013. Al-Qasr: the Roman castrum of Dakhleh oasis. In Bagnall et al.
2012, 205–16.
Kuhlmann, K.P. 1988. Das Ammoneion: Archäologie, Geschichte und Kultpraxis des
Orakels von Siwa. Mainz: Von Zabern.
Kuhlmann, K.P. 1998. Roman and Byzantine Siwa: developing a latent picture. In
Kaper 1998a, 159–80.
Kuhlmann, K.P. 2002. The ‘oasis bypath’ or the issue of desert trade in Pharaonic
times. In Jennerstrasse 8 (eds), Tides of the Desert: Contributions to the
Archaeology and Environmental History of Africa in Honour of Rudolph Kuper.
Köln: Heinrich-Barth-Institut, Africa praehistorica 14, 125–70.
Kuhlmann, K.P. 2005. Der Wasserberg des Djedefre (Chufu 01/1). Ein Lagerplatz mit
Expeditionsinschriften der 4. Dynastie im Raum der Oase Dachla. Mitteilungen des
Deutschen Archäologischen Institutes, Abteilung Kairo 61: 243–89.
Kuhlmann, K.P. 2013. The realm of two deserts: Siwa oasis between East and West.
In Förster and Riemer 2013, 133–66.
Kuper, R. 2002. Routes and roots in Egypt’s Western Desert: the Early Holocene
resettlement of the Eastern Sahara. In Friedman 2002a, 1–12.
Kuper, R. and Kröpelin, S. 2006. Climate-controlled Holocene occupation in the
Sahara: motor of Africa’s evolution. Science 313: 803–87.
Lane, P., and MacDonald, K. (eds). 2011. Slavery in Africa: Archaeology and
Memory. Oxford: Proceedings of the British Academy.
Lassus, J. 1969. Visite à Timgad. Algiers: Ministè re de l’Education Nationale.
Laurence, R., Esmonde-Cleary, S. and Sears, G. 2011. The City in the Roman West,
c.250 BC–c.AD 250. Cambridge: Cambridge University Press.
Leone, A. 2007. Changing Townscapes in North Africa from Late Antiquity to the
Arab Conquest. Bari: Epipuglia.
Lepelley, C. 1992. The survival and fall of the Classical city in Late Roman Africa. In
J. Rich (ed.), The City in Late Antiquity. London and New York: Routledge,
50–76.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 182 Anna Lucille Boozer

Liebeschuetz, J.H.W.G. 2000. The Decline and Fall of the Roman City. Oxford and
New York: Oxford University Press.
Liverani, M. (ed.). 2006. Aghram Nadharif. A Garamantian Citadel in the Wadi
Tannezzuft. Firenze: All’Insegna del Giglio.
Lomas, K. 1989. Roman imperialism and the city in Italy. In R. Laurence and
J. Berry (eds), Cultural Identity in the Roman Empire. London and New York:
Routledge, 64–78.
MacCormack, S. 1981. Art and Ceremony in Late Antiquity. Berkeley: University of
California Press.
Marlow, C.A. and Mills, A.J. (eds). 2001. The Oasis Papers. Proceedings of the First
International Symposium of the Dakhleh Oasis Project. Oxford: Oxbow Books.
Masonen, P. 2012. The Sahara as highway for trade and knowledge. In S.E. Alcock,
J. Bodel and R.J.A. Talbert (eds). Highways, Byways, and Road Systems in the
Pre-Modern World. Malden/Oxford: John Wiley & Sons, 168–75.
Mattingly, D.J. 1994. Tripolitania. Ann Arbor: University of Michigan Press.
Mattingly, D.J. 2013. The Archaeology of Fazzan. Volume 4, Survey and Excavations
at Old Jarma (Ancient Garama) Carried out by C.M. Daniels (1962–69) and the
Fazzan Project (1997–2001). London: Society for Libyan Studies, Department of
Antiquities.
Mattingly, D.J. and Hitchner, R.B. 1995. Roman Africa: an archaeological review.
The Journal of Roman Studies 85: 165–213.
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017. Trade in the Ancient Sahara and Beyond. Trans-Saharan
Archaeology Volume 1. Series editor D.J. Mattingly. Cambridge: Cambridge
University Press.
Mills, A.J. 1991. [Review of] Lisa L. Giddy, ‘Egyptian Oases: Bahariya, Dakhla,
Farafra and Kharga during Pharaonic Times’. Bibliotheca Orientalia 48: 132–35.
Mills, A.J. 1993. The Dakhleh oasis columbarium farmhouse. Bulletin de la societé
archéologique d’Alexandrie 45: 192–98.
Mills, A.J. 1999a. Dakhleh oasis: dynastic and Roman sites. In Bard 1999, 220–22.
Mills, A.J. 1999b. Pharaonic Egyptians in the Dakhleh oasis. In C.S. Churcher and
A.J. Mills (eds), Reports from the Survey of the Dakhleh Oasis, Western Desert of
Egypt: 1977–1987, (Dakhleh Oasis Project: Monograph 2). Oxford: Oxbow
Monograph 99, 171–78.
Mills, A.J. 2002. Deir el-Hagar, Ain Birbiyeh, Ain el-Gazzareen and el-
Muzawwaqa. In C.A. Hope and G.E. Bowen (eds), Dakhleh Oasis Project:
Preliminary Reports on the 1994–95 to 1998–99 Field Season. Oxford: Oxbow
Books, 25–30.
Mills, A.J. 2013. An Old Kingdom trading post at ‘Ain el-Gazzareen, Dakhleh oasis.
In Bagnall et al. 2012, 177–80.
Mills, A.J. and Kaper, O.E. 2004. Ain el-Gazzareen: developments in the Old
Kingdom settlement. In Bowen and Hope 2004, 123–29.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 183

Mitchell, F.C. 1992. Where was Putu-Iaman? Proceedings of the Seminar for
Arabian Studies 22: 69–80.
Mitchell, S. 2007. A History of the Later Roman Empire, AD 284–641: The
Transformation of the Ancient World. Malden, MA; Blackwell.
Morkot, R.G. 1996. The Darb el-Arbain, the Kharga oasis and its forts, and other
desert routes. In D.M. Bailey (ed.), Archaeological Research in Roman Egypt.
Ann Arbor, MI: Journal of Roman Archaeology Supplement, 82–94.
Mullins, P.R. 2011. The archaeology of consumption. Annual Review of
Anthropology 40: 133–44.
Nachtigal, G. 1974. Sahara and Sudan (translated by A.G.B. Fisher and H.J. Fisher).
London: Hurst & Co.
Newton, C., Whitbread, T., Agut-Labordière, D. and Wuttmann, M. 2013.
L’agriculture oasienne à l’époque perse dans le sud de l’oasis de Kharga
(Égypte, Ve-IVe s. AJC). Revue d’Ethnoécologie 4: 2–18.
O’Connor, D. and Reid, A. (eds). 2003. Ancient Egypt in Africa. London: UCL
Press.
Osing, J.R. (ed.). 1982. Denkmäler der Oase Dachla: aus dem Nachlass von Ahmed
Fakhry. Mainz am Rhein: Philipp von Zabern.
Pantalacci, L. 1985. Un décret de Pépi II en faveur des gouverneurs de l’oasis de
Dakhla. Bulletin de l’Institut Français d’Archeologie Orientale 85: 245–54.
Pantalacci, L. 2013. Broadening horizons: distant places and travels in Dakhla and
the Western Desert at the end of the 3rd millennium. In Förster and Riemer
2013, 283–96.
Parsons, P. 2007. City of the Sharp-Nosed Fish: Greek Papyri Beneath the Egyptian
Sand Reveal a Long-Lost World. London: Phoenix.
Reddé, M. 1990. Quinze années de recherches françaises à Douch. Vers un premier
bilan. Bulletin de l’Institut Français d’Archeologie Orientale 90: 281–301.
Reddé, M. 1999. Sites militaires romains de l’oasis de Kharga. Bulletin de l’Institut
Francais d’Archeologie Orientale 99: 377–96.
Reddé, M., Ballet, P., Barbet, A. and Bonnet, C. 2004. Kysis: fouilles de l’Ifao à
Douch, oasis de Kharga, 1985–1990. Cairo: Institut Français d’Archéologie
Orientale.
Riemer, H. 2004. News about the Clayton rings: long distance desert travellers
during Egypt’s Predynastic. In S. Hendrickx, T. Friedman, K. Cialowicz and
M. Chlodnicki (eds), Egypt at Its Origins. Studies in Memory of Barbara Adams.
Orientalia Lovaniensia Analecta 138. Leuven/Paris: Peeters, 971–89.
Riemer, H. 2009. A potsherd from northwest of Abu Minqar and the dispersal of
Sheikh Muftah pottery in the Western Desert of Egypt. Sahara 20: 57–62.
Riemer, H. and Förster, F. 2013. Ancient desert roads: towards establishing a new
field of archaeological research. In Förster and Riemer 2013, 19–58.
Riemer, H. and Kuper, R. 2000. ‘Clayton rings’: enigmatic ancient pottery in the
Eastern Sahara. Sahara 12: 91–100.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 184 Anna Lucille Boozer

Rodziewicz, M. 1984. Les habitations romaines tardives d’Alexandrie et la lumiere

des fouilles polonaises à Kom el-Dikka. Warsaw: ánstwowe Wydawnictwo
Naukowe.
Rodziewicz, M. 1991. A brief record of the excavations at Kom el-Dikka in Alexandria
(1960–1980). Bulletin de la societé archéologique d’Alexandrie 44: 1–70.
Rohlfs, G. 1875. Drei Monate in der Libyschen Wüste. Von Gerhard Rohlfs. Mit
Beiträgen von P. Ascherson, W. Jordan und K. Zittel, sowie einer Originalkarte
von W. Jordan, 16 Photographien nach Ph. Remelâe, 11 Steindruck-Tafeln und 18
Holzschnitten. Cassel: T. Fischer.
Ross, E. 2011. A historical geography of the trans-Saharan trade. In G. Krätli and
G. Lydon (eds), The Trans-Saharan Book Trade. Manuscript Culture, Arabic
Literacy and Intellectual History in Muslim America. Leiden/Boston: Brill, 1–34;
map on p. xv.
Rossi, C. 2000. Umm el-Dabadib, Roman settlement in the Kharga oasis: descrip-
tion of the visible remains with a note on ‘Ain Amur. Mitteilungen des Deutschen
Archäologischen Instituts, Abteilung Kairo 56: 335–52, Tafel 35–39.
Rossi, C. and Ikram, S. 2013. Evidence of desert routes across northern Kharga
(Egypt’s Western Desert). In Riemer and Förster 2013, 265–82.
Savage, E. (ed.). 1992. The Human Commodity: Perspectives on the Trans-Saharan
Slave Trade. London: Frank Cass.
Schacht, I. 2003. A preliminary survey of the ancient qanat system of the Northern
Kharga oasis. Mitteilungen des Deutschen Archäologischen Institutes, Abteilung
Kairo 59: 279–306.
Schild, R. and Wendorf, F. 1977. The Prehistory of the Dakhleh Oasis and Adjacent
Desert. Warsaw: Zaklad narodowy imienia ossolinskich wydawnictwo polskiej
akademii nauk.
Sidebotham, S.E., and Wendrich, W.Z. 1998. Berenike: archaeological fieldwork at
a Ptolemaic-Roman port on the Red Sea coast of Egypt: 1994–1998. Sahara 10:
85–96.
Smekalova, T.N., and Smekalov, S. 2005. Magnetic Survey. New York: Columbia
University.
Smekalova, T.N., and Smekalov, S. 2006. Geophysical Survey. New York: Columbia
University.
Soukiassian, G., Wuttman, M. and Schaad, D. 1990. La ville d’‘Ayn-Asil à Dakhla. État
des recherches. Bulletin de l’Institut Français d’Archeologie Orientale 90: 347–58.
Stanley, C.V.B. 1912. The oasis of Siwah. Journal of African Society 11: 290–324.
Swanson, J.T. 1975. The myth of trans-Saharan trade during the Roman era.
International Journal of African Historical Studies 8: 582–600.
Tacoma, L.E. 2006. Fragile Hierarchies. The Urban Elites of Third-Century Roman
Egypt. Leiden: Brill.
Thissen, H.-J. 2013. Donkeys and water: demotic ostraca in Cologne as evidence
for desert travel between Oxyrhynchos and the Bahariya oasis in the 2nd
Century BC. In Förster and Riemer 2013, 391–97.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 4 The Urbanisation of Egypt’s Western Desert under Roman Rule 185

Török, L. 2009. Between Two Worlds: the Frontier Region between Ancient Nubia
and Egypt, 3700 BC–500 AD. Leiden: Brill.
Valloggia, M. 1985. Les amiraux de l’oasis de Dakhleh. In Mélanges offerts à Jean
Vercoutter. Paris: Éditions recherche sur les civilisations, 355–64.
Verner, M. and Benesovska, H. 2008. Unearthing Ancient Egypt: Fifty Years of the
Czech Archaeological Exploration in Egypt. Prague: Charles University.
Vivian, C. 2002. The Western Desert of Egypt: An Explorer’s Handbook. Cairo:
American University in Cairo Press.
Wagner, G. 1987. Les Oasis d’Égypte à l’époque grecque, romaine et byzantine d´après les
documents grecs. Cairo: Institut Français d’Archéologie Orientale du Caire.
Wagner, G. 2000. Oasis Magna. In R.J.A. Talbert (ed.), Barrington Atlas of the
Greek and Roman World. Princeton: Princeton University Press, map 79.
Welsby, D. 1996. The Kingdom of Kush: The Napatan and Meroitic Empires.
London: British Museum Press.
Wendrich, W.Z. 1998. Fringes are anchored in warp and weft: the relations
between Berenike, Shenshef and the Nile Valley. In Kaper 1998a, 243–51.
Whitcomb, D.S. and Johnson, J.H. 1979. Quseir Al-Qadim 1978: Preliminary
Report. Cairo: American Research Center in Egypt, Inc.
Wilson, A.I. 2002. Urban production in the Roman world: the view from North
Africa. Papers of the British School at Rome 70: 231–73.
Wilson, A.I. 2006. The spread of foggara-based irrigation in the ancient Sahara. In
D.J. Mattingly, S. McLaren, E. Savage, Y. al-Fasatwi, and K. Gadgood (eds), The
Libyan Desert. Natural Resources and Cultural Heritage. London: Society for
Libyan Studies, 205–16.
Wilson, A.I. 2009. Foggaras in ancient North Africa or how to marry a Berber
princess. In Contrôle et distribution de l’eau dans le Maghreb antique et médiéval.
Rome: École française de Rome, 19–39.
Wilson, A.I. 2012. Saharan trade in the Roman period: short-, medium- and
long-distance trade networks. Azania: Archaeological Research in Africa 47.4:
409–49.
Wilson, A.I. and Mattingly, D.J. 2003. Irrigation technologies: foggaras, wells and
field systems. In D.J. Mattingly (ed.), The Archaeology of Fazzan. Volume 1,
Synthesis. London: Society for Libyan Studies, Department of Antiquities, 234–78.
Winlock, H.E. 1941. The Temple of Hibis in El Khargeh Oasis. III. New York:
Metropolitan Museum of Art.
Wright, J. 1988. Murzuk and the Saharan slave trade in the 19th century. Libyan
Studies 29: 89–96.
Wuttman, M. 1996. Premier rapport préliminaire des travaux sur le site de ‘Ayn-
Manâwîr (oasis de Kharga). Bulletin de l’Institut Français d’Archeologie
Orientale 96: 385–451.
Wuttman, M. 2001. Les qanats de ‘Ayn manāwīr. In P. Briant (ed.), Irrigation et
drainage dans l’Antiquité. Qanâts et canalisations souterraines en Iran, en Égypte
et en Grèce. Paris: Thotm Éditions (Persika 2), 109–36.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 186 Anna Lucille Boozer

Wuttman, M., Barakat, H., Bousquet, B., Chauveau, M., Gonon, T., Marchand, S.,
Robin M. and Schweitzer, A. 1998. ‘Ayn Manawir (oasis de Kharga). Deuxième
rapport préliminaire. Bulletin de l’Institut Français d’Archeologie Orientale 98:
367–462.
Wuttman, M., Gonon, T. and Thiers, C. 2000. The qanats of ‘Ayn-Manâwîr
(Kharga oasis, Egypt). Journal of Achaemenid Studies and Researches 1: 1–11.
Youssef, S. 2012. Qanats in the Dakhleh oasis. In 7th International Conference of
the Dakhleh Oasis Project: New Developments in the Archaeology of the Egyptian
Western Desert and Its Oases, 20–24 June 2012. Netherlands: Leiden University.

https://doi.org/10.1017/9781108637978.005 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern
Sahara
david j. mattingly, martin sterry,
muftah al-haddad and pol trousset

Introduction

This chapter focuses on the oases of the northern Sahara, both those close
to and in some cases incorporated within the frontiers of the Roman
provinces of Africa. With the exception of some outstanding contributions
from our co-author Pol Trousset, there has been little consideration of the
potential scale or on the ground reality of oasis development in the Roman
frontier region.1 This is only partly explicable in terms of the lack of
detailed archaeological work at these sites – as we shall demonstrate
there is quite a lot of fragmentary evidence to support the case for wide-
spread oasis development in pre-Islamic times. In large measure the lack of
recognition of the importance of oases here relates to the long-prevailing
myth that Rome was confronted in this frontier zone by nomadic (or at
best transhumant) peoples.2 It is hoped that what follows will provoke a full
re-evaluation of Rome’s African frontiers and what they were designed to
deal with. There are clearly major implications if, in place of a few scattered
bands of pastoralists, Rome was confronted by both pastoralists and settled
and populous oasis communities. The descriptive survey will pick up from
Chapter 3 by starting at Bu Nijim, the main oasis to the north of al-Jufra on
the main route to the Mediterranean, where a Roman fort was established
in AD 201. It will then proceed more or less westwards along the frontier
zone, finishing in the Ouled Nail mountains and the oases directly south of
there in the Mzab and Wargla groups.3 The very far western Algerian oases
1
Trousset 1986a; 1987a; 1987b; 1995; 2012a. The chapter is mainly written by the first two named
authors, but we worked with the third author on the section on Ghadames and have made use of
unpublished notes provided by the fourth author.
2
Trousset 1982b; 2012b. See also, Leveau 2018, for a sweeping historiographical overview.
3
For the geography of the Tunisian and Algerian oasis zones, see Despois and Raynal 1967,
421–49. 187

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 188 David J. Mattingly et al.

will be considered in the next chapter in its overview of the western

Sahara.4 Available radiocarbon dates are presented in Table 5.1.

The Tripolitanian Pre-Desert and the Hamada al-Hamra

The region (and later Roman province) of Tripolitana comprised western

Libya and southern Tunisia, covering a linear distance of over 1,000 km
from the Syrtic desert to the Chott Jarid (Fig. 5.1). For much of this
distance, the Roman provincial territory ran through, or was closely bor-
dered by, pre-desert or desert landscapes. The Hamada al-Hamra is
a massive and largely waterless rocky plateau area between the
Tripolitanian pre-desert and Fazzan.5 The two major Trans-Saharan routes
in Libya have traditionally skirted this hostile environment, one running
some way to the east of the Hamada through al-Jufra and on to Fazzan, the
other skirting its west flank through Ghadamis and Ghat.
However, despite the difficulties of travel at the hotter periods of
the year, the Hamada is not impassable. The campaign of Cornelius
Balbus c.20 BC appears to have traversed part of it in passing from
Ghadamis to Garamantian lands. In AD 70 the Roman army reduced the
time required to reach the Garamantian heartlands by using the ‘head of
the rock road’ (iter praeter caput saxi), almost certainly to be identified
with the direct route from Tripoli via Gharyan, Mizda, al-Qurayyat al-
Gharbia, Shwerif, to Brak in the Wadi ash-Shati.6 From Mizda to Brak the
route runs partly over and partly along the eastern edge of the Hamada.
The first site we shall examine is Bu Nijim on the eastern route, which
continued south to al-Jufra.7 The second part of this section will look at the
sites associated with the ‘central’ route via the Hamada al-Hamra, specifi-
cally Mizda and al-Qurayyat al-Gharbia. The third section will look at sites
on the western route north from Ghadamis to the Nefzaoua oases.

The Eastern Route

Bu Nijim
The Roman fort and garrison settlement at Bu Nijim is one of the best
explored locations of the Roman frontier in Africa, lying between the

4
See Sterry et al., Chapter 6, this volume. 5 Mattingly 1995, 6–7.
6
Pliny, Natural History, 5.38; Desanges 1980, 411–14, for discussion.
7
See Mattingly et al., Chapter 3, this volume.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 Table 5.1 Protohistoric and historic radiocarbon dates from the Northern Sahara

Area/Site Context Material Reference Date Calibrated

Tripolitanian Pre-Desert
Wadi Ganima Alluvial terrace Charcoal Q-656 614±100 BP calAD 1207–1470
Adzam (village) Midden Charcoal GrN-13539 435±20 BP calAD 1430–1472
Adzam (village) Midden Charcoal GrN-13538 405±25 BP calAD 1437–1618
Adzam (village) Midden Charcoal GrN-13540 390±20 BP calAD 1445–1618
Ben Telis (village) Midden Charcoal GrN-13526 555±30 BP calAD 1310–1431
Ben Telis (village) Midden Charcoal GrN-13525 390±30 BP calAD 1441–1631
Gh75 (qasr) Midden Charcoal GrN-13536 1770±25 BP calAD 143–342
Gh75 (qasr) Midden Charcoal GrN-13537 1765±25 BP calAD 145–378
Kh41 (qasr) Midden Charcoal GrN-13524 1510±70 BP calAD 412–652
Kh41 (qasr) Midden Charcoal GrN-13523 1380±90 BP calAD 430–875
Lm3 (qasr) Timber Wood HAR-5838 1840±70 BP calAD 23–376
Lm3 (qasr) Same sample as above Wood HAR-5689 1710±70 BP calAD 132–533

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

Lm3 (qasr) Same sample as above Wood HAR-5840 1720±80 BP calAD 126–535
Lm4 (farm) Occupation layer Charcoal GrN-13531 2075±35 BP 189 calBC–calAD 2
Lm4 (farm) Construction material Charcoal GrN-13533 1975±25 BP 41 calBC–calAD 74
Lm4 (farm) Abandonment layer Charcoal GrN-13535 1910±60 BP 41 calBC–calAD 237
Lm4 (farm) Occupation layer Charcoal GrN-13532 1835±30 BP calAD 86–246
Lm4 (farm) Destruction layer Charcoal GrN-13530 1770±30 BP calAD 138–345
Lm4 (farm) Destruction layer Charcoal GrN-13529 1710±25 BP calAD 253–395
Lm4 (farm) Abandonment layer Charcoal GrN-13534 615±45 BP calAD 1287–1410
Nf82 (farm) Midden Charcoal GrN-13527 1790±50 BP calAD 92–381
Nf82 (farm) Midden Charcoal GrN-13528 1750±50 BP calAD 138–394
Al-Qurayyat al-Gharbia Mortar floor Charcoal Erl-16639 1595±38 BP calAD 390–551
Al-Qurayyat al-Gharbia Buttress wall (mortar) Charcoal Erl-16637 1551±43 BP calAD 415–598
Al-Qurayyat al-Gharbia Column base (mortar) Charcoal Erl-16638 1538±42 BP calAD 420–605
 Table 5.1 (cont.)

Area/Site Context Material Reference Date Calibrated

Ghadamis
Ghadamis Not given Charcoal Gif-4581 2600±50 BP
Qasr Ghlul Occupation layer Charcoal OxA-34787 1501±23 BP calAD 435–625
Qasr Ghlul Occupation layer Plant remains OxA-34790 992±22 BP calAD 991–1150
Qasr Ghlul Occupation layer Plant remains OxA-34985 946±30 BP calAD 1025–1156
Qasr Ghlul Occupation layer Date stone OxA-34788 914±22 BP calAD 1034–1168
Qasr Ghlul Same sample as above Date stone OxA-34789 909±23 BP calAD 1037–1186
Southern Tunisia
El Menaguib Tumulus 1 Charcoal Pa-2352 4520±150 BP
El Menaguib Bazina 24 Human bone Pa-2368 1845±30 BP calAD 85–239
El Menaguib Tumulus 4 Human bone Pa-2352 1970±45 BP 88 calBC–calAD 130
Djebel Aziza Berber retreat Charcoal Hv-6979 415±130 BP calAD 1286– . . .
Djebel Aziza Berber retreat Charcoal Hv-6977 250±50 BP calAD 1483– . . .
Oued el Akarit Lithic scatter Shells Orsay 3100±190 BP 1869–850 calBC
Oued el Akarit Lithic scatter Shells GaK-3454 1470±190 BP calAD 135–963

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

Kebili Km 97 Hv-6976 295±40 BP calAD 1477–1791
Smidet Zine Hv-6974 130±120 BP calAD 1517– . . .
Les Perles Ostrich egg shell Mc-723 2400±100 795–232 calBC
Er Regaa Ostrich egg shell Hv-7974 2440±100 802–370 calBC
Southern Algeria
El Golea Hv-5620 3345±140 BP 2022–1302 calBC
El Golea Hv-9699 1930±260 BP 729 calBC–calAD 636
Ben Dahi Ostrich egg shell HV-7975 3435±100 2016–1505 calBC
Hassi el-Mejna Lac. shells Orsay 3380±150 2131–1305 calBC
Hassi Cheikh C Diat Orsay 3330±70 1862–1446 calBC
Hassi Cheikh C Diat Orsay 3230±200 2026–1001 calBC
Hassi Cheikh A Diat Orsay 3230±200 2026–1001 calBC
Hassi Cheikh B Diat Orsay 3220±200 2018–980 calBC
Hassi Cheikh Diat Orsay 3080±255 2016–788 calBC
 5 Pre-Islamic Oasis Settlements in the Northern Sahara 191

Figure 5.1. South Tunisia and Tripolitania, main regions and sites discussed in the text.

Greater Syrtes embayment and the al-Jufra oases along a natural seismic
corridor known as the Hun Graben. The fort is often starkly depicted in
relation to its desert environment, whether in Lyon’s famous painting of its
north gate or the aerial view of its buildings engulfed by sand dunes.8 The
fort is situated on the south-east side of a small L-shaped oasis. The
modern village lies on the west side of the oasis c.1.5 km from the fort,
a loose straggle of houses. The location is in a rather waterless area and
despite its small size the wells and oasis at Bu Nijim has been historically
important. For instance, it was the point on the nineteenth-century slave
route from Bornu where responsibility for the security of caravans passed
from Fazzan to Tripoli jurisdiction.9

8
Lyon 1821, f. p. 67; Mattingly et al. 2013, 86–89.
9
Mattingly 1995, 156. Nachtigal 1974, 47, gives a concise account of the depressed state of this
small oasis, reputedly the most northerly in the province of Fazzan in the late nineteenth
century.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 192 David J. Mattingly et al.

The excavations by Rebuffat in the 1960s–1970s provide a particularly

detailed record for the fort and garrison settlement, as well as establishing
the location of outlying temples and nearby outposts on the desert routes.10
Epigraphic evidence dates the foundation of the fort very precisely, to 24
January AD 201 and there is an abundance of documentation on the
activities of the garrison down to c.AD 263. The 1.28 ha fort was sur-
rounded by a substantial walled garrison settlement (c.14 ha) recently
truncated by an agricultural development.11 The buildings of the fort and
garrison settlement made extensive use of palm wood in the construction
of roofs and vaults, emphasising the site’s proximity to an active oasis.
Libyan inscriptions on the walls of a house in the garrison settlement could
relate to passing traders, or alternatively to the local Libyan community of
the oasis.12 Roman ostraca (texts written on pot sherds) attest to the
existence of a statio cameliorum here – presumably some sort of caravan-
serai attached to either the oasis or fort.13
There is no direct evidence of earlier oasis settlement at this location and
the nearest hillfort sites are 35–50 km to the north. However, near the oasis
there are two clusters of poorly preserved rectilinear features that are likely
the remains of oasis gardens and/or buildings. There are also a small
number of Roman period farms and cisterns around the wells in the
Wadi Bai al-Kabir (which eventually connects to al-Qurayyat ash-
Sharqiya.14

The Central Route

Al-Qurayyat al-Gharbia
The two small oases of al-Qurayyat al-Gharbia and al-Qurayyat ash-
Sharqiya are c.20 km apart, but are equally significant as rare spring-fed
oases in otherwise very inhospitable wadis. The fact that both were
marked by Roman military installations is a strong indication that the
oases pre-dated the frontier here. Had the oases been deliberately
created by the Roman army to serve the needs of its advanced garrison
post, you would imagine that resources would have been focused on
a single site, rather than on creating two distinct loci requiring surveil-
lance. Neither oasis is large, but the western oasis (al-Gharbia) covers

10
Rebuffat 1967; 1970a; 1970b; 1971; 1975b; 1975c; 1977a; 1980; 1982; 1989; 2004. Compare,
Goodchild 1954a; Mattingly 1995, 95–97.
11
Mattingly 1995, 134–35. 12 Rebuffat 1975a.
13
Marichal 1992, 100–6, 122, no. 5; compare, Guédon 2010, 83–85. 14 Rebuffat 1982.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 193

c.1,000 × 200 m and was the site of the major fort. The UNESCO
Libyan Valleys survey and more recently a team led by Michael
Mackensen have studied the Roman fort which dominates the oasis
from a rocky terrace above its eastern edge.15 There are indications of
a civil settlement around the fort and a trace on an early air-
photograph of a possible fortified enclosure on a spur overlooking
the western side of the oasis.16 A second rectangular enclosure with
playing card corners lies 700 m to the north-east of the fort and
settlement and is potentially the remains of an earlier fortlet. There
are also small patches of oasis cultivation in the parallel wadis 2.5 km
to the east and 1 km north-west of al-Qurayyat al-Gharbia. In both
cases there are traces of wadi cultivation and abandoned one- or two-
room houses along the flanks of the wadi as well as a scatter of burial
cairns on the plateaux above. Although these features cannot be dated,
it is further evidence of human investment at some point in time in
developing the cultivated landscape. Like Bu Nijim, the fort was an
early third century addition to the frontier deployment, presumably
implanted at an already existing oasis site. Radiocarbon, ceramic and
coin evidence from the most recent excavations suggests that occupa-
tion continued within the fort (with an 80-year hiatus) until the mid-
sixth century AD.17

Al-Qurayyat ash-Sharqiya
The Roman fortlet sits on a prominent spur well above the small oasis
(c.400 × 250 m) and although no inscription has been found, a military
interpretation seems certain.18 It is undoubtedly of Roman era construc-
tion and is associated with Roman pottery. There has been limited inves-
tigation of the oasis, though a decorated architrave near the fortlet may
relate to a mausoleum. There are some rock-cut houses and mudbrick
houses on the fringe of the cultivated area and this would be the area to
look for evidence of the suspected civilian population. Around 7 km to the
north-east of the oasis, the spur on the east side of the wadi is divided by
a wall (similar to hillforts like Zinkekra). There are a few possible buildings
500 m further north, but little evidence to suggest that this defended
outcrop ever developed into something more substantial.

15
Barker et al. 1996a, 114–16; 1996b, 98–103; Goodchild 1954a; Mackensen 2010; 2011; 2012;
Mattingly 1995, 92–95; Welsby 1988.
16
Barker et al. 1996b, 104–5. 17 Mackensen 2010; 2011; 2012.
18
Barker et al. 1996b, 125; Mattingly 1995, 104–5.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 194 David J. Mattingly et al.

Mizda
Mizda was an important staging point on the route north from the two
Qurayyat outposts. It lay at an important strategic location, just north of
the Hamada al-Hamra at the crossing of the plain of the Wadi Soffegin
close to the effective western limit of the zone of intensive floodwater
farming of the pre-desert wadis. Two Roman roads marked with mile-
stones converge on the location, surely indicating that Mizda was an
important centre, most likely at some point a frontier fort.19 The early
European travellers reported on a double oasis community, divided
between a north-western and a south-eastern village, separated by
a short distance.20 The Italian fort built here in the early twentieth
century incorporated quite a lot of Roman stonework as well as inscrip-
tions, some brought in from nearby sites, but some almost certainly
gathered in the immediate vicinity.21 The architecture of the ruins of
the Medieval/early modern oasis villages has many similarities with the
post-Roman Berber village that was built over the ruins of al-Qurayyat al-
Gharbia. Recent survey has recovered second–third century Roman pot-
tery sherds from the oasis around the old town and adds weight to the
suggestion that there was a Roman fort and civil settlement here.22 As
with other oasis forts, the likelihood is that the oasis predated the instal-
lation of the garrison.

The Libyan Valleys

Between the main routes running north from al-Qurayyat al-Gharbia and
Bu Nijim lay the Tripolitanian pre-desert zone, where a distinctive pattern
of sedentary settlement developed in the early centuries AD, based on
floodwater farming in the bottom of wadis (see Fig. 1.4c). This has been
one of the best studied examples of run-off farming in North Africa.23
Although the hydrological aspect of desert farming was very different here
to the groundwater dependent oases of Fazzan,24 there are nonetheless
some striking similarities in settlement types, especially the nature of
hillforts, villages and late Roman fortified sites (qsur) and those we have
seen in the Garamantian lands 500 km to the south.25

19
Mattingly 1995, 97 for the argument.
20
Barth 1857, 99–103, who reported extensive ruins extending into the plain; Rohlfs 2001, 102–5.
21
IRT p. 215 and numbers 883–84. 22 Schimmer 2012. 23 Barker et al. 1996a; 1996b.
24
Mattingly 2004; 2014. 25 Barker et al. 1996a, 116–18, 127–34.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 195

Ghadamis and Associated Oases

Ghadamis and Its environs
On the north-western side of the Hamada al-Hamra and close up against the
eastern limit of the Great Eastern Erg, is the remarkable and isolated oasis of
Ghadamis,26 another site for which we have ancient literary evidence of pre-
Roman origins. The site is dependent on a natural artesian spring, though
there is some evidence that this was enhanced through exploitation of
additional water sources at certain points. Pliny’s account of the campaign
of Cornelius Balbus in 20 BC mentioned a region known as Phazania.27 In
spite of the apparent similarity of name, Phazania does not correspond with
modern Fazzan, but appears to have related to the desert zone around
Ghadamis, separated from modern Fazzan by the Ubari Sand Sea and the
Hamada al-Hamra.28 The names of three urbes of the Phazanii – Alele,
Cilliba and Cidamus – were also recorded by Pliny and the latter can be
identified with the oasis settlement of Ghadamis. The western and northern
limits of Phazania were probably defined by the Grand Erg Oriental and by
the identification of Alele and Cilliba with the names of Roman frontier forts
at the small oases of Talalati (Ras al-Ain) and Tillibari (Remada) in southern
Tunisia.29
The epigraphic evidence for ancient Cidamus includes a series of Latin
inscriptions that indicate a Roman military presence in the third
century AD. There are also funerary texts in Latin and Latino-Punic
relating to civilians and several examples of Libyan inscriptions.30
The archaeological evidence for early Ghadamis is less abundant than
one might expect of a UNESCO World Heritage site, though the standing
buildings of the marvellously atmospheric early modern town are an
understandable distraction.31 Archaeological survey that we initiated in
2011 to investigate the broader archaeological context of the oasis was
prematurely interrupted by the Libyan Revolution of that year.32 The most
famous pre-Islamic remains are a group of late Roman era funerary
monuments (asnam) that stood just south of the oasis on the flanking

26
For the classic early modern accounts of Ghadamis, see Duveyrier 1864; Richardson 1848. For
medieval accounts, Thiry 1995.
27
Pliny, Natural History 5.35–38.
28
The best discussion of Pliny’s account of the campaign of Balbus is Desanges 1980; see also
Desanges 1957; 1978; Mattingly 1995; 2003.
29
Mattingly 1995, 25–30.
30
Mattingly and Sterry 2010, Appendix 1, 106–13 lists 20 inscriptions from Ghadames including
Latin, probable Latino-Punic and Libyan texts; IRT 907–9; Reynolds 1958.
31
Aymo 1958 for the famous Ghadamensi houses. 32 Mattingly and Sterry 2010.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 196 David J. Mattingly et al.

low rocky plateau.33 The asnam can be identified as of the same type as the
late Roman colonnaded tombs from the pre-desert site of Ghirza.34 In
addition to many architectural fragments and relief sculptures relating to
these tombs, the town has also yielded numerous other columns and
architectural elements demonstrating that there were major stone-built
buildings of Mediterranean inspiration there.35 There is certainly some
Roman imported pottery reported within the oasis gardens to confirm the
antiquity of the oasis cultivation, even if it is not possible to isolate the
footprint of the ancient site.36 It is most likely to have been built up close to
the spring. The spring basin at Ghadamis and the associated canal network
is evidently of long antiquity and is located c.100 m outside the central
square of the Early Modern town (Fig. 5.2; cf. also Fig. 1.4e).37
A number of other structures around the oasis are also suspected to
indicate pre-Islamic activity. Qasr Mqdul is a circular tower-like building
on the west side of the oasis gardens.38 There appears to be Roman tile built
into it and some associated ceramic finds again appeared Roman. Given
that the monument lies within the pre-Islamic cemetery zone, it may have
had a funerary function – it is certainly difficult to make sense of as
a dwelling. An isolated ‘tower’ structure on the north-east side of the
oasis gardens turned out on close inspection to be a small stone-built
mausoleum. Its location away from the Islamic cemetery areas strongly
suggests a pre-Islamic date.39 Slightly further out from the main oasis are
two satellite settlements associated with foggaras, at Tunin c.3 km west of
Ghadamis and at Qasr Glul, 6.5 km to the north.40 Qasr Glul is a rocky
fortress site, with complex defences and a deep well that had been bored
down through the centre of the outcrop it sits on. It is associated with
a now abandoned oasis area fed by the foggaras. Recently obtained radio-
carbon dates identify two possible phases of occupation with a sample from
the mortar of a defensive wall of calAD 435–625, suggesting the site was
established in late antiquity, and three further samples from the late tenth

33
Corò 1956; Mercier 1953; Mattingly and Sterry 2010, 62–79.
34
Brogan and Smith, 1984, 121–47.
35
Corò 1956, 23; Duveyrier 1864, 250 and Planche X, figure 1; Mattingly and Sterry 2010, 87–91;
Richardson 1848, I, 210.
36
Rebuffat 1969, 194–95; 1972, 322–23.
37
The most detailed study of the spring system is Eldblom 1968.
38
Corò 1956, 10–12; Mattingly and Sterry 2010, 83–84.
39
Mircher et al. 1863, illustrated on fold-out map, pl.3. Relocated by Ghadames Archaeological
Survey in 2011, Mattingly et al. 2011.
40
Mattingly and Sterry 2010, 91–101.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press
Figure 5.2. Proposed development of the Ghadamis oasis from Roman to Present (red circles - architectural fragments, brown lines - surveyed garden walls,
yellow polygons - Medieval to modern cemeteries).
 198 David J. Mattingly et al.

to mid-twelfth centuries.41 There are a foggara, abandoned fields and

Moslem burials associated with the site.

Darj
Located 92 km to the east of Ghadamis, Darj is a small collection of oases
that is an important stopping point on the trade routes. Although the oasis
is now greatly expanded by modern development, early writers recorded
four separate small oases here.42 Darj – with gardens totalling c.150 ha,
Terguddah – 2.2 km to the north, Madress – 9 km to the west and Fiffelt –
comprising 4–5 houses between Madress and Darj.43 Perugini described
Darj as measuring 1,500 × 400 m with 11,000 palms, 7 springs, 30 wells and
many distribution channels at the centre of which is a clustered village 230
× 230 m similar to that at the centre of Tunin. Prospection by Rebuffat
identified a polygonal fortified structure at Fiffelt and at Madress a circular
fort and a ‘Roman fortlet’.44 It is difficult to reconcile these observations
with sites noted from satellite imagery. An enigmatic circular feature is
visible 650 m north-west of Madress while the village itself is clustered on
a spur with the remains of further buildings on its east side. To the north of
Madress where the road crosses a line of hills there is a large cairn field that
winds along the edge for 2.5 km with c.500–1,000 cairns. The hill to the
west of Terguddah has a small outpost on its southern tip of unknown date
and there is a sizable area of disturbed ground that seems likely to have
been hut clearings and animal enclosures. To the north-west, on another
spur, there is a large polygonal enclosure, but with no signs of settlement,
just to the north of this is a small group of square huts. To the north and in
the wadi there is a square fortification 30 × 30 m, probably built of stone
that may be linked to the Roman military judging by the two-towered
gatehouse on its southern side. Like Ghadamis, Darj appears to have
a natural artesian spring; however, the strip of cultivation north of
Terguddah was originally fed by at least 20 foggara channels. None are
very long, but they range from very small to large shafts, perhaps suggest-
ing that they tap water reservoirs at different depths.

41
OxA-34787, 1501±23 BP, calAD 435–625; OxA-34790, 992±22 BP, calAD 991–1150; OxA-
34985, 946±30 BP, calAD 1025–1156; OxA-34988 and OxA-34989 R_Combine 912±16 BP, Χ2
test: df=1 T=0.0(5% 3.8) calAD 1039–1165.
42
See Richardson 1848, I, 255–56; Perugini 1929, 97.
43
Largeau 1881, 234 calls these Degoutta, Materes and Tfelfet. Rebuffat 1977b, 418 offers a slightly
different position for Fiffelt to the north-east of Madress.
44
Rebuffat 1972, 323–24.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 199

Chawan and Sinawin

Some 90 km north of Darj (140 km north-east of Ghadamis) are the small
oases of Chawan and Sinawin. At Chawan, Rebuffat recorded three for-
tifications: an oval fort on a hill and two rectangular ‘fortlets’ of third and
fourth century date.45 The old centre of Chawan is 140 × 115 m and
surrounded by sparse and scattered gardens, although modern construc-
tion has destroyed much of the area of the spring-fed oasis. Richardson
described Chawan as a small place sharing a population of 60 with
Sinawin.46 Most of the archaeological evidence is to the north of
Chawan. There are at least 12 small foggaras leading into the gardens
that stretch 1.5 km to the north-east and 500 m further north is a small
hilltop fortification that must be Rebuffat’s oval fort. It is not clear where
the fortlets were located, however 3.5 km to the north-east there is a small
scarp, below which there are four rectangular structures, three are very
similar to the open farms found in the Libyan Valleys to the east, the final
one is possibly a qasr; a Roman date for all four is likely. On top of the scarp
is a cairnfield of c.100 tombs.
Sinawin, 7 km to the north-east of Chawan, is a fortified village on top of
a small hill on the edge of a small oasis and area of sabkha. The oldest part
measures 126 × 85 m, but there has been significant modern building on all
sides across the whole of the oasis. When Richardson visited he described
30 to 40 inhabitants living in ‘half a dozen or more mud hovels’ (mudbrick
houses). The oasis consisted of four or five patches of cultivation protected
by two or three mud and stone round towers called burges.47

The Tripolitanian Jabal

The Tripolitanian Jabal is relatively well-watered at its eastern end, progres-
sively less so as one travels west towards the Tunisian border and then north
to the Nefzaoua.48 Although technically dry farming is possible in part of the
Jabal, the irregular pattern of rainfall favours floodwater farming and the
exploitation of springs that support a number of small oases. The early
Roman frontier seems to have been organised in relation to a road that ran
along the crest of the Jabal or just behind it. A number of Roman military
outposts along the limes road that followed the Jabal plateau were established

45
Rebuffat 1972, 323. 46 Richardson 1848, I, 80–81.
47
Richardson 1848, I, 79. In 2005 Mattingly noted some Roman pottery imports in and around
the old settlement.
48
For cartography, see Goodchild 1954b; Mattingly 2000a.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 200 David J. Mattingly et al.

alongside springs and oases: Ain Wif (Thenadassa)49 and Ain al-Auenia
(Auru).50
An interesting aspect of the limes Tripolitanus is that the hill range of the
Jabal marks a fairly sharp division between desert and the cultivated zone
with many of the passes through the mountain marked by short linear
barriers (conventionally referred to as clausurae) as far east as Madina
Ragda and with notable concentrations south of Remada and north of
Tatahouine. While these were accompanied by watch-towers and manned
gates through the barriers, supporting their identification as Roman fron-
tier works, some are adjacent to hillforts that may indicate an earlier period
of control of these passages obligés.51
There are many cairns in the Tunisian half of the Tripolitanian Jabal
recorded in the Atlas préhistorique de le Tunisie. The cairn cemetery of El
Menaguib has been excavated by Paris and Ghaki, who indicated that some
(perhaps the majority?) should be placed in the late first/early first
millennium BC/AD on the basis of radiocarbon dates and ceramic spot-
dates.52

Remada and Ras al-Ain

Remada is an oasis near the Tunisian/Libyan border on the main route
south towards Ghadamis. A Roman fort (Tillibari) was built here in
the second century AD (probably under Hadrian), with major renovation
being carried out in the Severan period.53 The fort was destroyed during
French colonial development, but Euzennat and Trousset were able to
reconstruct some key details.54 It is clear that there was also a civil settle-
ment and an oasis here. The closest range of hills, 7 km to the south-east
contains a small hillfort with a second 17 km from Remada, both lie close to
the modern road through the Jabal. Again, it is reasonable to question
whether the sequence was fort then oasis, or more likely that the oasis
already existed when the fort was constructed at this location.
A short distance to the south-east, Dehibat is on the site of a presumed
Roman road station and Roman ruins have been reported from the vicinity
of the modern oasis village.55 To the north of Remada in the edge of the
49
Mattingly 1982; 1995, 81, 99–102. 50 Mattingly 1995, 102; Reynolds and Simpson 1967.
51
Mattingly 1995, 106–15; Trousset 1984b; cf Mattingly et al. 2013, 70–80.
52
Paris and Ghaki 2010.
53
Desanges et al. 2010, 266–67 for the literary and epigraphic evidence.
54
See Euzennat 1973; Euzennat and Trousset 1975; Mattingly 1995, 90–92; Trousset 1974,
116–18.
55
Trousset 1974, 123–26.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 201

Great Eastern Erg along the Dahar lay the fortlet of Tisavar (Ksar Rhilane).
This well-preserved fortlet, often depicted in splendid isolation among the
dunes, was in fact located 3.3 km west of the small oasis of Bou Flidja, though
no certain evidence of ancient occupation has been reported there.56
There are also several smaller oases in the Tunisian sector of the Jabal to
the north of Remada that were marked by Roman military outposts.57 The
fort and associated garrison settlement at Ras al-Ain (Talalati) sits along-
side a small oasis c.10 km north-east of Foum Tatahouine.58

The Nefzaoua Oases

The Nefzaoua oases lie on the south-east side of the Chott Jarid and the south
side of the Chott Fjaj (Fig. 5.3).59 The salt flats are seasonal and so are the
direct routes from the north and north-west across them, making the alter-
native approach to the oases from the east either side of the Tebaga range very
important. To the south, the oases are screened by the Grand Erg Oriental,
but passage is possible from the direction of Ghadamis along the pre-desert
corridor known as the Dahar. In the early twentieth century there were c.40
small oases in Nefzaoua, based around numerous artesian wells and groves
containing c.500,000 palms,60 though the 200 active springs were many fewer
and less productive than those of the Jarid.61 There were also 57 identifiable
foggaras, known locally as khraig (sing. khariga), although more than half
have been destroyed by the modern expansion of the oases.62 Given the many
signs of activity in the Roman period, there is a very strong presumption in
this case that there was already pre-Roman occupation and cultivation. The
oases were a main centre of the people known as the Nybgenii and is a strong
candidate to have been part of the regio Arzugitana in late antiquity.63 The
Nybgenii were probably implicated in the Tacfarinan revolt (a major rebel-
lion of Gaetulian peoples in AD 17–23) and subsequently their territory was
subject to Roman land survey (delimitation) in AD 29. Their principal centre
56
Desanges et al. 2010, 268–69.
57
Donau 1906; Gauckler 1900; Heron de Villefosse 1894; Hilaire 1901; Lecoy de la Marche 1894.
58
Desanges et al. 2010, 225–26; Mattingly 1995, 98; Trousset 1974, 98–102.
59
On the Nefzaoua, see Mattingly 1995, 15–16, 30–31, 131–32; Mattingly 2000a; Moreau 1947,
53–111; Trousset 1974, 41–50, 143–48; 2012a, 5382–89.
60
Gendre 1908; Heywood 1926, 330; Moreau 1947, 13–42.
61
Martel 1965, I, 33. See also Baduel and Baduel 1980; Carton 1896/1897; Gauckler 1900/1912;
Trousset 1986a; 1986b; 1987a for general considerations of the importance of different
hydrological regimes in southern Tunisia.
62
Suter 1962 recorded 57 foggaras from maps and aerial photographs.
63
Ellefi 2015 has many interesting observations, though he does not fully appreciate the scale and
significance of the ancient oasis population here.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 202 David J. Mattingly et al.

Figure 5.3. Nefzaoua and surrounding oases: main regions and sites discussed in the
text.

was recognised as a civitas town by Rome in the Flavian period, becoming

a municipium under Hadrian. This site, known as Turris Tamalleni, has been
identified with the oasis village of Telmine.64

The Telmine Oasis

Modern Telmine is a small village, sitting on a low hill at the heart of dense
palm groves. The hill is today surrounded by a ring road and seems plausibly
identifiable as a large artesian spring mound on which the settlement grew
up. Although enclosing a small area (7–8 ha), it seems probable that the
annular road marks the limits of the original settlement (which perhaps was
demarcated by a wall or ditch). Two inscriptions referring to the municipium
64
Trousset 1974, 43–46; cf. Gascou 1972, 134–35. Desanges et al. 2010, 274–75 summarise main
literary, epigraphic and archaeological discoveries at Telmine and other main Nafzaoua oases.
On the Chott Jarid in antiquity, see Peyras and Trousset 1988.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 203

Figure 5.4. Photograph of Roman remains at the main spring of Telmine/Turris

Tamelleni.

were found at Telmine and it is generally agreed to have been the site of
Turris Tamalleni. The evidence for Roman construction at Telmine has
come from the area of the mound, with finds of numerous classical capitals
and columns from several ancient buildings, including 20 column shafts, 21
capitals and five bases relating to at least five or six different architectural
schema (Fig. 5.4).65 In the same palm grove as Telmine are two further
villages with significant ancient remains. Rabta has produced a column base
and two fragmentary Latin inscriptions among other traces of Roman
occupation.66 Manssoura had access to its own spring and there are again
ashlar blocks and architectural elements indicating Roman construction.
Numerous inscription fragments were found built into a ‘native fortification’
on a hill known as Torra above Manssourra.67 There are thus three or four
separate locations within the Telmine oasis where we have evidence of
Roman stone architecture being employed, but given the physical distance
between these locations (>1.5 km) it is implausible that these are all parts of
a single very large urban centre (Fig. 5.5). Rather, we seem to witness the
monumentalising of typical small oasis villages within different parts of
the palm grove, supplied by different springs, in much the same way that
65
Trousset 1974, 45; Toutain 1903, 289–303. 66 Trousset 1974, 43; Toutain 1903, 297, 301.
67
Trousset 1974, 42–43; Toutain 1903, 297–303.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 204 David J. Mattingly et al.

Figure 5.5. Areas of Roman materials in the vicinity of Telmine/Turris Tamelleni

(imagery: Google, GigitalGlobe).

the modern villages exist alongside each other. On the epigraphic evidence,
Telmine appears to equate to the municipal centre, but it is plausible that its
population subsumed the inhabitants of these neighbouring villages.
North of Telmine there is a group of ten foggaras along a 3.5 km front that
stop some 2 km short of the modern oasis.68 Their presence implies that the
palm groves of the Telmine oasis group originally had a major extension on
or close to its northern edge. Presumably this would have had its own small
settlement cluster(s) although these are not visible on satellite imagery today.

68
Today their remnants flow into a system of floodwater barrages, but this seems to be a later
adaption of a presumably failing system.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 205

The Western Nefzaoua

To the west of Telmine, a narrow ridge of higher ground projects out into the
Chott Jarid, with a line of villages along it and their palm groves and gardens
laid out predominantly on the ground sloping away to both the south and
north. Two villages in this part of the Nefzaoua have produced traces of
Roman buildings or activity. At the far west end of this linear group of oases,
the village of Bechri has yielded abundant dressed stone blocks, as well as
elements of columns and sculptural fragments. The site marked the south-
east end of the trans-Chott route to the Jarid (Kriz in the Jarid was at the
north-west end), so the presence here of a significant settlement and monu-
ments is not surprising.69 Traces of further structures and columns have been
noted at Oum Semaa a few kilometres to the east.70 Between Djezira and
Zaoura the oasis gardens were fed by 43 foggara systems which ran north and
south between the peaks of the western spur of the Tebaga range.

Kebili
Moving east from Telmine, the modern major town of Kebili contains
significant numbers of spolia from Roman buildings and fragments of
inscriptions. Although the provenance of some of this material is not
certain, the fact that early travellers accounts noted that Roman stonework
was built into house foundations in the old village of Kebili and in other
places on the fringe of the palm grove (Kasr Tebal, Dar al-Kaid) strongly
suggests Roman era settlement.71 Just to the east at Kedouat Johala,
a Roman era cremation cemetery was identified.72

Douz and Other Outlying Oases of Southern and South-Western

Nefzaoua
To the south of Kebili, further oases around Douz have also yielded Roman
material: Bechilli (walls), Jamna (column fragments and stone blocks), Douz
(coins) and Nouail (columns and mills).73 Other ruins noted by early travellers
are less certainly Roman, but a pre-Islamic origin cannot be excluded: al-Qalaa,
Ghidema, Sabria.74
The Roman land survey of AD 29 and the Flavian recognition of a native
civitas centre75 surely implies that oases were established here prior to the
69
Toutain 1903, 393–94; Trousset 1974, 42. 70 Trousset 1974, 42.
71
Toutain 1903, 394; Trousset 1974, 46–47. 72 Trousset 1974, 47–48.
73
Trousset 1974, 48–50; compare, Toutain 1903, 324, 335.
74
Trousset 1974, 49–50; compare, Toutain 1903, 49–50. 75 Trousset 1978b.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 206 David J. Mattingly et al.

Roman conquest, though it is certainly possible that the number and scale of
oasis settlements increased substantially in subsequent centuries. It is also
noteworthy that foggara irrigation, as at Ghadamis, al-Jufra and Fazzan, was
used in the Nefzaoua along with the canalisation of a number of important
artesian springs.76 The diversification of hydraulic techniques suggests
a growth in cultivated area and population at some point.
A number of Libyan inscriptions have been found in Nefzaoua, includ-
ing at Negga, Henchir Krannfir and Henchir al-Agareb, and these provide
strong proxy evidence for pre-Roman origins and complex societies among
the Libyan population.77 There are also some isolated mudbrick structures
on the edges of the oases. Morphologically these are not too dissimilar to
Roman-era farms and hint that there were rural establishments along with
the oasis centres.78

The Tripolitanian Coastal Oases

Oases appear to have been a common factor in the establishment of towns

and other settlements along the arid coastline of Syrtica and Tripolitania.79
Even today, driving along the coastal road, one is struck by the recurrence
of palm groves alongside many of the famous sites. In western Tripolitania,
the oases link to the same aquifers that feed the Jarid and Nefzaoua oases,
though with weaker flow rates as one moves eastwards. The springs tend to
occur close to the coast in the wadi beds, creating localised oases at sites
such as Gabes (see below), Terboubou, Kettana, Mareth, Zarat.80 In eastern
Tripolitania, the ancient towns of Sabratha and Oea (Tripoli) both had
significant oases adjacent, this was particularly extensive in the case of the
latter’s gardens (known as the menscia).81 Roman activity is attested at
many of these locations and references to two separate Roman road sta-
tions called ‘ad Palmam’ on the Tripolitanian coastal road implies the
existence of several more.82 The pre-Roman origins of oasis agriculture
at many of these sites seems assured by the prime example of Tacapae.

76
Capot Rey 1953, 327. 77 Chabot 1940; Trousset 1974, 42, 73, 79. 78 Guery 1986.
79
See Mattingly 1995, 6, 12–13; Mattingly 2000a; 2000b (Syrtica). On oasis cultivation in medieval
and more recent times, see Beechey and Beechey 1828, 33–112; Brett 1976; de Mathuisieulx
1912, 196–200; Franchi 1912.
80
Martel 1965, I, 30–33.
81
Mattingly 1995, 122; on the menscia, see Laing in Bovill 1964; Tully 1817, 51–54.
82
Mattingly 1995, 62–64; Tabula Peutingeriana VII.4, VIII.1.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 207

Tacapae
The Tunisian harbour city of Gabes (ancient Tacapae or Tacape) controlled
a passage obligé along the coastal plain. The oasis of Gabes is fed by a series of
springs located 10 km inland in the Wadi Gabes, creating a long linear palm
grove containing 200,000 palms and running c.6 km long and 1–2 km broad.
There is no suggestion in the sources that the creation of Tacapae was due to
Roman initiative, rather Pliny marvelled at the extraordinary productivity of
the alien oasis agriculture, where palms, fruit trees and cereals/legumes formed
several distinct layers of cultivation.83 The exact location of the ancient town
here is not clearcut, somewhat surprising for a town that achieved colonia
status by the early third century, but not that dissimilar in fact to some of the
other oases towns in Roman Africa.84 The early modern oasis at Gabes
comprised a series of ten villages spread across the oasis, of which the two
largest (Menzel and Jara) lay on the south bank of the wadi close to the sea and
the presumed ancient harbour. The concentration of ancient stonework in
Menzel and Jara seem to ensure early origins, as also at two other sites (Chenini
and the hill of Boulbaba).85 The seemingly multiple traces of settlement within
the oasis again does not accord with our expectation of a Roman municipium
or colonia, but is a recurrent feature of these oasis centres.

Other South Tunisian Oases

Having described the important oasis cluster of the Nefzaoua and the
coastal site of Tacapae above, in this section we discuss the other oasis
clusters of southern Tunisia.

From Tacapae to Capsa

Aquae Tacapitanae
Inland from Tacapae there was a Roman spa site on a hot spring at al-
Hamma (Aquae Tacapitanae) that again fed an important oasis cluster.86
The Roman road of AD 14 was driven straight from Tacapae to Aquae
Tacapitanae, strongly suggesting that the oasis already existed at this
date.87 In the early twentieth century there were two oasis villages in the al-
83
Pliny, Natural History 18.188; Trousset 1986a, 173–76; 1987a.
84
Desanges et al. 2010, 224 for the sources and epigraphic data.
85
Trousset 1986a, 167–68; cf Babelon et al. 1893/1913, feuille 147; Hilaire 1900; Monlezun 1885.
86
Heywood 1926, 330; Trousset 1974, 17–18. 87 Donau 1907.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 208 David J. Mattingly et al.

Hamma (al-Qasr and Debdaba) and the measured distance from Tacapae
confirms that the remains of a Roman bath complex exploiting the thermal
spring at al-Qasr correspond with Aquae Tacapitanae. There are also other
traces of Roman period remains adjacent to a secondary oasis to the west.

Capsa
Gafsa (ancient Capsa) depends on a spring at a prominent gap in the
mountain chain that forms the southern border of the Tunisian tell in
a region of generally scarce water resources. From this gap, two major
desert roads run respectively south-west to the Jarid oases and south-east
to the Nefzaoua. In more recent times Gafsa has been described as the ‘key
to the Jarid’.88 The oasis fans out to the south of the town, mainly on the
west side of a prominent wadi channel. There is a subsidiary oasis on the
east side of the wadi, corresponding to two villages (al-Ksar and Lalla),
suggesting the possibility of a secondary early oasis settlement also, though
no specific traces have been reported.89 Capsa corresponds with the centre
of the modern town on the west side of the wadi, where Roman baths are
known, and architectural fragments of Roman date were visible in the
mosques and private houses in the early twentieth century.90 In Roman
times the location was a low mound in the centre of the gap in the
mountain range. The clearest evidence for the antiquity of the oasis is the
testimony of Sallust and his account of the surprise attack on the town by
Marius in 107 BC.91 The town was described by Sallust as walled at this
date, but by making a forced march at night Marius arrived undetected and
successfully stormed the settlement when the gates were opened at day-
break to allow the population to access their gardens. Although according
to Sallust, the population was killed or enslaved and the town burnt, the
revival of such an important oasis was probably not long delayed. The town
remained under Numidian control until 46 BC, but was almost certainly
assigned a garrison in the early first century AD as it was a key location on
the AD 14 road from Tacapae to the main base of the legio III Augusta at
Ammaedara (Haidra). The site became a municipium and later an

88
Borderaux 1907, 177.
89
Lahbib 2010 links the presence of springs here to a fault that runs from Lalla westwards, which
suggests that the origin of the Lalla oasis, fed by a famous spring, ought to also be ancient, see
Douglas 1912, 42–43.
90
Borderaux 1907, 57–58; Douglas 1912, 12 noted the thousands of Roman blocks built into later
walls and that the two main springs rose in the Kasbah and in the Roman baths, known locally
as the termid; Heywood 1926, 324; Mrabet 2004, 20–24; Saumagne 1962, 519–23.
91
Sallust, Bellum Jugurthinum 89–91. See Berthier 1981, 71; Borderaux 1907, 181–82.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 209

honorary colony under Rome. At 17 km to the east of Gafsa there is a line of

foggaras that underlies the modern villages of Lortess and El Guettar – 12
are still visible and some of them appear to be of substantial size.

Minor Oases between Capsa and Nafzaoua (Seftimi and Limagues)

Seftimi and Limagues are very small oases on the northern flank of the
Jabal Tabaga, close to the southern shore of the Chott Fajaj at the end of
crossing points of the Chott. Both these oases have yielded traces of Roman
activity.92

The Jarid
Located c.60–100 km south-west of Gafsa, the oases of the Jarid were
known in the Islamic era as Kastiliya, after their fortified character, perhaps
ultimately derived from the Latin term castella.93 There is certainly ample
evidence to attest pre-Islamic origins for all four of the main oasis clusters
in the Jarid (al-Wadian, al-Hamma, Tozeur and Nafta), situated along
a narrow and low-lying isthmus between two great chotts (Jarid and
Rharsa).94 The term ‘Jarid’ means ‘branch of the palm’ or region of the
palm.95 The palms of the Jarid have a high reputation in Tunisia and there
are numerous artesian springs (some thermal) that feed the complex
irrigation regime, which is evidently of considerable antiquity.96 Both
Tozeur and Nafta, the two biggest oases in the early twentieth century,
possessed about 200,000 palms.97 Many local customs and festivals also
evoke pre-Islamic origins, as is the case in the ‘Festival of Spring’.98
Roman activity in the Jarid is attested from the Flavian period, with
a boundary stone of the governor Javolenus Priscus (AD 83) indicating the
north-eastern limits of the lands of the civitas Thigensium.99 In the reign of
Nerva, a fort (castellum Thigensium) was built at Gouifla on the Capsa-
Thiges road close to the northern entry point to the oasis proper.100
The literary and epigraphic evidence strongly suggests that three of the
oasis settlements here were eventually accorded full urban status (Thiges,
92 93
Trousset 1974, 50. Encyclopedia Islam, sv Kastiliya; Trousset 1986a. Cf. Euzennat 1972.
94 95
Heywood 1926, 325–26. Guerin 1862, 250.
96
Douglas 1912, 83; Gendre 1908; Martel 1965, I, 30–33; Trousset 1986a; 1986b; 1987a; Vuillot
1893, 89–113. An interesting feature of the system of water division at Tozeur is that one of the
main dividers of an important branch was known as ‘Garahman’, Trousset 1986a, 183, figure 8.
97 98
Heywood 1926, 326. Payre 1942a; 1942b; Trousset 1995.
99
Trousset 1978a, 28; CIL 8.23165.
100
CIL 8.23166; Euzennat 1971; Trousset 1978a, 28–29, for the localisation of the inscription.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 210 David J. Mattingly et al.

Thusuros and Nepte). Thiges was named on an inscription as a civitas.101

The names of Thiges, Thusuros and Nepte also featured in the ancient
itineraries.102 Bishops of Thusuros and Nepta were included in the episco-
pal list for Byzacena.103

Kriz and the al-Wadian oasis

Thiges is now generally equated with the oasis of al-Wadian and in
particular with the vicinity of Kriz, where there are extensive ancient
ruins at al-Guebba and an inscription of Silvanus Mercury on
a neighbouring hill, close to the north-western terminus of an important
seasonal crossing route of the Chott that provided a link between the Jarid
and the Nefzaoua.104 Toussaint described extensive traces of ancient set-
tlement and noted much ancient pottery at al-Guebba.105 There was also
a hot spring at Kriz and the villages of Cededa and Oulad Majed within the
Kriz oasis were also locations with signs of ancient activity.106

Al-Hamma
On the northern flank of the Kriz oasis group, al-Hamma of the Jarid has
numerous traces of ancient activity including a pool and irrigation
channel.107

Tozeur
Thusuros is the ancient name for Tozeur and early travellers noted extensive
reuse of ancient stonework in more recent houses.108 Dressed stone of
Roman date was visible in the foundations of many houses in the south-
west quarter of Tozeur in the early twentieth century.109 In addition, recon-
struction of the main mosque in a southern suburb at Blad al-Hadar, near

101
CIL 8.23165; Trousset 1990. 102 Tabula Peutingeriana V.4–5.
103
Lancel 1972/1991, I, 140; Lepelley 1979 ; 1981. 104 Trousset 1982a.
105
Toussaint 1907, 310.
106
Duveyrier 1915, 53; Guérin 1862, 252–54; Toussaint 1907, 310; Trousset 1978a, 29–31; 1986a,
170–72. See also Desanges et al. 2010, 246, for a summary of literary and epigraphic evidence.
107
Toussaint 1907, 309; Vuillot 1893, 89–95.
108
Duveyrier 1915, 86 (stones and columns in gardens round Tozeur); Guérin 1862, 258;
Toussaint 1907, 310 (noted particular concentration of ancient materials in the south-west
quarter of the town). See also Vuillot 1893, 95–106.
109
Douglas 1912, 89 reports seeing Roman stonework, fragments of cipollino columns reused as
door sills and a gold coin of Gordian.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 211

the centre of the oasis, revealed Roman period remains. This includes
a substantial foundation in ashlar blocks reused as the foundation of the
mudbrick minaret.110 There are additional traces of Roman ruins at Bir
Kastilia and at Draa al-Barhla (Henchir Helbla) on the north-east side of the
oasis.111

Nafta
Ancient Nepte or Agarsel Nepte is equated with the oasis town of Nafta,
where early travellers again noted ancient stonework and capitals reused,
though in lesser quantity than at Tozeur.112 There are traces of Roman
columns and other stonework at Garat bel-Adda, midway between Nafta
and Tozeur.113 The oasis of Nafta centres on a north-south ravine, where
numerous artesian springs well up.114 The abundance and power of the
springs of Nafta and Tozeur made these ideal locations for precocious oasis
development.

Oases along the Roman Frontier in Eastern Algeria

The next group of oases straddles the modern Tunisian-Algerian border,

but are best treated as a group.

Oases of the Nemencha

Chebika and Tamerza
To the west of the Jarid oases, the Chott al-Rharsa deflects the Roman
frontier road towards the north-west, where it runs along the southern
flank of an east-west range, with relatively few south-north access
routes. Several small oases associated with Roman ruins or placenames
exist along this Nemencha range, often at points where seasonal wadis
issue from the mountain front, most notably at (or near) Chebika
(Speculum), Tamerza (ad Turres) and Mides along the major route
north towards Tebessa.115

110
Mrabet 2004, 33–34, photo of Bled el-Hachem minaret; Trousset 1995.
111
Desanges et al. 2010, 275; Toussaint 1907, 310; Trousset 1978a, 29; 1986a, 184.
112
Desanges et al. 2010, 187; Toussaint 1907, 310. 113 Toussaint 1907, 310.
114
Douglas 1912, 94 provides a vivid description; Heywood 1926, 326.
115
Desanges et al. 2010, 215; Trousset 1978a, 26–28; 1980b.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 212 David J. Mattingly et al.

Negrine
At c.45 km west of Chebika across the Algerian border, the Roman fort at
Besseriani (Nigrenses Maiores)116 lies 5 km south of the oasis of Negrine,
which is tucked back into the mountain valley.117 Immediately to the west of
the fort are extensive traces of ancient irrigation systems, while a substantial
palm grove still exists in the valley in which Negrine is located.118 The fort
was surrounded by a large civil settlement of c.20 ha, eventually walled and
promoted at some point to municipium status. That is, this was a properly
constituted Roman town, notwithstanding the fact that it retained a garrison
post at its centre until late Roman times.119 However, ancient activity here
was not limited to the immediate vicinity of the fort. The name Nigrenses
Maiores implies a connection with a people called the Nigrenses and the
modern settlement of Negrine and an older abandoned settlement called
Negrine el-Khadim (c.3 km north-west of Negrine) clearly evoke the same
name. At both of these locations significant evidence has been found of pre-
Islamic activity. At the abandoned mudbrick site of Negrine al-Khadim,
both Roman and Islamic era pottery has been noted and excavations in 1960
uncovered a Roman mosaic.120 There are also a number of suggestive traces
from around the village of Negrine itself, including a fortified promontory
site (undated), a reported find of coins and another of a Byzantine ostracon
mentioning olive oil production.121

Other Sites South of the Aures Mountains

There were further Roman military stations to the west of Nigrenses
Maiores, occupying similar topographic positions at the outlet of mountain
passes, where oases now exist. Mdlia, noted by Baradez, sits within an
extensive area of abandoned fields, settlements and qsur reminiscent of the
Garamantian landscapes of Fazzan.122 Ad Medias is generally identified

116
The site was recorded as Ad Maiores in the Tabula Peutingeriana, but the name has recently
been plausibly expanded and restored as Nigrenses Maiores, see Laporte and Dupuis 2009,
57–58. The important thing to note is that this name indicates that the settlement combined
a fort and a chief settlement of an indigenous people, the Nigrenses.
117
See now Laporte and Dupuis 2009, for a detailed recapitulation of all the evidence from the
group of sites here. Also Desanges et al. 2010, 96.
118
Birebent 1962, 118, 121 for the ancient irrigation systems; Gueneau 1907 for an early report.
119
Laporte and Dupuis 2009, 58–62, 68–74; Mattingly 1986.
120
Laporte and Dupuis 2009, 74–75.
121
Laporte and Dupuis 2009, 75, 90–91; cf. Fentress 1979, 97; Ragot 1874, 298.
122
Baradez 1949a, 147–48; see also Baradez 1957; Birebent 1962 on the field systems and
irrigation works.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 213

with an undated fortified site ‘en berbere’ at Henchir Taddert.123 Roman

Badias, possibly to be identified with a rectangular tell-like mound on the
edge of the modern settlement of Bades, also sat on an irrigated fan just
south of the Aures. The irrigation here seems to have in part exploited
seasonal runoff from the mountain valley. There are epigraphic and writ-
ten records indicating that this was a Roman frontier post, but modern
buildings cover the site.124

Wadi Jedi and the Ziban

The region to the south-west of the Aures mountains comprises
another concentrated area of oases in close connection with the
Roman frontier (Fig. 5.6).125 The pre-Roman origins of some at least

Figure 5.6. Eastern Algeria, main regions and sites discussed in the text.

123
Desanges et al. 2010, 96–97. 124 Fentress 1979, 98; Ragot 1874, 294.
125
Meouak 2018, for a wide-ranging review of the Medieval sources, the potential pre-Islamic
origins and the environmental setting of the Ziban.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 214 David J. Mattingly et al.

of these oases can be argued by a plausible interpretation of Pliny’s

account of the African campaign of Cornelius Balbus in 20 BC. Among
the list of sites that Pliny states were paraded on placards at the
triumphal procession in Rome relating to the Garamantian campaign,
there was a group of names that seem to relate to a different theatre of
war (Tabudium oppidum, Milgis Gemella oppidum, Thuben oppidum,
Viscera).126 These names appear remarkably similar to the later Roman
placenames from the Algerian frontier/Ziban oasis zone (Thabudeos,
Gemellae, Tubnae and Vescera).

Sidi Uqba
Thabudeos has for long been identified with the oasis of Sidi Uqba, c.15 km
south-east of Biskra, where Uqba bin Nafi was buried after his death in battle
in AD 683. Two possible Roman fortifications have been recognised c.6 km
north of the modern settlement of Sidi Uqba, a trapezoidal late Roman
enceinte with projecting towers and a rectangular tell-like mound that prob-
ably represents successive villages built up over a second-century Roman fort
platform.127 Roman pottery was also noted at the site in 1980.128 In the
irrigated fan around the site are numerous remains of qsur and field systems.

Biskra
Biskra is the capital of the Ziban district and the largest of the oases
(150,000 palms in 1926) stretching for 5 km along the Wadi Biskra.
Traces of ancient Vescera have been obliterated in modern times as the
town has grown extensively. However, traces of a fort on the west bank and
a civil settlement on the east bank of the Wadi Biskra are indicated in
nineteenth-century reports.129 The oasis developed on both sides of the
Wadi Biskra, fed by canals carrying water from two groups of springs
located a short distance north, with the larger palm groves on the west bank
including several old villages and smaller palmeries on the east bank at
Lahlia and Filiach.130 The Ziban (the name means the ‘area of villages’) is in
126
Pliny, Natural History 5.37; Desanges 1980, 400–10.
127
Baradez 1949a, 279–87; Benseddik 2018; Fentress 1979, 88.
128
The site was visited along with others in the Wadi Jedi, Ziban and Ouled Nail districts by the
lead author with Barri Jones, Charles Daniels and Rob Burns in December 1980.
129
Fentress 1979, 88; Meouak 2018, 216–21; Ragot 1874, 278.
130
Vuillot 1893, map as frontispiece; Daumas 1845, 103–5 described Biskra as less a town, more
a reunion of seven villages in the palm groves. He also confirms the presence of Roman
structures at Filiach.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 215

itself highly suggestive of the oasis character of most of the main settle-
ments. The oases benefit both from springs and from the seasonal flood-
waters carried down from the Zab and Ouled Nail mountains into the
Wadi Jedi and its tributaries.

Wadi Jedi Area

Exploring the Ziban region south-west of Biskra, Hamy reported a number
of sites with a Roman appearance (including ashlar stone, columns etc.), as
well as circular drum cairns and megalith burials of Protohistoric
tradition.131 A more detailed study of the Protohistoric burials of this
region was carried out by Roffo, confirming a Roman date, but profoundly
Saharan burial traditions.132 Daumas reported Roman remains in at least
14 oasis villages of the Ziban.133
The Roman fort of Gemellae and the nearby oasis of Mlili provide more
conclusive evidence of impressive Roman era activity (Fig. 5.7). The fort is
located on the south side of the Wadi Jedi, while the oasis of Mlili lies over
1 km away on the north bank and seems to divide into two main zones (which
may account for the ‘twin name’ = Gemellae). Mlili presumably is derived
from the original Libyan name for the pre-Roman oasis, which was rendered
as Milgis by Pliny. As at Nigrenses Maiores, the fort is surrounded by
a substantial walled garrison settlement, which gained municipium status.134
The greatest density of Roman elements seen by Hamy were around the
Roman forts of Gemellae and Doucen, but he also noted the remains of
a stone building in Mlili oasis itself, reused as the foundations of a modern
house.135
Doucen is another oasis location with evidence for a sequence of Roman
forts identified by Baradez, which indicate an early origin of the oasis prior
to its nineteenth-century refoundation.136 Satellite remote-sensing has
identified another probable fort, 17 km to the south-east of Doucen on
the north bank of the Wadi Jedi. Also on the north bank, a major fortified
settlement, with several additional rectangular fortified compounds, was
located a few kilometres to the south-west towards Ouled Jellal.137 It is
highly likely that there were further early settlements associated with the
131
Hamy 1896. 132 Roffo 1938. 133 Daumas 1845, 105, 148.
134
On the fort, see Baradez 1948; 1949a; 1949b; 1966a; 1966b; 1967; Fentress 1979; Trousset 1977.
135
Hamy 1896.
136
Capot-Rey 1953, 237. On the Roman forts, see Baradez 1949a, 116, 123; Fentress 1979, 85.
137
The site is difficult to locate today and is probably destroyed. Based on a photo reprinted in
Baradez 1949a, 124, the site lay adjacent to a stretch of the wadi that supported substantial oasis
gardens and the morphology suggests pre-Islamic origins.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 216 David J. Mattingly et al.

Figure 5.7. The headquarters building in the centre of the fort and vicus of Gemellae.

dense oasis zone that stretches along this part of the Wadi Jedi, encom-
passing Ouled Jellal and Sidi Khaled, c.7 km to its west.
Another intriguing aspect of the Roman frontier disposition in the Wadi
Jedi area is that a linear barrier (fossatum) was constructed to the south of
the wadi, running east-west for c.50 km and equipped with watch towers,
gates and supporting small outposts and fortlets.138 The positioning of the
fort of Gemellae south of the wadi placed it in a good position to co-
ordinate this frontier sector. But what was the purpose of this complex
arrangement? The French colonial view interpreted it as an attempt by the
Roman authorities to protect new agricultural investments to the north
from pastoral movements and nomadic attacks from the south. An alter-
native proposition is that it created a line of Roman control between the
densely populated oasis settlements of the Zab and Ziban, and other oases
and potential population centres further to the south (see below, Southern
Algerian Oases Beyond the Frontier).

Hodna and the Saharan Atlas

From Viscera, the Roman frontier continued as a line of forts and outposts
running north towards the Chott al-Hodna. Along part of this sector there

138
Baradez 1949a; 1967; Daniels 1987; Mattingly et al. 2013. For some insightful discussions of the
nature of the African frontier, see Trousset 1980a; 1984a; 1984b.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 217

was another sector of linear barrier (the Mesarfelta fossatum sector) and
there were a number of forts and minor military outposts at the small oases
that controlled the main western passes through the Aures mountains to
the north-east, such as Ad Calceum Herculis and Burgus Speculatorum
(al-Kantara).139 The fossatum appears to have terminated close to the
Chott al-Hodna at the substantial fortified town of Tubna (Tobna),
where the Roman fort is overlain by Byzantine and Islamic era
fortifications.140 The plain is irrigated by seasonal wadis and its develop-
ment may be considered as a quasi-oasis. Slightly to the north-east lay the
site of Zarai, where a Roman fort was placed on a route leading towards
Cirta and where a famous customs edict has been found.141
The Saharan Atlas forms an imposing salient to the south-west of the
Chott al-Hodna. It is made up of several main blocks (the Zab, the Ouled
Nail and the Jabal Amour). The heart of the Saharan Atlas comprises linear
ranges aligned south-west to north-east, rising to over 1,500 m in places,
with high plateau plain lands in between. The major wadis issuing from the
mountains flow into the areas of Chotts to the north-east (Hodna) and the
south-east (Melrhir). At the fringes of the mountains there are a number of
substantial oases in these wadis, notably Bou Saada in the north-east and
the line of sites along the Wadi Jedi from Laghouat to Gemellae, including
Messaad (site of the Roman fort of Castellum Dimmidi),142 Sidi Khaled,
Ouled Jellal. No certain traces of Roman settlement have been found at
Laghouat, but this oasis sits at a key junction point between Saharan routes
that run east-west (connecting the Biskra area with southern Morocco) and
north-south via Ghardaia.143 It would be surprising in light of the other
regional oasis developments if this was not also a precocious development.
The Roman frontier disposition in the Saharan Atlas has long been
remarked on as unusual. The frontier of the province of Mauretanian
Caesarensis lies well to the north and the Numidian frontier well to the
east, running from Biskra up to Zarai on the east side of the Chott al-
Hodna. However, from the second century onwards it is clear that there
139
Baradez 1949a, 229–42. Though Baradez was sceptical that the fort of Ad Calceum Herculis
was to be sought in the oasis of al-Kantara itself, he does note the substantial collection of
Roman dressed stone from the three villages of the al-Kantara oasis (331–32). Material from
al-Kantara includes inscriptions, architectural elements, altars, sculpted stone, tombstones,
etc. and the largest concentration was from a spot on the right bank of the wadi just south of
the oasis.
140
Baradez 1949a, 293; Fentress 1979, 92.
141
For the Zarai tariff, see France 2014; Guédon 2014; Trousset 2003.
142
Fentress 1979, 87–88; Picard 1944. See also the sites identified by Morizot 1999.
143
Heywood 1926, 172–74. There is a tradition of textile working in the oasis, based on wool
production in the Saharan Atlas.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 218 David J. Mattingly et al.

was a substantial investment made in garrisoning the Saharan Atlas, with

many of these forts placed at springs at the key access/egress points: Ausum
(Sadouri), al-Ghara, Ain Rich.144

Southern Algerian Oases Beyond the Frontier

The oases in this section (Mzab, Tuggurt, Wargla, Suf) have not yet
produced direct evidence of Protohistoric activity and clearly lay well
outside the Roman frontier. The lack of recognisable Roman structures is
not surprising if we accept that the first French investigators associated
Rome with cut-stone buildings.145 However, a number of finds of Roman
artefacts have been reported from the general area of the Mzab, Tuggurt
and Wargla oases by Mauny and Salama, including coins and Roman
pottery.146 A number of authors have picked up on the possibility that
these oases could be the location from where the Pharusii (mentioned by
Strabo) set off across the desert on horseback (with waterskins slung below
their mounts) passing the area of the ‘lakes’ (chotts?) to reach the high
plains around Cirta.147 The case for pre-Islamic origins is suggestive, if
unproven at present. Given the demonstrable extent of oasis development
in the Jarid, Wadi Jedi and Ziban zones along the Roman frontier to the
north of them it is a moot point whether these rare locations of springs and
a high water table were entirely neglected until the Islamic era. These oases
should be a priority for future investigation.

Tuggurt and the Wadi Rhir

The Tuggurt oasis is the major centre in a line of c.25 oases that follow the
north-south line of the Wadi Rhir, demarcating the western edge of the Great
Eastern Erg.148 In the early colonial period it was seen as the start of the
pestilential desert, because of the prevalence of malaria there.149 This is an area
144
Fentress 1979, 114; Faure and Leveau 2015; Laporte 2004. 145 Berbrugger 1862, 31.
146
Mauny 1956; 1978; Salama 1981, 514–15, sites 7–9 (Ghourd el-Oucif, Hassi el-Hadjar and Fort
Miribel).
147
Strabo, Geography 17.3.7; Romey 1992, 20; Rouvillois-Brigol 1975, 9–10.
148
For the hydrology see Berbrugger 1862; Durand and Guyot 1955; Savornin 1947; 1950. For
summary descriptions of oases, see Berbugger 1862, 36–43; Heywood 1926, 254–57; Nesson in
Nesson et al. 1973.
149
Berbrugger 1862, 30–31; compare, de Villiers and Hirtle 2002, 115–16 for the observation that
at many of the south Algerian oases the additional artesian wells dug by the French increased
the pools of surface water for mosquitoes to breed in, greatly increasing the prevalence of
malaria.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 219

of artesian hydrology and the first springs must have come to the surface
naturally – almost certainly still feeding the 16 small lakes/pools (bahar) that lie
at the centre of some of the palmeries. These seem to have been the focal points
around which the local people have tried to dig further artesian wells to
augment the supply.150 Some unusual circular or radiating palmeries can be
recognised on air-photographs and satellite imagery and these seem to repre-
sent the traditional layout of gardens around the major or primary springs.151
The artesian water lies at considerable depth (average 45 m), sometimes
beneath clay, which has a tendency to subside after a few years, and sometimes
beneath many metres of impermeable rock. A special class of well diggers
existed among these oasis communities in more recent times, capable of
undertaking the difficult and dangerous work.152 Prior to the French colonial
administration making substantial additions to the number of artesian wells
with modern drilling equipment, the oases were much smaller.153
Nevertheless, there were still 340,000 palms prior to the establishment of the
new wells. Little is known about the archaeology of any of the oases here and
the history has been somewhat romanticised.154
Tuggurt is c.200 km south of Biskra, but the northernmost outposts of
this oasis chain are only about 50 km from Biskra and 20–30 km south of
the fossatum. Uqba ibn Nafi evidently initiated a campaign against the
people of the Wadi Rhir, but, having reached the hills of ad-Dour that mark
the northern limits of the valley, he turned back instead.155 If there was
well-established oasis agriculture to the north and naturally occurring
artesian springs in the Wadi Rhir, it seems implausible that some oasis
development was not underway here at an early date.
Between Tuggurt and Sidi Amrane, a small number of settlements are visible
along a now unused part of the wadi. At least one of these has a square qasr
similar to those found further north in the Roman period. To the immediate
north-west of Tuggurt are a cluster of large abandoned settlements and oasis
gardens, though these seem likely to be Medieval in date. The circumstantial

150
Berbrugger 1862, 20–27 lists the following bahar springs: ‘el-Salehin and Merzioui
[Merdjadja], Tebaiche, el-Ourir, Nsir’a, Bakhbakha, Oriana [x 2], Djama, Ain Zerga
[Tineguidin], Sidi Amran, Sidi Yahya, Orlan and el-Arais [Tamerna-Djedida], Tattouin [x 2].
151
Nesson 1965, 92–104.
152
Berbrugger 1862, 44–83, for a detailed account of the digging of a traditional artesian well. Due
to the tendency for intervening perched water tables to flood the shafts, the task of releasing the
artesian layer of pressurised water involves the diggers working under water for prolonged
periods, with exceptional danger at the moment of breaking through into the artesian layer.
153
For early accounts of the Tuggurt oasis villages, see Daumas 1845, 121–42; Hurabeille 1899.
154
Fontaine 1952. Féraud 2012, 8–125 is a more useful if disjointed compendium of information
on the history of the oases.
155
Berbrugger 1862, 28–31.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 220 David J. Mattingly et al.

case for Roman era activity in this oasis group is particularly strong. If oasis
cultivation was well established at Biskra and other oases of the Ziban in the
pre-Roman era it would be surprising if the similar potential of the artesian
springs of the Wadi Rhir and Tuggurt area had not also been developed.
Likewise, the elaborate fossatum control line makes much more sense if it
was designed to regulate contact with both pastoral groups and substantial oasis
communities south of the Roman frontier.

The Suf
To the north-east of Tuggurt in the direction of the Jarid are the challenging
dunes of the Great Eastern Erg. However, midway along this difficult route
lies the extraordinary group of oases known as the Suf, with al-Wad the
largest among them.156 Unlike most of the sand seas, which support only
limited oasis development at their heart, due to a very high water table, the
Suf communities have established gardens and micro-catchments in the
interdune basins (supporting 290,000 palms). Textile production has been
a feature of these oases in the past.157 The domed architecture and overall
character of these villages and the associated cultivation practices is very
distinctive and seemingly entirely Islamic in character. However, it is worth
bearing in mind that if the Tuggurt area did see some pre-Islamic develop-
ment, this could also be a possibility for the Suf, given its geographical
position between other oases. There have been a number of Roman coin
finds reported in the Great Eastern Erg, including some from the Suf itself.158
Routes lead out of the sands to the Jarid c.100 km to the north-east, towards
Negrine c.130 km to the north, towards Biskra c.230 km to the north-west
and to Tuggurt at 95 km to the south-west.

Wargla
Wargla is a separate cluster of oases c.160 km south-west of Tuggurt.159
The oasis has long had a reputation as being among the earliest of Algerian

156
Battailon 1955, especially 27–32 for history and 33–46 for recorded settlements, though these
are evidently of Medieval or early modern date. See also Daumas 1845, 189–94, for names of
villages and modern population groups; Féraud 2012, 126–44.
157
Heywood 1926, 258–60.
158
Féraud 2012, 369; Mauny 1956, 252 specifies a cache of 60 silver coins from c.40 km south-west
of al-Wad; Salama 1981, 514–15.
159
See Rouvillois-Brigol 1975; also her section in Nesson et al. 1973; Lethielleux 1983; Romey
1992. The detailed study of the most famous Medieval site, Sedrata, also contains much
information on the wider zone of Wargla, Aillet et al. 2017.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 221

oases.160 Ibn Khaldun implied its first occupation dated back to the ninth
century and it was certainly an early centre of Ibadite resistance in the tenth
century AD. There are also oral traditions that it was involved in the
seventh-century resistance to Uqba ibn Nafi,161 and that it had associations
with Jewish Berbers. Numerous ‘Berber ruins’ are reported, but are mostly
undated.162 In fact numerous Roman coin finds have been reported from
the Wargla area,163 as well as Carthaginian gold coins from Hassi al-Hajira
between Wargla and al-Golea in 1920,164 suggesting that some activity
could date considerably earlier than the Islamic tradition. The palm groves
(500,000 palms in the early twentieth century) are sustained by the sub-
terranean waters of the Wadi Mya and c.200 artesian wells/springs.165 The
remains of a small number of foggaras are also to be found on the west side
of the oasis. From Wargla routes run south across hostile terrain towards
the Ahaggar and Tamanrasset (1,500 km) and west to the Mzab.166

Sedrata
Largely buried below the dunes immediately to the south of Wargla, the
abandoned early Islamic town (tenth century) of Sedrata is exceptionally
large for an oasis settlement (c.30 ha, Fig. 5.8). There are also visible traces
of an extensive area of gardens.167 There is no evidence of pre-Islamic
origins of settlement of the Ibadite town, though some Protohistric activity
in the vicinity is possible.168 Beyond them there are ruins of a number of
rectangular buildings of unknown date on top of the isolated flat-topped
160
Daumas 1845, 72–79; Lethielleux 1983, 5.20; Rouvillois-Brigol 1975, 9–12.
161
Rouvillois-Brigol 1975, 12.
162
Bensaad 2013; Tarry 1884 (though most of the published site plans seem to have Islamic
elements like mosques). On the other hand, Rouvillois-Brigol 1975, 65 notes the lack of
certainty over dating: ‘le problème de la foundation des ksar et donc leur age reste toujours
posé’. As we have seen in Chapter 2, these types of fortified sites are well attested in the
Garamantian kingdom and Wargla is about the same distance from Gabes as Ghadamis.
Romey 1992, 22–23 concludes that pre-Islamic origins for the Wargla group are probable.
163
Rouvillois-Brigol 1975, 11, following Féraud 2012, 369, mentions ‘nombreuses monnaies
romaines’ found in the region south of the Roman frontier and bought by French officers,
including 12 late Roman coins of Constantine and Maximian. There must have been at least 15
coins in total as Féraud says he bought 12 and other officers in the 1871 military column with
him also bought some. This contradicts Mauny’s assertion (1956, 251) that Roman coins have
never been found in Wargla, al-Golea etc.
164
Lethielleux 1983, 9. 165 Féraud 2012, 358–67. 166 Heywood 1926, 257–58.
167
Van Berchem 1953a; 1953b. See now Aillet et al. 2017, for a detailed volume on the site and its
environs (fig. 1b between pages 4 and 5 is the best plan of the urban site/palace complex and
the related gardens).
168
Aillet et al. 2017, 39–41, review the evidence for the ‘black hole’ of the pre-Islamic era, but are
ultimately pessimistic that earlier oasis foundation can be corroborated at present.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 222 David J. Mattingly et al.

Figure 5.8. Satellite image of Sedrata (imagery: Google, DigitalGlobe).

hillfort of Gara Krima (5.2 ha), with an 80 m+ deep well dug through solid
rock to reach the water table.169 The visible elements at Gara Krima may be
Medieval and early modern, but earlier origins of the site seem probable.

Ghardaia and the Mzab

Ghardaia is the main centre of the Mzab,170 a small cluster of unusual
oases, located to the west of the main north-south depression of the Wadi
Righ and Wargla oases. These are located not in a large depression, but in
the upper reaches of a series of deep gorge-like valleys. Ghardaia lies
c.170 km south-east of Laghouat and 180 km south of Ouled Jellal. It is
also accessible via the Wargla oases c.190 km to the east along the Wadi

169
Aillet et al. 2017, 345–56, for various observations of Gara Krima.
170
Amat 1888; Doumone 2010; cf. d’Armagnac 1934; Roche 1970.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 223

Mzab. The main oasis group in the Mzab comprises five towns (Ghardaia,
Melika, Bou Noura, Beni Izguen and al-Atteuf) in a c.7 km stretch of the
Wadi Mzab, where a high water table is augmented by periodic flash floods
in the dry river bed.171 The irrigation is mainly from shallow wells, but the
irregular wadi floods are exploited by means of a series of four barrages in
the wadi.172 At least one foggara is reported here, but none have been
located on satellite imagery. Foggaras are also known in the Wargla, at In
Salah (and the Tidikelt and Tuwat oases to its west).173 The Wadi Mzab
terminates in a chott on the north side of Wargla. The history of the Mzab
oases is built around the usual foundation myths, but limited archaeologi-
cal investigation has taken place, mainly focused on the Medieval and early
modern architecture – which is quite distinctive. In the early modern era,
the Mzab population had a reputation for being nonconformist (Ibadite
Berbers), with a sizeable Jewish minority. The morphology of the various
settlements is somewhat varied and their enceintes as reported in the
nineteenth century comprised multiple phases of differing construction,
pisé, mudbrick and stone. The core areas of the towns are built up, with
a generally oval shape, potentially representing a long urban history. No
obvious pre-Islamic tombs have been recorded, but given the constraints of
the narrow valley this is a location where successive phases of settlement
and cemeteries are likely to have occupied the same physical space.
A route leads from Ghardaia down the east edge of the Great Western
Erg to al-Golea c.270 km to the south. Beyond al-Golea, the route continues
south to In Salah (400 km) and then Tamanrasset and the Ahaggar
(650 km).

Conclusions

This section of our survey has demonstrated that there is compelling and
widespread evidence for Protohistoric oasis development in north-western
Libya and southern Tunisia and Algeria in and close to the Roman frontier
zone. In desert areas, the placement of Roman forts and minor installations
often corresponded with water sources. In the past, this has been portrayed
as an attempt primarily to supervise the movement of pastoral groups by
controlling access to the major wells. Yet, the fact that many military
171
Amat 1888, pl. II and Roche 1970, for maps showing the wadi and its five distinct centres, now
totally transformed into a single urban sprawl. Sèbe 1989 includes some spectacular aerial
views of the Ghardaia oasis towns.
172
Amat 1888, 52–63 on the hydrology. 173 Capot-Rey 1953, 320 Carte VII; Cornet 1952.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 224 David J. Mattingly et al.

installations were placed at, or in close proximity to, oasis locations

suggests that there were other concerns. A significant proportion of forts
along this stretch of desert frontier were situated at oases: Bu Nijim, al-
Qurayyat al-Gharbia, Mizda, Ghadamis, Remada, Ras al-Ain, Bezeroes,
Besseriani, Biska, Sidi Okba, Gemellae, Doucen, Castellum Dimmidi, as
were many of the smaller fortlets like al-Qurayyat ash-Sharqiya, Ain Wif,
Ain al-Avenia, Tisavar (Qasr Rhilane), etc. (Fig. 5.9). This is a clear and
hitherto underappreciated pattern and it raises the chicken and egg ques-
tion of which came first, fort or oasis? These Northern Saharan locations,
including pre-desert margins, correspond with the areas where our ancient
sources recurrently referred to peoples known as Gaetuli, conventionally
interpreted in modern times as scattered pastoral groups.174 The identifi-
cation of a strong association between the Gaetuli and early oases is thus of
considerable significance.
The archaeological evidence is too unrefined to offer a clear guide on the
foundation dates of these oases, though finds of Roman material culture,
including stone architectural elements, sometimes Latin inscriptions, etc.,
confirm the activity in the early centuries AD at many sites, including
numerous oases where no military installations have been recorded.175
There are compelling reasons to argue that many of these oases originated
in the first millennium BC and were thus already established centres of the
Gaetulian peoples at the time of first contact with the Roman army. When the
Roman Empire extended its direct control into these desert regions, it placed
its garrisons at locations where water and food supplies were readily available
and close to the major population centres of the region. The campaign of
Caius Marius against Capsa in 110 BC was explicitly made against an oasis
town, while the famous coastal oasis of Tacapae has yielded evidence of
Libyphoenician stelae. When Cornelius Balbus made his celebrated campaign
against a series of desert peoples in 20 BC, the list of locations captured (as
displayed at his triumph in Rome and reported by Pliny), plainly extended
well beyond the Libyan oasis sites of Ghadamis and Jarma. A significant group
of names correspond to the Algerian oases centres around Biskra and
Gemellae, as noted above. The importance of the palm iconography in
Punic artworks and coins is a further demonstration that the oasis was
a well-established part of the landscape by the last centuries BC. It is of course

174
The Gaetuli are long overdue a thorough review in light of the sort of evidence gathered in this
and the next chapter. For a range of views, see Desanges 1962; 1980, 342–46; Hamdoune 2018,
63–64; 69–70; Leveau 2018, 91–93; Mattingly 1995, 26, 29–32.
175
The systematic collation of evidence here allows us to go a lot further than previous
speculations, see for instance Marouf 1980.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 225

conceivable that there was significant growth in the numbers of oasis settle-
ments in the Roman era beyond the number that existed at first contact.
Nonetheless, the conclusion that the oases of the Roman frontier zone were
starting to be developed from the later first millennium BC (if not earlier)
seems inescapable.
As concerns the oases beyond the frontier in southern Algeria, the
current state of knowledge reflects more an absence of evidence than
evidence of absence. Apart from a few isolated finds of Roman material
culture (mainly coins), there is little hard information. Oases that were
throughout the Roman period outside the frontier were less likely to
receive an abundance of Roman material culture or to adopt monumental
stone architecture on the Roman model. The Garamantes and Cidamensi
are the main exceptions to this rule. Yet, if oases were well established in
the frontier zone itself, it is entirely plausible that they were also developed
at some of the key locations beyond the frontier where hydraulic condi-
tions were favourable. The Roman frontier deployment in the Gemellae/
Biskra sector, fronted by a long linear earthwork (fossatum) looks to be
designed to manage contact with people coming from the direction of the
Wadi Rhir and Tuggurt. A recent study of the customs tariff of Zarai from
the Algerian sector of this frontier by Guédon has identified that part of the
text seems to concern textiles traded across the frontier.176 Textile produc-
tion has been a traditional mainstay of oasis communities in this region,
one that emphasises the essential links between oasis cultivators and
pastoral groups of the desert. More intensive work on the Garamantes
has identified significant textile production and has highlighted this and
other manufacturing activity in the oases as a key element of ancient
Saharan trade.177 For these reasons, we suggest that the default assumption
in future work should be that some of the initial oasis development in the
Wadi Rhir, Wargla, Ghardaia, Mzab and perhaps the Suf must have
occurred by the early first millennium AD at the latest.
The sort of oasis sites that developed in the Roman frontier zone are
distinctive from those of the Western Egyptian Desert, or the Garamantian
heartlands.178 They also differ from the normal patterns of urban devel-
opment in Roman Africa.179 There are probably two main reasons for this,
one relating to the nature of Roman garrison settlements, the other to the
adaptation of a Roman Mediterranean urban model by oasis societies. The
nature of the military-focused communities that coalesced around army

176
Guédon 2014. 177 Guédon 2017; Mattingly 2010, 523–30; Mattingly and Cole 2017.
178
See Chapters 2–4, this volume. 179 Wilson, Chapter 10, this volume.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 226 David J. Mattingly et al.

Figure 5.9. Comparative plans of Oasis forts and settlements (Bu Nijim, al-Qurayyat al-Gharbia, Ras al-
Ain, Negrine, Badias, Gemellae) (imagery: Google, DigitalGlobe).

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 227

bases was likely to differentiate such locations from neighbouring oases.

Garrison towns in many ages have a certain character that sets them apart
from civilian towns. In the case of the African frontier, there are indica-
tions that the major military bases were built on the fringe of established
oases, with the fort centrally positioned within a defined garrison settle-
ment (vicus), as illustrated by the case studies of Bu Nijim, Nigrenses
Maiores and Gemellae (Fig. 5.9). The second and third named sites both
went on to acquire full urban status as municipia, but lack many of the
standard amenities of Roman towns in Africa, while long retaining the
central focus of a garrison post.
The civitas of the Nybgenii at Telmine in the Nefzaoua (and perhaps also
the major centres in the Jarid) reveals another type of urban development.
The apparently small size of the core urban settlement reflects what may be
a normative pattern in oasis societies in Roman Africa, with the population
split between multiple villages and small towns, each with associated
gardens, rather than to coalesce into a single large centre. It is a moot
point whether the Roman municipium (as Telmine became under
Hadrian) actually designated a single location or rather encompassed
a number of villages in the immediate area of Telmine. The major city at
Tacapae (Gabes) has proved elusive to trace on the ground, perhaps
because in reality it too was formed of several adjacent nuclei within the
oasis. Although there is evidence of an engagement with Classical archi-
tecture and institutions at the urban oases of the frontier zone – as
illustrated by finds at Ghadamis, Telmine, Tozeur, Besseriani, Gemellae,
for instance – it seems inherently unlikely in terms of what survives and
what is currently known of these sites that they were miniature versions of
the typical urban centres of northern Tunisia, Libya or Algeria. Again,
future work is needed to shed more light on the typical amenities and
layout of such oasis towns. This pattern stands in contrast to the larger size
of some of the major towns in the Egyptian oases, though there also we may
note the presence in some districts of multiple village scale settlements.180
The wider implications of the existence of a world of oases in the African
frontier zone lie beyond the scope of this volume and will be developed
elsewhere.181 The negative stereotyping of the peoples of the northern

180
See Boozer Chapter 4; Mattingly et al. Chapter 3, this volume.
181
On the Roman frontier in Africa, see now the recent reviews of Guédon 2018a; 2018b. Further
work is needed to take full account of the different model proposed here for the partial
sedentarisation of Saharan populations in and beyond the frontier zone.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 228 David J. Mattingly et al.

Sahara in Roman sources also needs more critical analysis.182 So too does
the pernicious afterlife of a deeply prejudiced modern colonial discourse
on Saharan populations and the recognition that oasis communities had
the capacity to develop into urban societies is an important step in this
regard.183 However, it is apparent that traditional models that focus either
on ideas of warlike nomadic people184 or assimilatable transhumant
pastoralists185 both require revision to take account of the new reality of
the substantial population bodies and productive capacities of the many
oases that existed alongside pastoral groups in the northern Sahara.

References

Aillet, C., Cressier, P. and Gilotte, S. (ed.). 2017. Sedrata. Histoire et archéologie
d’un carrefour du Sahara médiéval à la lumière des archives inédites de
Marguerite van Berchem. Madrid: Casa de Velázquez.
Amat, C. 1888. Le M’zab et les M’zabites. Paris: Challamel editions.
Aroca, A. 1942. Uau el-Chebir. L’oasi della redenzione. Milan: Ministero
Dell’Africa Italiana.
Aymo, J. 1958. La maison ghadamsie. Travaux de l’institut de recherches
Sahariennes de l’Université d’Alger 17: 157–94.
Babelon, E, Cagnat, R and Reinach, S. 1893/1913. Atlas archéologique de la Tunisie.
Paris: Ministère de la guerre.
Baduel, A. and Baduel, P. 1980. Le pouvoir de l’eau dans le sud-Tunisien. Revue de
l’Occident Musulman et de la Méditerranée 30.2: 101–34.
Baradez, J. 1948. Gemellae, un camp d’Hadrien et une ville des confins sahariens
aujord’hui ensélevis sous les sables. Comptes Rendues à l’Académie des
Inscriptions et Belles Lettres 1948: 390–95.
Baradez, J. 1949a. Vue aérienne de l’organisation romaine dans le sud Algérienne.
Fossatum Africae. Paris: É d. arts et métiers graphiques.
Baradez, J. 1949b. Gemellae un camp d’Hadrien et une ville des confins sahariens
aujord’hui ensélevis sous les sables. Revue Africaine 93: 5–24.
Baradez, J. 1957. Travaux hydrauliques romaines révéles par photographies
aériennes dans un région aujord’hui steppienne. In Actes du 79th Congrès
National des Societés Savants, Alger 1954. Paris: 273–75.

182
Desanges 1962, for the best collation of the source material. While there is an entry on a people
referred to in the Classical sources as Oasitae and many entries on key oasis groups in the
Encyclopédie berbère, there is tellingly no entry for ‘oasis’.
183
See, for example, Aroca 1942, 217–25 on the ‘etno criminalità libica’ and 281–90 ‘la criminalità
dei nomadi’. For an analysis of such crude models, see Leveau 2018; Mattingly 2011, 43–72.
184
Cagnat 1914; Guey 1939; Leschi 1942; Toutain 1896.
185
Trousset 1984a; 1986b; Whittaker 1978.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 229

Baradez, J. 1966a. Deux amphithéatres inédits du ‘limes’ de Numidie: Gemellae et

Mesarfelta. In Mélanges d’archéologie, d’épigraphie et d’histoire, offerts à Jerome
Carcopino. Paris: Hachette, 55–63.
Baradez, J. 1966b. Les thermes légionnaires de Gemellae. In Corolla Memoriae
Erich Swoboda Dedicata. Köln: Böhlau Verlag, 14–22.
Baradez, J. 1967. Compléments inédits au ‘Fossatum Africae’. In Studien zu den
Militargrenzen Roms: Vortrage des 6. Internationalen Limes Kongresses in Süd-
deutschland. Köln: Böhlau Verlag, 200–10.
Barker, G.W., Gilbertson, D, Jones, B. and Mattingly, D. 1996a. Farming the Desert.
The UNESCO Libyan Valleys Archaeological Survey, vol. 1 Synthesis (editor
G. Barker). Paris: UNESCO.
Barker, G.W., Gilbertson, D, Jones, B. and Mattingly, D. 1996b. Farming the Desert.
The UNESCO Libyan Valleys Archaeological Survey, vol. 2 Gazetteer and Pottery
(editor D. Mattingly). Paris: UNESCO.
Barth, H. 1857. Travels and Discoveries in North and Central Africa (reprint 1965).
London: Ward Lock and Co.
Battailon, G. 1955. Le Souf. Etude de geographie humaine. Alger: Memoires de
l’Institut de recherches sahariennes 2.
Beechey, F.W. and Beechey, H.W. 1828. Proceedings of the Expedition to Explore
the North Coast of Africa from Tripoly Eastward. London: John Murray.
Bensaad, A. 2013. Ouargla. Encyclopédie berbère fasc. 36. Aix-en-Provence: Edisud,
5899–910.
Benseddik, N. 2018. Sidi Okba ou Thouda-Thabudeos: Un nouveau milliaire. In
Guédon 2018, 207–14.
Berbrugger, A. 1862. Les puits artesiens des oasis meridionales de l’Algerie. Paris:
Hachette.
Berthier, A. 1981. La Numidie. Rome et la Maghreb. Paris: Picard.
Birebent, J. 1962. Aquae Romanae. Algiers: Bacournies
Bisson, J. 1957. Le Gourara. Etude de géographie humaine. Alger: Memoires de
l’Institut de recherches sahariennes 3.
Borderaux, P. 1907. La Capsa ancienne. La Gafsa moderne. Paris: Challamel.
Bovill, E.W. (ed.). 1964. Missions to the Niger Vol. I The Journal of F. Hornemann’s
travels from Cairo to Murzuk in the years 1799–1798. The Letters of Alexander
Gordon Laing 1824–26. Cambridge: Hakluyt Society.
Brett, M. 1976. The journey of al-Tijani to Tripoli at the beginning of the four-
teenth century A.D./eighth century AH. Libyan Studies 7: 41–51.
Brogan, O. and Smith, D.J. 1984. Ghirza: A Romano-Libyan Settlement in
Tripolitania. Tripoli: Department of Antiquities, Libyan Antiquities Series 1.
Cagnat, R. 1914. La frontière militaire de la Tripolitaine à l’époque romaine.
Mémoires de l’Institut National de France, Académie des Inscriptions et Belles
Lettres 39: 77–109.
Capot-Rey, R. 1953. Le Sahara Français (L’Afrique Blanche Française II). Paris:
Presses universitaires.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 230 David J. Mattingly et al.

Carton, L. 1896/1897. Etudes sur les travaux hydrauliques des romains en Tunise.
Revue Tunisienne 3: 373–85, 530–564; Revue Tunisienne 4: 27–85.
Chabot, J.B. 1940. Recueil des inscriptions libyque. Paris: Imprimerie nationale.
Cornet, A. 1952. Essai sur l’hydrogéologie du Grand Erg Oriental et des régions
limitrophes (les foggaras). Travaux de l’Institut de Recherches Sahariennes 8:
71–122.
Corò, F. 1956. Ghadames archeologica. Storia degli studi delle esplorazioni e dei
risultati su alcuni fra i più tipici antichi monumenti dell’oasi famosa, Libia.
Rivista trimestriale di studi Libici 4.34: 3–26.
Daniels, C.M. 1987. Africa. In J. Wacher (ed.), The Roman World, vol. 1. London:
Routledge, 223–65.
D’Armagnac, Lt. 1934. Le Mzab et les pays Chamba. Alger: Editions Baconnier.
Daumas, E. 1845. Le Sahara algerien. Etudes geographiques, statistiques et histor-
iques sur la region au sud des etablissments français en Algerie. Paris: Langlois et
Leclerc.
De Mathuisieulx, H.M. 1912. La Tripolitaine d’hier et de demain. Paris: Hachette.
Desanges, J. 1957. Le triomphe de Cornelius Balbus (19 B.C.). Revue Africaine
101: 5–43.
Desanges, J. 1962. Catalogue des tribus africaines de l’antiquité classique à l’ouest du
Nil. Dakar: Université de Dakar.
Desanges, J. 1978. Recherches sur l’activité des Méditerranéens aux confins de
l’Afrique. Rome: Collection de l’École française de Rome 38.
Desanges, J. 1980. (Pline l’ancien) Histoire naturelle Livre V.1–46 (L’Afrique du
nord). Paris: Les Belles Lettres.
Desanges, J., Duval, N. and Lepelley, C. 2010. Carte des routes et des cités de l’est de
l’Africa à la fin de l’antiquité. Paris: Bibl. Antiquité tardive 17.
Despois, J. and Raynal, R. 1967. Géographie de l’Afrique du Nord-Ouest. Paris:
Payot.
Donau, R. 1906. Notes sur des ruines du sud-Tunisien. Bullétin archéologique du
comité des travaux historiques et scientifiques 1906: 113–22.
Donau, R. 1907. Etude sur la voie romaine de Tacapae à Turris Tamalleni. Bulletin
de la Societé archéologique de Sousse 1907: 52–67; 173–90.
Douglas, N. 1912. Fountains in the Sand. London: Secker.
Doumane, S. 2010. Mzab. Encyclopédie berbère fasc. 32. Aix-en-Provence: Edisud,
5173–81.
Durand, J.H. and Guyot, J. 1955. L’irrigation des cultures dans l’oued Righ.
Travaux de l’Institut de Recherches Sahariennes 13: 75–130.
Duveyrier, H. 1864. Les Touareg du nord. Exploration du Sahara. Paris: Challamel.
Duveyrier, H. 1915. Sahara algérien et tunisien. Journal de route de Henri Duveyrier
(edited by C. Mannoir and H. Schizner). Paris: Challamel.
Eldblom, L. 1968. Structure foncière. Organisation et structure sociale. Une étude
comparative sur la vie socio-économique dans les trois oasis libyennes de Ghat,
Mourzouk et particulièrement Ghadamès. Lund: Uniksol.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 231

Ellefi, M. 2015. L’Arzugitana: Recherches d’histoire et d’archéologie sur une entité

africaine de l’antiquité tardive. Unpublished PhD thesis, University of Sousse.
Euzennat, M. 1971. Le castellum Thigensium (région de Metlaoui, Tunisie).
Bullétin archéologique du comité des travaux historiques et scientifiques ns 7B:
229–39.
Euzennat, M. 1972. Quatre années de recherches sur la frontière romaine en
tunisie Méridionale. Comptes Rendues à l’Académie des Inscriptions et Belles
Lettres 1972: 7–27.
Euzennat, M. 1973. Tillibari, forteresse du ‘limes Tripolitanus’. Bullétin
archéologique du comité des travaux historiques et scientifiques 1973: 143–44.
Euzennat, M. and Trousset, P. 1975. Le camp de Remada, fouilles inédites du
commandant R. Donau. Aix-en-Provence (privately circulated) = Africa 5–6,
1978: 111–89.
Faure, P. and Leveau, P. 2015. Une nouvelle inscription séverienne du Djebel Bou
Kahil (Ouled Naïl, Djelfa, Algérie) et son apport à la connaissance des confins de
la Numidie. Antiquités africaines 51: 119–42.
Fentress, E.W.B. 1979. Numidia and the Roman army. Social, military and eco-
nomic aspects of the frontier zone. Oxford: BAR S53.
Féraud, L.-C. 2012. Les Ben-Djellab, sultans et Touggourt: Notes historiques sur le
sud-Constantinois Ued Righ. Souf, Ziban. Alger: Livres editions.
Fontaine, P. 1952. Touggourt, capitale des oasis. Paris: Eds Dervy.
France, J. 2014. Normes douanières et réglementation des échanges, trois questions
simples sur le tarif de Zaraï (Numidie). Antiquités Africaines 50: 93–110.
Franchi, S. 1912. Richerche e studi agrologici sulla Libia. I, La zona di Tripoli.
Bergamo: Istituto italiano d’arti grafiche.
Gascou, J. 1972. La politique municipale de l’Empire romain en Afrique
Proconsulaire de Trajan é Septime Sévère. Paris: Publications de l’École
française de Rome.
Gauckler, P. 1900/1912. Enquête sur les installations hydrauliques romaines en
Tunisie, 2 vols. Paris: Imprimerie rapide.
Gauckler, P. 1900. Notes sur les fouilles executées dans le Sahara tunisien. Comptes
Rendues à l’Académie des Inscriptions et Belles Lettres 1900: 541–47.
Gendre, F. 1908. De Gabes à Nefta (Le Nefzaoua et le Djerid). Revue Tunisienne 15:
383–411, 499–520.
Goodchild, R.G. 1954a. Oasis forts of ‘legio III Augusta’ on the routes to the
Fezzan. Papers of the British School at Rome 22: 56–68.
Goodchild, R.G. 1954b. Tabula Imperii Romani: Lepcis Magna (Sheets H.33 1.33).
Oxford: Society for Antiquaries.
Guédon, S. 2010. Le voyage dans l’Afrique romaine. Bordeaux: Ausonius.
Guédon, S. 2014. La ‘lex uestis peregrinae’ dans le tarif de Zaraï. Antiquités
Africaines 50: 105–17.
Guédon, S. 2017. Circulation and trade in textiles in the southern borders of
Roman Africa. In Mattingly et al. 2017, 259–84.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 232 David J. Mattingly et al.

Guédon, S. 2018a. La frontière romaine de l’Africa sous le Haut-Empire. Madrid:

Casa de Velazquez.
Guédon, S. (ed.). 2018b. La frontière méridionale du Maghreb. Approches croisées
(Antiquité-Moyen Âge). Bordeaux: Ausonius.
Gueneau, Capt. 1907. Ruines de la région de Negrine. Bullétin archéologique du
comité des travaux historiques et scientifiques 1907: 314–35.
Guérin, V. 1862. Voyages archéologiques dans la Regence de Tunis. Paris: Plon.
Guery, R. 1986. Chronologie de quelques établisements de la frontière romaine du sud
tunisien, à partir de la céramique collectée sur les sites. In Akten des 13
Internationalen Limeskongresses Aalen, Sept. 1983. Stuttgart: Theiss Verlag, 600–4.
Guey, J. 1939. Note sur le ‘limes’ romain de Numidie et le Sahara au IVe siècle.
Mélanges de l’école française de Rome 56: 178–248.
Hamdoune, C. 2018. Ad fines Africae Romanae. Les mondes tribaux dans les
provinces maurétaniennes. Bordeaux: Ausonius.
Hamy, E.-T. 1896. Note sur de nouvelles observations archéologiques recueillies
entre El-Alia et Biskra. Comptes rendus des séances de l’Académie des Inscriptions
et Belles-Lettres 40.1: 10–15.
Heron de Villefosse, 1894. Rapport sur la mission du Lieutenant d’artillerie,
H. Lecoy de la Marche, dans le sud Tunisien. Comptes Rendues à l’Académie
des Inscriptions et Belles Lettres 1894: 469–81.
Heywood, C. 1926. Algeria and Tunisia. Paris and London: Librarie Hachette.
Hilaire, Capt. 1900. L’emplacement de Tacape. Bullétin archéologique du comité des
travaux historiques et scientifiques 1900: 115–25.
Hilaire, Capt. 1901. Note sur la voie stratégique romaine qui longeait la frontière
militaire de la Tripolitaine. Essai d’identification des gîtes d’étapes de la portion
de cette voie comprise entre ad Templum et Tabuinati. Bullétin archéologique du
comité des travaux historiques et scientifiques 1901: 95–105.
Hurabeille, J. 1899. Au pays du bleu. Biskra et les oasis environnants. Paris:
Challamel.
Lahbib, B.A.M. 2010. Etude de rationalisation d’eau de l’ oasis historique de Gafsa.
ASM/GIAHS. Rome: FAO.
Lancel, S. 1972/1991. Actes de la conférence de Carthage en 411. Paris: Ed du Cerf:
SC 194, 195, 224, 383.
Laporte, J.-P. 2004. Trois sites militaires sévériens en Algérie moyenne: Grimidi,
Tarmount (Aras), El Gahra. L’Africa romana 15: 439–77.
Laporte, J.-P. and Dupuis, X. 2009. De Nigrenses Maiores à Négrine. Antiquités
africaines 45: 51–102.
Largeau, V. 1881. Le Sahara Algerien. Les deserts de l’Erg. Paris: Libraire Hachette.
Lecoy de la Marche, H. 1894. Recherche d’une voie romaine du Golfe de Gabes vers
Rhadames. Bullétin archéologique du comité des travaux historiques et scientifi-
ques 1894: 389–413.
Lepelley, C. 1979. Les cités de l’Afrique romaine au Bas-Empire. I, La permanence
municipal. Paris: Etudes augustiniennes.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 233

Lepelley, C. 1981. Les cités de l’Afrique romaine au Bas-Empire. II, Notices d’histoire
municipal. Paris: Etudes augustiniennes.
Leschi, L. 1942. Rome et les nomades du Sahara central. Travaux de l’Institut de
Recherches Sahariennes I: 47–62.
Lethielleux, J. 1983. Ouargla. Cité saharienne des origines au début du XXe siècle.
Paris: Paul Geuthner.
Leveau, P. 2018. Climat, sociétés et environment aux marges sahariennes du
Maghreb: Une approche historiographique. In Guédon 2018b, 19–106.
Lyon, G.F. 1821. A Narrative of Travels in Northern Africa in the Years 1818–1819
and 1820. London: John Murray.
Mackensen, M. 2010. Das severische Vexillationskastell Myd(–)/Gheriat el-Garbia
am limes Tripolitanus (Libyen) – Bericht über die Kampagne 2009. Mitteilungen
des Deutschen Archäologischen Instituts, Römische Abteilung 116: 363–458.
Mackensen, M. 2011. Das severische Vexillationskastell Myd(–) und die spätantike
Besiedlung in Gheriat el-Garbia (Libyen). Bericht über die Kampagne im
Frühjahr 2010. Mitteilungen des Deutschen Archäologischen Instituts, Römische
Abteilung 117: 247–375.
Mackensen, M. 2012. New fieldwork at the Severan fort of Myd(–)/Gheriat el-
Garbia on the limes Tripolitanus. Libyan Studies 43: 41–60.
Marichal, R. 1992. Les ostraca du Bu Njem. Tripoli: Department of Antiquities.
Marouf, N. 1980. Lecture de l’espace oasien. Paris: Sinbad.
Martel, A. 1965. Les confins Saharo-tripolitains de la Tunisie (1881–1911), 2 vols.
Paris: Presses universitaires de France.
Mattingly, D.J. 1982. The Roman road-station at Thenadassa (Ain Wif). Libyan
Studies 13: 73–80.
Mattingly, D.J. 1986. Soldier or civilian? Urbanisation on the frontiers of Roman
Africa. Popular Archaeology Dec 1985/Jan 1986: 61–66.
Mattingly, D.J. 1995. Tripolitania. London: Batsford.
Mattingly, D.J. 2000a. Tripolitana. In R. Talbert (ed.), Barrington Atlas of the Greek
and Roman World. Princeton: map 35 and Map-by-Map Directory pp. 529–44.
Mattingly, D.J. 2000b. Syrtica. In R. Talbert (ed.), Barrington Atlas of the Greek and
Roman World. Princeton: map 37 and Map-by-Map Directory pp. 552–57.
Mattingly, D.J. (ed.). 2003. The Archaeology of Fazzan. Volume 1, Synthesis.
London: Society for Libyan Studies, Department of Antiquities.
Mattingly, D.J. 2004. Surveying the desert: From the Libyan Valleys to Saharan
oases. In M. Iacovou (ed.), Archaeological Field Survey in Cyprus. Past History,
Future Potentials. Proceedings of a Conference Held at the University of Cyprus,
1–2 December 2000. London: BSA monograph 11, 163–76.
Mattingly, D.J. (ed.). 2010. The Archaeology of Fazzān. Volume 3, Excavations
Carried out by C.M. Daniels. London: Society for Libyan Studies, Department of
Antiquities.
Mattingly, D.J. 2011. Imperialism, Power and Identity. Experiencing the Roman
Empire. Princeton: Princeton University Press.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 234 David J. Mattingly et al.

Mattingly, D.J. 2014. Castles in the desert: Living in Libyan landscapes. In K. Boyle,
R. Rabett and C. Hunt (eds), Living in the Landscape. Essays in Honour of
Graeme Barker. Cambridge: McDonald Institute Monographs, 299–311.
Mattingly, D.J. and Cole, F. 2017. Visible and invisible commodities of trade: The
significance of organic materials in Saharan trade. In Mattingly et al. 2017,
211–30.
Mattingly, D.J. and Sterry, M.S. 2010. Ghadames Archaeological Survey. Phase 1
Desk-top Report. Unpublished consultancy report produced for Libyan
Department of Antiquities, Ghadames Development Authority and BP.
Mattingly D.J., Ahmed, M., Sterry, M., Fenwick, C. and Moussa, F. 2011. The
Ghadames Archaeological Survey Phase 2 Work February 2011. Preliminary
Report on Work Accomplished Prior to Suspension of Project. Unpublished
consultancy report produced for Libyan Department of Antiquities,
Ghadames Development Authority and BP.
Mattingly, D.J., Rushworth, A., Sterry, M. and Leitch, V. 2013. The African
Frontiers. Frontiers of the Roman Empire series, general editor, D. Breeze.
Edinburgh: Hussar.
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017. Trade in the Ancient Sahara and Beyond. Trans-Saharan
Archaeology Volume 1. Series editor D.J. Mattingly. Cambridge: Cambridge
University Press.
Mauny, R. 1956. Monnaies antiques trouvées en Afrique au sud du limes romain.
Libyca 4: 249–61.
Mauny, R. 1978. Trans-Saharan contacts and the Iron Age in West Africa. In J.
D. Fage (ed.), The Cambridge History of Africa, vol. 2. Cambridge: Cambridge
University Press, 272–341.
Mercier, R. 1953. Les idoles de Ghadames. Revue Africaine 97: 17–47.
Meouak, M. 2018. Biskra et ses oasis au Moyen Âge, marge aurésienne, marge
saharienne? Notes préliminaires. In Guédon 2018b, 215–42.
Mircher, H., Polignac, L. de, Vatonne, F. and Hoffman. 1863. Mission de Ghadames
(septembre, octobre, novembre et décembre 1862). Rapports officiels et documents
à l’appui. Alger: Typographie Duclaux.
Monlezun, Capt. 1885. Les ruines de Tacape (Gabes). Bullétin archéologique du
comité des travaux historiques et scientifiques 1885: 126–31.
Moreau, P. 1947. Des lacs du sel aux chaos de sable. Le pays de nefzaouas. Tunis:
Bascone & Muscat.
Morizot, P. 1999. La présence romaine dans le Djebel Amour (Algérie): Apport des
images Spot et de la photographie aérienne. In Frontières et limites géographiques
de l’Afrique du Nord antique, Hommage à Pierre Salama, Actes de la Table ronde
réunie à Paris les 2 et 3 mai 1997. Paris: Publications de la Sorbonne, 185–94.
Mrabet, A. 2004. La Tunisie du sud. Sites et monuments. Tunis: Simpact.
Nachtigal G. 1974. Sahara and Sudan, vol I, Tripoli and Fezzan, Tibesti or Tu,
translated by A.G.B. and H.J. Fisher. London: C. Martin.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 235

Nesson, C. 1965. Structure agraire et evolution sociale dans les oasis de l’oued Righ.
Travaux de l’Institut de Recherches Sahariennes 24: 85–127.
Nesson, C., Rouvillois-Brigol, M., and Vallet, J. 1973. Oasis du sahara algérien.
Paris: Institut géographique national.
Paris, F. and Ghaki, M. 2010. Les monuments mégalithiques du sud tunisien: État
de la question. Les nouvelles de l’archéologie 120–121: 71–74.
Payre, G. 1942a. Une fete du printemps au Djerid. Revue Tunisienne 1942: 171–77.
Payre, G. 1942b. Amines d’oasis au Djerid. Revue Tunisienne 1942: 335–39.
Perugini, Ilo. 1929. Gadames. Monografia del teritorio. Tripoli: Comando R.C.T.
C. della Tripolitania.
Peyras, J. and Trousset, P. 1988. Le ‘lac Tritonis’ et les noms antiques du Chott el
Jerid. Antiquités africaines 24: 149–204.
Picard, G.C. 1944. Castellum Dimmidi. Algiers: de Boccard.
Ragot, W.R. 1874. Le Sahara de la Province de Constantine. Récueil des notices
et mémoires de la société archéologique de Constantine 16: 91–299.
Rebuffat, R. 1967. Bu Njem 1967. Libya Antiqua 3–4:49–137.
Rebuffat, R. 1969. Deux ans de recherches dans le sud de la Tripolitaine. Comptes
Rendues à l’Académie des Inscriptions et Belles Lettres 1969: 189–212.
Rebuffat, R. (with contributions by others) 1970a. Bu Njem 1968. Libya Antiqua
6–7: 9–105.
Rebuffat, R. 1970b. Bu Njem 1970. Libya Antiqua 6–7: 107–65.
Rebuffat, R. 1971. Recherches en Tripolitaine du sud. Revue Archéologique 1971
ns2: 177–84.
Rebuffat, R. 1972. Nouvelles recherches dans le sud de la Tripolitaine. Comptes
Rendues à l’Académie des Inscriptions et Belles Lettres 1972: 319–39.
Rebuffat, R. 1975a. Graffiti en Libyque de Bu Njem. Libya Antiqua 11–12: 165–87.
Rebuffat, R. 1975b. Bu Njem 1971. Libya Antiqua 11–12: 189–242.
Rebuffat, R. 1975c. Trois nouvelles campagnes dans le sud de la Tripolitaine.
Comptes Rendues à l’Académie des Inscriptions et Belles Lettres 1975: 495–505.
Rebuffat, R. 1977a. Bu Njem 1972. Libya Antiqua 13–14: 37–77.
Rebuffat, R. 1977b. Dix ans de recherches dans le prédesert de Tripolitaine. Libya
Antiqua 13–14: 79–91.
Rebuffat, R. 1980. A propos du ‘limes Tripolitanus’. Revue Archéologique 1980.1:
105–24.
Rebuffat, R. 1982. Recherches dans le désert de Libye. Comptes Rendues à
l’Académie des Inscriptions et Belles Lettres 1982: 188–99.
Rebuffat, R. 1989. Le camp romain de Gholaia (Bu Njem). Libyan Studies 20:
155–67.
Rebuffat, R. 2004. Les romains et les routes caravanieres africaines. In M. Fantar
(ed.), Le Sahara. Lien entre les peuples et les cultures. Tunis: Université de Tunis,
221–60.
Reynolds, J.M. 1958. Three inscriptions from Ghadames in Tripolitania. Papers of
the British School at Rome 26: 135–36.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 236 David J. Mattingly et al.

Reynolds, J.M. and Simpson, W.G. 1967. Some inscriptions from el-Auenia, near
Yefren in Tripolitania. Libya Antiqua 3–4: 45–47.
Richardson, J. 1848. Travels in the Great Desert of Sahara in 1845–46, 2 vols.
London: R. Bentley.
Roche, M. 1970. Le Mzab: Architecture ibadite en Algérie. Bellegarde: Antaud.
Roffo, P. 1938. Sépultures indigènes antéislamiques en pierres sèches, étude sur
trois nécropoles de l’Algérie centrale. Revue Africaine 82: 197–242.
Rohlfs, G. 2001. Voyages et explorations au Sahara. Tome II, Tripoli – Rhadamès –
Fezzan – Kaouar – Bornou 1865–1867 (translated by J. Debetz). Paris: Karthala.
Romey, A. 1992. Histoire, mémoire et sociétés: l’exemple de N’Gouss, oasis
berbèréphone du Sahara (Ouagla). Paris: Awal/l’Harmattin.
Rouvillois-Brigol, M. 1975. Le pays de Ouargla (Sahara algérien). Variations et
organisation d’un espace rural en milieu désertique. Paris: Université Paris-
Sorbonne.
Salama, P. 1981. The Sahara in classical antiquity. In G. Mokhtar (ed.), UNESCO
General History of Africa, II Ancient Civilisations of Africa. Paris: UNESCO,
513–32.
Saumagne, C. 1962. Capsa, les vestiges de la cite latine de Gafsa. Cahiers de Tunisie
37–40: 519–23.
Savornin, J. 1947. Le plus grand appareil hyrdraulique du Sahara (nappe artésienne
dite de l’Albia). Travaux de l’Institut de Recherches Sahariennes 4: 25–66.
Savornin, J. 1950. Le Bas-Sahara (appareil artésienne le plus simple du Sahara).
Travaux de l’Institut de Recherches Sahariennes 6: 45–62.
Schimmer, F. 2012. New evidence for a Roman fort and vicus at Mizda
(Tripolitania). Libyan Studies 43: 33–39.
Sèbe, A. 1989. Moula-Moula. Le Sahara à vol d’oiseau. Vidauban: Collection
Tagoulmoust.
Suter, K. 1962. Über Quelltöpfe, Quellhügel und Wasserstollen des Nefzaoua
(Südtunesien). Vierteljahrsschr. Nat. Ges. Zürich, Jg. 107.2: 49–64.
Tarry, H. 1884. Les villes berberès de la vallée de l’oued Mya. Revue d’ethnographie
3: 1–44.
Thiry, J. 1995. Le Sahara Libyen dans l’Afrique du Nord Medievale. Leuven:
Orientalia Lovaniensia Analecta, 72.
Toussaint, P-M. 1907. Résumé des reconnaisances archéologiques exécutées par les
officiers des Brigade Topographiques d’Algérie et de Tunisie. Campagne de
1905–1906. Bullétin archéologique du comité des travaux historiques et scientifi-
ques 1907: 308–14.
Toutain, J. 1896. Les Romains dans le Sahara. Mélanges de l’école française de Rome
16: 63–77.
Toutain, J. 1903. Notes et documents sur les voies stratégiques et sur l’occupation
militaire du sud Tunisien à l’époque romaine. Bullétin archéologique du comité
des travaux historiques et scientifiques 1903: 272–409.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 5 Pre-Islamic Oasis Settlements in the Northern Sahara 237

Trousset, P. 1974. Recherches sur le ‘limes Tripolitanus’ du chott el-Djerid à la

frontière tuniso-libyenne. Paris: Editions CNRS.
Trousset, P. 1977. Le camp de Gemellae sur le ‘limes’ de Numidie d’après les
fouilles de Colonel Baradez (1947–50). In J. Fitz (ed.), Akten des Internationalen
Limeskongresses, 1976. Budapest: Akadémiai Kiadó , 559–76.
Trousset, P. 1978a. Reconnaisances archéologiques sur la frontière saharienne de
l’Empire romain dans le sud-ouest de la Tunisie. Actes 101e Congrès National des
Societés Savantes, Lille. Archéologie. Paris: Bibliothèque national, 21–33.
Trousset, P. 1978b. Les bornes du Bled Segui. Nouveaux aperçus sur la centuriation
romaine du sud Tunisie. Antiquités africaines 12: 125–78.
Trousset, P. 1980a. Signification d’une frontière: nomades et sédentaires dans la
zone du ‘limes’ d’Afrique. In W.S. Hanson and L.J.F. Keppie (eds), Roman
Frontier Studies 1979. Papers Presented to the 12th International Congress of
Roman Frontier Studies. Oxford: BAR S71, 931–43.
Trousset, P. 1980b. Les milliaires de Chebika (sud Tunisien). Antiquités africaines
15: 135–54.
Trousset, P. 1982a. Le franchissement des chotts du sud Tunisien dans l’antiquité.
Antiquités africaines 18: 45–59.
Trousset, P. 1982b. L’image du nomade saharien dans l’historiographie antique.
Production pastorale et sociéte 10: 97–105.
Trousset, P. 1984a. L’idée de frontière au Sahara d’après les données
archéologiques. In P. Baduel (ed.), Enjeux Sahariens. Paris: CNRS, 47–78.
Trousset, P. 1984b. Note sur un type d’ouvrage linéaire du ‘limes’ d’Afrique.
Bullétin archéologique du comité des travaux historiques et scientifiques ns 17B:
383–98.
Trousset, P. 1986a. Les oasis présahariennes dans l’antiquité: partage de l’eau et
division du temps. Antiquités africaines 22: 161–91.
Trousset, P. 1986b. Limes et ‘frontière climatique’. In Histoire et archéologie de
l’Afrique du Nord, 3e Colloque International Montpellier, 1985. Paris: CNRS,
55–84.
Trousset, P. 1987a. L’organisation de l’oasis dans l’antiquité (exemples de Gabés et
du Jérid). In A. de Réparaz (ed.), L’eau et les hommes en Méditerranée. Paris:
CNRS, 25–41.
Trousset, P. 1987b. De la montagne au désert: Limes et maitrise de l’eau. Revue de
l’Occident Musulman et de la Méditerranée 41–42: 90–115.
Trousset, P. 1990. Thiges et la civitas Thigensium. In L’Afrique dans l’occident
romain (i er siècle av. J. -C.-iv e siècle ap. J.-C). Rome: Publications de l’École
française de Rome, 143–67.
Trousset, P. 1995. Djerid. Encyclopédie berbère, fasc 16. Aix-en-Provence, Edisud:
2461–65.
Trousset, P. 2003. Le tarif de Zaraï: Essai sur les circuits commerciaux dans la zone
présaharienne. Antiquités Africaines 38–39: 355–73.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 238 David J. Mattingly et al.

Trousset, P. 2012a. Nefzaoua. Encyclopédie berbère, fasc 33. Aix-en-Provence,

Edisud: 5382–89.
Trousset, P. 2012b. Nomadisme (Saharien en Afrique du nord dans l’antiquité.
Encyclopedie berbère, fasc 34. Aix-en-Provence, Edisud: 5578–89.
Tully, R. 1817. Narrative of a Ten Years’ Residence at Tripoli in Africa. London:
H. Colburn.
Van Berchem, M. 1953a. Deux campagnes de fouilles à Sedrata 1951–52. Travaux
de l’Institut de Recherches sahariennes 10: 123–38.
Van Berchem, M. 1953b. Sedrata, une ville au moyen âge ensevelie dans les sables du
Sahara algérien. Alger: Documents algériens, série monographie 11.
Villiers, M. de and Hirtle, S. 2002. Sahara. A Natural History. New York: Walker
and Company.
Vuillot, P. 1893. Des Zibans au Djerid par les Chotts Algériens. Paris: Oberthus.
Welsby, D. 1988. The defences of the Roman forts at Bu Ngem and Gheriat el-
Garbia. In P. Bidwell, R. Miket and B. Ford (eds), Portae Cum Turribus. Studies
of Roman Fort Gates. Oxford: BAR S206, 63–82.
Whittaker, C.R. 1978. Land and labour in North Africa. Klio 60.2: 331–62.

https://doi.org/10.1017/9781108637978.006 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the
North-Western Sahara
martin sterry, david j. mattingly
and youssef bokbot

Introduction

This chapter focuses on the northern oases of the Western Sahara (Fig. 6.1).
As with the Northern Sahara, the North-western Sahara has also too often
been dismissed as wholly nomadic before the development of Arab trading
networks. There have been only limited attempts to view the region as
a whole, as opposed to restricted studies of the different oasis groups. In
this chapter we start with the western end of the Saharan Atlas and move west
towards the Atlantic through the Wadi Gir, Wadi Saoura, Tafilalat, Wadi
Draa and Wadi Noun. We conclude with a diversion to some outlying oases
to the south, Gourara and Tuwat at the southern end of the Wadi Gir and
Tabalbala c.300 km to the west of Tuwat. This broad zone also needs to be
considered in relation to the south-western Saharan sites that are presented in
Chapter 7. The Western Sahara as a whole has been structured around several
extended north to south trading routes and is known for its famous trading
towns. The rich mineral resources, especially silver and copper in the High
Atlas, Anti-Atlas mountain ranges and the Adrar Plateau were another
distinctive aspect of the human exploitation of the Western Sahara. The
Protohistoric archaeology of the region has been somewhat underexplored,
though as we shall see has great potential.1 In the Medieval and early modern
periods the kingdoms of the Almoravids and Saadi in Morocco and the
kingdoms of Ghana, Mali and Songhay in the Sahel repeatedly made efforts
to extend their control over the region and its lucrative trade.
The Roman sources mention a number of indigenous peoples lying to
the south of the Roman provincial territory of Mauretania Tingitana in
northern Morocco and these were evidently considered to be distinct from

1
Bokbot 2015. 239

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 240 Martin Sterry, David J. Mattingly and Youssef Bokbot

the Maures of the north.2 The population of the Wadi Draa are specifically
referred to as Gaetuli.3 As we noted in the previous chapter, references to
Gaetuli elsewhere along the frontier zone correlate broadly with groups
who we suggest combined pastoral activity and sedentary oasis farming.
While early oasis settlements have hitherto been elusive archaeologi-
cally, there are some important proxy elements of Protohistoric activity in
the area, notably large funerary landscapes of pre-Islamic burials4 and rock
art featuring horsemen, chariots and camels.5 There are some dense con-
centrations of this sort of evidence either adjacent to oasis zones, or
marking the close approach routes to the oasis. We shall allude to this
evidence where relevant below, but want to stress at the outset that whether
these related to sedentary or pastoral groups, we suspect that their dense
clustering close to oasis locations suggests a connection with the initial
stages of sedentarisation and oasis creation.6
Available radiocarbon dates are presented in Table 6.1.

Oases of North-Western Algeria and Southern Morocco

The development of oases in southern Morocco and the extreme west of
Algeria has been of a rare and distinctive type compared to much of the rest
of the Sahara. The key hydrological feature in this region is several sig-
nificant wadi systems with available ‘surface water’ as a result of snow melt
and rainfall on the High Atlas and Anti-Atlas ranges. These run far out into
the desert to the south: the Wadi Gir/Saoura, the Wadi Ziz and the Wadi
Draa, as well as the Wadi Noun which flows into the Atlantic.7 The
distribution of date palms in Morocco highlights the existence of two
predominant oasis zones: the Wadi Draa and the Erfoud/Rissani area
(Tafilalat) on the Wadi Ziz. There are in addition numerous much smaller
palmeries in the Wadi Gir, the Wadi Noun and in other wadis on the
southern and south-eastern sides of the Atlas ranges fed from run-off from
the mountain face.8 There are no significant stands of palms in the moun-
tains or to their north side. The greater size of the palmeries in the Draa and

2
See now Hamdoune 2018, for a detailed overview of the ancient indigenous peoples of the
Western Maghrib and the Saharan margins.
3
Pliny, Natural History 5.9–10 (Gaetulos Daras), though there were evidently also black Africans
nearby (Aethiopas Darathitas).
4
Bokbot 2019. 5 Gauthier and Gauthier 2011; 2015; Simoneau 1977.
6
As also argued in other volumes of this series, see Mattingly 2017, 22–24; Mattingly et al.
2019, 6–17.
7
For the geographical setting, see Despois and Raynal 1967, 458–73; Gautier 1970, 76–87.
8
Joly 1948, 204, 209 for maps.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 241

Figure 6.1. North-west Sahara, main regions and sites discussed in the text.

the Ziz indicates the perennial or near perennial nature and greater flow of
these wadis, a unique occurrence in the Sahara, where the only other
permanent rivers relate to Sub-Saharan catchments (Niger and Nile).
The Gir/Saoura is notable on the other hand for the depth of penetration
of its more seasonal floodwaters, extending as far south as Tuwat.
While the historical data on the Moroccan oases mostly dates to the
Islamic era, there are a number of significant hints at pre-Islamic origins.9
A most compelling argument for an early date of the first activity in the two
major oases is that they are fed by perennial rivers and are thus among the
most obvious and simplest of hydrological situations to exploit.10
The first Roman encounter with the people of the zone to the south of
the Atlas can be precisely dated to the early AD 40s, when two generals

9
Jacques-Meunié 1982 (especially 159–89 for discussion of pre-Islamic evidence).
10
Rohlfs 2001, for a mid-nineteenth-century account of both the major Moroccan oases. See also
Capel, Chapter 16, this volume for discussion of the difficulties of irrigation in the Tafilalat.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 Table 6.1 Protohistoric and historic radiocarbon dates from the Western Sahara

Area/Site Context Material Reference Date Calibrated

Wadi Gir
Jorf Torba Tomb 19 Human bone Pa-1252 2230±60 BP 403–120 calBC
Jorf Torba Tomb 28 Human bone Pa-1257 1760±60 BP calAD 128–402
Jorf Torba Tomb 27 Human bone Pa-1260 1700±60 BP calAD 143–534
Tafilalat
Ahshilifa Barrage Dam #5 Charcoal Not given 350±70 BP calAD 1432–1664
Ziz Barrage At head of Ziz Charcoal Not given 650±35 BP calAD 1279–1396
Sijilmasa T41.D014 Wood Not given 1360±70 BP calAD 545–863
Sijilmasa T21A.D035 Charcoal Not given 990±90 BP calAD 780–1252
Sijilmasa T28.D155 Charcoal Not given 710±90 BP calAD 1058–1421
Sijilmasa T50.D014 Phase 0/I Charcoal Not given 1270±50 BP calAD 660–876
Sijilmasa T52.D019 Phase I Wood Not given 1700±60 BP calAD 138–527
Sijilmasa T42.D096 Phase I Charcoal Not given 1480±70 BP calAD 424–660

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

Sijilmasa T21A.D059 Phase I Charcoal Not given 1350±180 BP calAD 258–1030
Sijilmasa T41.D013 Phase I Wood Not given 1280±40 BP calAD 657–864
Sijilmasa T52.D017 Phase I Charcoal Not given 1230±40 BP calAD 683–887
Sijilmasa T21A.D055 Phase IA Charcoal Not given 1260±70 BP calAD 648–947
Sijilmasa T21A.D062 Phase IB Charcoal Not given 1410±100 BP calAD 413–863
Sijilmasa T52.D012 Phase I/II Wood Not given 1430±40 BP calAD 558–663
Sijilmasa T21A.D042 Phase I/II Charcoal Not given 1260±120 BP calAD 576–1016
Sijilmasa T40.D018 Phase I/II Charcoal Not given 1210±50 BP calAD 678–952
Sijilmasa T21A.D038 Phase I/II Charcoal Not given 1010±80 BP calAD 779–1217
Sijilmasa T32.D091 Phase II Charcoal Not given 1240±80 BP calAD 654–986
Sijilmasa T32.D104 Phase II Charcoal Not given 1010±60 BP calAD 894–1162
Sijilmasa T21A.D040 Phase IIA Charcoal Not given 1140±90 BP calAD 675–1030
Sijilmasa T21.D066 Phase IIA Charcoal Not given 910±60 BP calAD 1019–1245
Sijilmasa T38.D017 Phase IIB Charcoal Not given 1190±70 BP calAD 682–981
 Sijilmasa T21.D029 Phase IIB Charcoal Not given 1080±120 BP calAD 681–1184
Sijilmasa T21.D047 Phase IIB Charcoal Not given 1020±90 BP calAD 777–1212
Sijilmasa T32.D083 Phase II/III Charcoal Not given 1210±60 BP calAD 675–966
Sijilmasa T40.D016 Phase II/III Charcoal Not given 1090±130 BP calAD 670–1185
Sijilmasa T32.D061 Phase III Charcoal Not given 1000±80 BP calAD 782–1220
Sijilmasa T31.D009 Phase III/IV Charcoal Not given 1250±60 BP calAD 657–939
Sijilmasa T54.D005 Phase III/IV Wood Not given 480±50 BP calAD 1315–1616
Sijilmasa T54.D005 Phase III/IV Wood Not given 450±50 BP calAD 1398–1630
Sijilmasa T54.D005 Phase III/IV Wood Not given 410±40 BP calAD 1427–1632
Sijilmasa T34.D012 Phase IV Wood Not given 970±80 BP calAD 894–1224
Sijilmasa T32.D053 Phase IV Charcoal Not given 360±100 BP calAD 1328–1950
Sijilmasa T34.D012 Phase IV/V Charcoal Not given 1220±130 BP calAD 566–1117
Sijilmasa Bab al-Monsouria Not given Beta-431101 390±30 BP calAD 1441–1631
Sijilmasa Bab al-Monsouria Not given Beta-431103 390±30 BP calAD 1441–1631
Sijilmasa Bab al-Monsouria Not given Beta-431102 340±30 BP calAD 1470–1640
Tafilalat P08-03 Charcoal Not given 1123±23 BP calAD 880–987

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

Tafilalat P08-02 Charcoal Not given 964±27 BP calAD 1019–1155
Wadi Draa
BOU001 From mudbrick wall of building Plant remains (grass) OxA-33783 933±31 BP calAD 1025–1164
BOU002 From SE wall of qasr Charcoal (cf. OxA-33784 1196±32 BP calAD 713–945
tamarix sp.)
BOU003 From SE wall of qasr Charcoal (cf. OxA-33823 870±26 BP calAD 1046–1244
tamarix sp.)
BOU009 From mudbrick wall of building Plant remains OxA-33824 810±27 BP calAD 1170–1270
BOU022 From mudbrick wall of building Plant remains (grass) OxA-33718 969±37 BP calAD 997–1159
BOU033 Outer wall of qasr Charcoal (cf. OxA-35336 1050±25 BP calAD 901–1025
tamarix sp.)
BOU036 From mudbrick wall of building Charcoal (cf. OxA-35361 899±28 BP calAD 1040–1211
tamarix sp.)
 Table 6.1 (cont.)

Area/Site Context Material Reference Date Calibrated

BOU038 From mudbrick wall of building Charcoal (cf. OxA-35362 894±26 BP calAD 1042–1214
tamarix sp.)
BOU051 From mudbrick wall of building Wood OxA-35447 331±25 BP calAD 1482–1642
BOU051 From mudbrick wall of possible Charcoal OxA-35363 287±29 BP calAD 1493–1664
mosque
BOU051 From mudbrick wall of building Plant remains (grass) OxA-35647 174±23 BP calAD 1662–present
BOU053 From mudbrick wall of qasr Date stone OxA-35364 535±27 BP calAD 1320–1438
BOU053 From mud bonding of stone wall Charcoal (cf. OxA-35365 450±26 BP calAD 1418–1469
tamarix sp.)
BOU054 From clay-lined pit Barley grain OxA-35367 1338±28 BP calAD 646–767
T3 (3111)
BOU054 From clay-lined pit Charcoal (cf. OxA-35366 1318±27 BP calAD 655–767
T1 (3104) tamarix sp.)
BOU057 Hearth within chapel-tomb Charcoal OxA-35303 2710±80 BP 1107–768 calBC

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

BOU057 Floor of chapel-tomb Charcoal OxA-35403 1820±29 BP calAD 92–320
BOU057 Above floor of chapel-tomb Charcoal (cf. OxA-35185 1751±27 BP calAD 230–381
tamarix sp.)
BOU064 Burnt clay feature Charcoal OxA-35368 890±40 BP calAD 1034–1220
BOU071 From mudbrick wall of building Charcoal OxA-35369 802±27 BP calAD 1186–1274
BOU073 From mudbrick wall of building Charcoal (cf. OxA-35370 874±26 BP calAD 1045–1224
tamarix sp.)
BOU076 From mudbrick wall of building Charcoal OxA-35371 874±26 BP calAD 1045–1224
BOU092 Burnt clay feature Wood OxA-35603 833±26 BP calAD 1163–1258
BOU105 From mudbrick wall of building Charcoal (cf. OxA-35372 957±26 BP calAD 1022–1155
tamarix sp.)
BOU110 From mudbrick wall of building Charcoal (cf. OxA-35373 971±27 BP calAD 1016–1155
tamarix sp.)
BOU117 Neolithic pottery scatter Ostrich Egg Shell OxA-35225 8147±39 BP 7302–7057 calBC
 BSM001 Mortar from mosque Charcoal OxA-35375 988±26 BP calAD 991–1152
BSM001 Mortar from qasba Charcoal (cf. OxA-35374 900±26 BP calAD 1039–1210
tamarix sp.)
BSM002 Pisé wall Charcoal OxA-35376 472±26 BP calAD 1414–1451
LAR001 Cemetery Human bone OxA-33338 2649±28 BP 894–791 calBC
LAR001? Cemetery Bone Gif-2912 2369±250 BP 1082–129 calBC
LAR002 Trench 1, context 1015 Charcoal OxA-32373 1504±28 BP calAD 432–633
LAR002 Same sample as above Charcoal (cf. OxA-33825 1616±27 BP calAD 389–536
tamarix sp.)
LAR002 (same sample as above) (same sample as above) OxA-33826 1676±28 BP calAD 260–422
LAR002 Trench 1, context 1019 Charcoal OxA-33827 1638±27 BP calAD 339–534
LAR002 Trench 2, context 1021 Charcoal (cf. OxA-33828 1121±28 BP calAD 778–994
tamarix sp.)
LAR002 Trench 2, context 1021 Barley grain OxA-33829 1377±27 BP calAD 612–679
LAR003 Trench 1, context 2006 Charcoal OxA-33719 823±28 BP calAD 1166–1263
LAR004 From pisé of S enclosing wall Charcoal OxA-32374 878±25 BP calAD 1046–1221

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

LAR005 Trench 1, context 1014 Charcoal (cf. OxA-33720 1589±30 BP calAD 406–542
tamarix sp.)
LAR006 From external wall (intrusive) Plant remains OxA-33898 104±28 BP calAD 1682–1935
LAR031 Surface of offering structure Calcined bone OxA-33899 1344±30 BP calAD 641–765
TAG008 Mortar from tower of enceinte Charcoal (cf. OxA-35378 1082±27 BP calAD 895–1017
tamarix sp.)
TAG008 Mortar from tower of enceinte Charcoal (cf. OxA-35377 1057±27 BP calAD 900–1024
tamarix sp.)
TAG009 From pisé wall Textile OxA-35604 993±24 BP calAD 990–1150
TAG010 From mudbrick wall of building Plant remains OxA-35448 409±21 BP calAD 1437–1615
TAG014 Occupation layer Charcoal (cf. OxA-35187 1658±28 BP calAD 263–526
Tr1 (2033) tamarix sp.)
TAG014 Occupation layer Charcoal (cf. OxA-35186 1544±27 BP calAD 426–577
Trench 1 (2032) tamarix sp.)
 Table 6.1 (cont.)

Area/Site Context Material Reference Date Calibrated

TAK001 From a mudbrick wall Plant remains OxA-33900 129±29 BP calAD 1675–1942
TAK004 From mortar of outer wall Charred seeds OxA-35652 943±26 BP calAD 1028–1155
TAK004 From S boundary wall (intrusive) Sheep/goat pellet OxA-33850 Modern± BP 0
TAK006 Trench 1 (2010) Charcoal (cf. OxA-33851 1560±26 BP calAD 423–557
Tamarix sp.)
TAK006 Trench 2 (2012) Charcoal OxA-33852 1586±28 BP calAD 411–542
TAK013 From mortar of outside wall Plant remains OxA-33853 365±26 BP calAD 1450–1634
TAK016 From pisé wall Charcoal OxA-33854 243±24 BP calAD 1528– . . .
TAM001 Trench 1 (3028) Charcoal OxA-33966 1512±27 BP calAD 430–615
TIN001 Trench 4 (3030) Charcoal OxA-35188 1490±27 BP calAD 475–643
TIN005 From pisé wall Wood OxA-35449 768±22 BP calAD 1222–1279
TIN015 Occupation layer, Trench 2 (1054) Wheat grain OxA-35192 1631±27 BP calAD 345–535
TIN015 Fill layer Trench 2 (1028) Charcoal OxA-33855 1624±25 BP calAD 382–536
TIN015 Occupation layer Trench 2 (1053) Barley grain OxA-35190 1607±29 BP calAD 396–538

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

TIN015 Occupation layer Trench 4 (1051) Barley grain OxA-35189 1601±26 BP calAD 402–537
TIN015 Fill layer Trench 2 (1028) Wheat grain OxA-35191 1594±27 BP calAD 405–539
TIN033 From pisé wall Date stone OxA-35648 633±24 BP calAD 1287–1396
TIR006 From wall of qasr Plant remains OxA-32432 919±25 BP calAD 1030–1169
TIR008 Occupation layer Trench 1 (3116) Charcoal (cf. OxA-35649 1500±24 BP calAD 437–631
Tamarix sp.)
TIR013 Occupation layer Trench 1 (3119) Wheat grain OxA-35194 1604±28 BP calAD 399–537
TIR013 Occupation layer Trench 1 (3118) Wheat grain OxA-35193 1588±29 BP calAD 408–542
TIS007 Occupation layer Trench 1 (2067) Charcoal OxA-35650 911±23 BP calAD 1035–1184
TIS008 Metalworking debris Trench 1 Charcoal (cf. OxA-35196 1812±27 BP calAD 128–320
(2068) Tamarix sp.)
TIS008 Same sample as above Same sample as above OxA-35195 1795±29 BP calAD 133–326
ZAG001 Trench 2, context 3021 Date stone OxA-32372 792±25 BP calAD 1210–1275
ZAG001 Trench 1, context 3004 Date stone OxA-33856 688±26 BP calAD 1269–1387
 ZAG001 Same sample as above Charcoal OxA-33857 946±26 BP calAD 1026–1155
ZAG001 From mortar of S enclosing wall Charcoal OxA-33858 974±25 BP calAD 1015–1155
ZAG001 From mortar of citadel Date stone OxA-35651 908±26 BP calAD 1035–1203
ZAG002 From mortar of S enclosing wall Charcoal OxA-32521 989±26 BP calAD 991–1152
ZAG002 Trench 1, context 1007 (fabric of Charcoal OxA-33859 1113±27 BP calAD 878–995
mudbrick wall)
ZAG002 From mortar in W enclosing wall Charcoal (cf. palm) OxA-33860 975±26 BP calAD 1014–1155
ZAG002 From mortar of W wall of tomb Charcoal OxA-33861 992±26 BP calAD 989–1151
ZAG002 From secondary phase of Charcoal (cf. OxA-33862 972±25 BP calAD 1016–1155
S enclosing wall tamarix sp.)
ZAG002 From mortar of E gate Charcoal (cf. palm) OxA-33863 1023±26 BP calAD 971–1115
ZAG002 From tower in E wall Charcoal (cf. OxA-33864 983±26 BP calAD 995–1153
tamarix sp.)
ZAG022 Trench 1 (intrusive) Seeds (geraniaceae OxA-33865 Modern± BP -
family)
ZAG040 From pisé wall Charcoal OxA-33866 307±24 BP calAD 1493–1649

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

ZAG042 From pisé wall Date stone OxA-33867 205±24 BP calAD 1649– . . .
ZAG046 From pisé wall Plant remains (grass) OxA-33967 234±25 BP calAD 1639– . . .
ZAG048 From pisé wall Plant remains OxA-33868 128±24 BP calAD 1678–1940
ZAG048 From pisé wall Charcoal (palm) OxA-33901 195±29 BP calAD 1648– . . .
Wadi Noun
Adrar Zerzem Tumulus 2, burial Human bone Hela-14884 1430±35 BP calAD 565–660
Agwidir From underneath the building, Charcoal Beta-259285 Data missing calAD 900–1030
layer 21, STC 35
Agwidir From underneath the building, Wood Beta-259286 Data missing calAD 1000–1160
layer 21, STC 35
Agwidir From wall construction, STC 35 Wood Beta-259287 Data missing calAD 1020–1200
Agwidir End of occupation, layer 71, Charcoal Beta-259282 Data missing calAD 1030–1220
STC 35
Agwidir Wall of the tabiya Charcoal Beta-259290 Data missing calAD 1040–1260
 Table 6.1 (cont.)

Area/Site Context Material Reference Date Calibrated

Agwidir North-west tower Charcoal Beta-259284 Data missing calAD 1160–1280

Agwidir Roof of the stable, layer 59, STC 35 Charcoal Beta-259281 Data missing calAD 1170–1280
Agwidir Occupation of house, layer 43, Charcoal Beta-259283 Data missing calAD 1280–1410
STC 35
Agwidir From foundation trench of wall Charcoal Beta-259289 Data missing calAD 1440–1640
Agwidir Sondage A, layer 5 Charcoal Beta-259288 Data missing calAD 1480–1660
Foum Asca Layer at base of trench 1 Shell Beta-295799 2880±40 BP 1207–931 calBC
Foum Asca Layer at base of trench 3 Shell Beta-295800 1950±40 BP 41 calBC – calAD 129
Tuflit el-Harratin Well, STC41, sondage 1, silo 1, Charcoal Beta-259291 Data missing calAD 690–980

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

context 2
Tuflit el-Harratin Furnace, next to copper ore, slag Charcoal Beta-259291 Data missing calAD 710–980
and glazed lamp, STC41, son-
dage 2, context 15
Tuwat
Between Tamentit and Wall of qasr Wood Hv-9702 600±55 BP calAD 1286–1421
Adrar, Complex 1
Between Tamentit and Wall of mosque Wood Hv-9703 470±65 BP calAD 1310–1632
Adrar, Complex 2
 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 249

were sent in succession over the Atlas.11 These expeditions took place 60
years after the famous campaigns of Balbus against the oases of the Ziban,
Ghadamis and Fazzan, but seem to conform to a similar pattern. Although
some commentators have argued that the Roman troops were pursuing
rebels from north of the Atlas southwards, Dio’s comments indicate that
their opponents were familiar with the desert environment and conditions.
Although they have generally been assumed to have been nomads,12 just as
has proved the case with the peoples of the Ziban and Fazzan, it is plausible
that they included oasis dwellers. The corollary of the campaigns is that the
account in Ptolemy’s Geography relating to interior Africa south of the
Atlas contains reference to numerous settlements, some of which are
specifically designated as ‘towns’.13 It would be hazardous to attempt to
reconstruct the Saharan world in detail from Ptolemy’s lists of names and
co-ordinates; indeed the geographical errors and imprecisions are well
established.14 Ptolemy’s account of the Sahara, although clearly geographi-
cally flawed by a tendency to join the dots between widely separated
phenomena, nonetheless indicates that Rome had acquired knowledge of
a large number of peoples, rivers, lakes and significant settlements deep in
the Sahara. In particular, he mentions two ‘great rivers of the interior’ the
Gir and the Niger and it is clear from the text that these were at least
seasonally active streams. These major wadis were evidently a focus of
population including ‘urban-scale’ settlements at some locations. The
logical conclusion is that this refers to some of our major oases zones –
and indeed the Garamantian capital Garama (Jarma) is listed as
a metropolis along with two other locations: Nigira, on the north bank of
the Niger (probably the Algerian Wadi Jedi in this context) and Gira, below
the Gir river (Wadi Gir/Saouara).15
In the Medieval and Early Modern world, the oases of the Western
Sahara were among the most prominent, populous and influential, with
a profound importance for Saharan trade and for the security of the
Muslim states of the Maghrib.16 The Bilad as-Siba, land of disorder, as
11
Pliny, Natural History 5.14–16; Dio, Histories 60.9; de la Chappelle 1934, for general discussion.
12
De la Chappelle 1934, 110.
13
Ptolemy, Geography 4.6; cf. Martin 1908, 30–31 who identifies 17 names in Ptolemy’s account
with modern Berber toponyms. The plausibility of specific name suggestions seems less
significant than the overall observation that Ptolemy’s account reproduces Berber toponyms
and attributes them to a Saharan locale, where the only urban-scale sites must have been oases.
14
On the toponyms, population groups and Roman knowledge of the western Sahara, see
especially Desanges 1962; 1978; 1980.
15
Ptolemy, Geography 4.6.
16
On the rising importance of the trade connections of the Western Sahara in the Medieval era,
see inter alia, Lydon 2009; MacDonald 2011; Mauny 1978; Vanacker 1973.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 250 Martin Sterry, David J. Mattingly and Youssef Bokbot

the mountainous and desert south of Morocco was known, was a recurrent
source of insurrection against the Bilad al-Mahkzan, land of order, com-
prising the northern plains of Morocco.17 Yet it was from the Bilad as-Siba
and ‘especially the land of desert and palm oasis lying to the east of the
Atlas that almost every dynasty of Sultans rose to conquer and replace the
last’.18 Several factors help explain how the Western Sahara could be such
a potent threat to successive states in the Western Maghrib. Firstly, the
desert settlements were shielded to the north by the imposing Atlas ranges,
with limited passes over which armies could be taken and which are closed
at times in winter – making continuous control of the south by the powers
in the north extremely difficult. Even the French colonial regime in the
twentieth century chose to operate hegemonically through the intermedi-
ary ‘lords’ of the Atlas passes.19 The second factor is that the oases,
especially of the Draa and Tafilalat, are among the largest and most
productive in the Sahara and have supported substantial populations.20
Rohlfs reported hundreds of fortified villages in the Draa and Tafilalat
oases, with an alleged combined population of 350,000 in the mid-
nineteenth century, likewise Ibn Khaldun describes these regions along
with Figuig, the Gourara and Tuwat as each having hundreds of qsur or
villages.21 Even if a little inflated, this was a very large manpower reserve
and in combination with additional pastoral groups of the desert, could
pose a potent military threat, albeit that they were as often engaged in
internecine warfare with each other.22 The Almoravid, Almohad, Merinid,
Saadian and Alouite dynasties all originated, or gained critical strength, in
the oases of Draa and Tafilalat.23
Transposing this sort of pattern into the remoter past represents another
thing altogether, of course. However, if pre-Islamic oasis settlement and
Saharan trade can be demonstrated, it is plausible that the same combination
of their relative isolation, the mass of the oasis populations and the devel-
opment of trade in the Moroccan Sahara had an influence on the pattern of
Roman provincial government in Mauretania Caesariensis and Tingitana.
We have already noted the evidence of Roman campaigns beyond the Atlas
as a sequel to the annexation of the Mauretanian kingdom in the AD 40s. Yet
most commentators on Roman Morocco have concluded that the mountains
17
For the best and most detailed account of the history of the Moroccan Sahara, see Jacques-
Meunié 1982. For the explanation of the Bilad as-Siba, see Maxwell 2004, 29–31; Dunn 1977.
18
Maxwell 2004, 30. 19 Maxwell 2004. 20 Joly 1948.
21
Levtzion and Hopkins 2000, 325; Rohlfs 2001, 56, 174.
22
On the inter-oasis raiding between Draa, Tafilalat and Guir, see Landau 1969, 92; Rolhfs 2001,
168–69, 178.
23
Jacques-Meunié 1982; Lugan 2000.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 251

were essentially a barrier to the south and, if Rome had occasional problems
with indigenous peoples thereafter, it is commonly assumed that these were
montane groups.24 On the other hand, it is striking how the geography of
Roman Tingitana reflects some key aspects of the later Islamic territory. The
triangular northern tip of the country is defined by two main routes running
south and south-east from Tingi/Tangiers. In the Roman period, the western
coastal route ended at Sala/modern Rabat, while the eastern inland route led
to Volubilis at the foot of the Middle Atlas, in relatively close proximity to
successive Islamic cities of Moulay Idris, Meknes and Fez. All of these south-
eastern sites are located in a position to serve as a western terminus of a land
route from the rest of the Maghrib to the east of the Mouloya river, but they
were also the northern end points of routes across the High Atlas to the
Tafilalat. The most notable difference between the Roman and Medieval
periods is that in the latter era further large cities were established to the
south, such as Marrakech and Sijilmasa.
To approach this from another direction, the town of Volubilis was an
extraordinary urban creation, and one already well established under the
Mauretanian kingdom that preceded the Roman province. Traditional
accounts have not satisfactorily explained the reasons for its prominence
in this inland location. The possibility that this was a centre from which
connections were maintained with both mountain and desert peoples to
the south and south-east bears serious consideration.25 As we shall see, the
current evidence for oasis development in this part of the Sahara suggests it
may have begun a little later than the central Saharan zone dominated by
the Garamantes.
The prime aim of this chapter is to assess whether any evidence exists to
demonstrate early oasis development in the Western Sahara. There has
been little archaeological research on these oases hitherto, but we present
below some significant pointers to their initial development by the early
first millennium AD. The order of discussion below will start in the north-
east and move gradually south-west along the southern flank of the Atlas
mountains. The following sections will then turn to other oases of the
western Algerian Sahara, the far south of Morocco and Mauritania.
Although the arguments we adduce for development of these oases suggest
24
For traditional accounts of Mauretania, see Carcopino 1943; Châtelain 1944. On the frontier,
see Euzennat 1989; Mattingly et al. 2013, 40–62; Rebuffat 1999.
25
The famous altars of peace erected at Volubilis, celebrating treaty relations between Rome and
a people known as the Baquates has led to much speculation, but no great clarity, as to the
whereabouts of the lands of the Baquates. See Rebuffat 1999, 276–82; Shaw 1987. The possibility
that the geo-political context of these treaties spread beyond the Atlas mountains has not been
seriously considered.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 252 Martin Sterry, David J. Mattingly and Youssef Bokbot

a somewhat later date than is supported by the evidence presented above

for oases further east in the Sahara, we would emphasise that the hydro-
logical conditions in the Moroccan Sahara were well suited for an early
evolution of oasis farming. The volume of water running off the High Atlas
mountains into the desert lands bordering their southern and eastern
flanks not only directly supports oasis cultivation, but also recharges the
water table in these areas.

Wadi Gir/Wadi Saoura

To the west of Laghouat, at the west end of the Saharan Atlas, there is
a group of oases that lie c.250 km south of the Roman frontier forts of
Mauretania Caesariensis. These are Moghrar, Boussemghoune, Ain Sefra,
Figuig, Ich and Bechar, and they are known for the remains of fortified
villages of presumed Medieval date. There are also foggaras known at all
these sites except Ain Sefra.26 What is particularly intriguing about these
oasis clusters is that the route that connects them is a key pass through the
Saharan Atlas, which, if projected north, passes close by the westernmost
forts of the Roman frontier of Mauretania Caesariensis at Numerus
Syrorum and Pomaria.27 To the south-west, the route follows the Wadi
Zousfana which skirts the north-western edge of the Great Eastern Erg to
merge with the Wadi Gir and form the Wadi Saoura, carrying seasonal
floodwater from the Atlas ranges deep into the desert.28 The western side of
the Wadi Saoura comprises a series of rocky escarpments, while the east
side is defined by the limits of the sand sea. Remarkably, until the con-
struction of the modern dam at Jorf Torba on the Gir, floods were regularly
carried down this normally dry valley as far as the Tuwat oases (see below),
a total distance of c.1,000 km. Ptolemy’s description of the Gir states that it
was one of two major rivers of the Sahara and that it united ‘as it were, the
Garamantian defile and the Usrgala mountains’.29 If this was the Gir/
Saoura, as seems entirely plausible, we might understand this comment
as denoting a river valley running for a great distance south-east from the
Atlas, in the general direction that the Romans knew the Garamantes lay,
rather than absolutely to their door. Ptolemy also lists a settlement called
Gira as a metropolis and numerous other poleis on the north (left) bank of
the Gir, which would correspond well topographically with the qsur along
26
Châtelain 1896; Heywood 1926, 94–100; Homo and Hugonot 1953; Remini et al. 2014.
27
Laporte 2011 provides a useful overview of the western area of Mauretania Caesariensis, but
does not consider the potential existence of oasis communities.
28
Gautier 1970, 77–87. 29 Ptolemy, Geography 4.6; Stevenson 1991, 105–6.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 253

the Souara in Tuwat. That some genuine information underlies Ptolemy’s

notion of rivers in the desert is confirmed by his references to rivers that
disappeared at the surface, but which continued to run underground and
to these desert rivers ending in seasonal lakes (chotts).
Even when surface flow is limited, the transference of sub-surface waters
down this wadi is significant. Looking down the course of the Wadi Gir/
Saoura today, it is evident that numerous small oases have been developed
along it. Some of these lie close to significant cemeteries of pre-Islamic
burial monuments, some of exceptional scale or decorative features (as at
Jorf Torba, see below). The irregular floodwaters were also augmented by
the use of artificial means. Between Figuig and Adrar at the north of the
Tuwat oases are a number of small oases that lie on the banks of the Wadi
Saoura, many of which are foggara fed. Settlements and hillforts are visible
on the escarpments overlooking Mazzer and Taghit which appear to
encompass a range of periods. At Igli, where the Wadi Zousfana and
Wadi Gir unite to form the Saoura, there is a low hillock, now much
damaged, on which a settlement has stood. Another hillfort is visible
opposite the oasis of Beni Abbas.
As noted already, the first Roman campaign in AD 41 crossed the Atlas
and proceeded to a river valley called Ger, and this can surely be equated
with the Wadi Gir.30 The second campaign clearly went further, with the
commander winning two major victories (presumably in two separate
oasis areas), but also passing through hostile and waterless terrain. The
headwaters of the Gir are in the High Atlas, close to the Algerian border of
Morocco, but the Roman struggle to find water at a crucial point in the
campaign suggests that this column marched a long way down the Gir/
Saoura towards Tuwat. The miracle of a rainstorm that allegedly came to
the rescue of the second Roman army may well be a literary trope, but it
might equally be an allusion to the sudden and astonishing appearance of
a flood in a wadi deep in the Sahara.31 The connectivity between the Atlas
and the Tuwat oases along the Gir/Saoura is important because it would
more or less guarantee that oasis technologies and crops were shared along
this line once they were present at either of the ends.
An example of pre-Islamic connectivity comes from the distribution of
burial monuments with ‘funerary chapels’ (tombes à chapelles) along the

30
Dio, Histories 60.9; Pliny, Natural History 5.14–16; Ptolemy, Geography 4.6.
31
A flood of this kind took place most recently after the rains in November 2014. The Saoura
eventually branches, with part of its flow ending in a seasonal lake and a second branch running
on south to the Tuwat (Gautier 1970, 80–82), perhaps a distant echo of Ptolemy’s garbled
account (Geography 4.6) of the tributary streams, lakes and marshes associated with the Gir.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 254 Martin Sterry, David J. Mattingly and Youssef Bokbot

Gir, in the Tafilalat and the Draa. This has rarely been considered in the
context of sedentary oasis cultivators, yet equally, the default assumption
that these were tombs of pastoralists is unproven and the monuments are
clustered in areas that were suited to oasis development.32 Of particular
interest are a corpus of carved and painted stele that were found in
association with a number of tombs at Jorf Torba on the upper Wadi
Gir.33 The images include many horses and horsemen (something also
paralleled on two early-mid-first millennium AD settlement sites in the
Wadi Draa), people associated with a Libyan inscription and most notably
a scene with six people represented frontally, including a man and a boy
armed with spears and four women, two bearing small ‘crosses’.34 All the
figures evidently wore woven garments, some with geometric patterning.
As we shall see, there is now a parallel for this sort of decoration in a tomb
in the Draa in an area where we do seem to have evidence of pre-Islamic
oasis development. An approximate dating would place these images
somewhere in the late antique era.

Tafilalat (Erfoud and Sijilmasa)

The Tafilalat is the second most extensive oasis in Morocco, after the Wadi
Draa (see below), with c.1,000,000 palms in the 1950s.35 This area, centred
on Erfoud, utilises both seasonal run-off waters from the mountains and
some limited groundwater sources to support an extensive oasis zone.36
Although there is no perennial flow from the Atlas into this basin, a broad
extent of plain land is irrigated by a combination of a low volume flow in
part of the year (at most 5–6 months), augmented by the exploitation of
violent seasonal floods spilling out from the mountain front. There are two
major seasonal rivers here, the Wadi Ziz (Sis) and the Wadi Rheris
(Gheris), along with a number of smaller tributaries. Although sometimes

32
Bokbot 2019.
33
See Camps 1984, 208–12, especially figure 10; Lihoreau 1993; Reygasse 1950, 104–8, figures
158–161. Radiocarbon dating on three of these monuments place them in the late first
millennium BC/early-mid-first millennium AD.
34
These crosses have conventionally been seen as possibly Christian crucifixes, Camps 1984,
209–11, but they could just as plausibly be something more prosaic such as an item of female
equipment to match the arms of the males – could these be spindle whorls and distaffs depicted
as they would be carried? See Lihoreau 1993, 109–15 for a detailed description of the painted
slabs recovered from Tomb 5 within an extensive cemetery. The slabs seem to have been placed
in the inner recess of a ‘funerary chapel’ attached to the east side of Tomb 5 (1993, 70–74). The
other painted slabs are believed to derive from two further large ‘tombes a chappelles’, nos 1 and
6 (1993, 24–25, 27–32).
35
El-Khoumsi et al. 2017, 46. 36 El-Khoumsi et al. 2017; Gaucher 1948; Ricard 1925, 377–80.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 255

used in the more geographically limited sense of one of the sub-groups of

oases in this plain, the Tafilalat also functions as the collective name for all
the oases here. The Tafilalat oases exploit the wide alluvial deposits
dumped in front of the mountains, into which the wadis have incised
channels, making it more difficult to divert water out of the wadis onto
the neighbouring plains.37
The upper courses of the two wadis contain a number of smaller oases,
fed by more or less perennial flow and springs. In the Tafilalat proper,
starting from the point where the wadis debouche from gorges into a wide
plain, the hydrology becomes more complicated and difficult. The irriga-
tion largely depends on control of the floods and the diversion of waters
into canals and a large artificial channel (the ‘Ziz filali’), which has effec-
tively diverted the main flow of the Ziz into a man-made system. The Wadi
Rheris eventually joins the Ziz in the southern Tafilalat, but the two wadis
run in parallel a comparatively short distance apart (c.30 km average) for
a considerable distance through the oasis belt. As the bed of the Ziz is
10–12 m higher than the Rheris, most of the irrigation water is taken from
the Wadi Ziz. However, at one point the wadis come within 2.5 km of each
other and given the strength of some of the peak flood incidents, there is
a serious danger that the Rheris could at some point capture the Ziz much
higher up, at a stroke depriving about 75 per cent of the oasis zone of its
irrigation.38 Foggaras have also been constructed at some point in time in
considerable numbers in the Tafilalat to supplement the water in the wadis,
with large numbers identified on the west side of the Wadi Rheris (see Fig.
1.4d), but others also to the east of the Ziz.39
The oasis comprises a number of distinct zones. Along the Wadi Ziz
from north to south are the oasis districts of Medarhera, Reteb, Tizimi,
Erfoud, Rissani/Tafilalat; along the Rheris are the palmeries of Goulmina,
Tiliouine, Touroug, Fezna and Jorf. Beyond Tafilalat, the two wadis com-
bine as the Wadi Daoura.40 In the 1860s, Rohlfs listed 150 fortified villages
spread across the various palmeries of the Tafilalat.41 From Sijilmasa,
routes ran south-west to the Draa at Zagora (c.200 km) and north up the
gorge of Wadi Ziz to the Atlas passes of Tizi n’Talghemt and Tizi N’Zad,
leading across the mountains to Fez.42
37
Gaucher 1948; Jacques-Meunié 1982, 146–54 for a detailed explanation of the topography and
hydrology of the Tafilalat.
38
Jacques-Meunié 1982, 150–52. Much effort has been expended in modern times to try to avert
this danger.
39
Castellani 1999, 231–38.
40
Rohlfs 2001, 59–61, 163–75, for descriptions of the Tafilalat in the mid-nineteenth century.
41
Rohlfs 2001, 174. 42 Jacques-Meunié 1982, 400–1 (map).

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 256 Martin Sterry, David J. Mattingly and Youssef Bokbot

Sijilmasa
Little is known about the earliest archaeology of the oasis, with most attention
directed to the study of the celebrated Islamic site, Sijilmasa.43 The legendary
account of al-Bakri specifies that the town was founded in AD 757–758, by the
Bani Midrar following up on a tip from one of a group of pastoralists who had
allegedly hitherto only used the potential of the location for pasturing their
flocks.44 This traditional account of an Arab foundation of the oasis and
canonical dating is still followed in much recent work, despite the fact that
excavations and a series of radiocarbon dates suggest that the activity at the
location of Sijilmasa prior to the mid-eighth century may have amounted to
more than some seasonal pastoral visits.45 Beneath the main mosque at the
centre of Sijilmasa, there was an elite residence and beneath that a hard-packed
floor and hearth which has been dated to calAD 138–527 (2-sigma). A further
sample from beneath the mosque has a date of calAD 413–863 and excavations
to the west of the mosque dated another fire pit to calAD 545–863.46 The logic
of the three dates is that Sijilmasa existed as a site prior to the mid-eighth
century, though the character of its early occupation requires further investi-
gation. There are certainly some substantial pre-Islamic cemeteries in this area
and a number of hillfort sites in the periphery of the oasis again hint at pre-
Islamic origins of sedentary settlement.47 It is quite plausible that the Islamic
foundation myth has served to misdirect us from recognising a pre-existing
oasis farming community here.

Rock Art and Protohistoric Settlement

Recent studies by Yves and Christine Gauthier have mapped the location of
chariot representations across the Saharan regions, with a particular cluster
in southern Morocco.48 The very densest concentration of chariot imagery
43
For the history, cf. Jacques-Meunié 1982. For the archaeology, see Lightfoot and Miller 1996;
Messier 1997; Messier and Miller 2015.
44
Levtzion and Hopkins 2000, 65 for al-Bakri’s classic account of the foundation and the state of
the town in the mid-eighth century. See Capel, Chapter 16, this volume for a full description
surrounding the founding of Sijilmasa.
45
Messier and Fili 2011, 129 present the new dates, but stick with the orthodox interpretation.
This probably needs revisiting in light of the information emerging from the Draa (see below).
See also Messier 1997, 81–82.
46
Messier and Fili 2011, 129.
47
On the cemeteries, Bokbot 2005; 2019; on the hillforts, see Capel, Chapter 16, this volume.
48
Gauthier and Gauthier 2011. Additional work, reported on in Gauthier and Gauthier 2015
(with map at 69), adds further weight to the Moroccan cluster and many of these discoveries are
in the southernmost mountains, bordering the open desert and relatively close to the major
Moroccan oases.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 257

occurs at three closely linked sites about 30 km south of the Tafilalat on the
Wadi Ziz at a place known as Taouz, a topographic control point for people
moving towards the oasis from the south. The number of individual chariot
images recorded here is an unprecedented 323 (22 per cent of the total of
1,424 scenes in their database for the entire Sahara).49 The vast majority
were carved on rocks alongside the defended southern side of a hillfort that
controlled a bend in the Wadi and from where extensive views of people
approaching could be had.50 The other chariot engravings at Taouz were
closely associated spatially with some very large pre-Islamic tombs.51 While
the contemporaneity of hillfort (with a mix of rectangular and sub-
rectangular buildings) and the large tombs with the engravings is not
certain, the geographical co-location is very suggestive as is the likelihood
that the people here were at last partly sedentarised. Although there are no
date palms here, there is some modern cultivation of the wadi alluvium,
especially in the less volatile side channels. Overall, Taouz could be viewed
as a portal from the desert into the main oasis zone to the north. There are
certainly reasons to believe that spread of horses and chariots was closely
linked with the establishment of oasis communities and trade networks.52

Wadi Draa
New information on the Draa valley c.200 km to the west of Tafilalat adds
weight to the argument for Protohistoric origins. The Wadi Draa is one of the
most extraordinary and largest of all Saharan oases, spread along a 900 km
long valley, with an estimated 2,000,000 date palms.53 The source of the wadi
lies in the High Atlas above Ouarzazate and for a substantial part of its initial
course, running broadly north-west to south-east, the wadi is a perennial river,
though of highly variable flow across the year and between years.54 In Roman
times it allegedly (but plausibly) still had a population of crocodiles.55 South of

49
Gauthier and Gauthier 2015, 13–32.
50
Gauthier and Gauthier 2015, 13 and 69 for maps and 14–18 for discussion of location.
51
Gauthier and Gauthier 2015, 18–27.
52
See further, Mattingly 2017, 22–24. This is the same site called Jabal Afilal and illustrated in
Figures 16.4–16.6 by Capel, Chapter 16, this volume.
53
Jacques-Meunié 1982, 142–46 (for a detailed explanation of the topography and hydrology of
the Draa); Ricard 1925, 300–1 (brief and lacunous, reflecting poor state of knowledge at that
date); Richet 1920 (covers the lower Draa only); Zaïnabi 2004, 30 (for the number of palms).
54
Jacques-Meunié 1982, 145.
55
Pliny, Natural History 5.9: ‘flumen Darat in quo crocodilos gigni’; cf. Desanges 1980, 114–15,
117–18 for commentary. Crocodiles have been reported in the lower river Draa as recently as
1950 – Monteil 1951. See Brito et al. 2011 for a general discussion of Saharan crocodiles from
the Holocene to the present.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 258 Martin Sterry, David J. Mattingly and Youssef Bokbot

Ouarzazate the river runs through a narrow defile before opening into a broad
valley bordered by mountains on either side. This section of the valley (the
Middle Draa) is divided into a series of oasis zones: from north-west to south-
east these are Agdz/Mezguita (30 × 1.5 km and 16 × 0.5 km palmeries),
Tinzouline (20 × 1.5 km), Ternata (20 × 5–10 km), Fezwata (30 km × 3 km),
and Ktawa (20 × 6–8 km).56 Only after making a turn to run east to west (the
‘coude du Draa’), does the wadi finally disappear underground beyond the
final major oasis zone of Mhamid (20 × 4 km). The dates of the Wadi Draa
have a high reputation and the productive capacity is large.57
To the west, there are sporadic palm stands on the many tributaries (but not
on the main channel) along its course to the Atlantic seaboard.58 Only in the
most exceptional flood conditions do the waters from the high Atlas get carried
all the way down the Draa to the Atlantic.59 More commonly, the conditions
along the lower Draa are not conducive for sedentary settlement.60
The Agdz to Mhamid sector of the valley thus constitutes the Draa oasis
proper, but it is by no means uniform. Year-round flow has tended to reach
only as far as Zagora and then rather sluggishly. Irrigation in the Draa
valley depends on diverting water from the river when the water is high
into canals that feed garden areas laid out parallel to the wadi downstream
of the diversion dam (Fig. 1.4a/b). The upper sections of the Middle Draa
have thus traditionally been much better served by water than the region to
the south-east of Zagora. In average years the main irrigation for Fezwata,
Ktawa and Mhamid came from three or four extended flood incidents of
15–20 days duration, when a substantial flow was carried beyond Zagora.
In poor years, when only one or two floods of a brief duration have
occurred, this did little more than keep alive the palm groves and ruled
out cultivation of cereals.61 Irrigation has been supplemented by wells, but
there are only limited traces of foggara systems in the Draa.
A singular aspect of the topography of the Middle Draa is provided by the
steep escarpments of Jabal Bani, which from certain directions limited access
into and out of the wadi. Jabal Bani runs west to east south of and parallel to
the Anti-Atlas range, with two branches crossing the Draa valley and dividing
it into distinct basins. On the east bank of the Draa the Jabal Bani turns and
runs north-west. The presence of Jabal Bani thus accounts for the strange
56
Rohlfs 2001, 53–59; Zaïnabi 2004, 17–24. 57 Rohlfs 2001, 55.
58
Jacques-Meunié 1982, 143–44; cf. also Richet 1920; Riser 1996.
59
Jacques-Meunié 1982, 144 records that this happened in two exceptional floods in 1949 and
1950.
60
See Tayler 2003 for a revealing account of the rigours of journeying down the Draa from near
Ouarzazate to the Atlantic, with the difficulties escalating significantly beyond Mhamid.
61
Jacques-Meunié 1982, 146.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 259

dog-leg course of the Wadi Draa – once it breaks through the southern
branch of Jabal Bani at Foum Larjam, the wadi turns west along the south
side of the range towards the Atlantic.62 The principal routes out of the valley
to the east run north-east from Zagora to Sijilmasa in the Tafilalat, north-
west of Agdz across the mountains towards Ouarzazate and Marrakech
across the High Atlas passes, south-east from Foum Larjam towards
Tabalbala. At two points along its course, the valley narrows and the river
flows through defiles or ‘passages obligés’ in the Jabal Bani, at the Imi n’Takat
pass between the Fezwata and Ktawa, and at Tidri controlling the key pass
between Ktawa and Mhamid. An additional significant topographic control
point in the valley is marked by Jabal Zagora, an imposing mountain that
dominates the valley at the division between the Ternata and Fezwata oases.
A huge area of the Middle Draa (where the flow is north-west to south-
east) is irrigated by diverting water from the river into irrigation channels
and gardens created along its course. There is an abundance of archaeology
visible on satellite imagery in the wadi: hillforts, fortified settlements,
extensive cemeteries of pre-Islamic tombs, and so on (Fig. 6.2). Since
2015, we have been conducting fieldwork along this section of the Draa,
researching the abundant evidence of pre-Islamic cairn cemeteries and
abandoned hillforts and villages. The areas of the oasis that are still
cultivated are studded with hundreds of minor villages, mostly of the
fortified qasba type and probably dating to the last few hundred years.
The earliest settlements within the oasis belt itself are buried or masked by
the later phases of settlement and cultivation, so our initial survey work has
concentrated on structures that lie at or close to the fringes of the palm
groves, often on the higher ground bordering the oasis.
We have obtained a number of radiocarbon dates that attest to sedentary
settlement in the first half of the first millennium AD. Ten hillfort sites and
one oasis site date to the first half of the first millennium AD, with evidence
of numerous fragments of rotary quern stones and a distinctive cordoned
pottery tradition (Fig. 6.3).63 The sites are all associated with pre-Islamic
burials (in three cases the cairns overlie the settlement) and in two instances
also with horse-dominated rock art scenes. At three of the sites we have
recovered wheat and barley grains that are dated to the same period.64
A classic example of a Protohistoric hillfort is TIN001, situated on top of
the Foum Chenna cliff face that holds a dense concentration of rock art and
62
Jacques-Meunié 1982, 116–17.
63
Mattingly et al. 2017; 2018. The hillfort sites are LAR002, LAR005/6, TAK006, TAM001,
TIN001, TIN015, TIR008, TIR013, TIS007 and TIS008, the oasis site is TAG014.
64
Sites TIN015, TIR013 and LAR002.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 260 Martin Sterry, David J. Mattingly and Youssef Bokbot

Figure 6.2. Protohistoric settlements and funerary zones in the Wadi Draa.

Libyan inscriptions (TIN012).65 The hilltop site was protected by a series of

substantial drystone walls, each with a single gateway. The walled core of the
site comprises a series of enclosures and c.8–10 buildings leading off a central
alley that led to the highest point of the hill, which was left empty apart from
a cairn at the top. There are at least two phases of construction visible: the
earliest consists of loosely coursed boulders and the second of drystone slab
masonry. A number of horse engravings were found built into the structures
of the settlement implying a close relation between the rock art and the people
of this site. Another nearby site (TIN015) shares some similarities of layout,
with a lower outer enclosure (that contained a few vestigial traces of struc-
tures), a middle zone entered through a gate in a second wall, where most of
the habitation seems to have been concentrated. A central road led up through

65
See Pichler 2000; Simoneau 1972a/b.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 261

Figure 6.3. Comparative plans of protohistoric sites in the Wadi Draa.

the settlement to a third area on the highest part of the site that was empty of
structures, but which was covered in rock art on natural rock surfaces and
many boulders. The vast majority of the images relate to horses (or to crudely
engraved quadrupeds which are thought likely to represent horses). Once
again, horses (both riderless and with riders), were engraved on blocks built
into the walls of dwellings as well as natural boulders and bare rock surfaces.
The highly structured placement of the rock art imagery in relation to
both these settlements strongly supports the view that the two activities were
at least in part contemporaneous. Both hillforts have produced pre-Islamic
AMS dates (see below) and TIN015 has yielded both wheat and barley grains
of fourth- to sixth-century AD date. The sites share a consistent handmade
ceramic assemblage that is characterised by many cordon decorated jars and
an absence of wheel-made or glazed material. Rotary querns are another
conspicuous find on these sites. Overall the indications are of a society that
was in part at least sedentarised and part of or closely linked to an incipient
oasis community. Although no chariots have been noted here, the emphasis
on warriors on horseback finds parallels with the Garamantian heartlands
and a close association between the spread of oases and horses is probable.66

66
For Garamantian horse (and chariot) imagery from key topographic locations, see Mattingly
2003, 287, 297, 314–17; 2007, 101–2, 146, 192–93.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 262 Martin Sterry, David J. Mattingly and Youssef Bokbot

Another interesting hillfort site (LAR002) lies in the region of Foum

Larjam, just on the west side of the point where the Draa breaks through
the final ridge of hills to enter the open desert and make its turn to the west.
This site has so far yielded five dates that indicate occupation from at least
the fifth to the eighth centuries.67 To the south and south-west of this
settlement on the opposite bank of the Draa there is a large area of
abandoned gardens, irrigation canals and fortified settlements (qsur)
built in mudbrick rather than the ubiquitous pisé of the early modern
qasabas. Initial dating of these sites seems to indicate they were constructed
in a major phase of expansion of the oasis in the eleventh to twelfth
centuries AD. However, two dates from an area with clay-lined pits and
building foundations (BOU054: calAD 646–765 and calAD 650–765)68 are
contemporary with barley cultivation at LAR002: calAD 612–67969 while
one of the qsur has a slightly later date of the eighth to tenth centuries.70 On
the same side of the wadi we have identified a group of large tumuli with
painted funerary chapels, providing a very close parallel to those excavated
at Jorf Torba, c.400 km to the north-east.71 Radiocarbon dates from one of
the painted tombs in the Draa suggests it was built in the fourth
century AD or earlier.72 These dates from the terrace on the south side of
the wadi, where later oasis gardens were established, is a strong hint that
there was also pre-Islamic farming activity on this land (whether proto-
oasis cerealiculture or fully developed palm oasis we cannot say at this stage
of research).
There are dozens of these Protohistoric settlements identifiable along
the Middle Draa and the nature of the settlement is strikingly similar to
early oasis farming sites, like Zinkekra in Fazzan, but evidently originating
about 1,000 years later. There are large cemeteries of pre-Islamic drum
cairns and corbelled cairns associated with Foum Larjam.73 A fragment of

67
OxA-33826, 1676±28 BP, calAD 260–422; OxA-33827, 1638±27 BP, calAD 339–534; OxA-
33825, 1616±27 BP, calAD 389–536; OxA-33829, 1377±27 BP, calAD 612–679; OxA-33828,
1121±28 BP, 778–994.
68
OxA-35367, 1338±28 BP, calAD 646–765; OxA-35366, 1318±27 BP, calAD 650–765.
69
OxA-33829, 1377±27 BP, calAD 612–679. 70 OxA-33784, 1196±32 BP, calAD 713–945.
71
Painted scenes had been previously photographed from inside one of the large tombs (Zaïnabi
2004, 40), but we were able to excavate a second funerary chapel in 2016. Many of the images are
similar to the well-known scenes from the late antique tomb of Jorf Torba (above).
72
Use of the painted tomb BOU057.T1 is dated to calAD 92–320 (OxA-35403, 1820±29 BP) and
calAD 230–381 (OxA-35185, 1751±27 BP), contemporary with, or slightly earlier than the
nearby hillforts and similar tombs from Jorf Torba (above).
73
The cemeteries of Foum Larjam are also important because some tombs were excavated in the
1950s, (Jacques-Meunié 1958). The recent radiocarbon dates suggest that these pre-Islamic
burial monuments were constructed from the early first millennium BC until late antiquity.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 263

bone on the surface of one of the large tombs c.1 km south of LAR002 has
produced an AMS date of 894–791 calBC, though for the moment this date
is far earlier than other Protohistoric dates.74 Two hillfort sites have yielded
evidence of copper working, perhaps in the early centuries AD, and the
mineral resources of the region are another important factor that may have
underlain the Protohistoric settlement in the region.
While the literary evidence and the early burial mentioned above may hint
at origins in the first millennium BC, the AMS dates from the Draa at present
highlight a significant phase of development of settlement associated with
oasis agriculture in the third–sixth centuries AD. One point to bear in mind,
though, is that to this stage most of our investigation has taken place in the
central and southern areas of the Middle Draa valley, whereas the part of the
Draa that received the highest and most reliable water supply was the upper
half of the valley closer to Agdz. That is perhaps the area where we may
ultimately find the earliest oasis development in Morocco.75
There seems to have been subsequent and very significant growth in the
oasis in the Medieval period, especially in the vicinity of the major control
points along the valley, as at Zagora, which was a major town in the
eleventh century. Other urban-scale settlements of Medieval date have
been recorded at Imi n’Takat and Tidri and there are further areas of
abandoned oasis close to both of these sites, with traces of gardens, irriga-
tion canals and settlements. Certainly, this area (like the Tafilalat) has
a long and complex Medieval history too, but the key conclusion of the
new archaeological work here is that there were pre-Islamic origins for at
least some parts of the oasis.76

Wadi Noun and Tamdult

The Wadi Noun is the southernmost perennial stream in north-west
Africa, running east to west into the Atlantic beyond Guelmim. The area
was surveyed by Bokbot et al. who identified a range of archaeological
remains from Neolithic to present.77 Many cairn cemeteries were identi-
fied, including Adrar Zerzem where three tombs were excavated and dated
to calAD 565–660 and a number of examples of chariot rock art78 The

74
OxA-33338, 2649±28 BP, 894–791 calBC. A similar date was recovered from a tomb in the
vicinity in the 1970s – Gif-2912, 2369±250 BP, 1082–129 calBC.
75
A brief reconnaissance in January 2018 identified a number of Protohistoric and Medieval sites
in this region.
76
Bokbot 2019. 77 Bokbot et al. 2011; 2013a; 2013b.
78
Bokbot 2019; Bokbot et al. 2008; El Graoui et al. 2010.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 264 Martin Sterry, David J. Mattingly and Youssef Bokbot

Figure 6.4. Satellite image of Tamdult (imagery: Google, DigitalGlobe).

palmeries are in small clusters, often at the junction of wadis. Many of these
oasis areas have evidence of foggara cultivation and there are several
important Medieval settlements, most prominent of which is the tenth–
thirteenth century Agwidir fortress, potentially linked to Nul Lamta, the
site of a gold mint in the Almoravid period.79 Protohistoric settlement is
less visible than in the Draa, but may underlie some of the Medieval hilltop
sites. Along the Atlantic coast, just to the north of the Noun, there is
evidence of Protohistoric purple dye production at Foum Asca.80 Here
deposits of shellfish including stramonita haemastoma shells (from a type
of predatory sea snail), a key ingredient of purple production, have been
uncovered that date to the first millennium BC to first centuries AD.81
In the area between the Noun and the Middle Draa (and bordered to the
south by the lower Draa) there are many more small oases with small
palmeries often located where a wadi moves through a pass in the moun-
tains (thus concentrating the flow of water).82 These oases have seen little
exploration, but of particular interest is the early Medieval town of
Tamdult, which survives as a tell of some 16 ha that is located 13 km to
the south-west of Akka (Fig. 6.4).83 The ninth-century geographer al-
Yaqubi described it as dominated by a fort with deposits of gold and silver
nearby.84

79
Bokbot et al. 2013a; 2013b. 80 Onrubia Pintado et al. 2016.
81
Beta-295799, 2880±40, 1207–931 calBC; Beta-295800, 1950±40, 41 calBC–calAD 129.
82
Richet 1920. 83 Rosenberger 1970. 84 Levtzion and Hopkins 2000, 22.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 265

Oases of Western Algeria

Gourara, Tuwat and Tidikelt

The Gourara, Tuwat and Tidikelt oases (Fig. 6.5) are three closely related
zones of dense oases and are among the best researched for their irrigation
works (foggaras).85 The northerly cluster is Gourara, a roughly triangular
area of 200 km side centred on the oasis of Timimoun, set back about 50 km
east of the lower course of the Wadi Saoura. The oases in the western part of
Gourara are aligned along north-south valleys, but there is an east-west
spread along the southern edge of the Great Western Erg. Tuwat comprises
a north-south line of oases spread over c.140 km along the most southerly
sector of the Wadi Saoura, now with the Chech sand sea extending to the
west and the Tademaït plateau to the east. The Wadi Gir/Saoura finally
peters out in the sabkha flanking the west side of the Tuwat oases, c.1,000 km
from its source in the Atlas (see above, pp. 252–54). The Tidikelt is a plain
running for about 300 km west to east from Reggane at the southern end of
the Tuwat to In Salah. All three of these oases zones are notable locations to
find foggara irrigation and numerous fortified villages (qsur).86 Rohlfs
recorded the names of more than 175 villages and fortified sites across the
Gourara and Tuwat oases, recognising Timimoun and Tamentit as towns or
principal markets.87 In the 1880s, a French study of the Saharan frontier
identified the western Algerian oases as a particular threat, listing 258 qsur
(Gourara 164; Tuwat 60, Tidikelt 34).88 The total palms spread across
Gourara, Tuwat and Tidikelt in the late nineteenth century was recorded
as 1,429,652,89 and the population recorded by the French census around
this time was c.50,000, with 20,000 in the Gourara, 18,000 in Tuwat and
12,000 in Tidikelt.90 South of Reggane there is a long and difficult (but fairly
straight) north-south route that eventually reaches the Niger at Gao.
However, the better-watered and more travelled route in pre-modern
times crossed the Tidikelt, then turned south-south-east from In Salah
towards the Ahaggar.
85
All three oases groups are covered in Gabriel 1984; Gautier 1905; Martin 1908; Rohlfs 2001,
189–245; see also Gourara: Bisson 1957; 1989; 1999; Capot-Rey 1956; Cornet 1952; Echallier
1972; 1973; Tuwat: Daumas 1845; Capot-Rey and Damade 1962; Tidikelt: Chaintron 1957/
1958; Lehuraux 1943; Lô 1953/1954; Voinot 1909.
86
Capot-Rey 1953, 324–26 (1,520 km of subterranean channels); Cornet 1952; Echallier 1972;
Gabriel 1984; Lô 1953/1954.
87
Rohlfs 2001, 216–19. 88 Rinn 1886, especially the fold-out map.
89
Martin 1908, 301 (Gourara: 644,447 palms; Tuwat 453,490; Tidikelt: 331,715), but note that
c.720,000 of the total palms were immature or in declining yield or fallen out of production.
90
Gautier 1905, 18.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 266 Martin Sterry, David J. Mattingly and Youssef Bokbot

Figure 6.5. Distribution of fortified sites in the Gourara, Tuwat and Tidikelt oases.

The initial French investigations of the Gourara/Tuwat/Tidikelt area were

quite open to the idea that some of the hundreds of abandoned villages and
foggara-fed gardens might be pre-Islamic.91 Martin correlated a number of
toponyms in Ptolemy’s Geography (associated with the Niger rather than the
Gir) with the Tuwat.92 However, a lack of recorded finds of Roman material
culture and a failure to follow up some excellent work creating a typology of
the older settlements with radiocarbon dating has left the chronology

91
Echallier 1972, 51–59; Martin 1908, 25–55; Vallet in Nesson et al. 1973.
92
Martin 1908, 30–31. cf. Ptolemy, Geography 4.6, but if the Souara was in fact Ptolemy’s Gir
rather than the Niger, the following sites were associated with it: Gira, Thycimath, Geva,
Badiath, Ischeri, Turcumanda, Thuspa, Artagara, Rubune, Lynxama.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 267

floating. Echallier preferred a date of c.AD 1000–1500, and two radio-

carbon dates collected by Gabriel gave dates in this range.93 Many spe-
cialists have preferred to accept the idea of Islamic-era origins. The first
(certain) toponym to appear in the historical record is Buda which was
visited by Ibn Battuta on his way from Takadda to Sijilmasa in 1353.94
Similarly, the local traditions, preserved in written Arabic chronicles
recorded by the French in the nineteenth century, seem to indicate
a wave of village foundations from the tenth to thirteenth centuries.95
However, there is also a strong tradition of ‘Berber’ and ‘Jewish’ founda-
tion stories in this region.96 Gautier suggested that the initial migration of
Berbers and Jews pre-dated the Islamic conquest of the Maghrib, but that
it related to the Vandal or Byzantine period, rather than earlier.97 Many
of the abandoned qsur and foggaras in the late nineteenth century had
strong Jewish associations.98 A Hebrew inscription from a tombstone
found in the Buda region near Qasr Ghormali (north-east of Adrar)
records a date of 5089 AM corresponding with AD 1329 and there is
correspondence between the oases and rabbis in Algiers from the late
fourteenth and early fifteenth century that mention seven ‘wardens’.99
In terms of earlier evidence, the nearest pre-Islamic ceramic evidence is
a fragment of a long-necked Byzantine lamp that was recovered from Fort
Miribel 400 km to the east in the Tademaït.100 However, Wilson has
convincingly argued that the foggaras may predate the arrival of Arabs in
the region.101 Echallier has also noted that there is a strong correlation
between finds of Libyan inscriptions and his Type II qsur, again hinting at
a pre-Islamic date for some of the sites.102 Given that the Fazzan work has
now demonstrated that there are Garamantian antecedents and parallels
for virtually all the architectural forms that Echallier recorded in Gourara
and Tuwat, the lack of radiocarbon dates for these sites is critical. It is by no

93
Echallier 1972; Gabriel 1984, 104–7 – the dates are from two mudbrick sites between Tamentit
and Adrar, Hv-9702, wood splinters from mud mortar, 600±55 BP, calAD 1286–1421 and Hv-
9703, wood set in the masonry of a ‘mosque’, 470±65, calAD 1310–1632.
94
Levtzion and Hopkins 2000, 304.
95
Daumas 1845, 273–303; Echallier 1972, 51–52; Lehuraux 1943; Martin 1908, 49–59.
96
Gautier 1905, 24–28; Martin 1908, 37–47.
97
Gautier 1905, 30: ‘il semble que les Berbères ont peuplé les oasis à une époque postérieuer à
l’empire Romain . . . les souvenirs indigènes . . . placent la première migration au VI siècle à
l’époque des Vandales et de Justinian’.
98 99
Gautier 1905, 24–25; Oliel 1994. Gautier 1905, 25; Hirschberg 1963, 323.
100
Salama 1981, 514–15.
101
Wilson 2005, 232–33; 2006, 213–14; 2009, 30–32. See also, Wilson et al. forthcoming.
102
Echallier 1972, 56–58.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 268 Martin Sterry, David J. Mattingly and Youssef Bokbot

means implausible that systematic dating of these sites could backdate the
origins of some at least to the mid-first millennium AD or earlier. At any
event, the often assumed Islamic origin should not be considered as proven
on the currently available evidence.
Other evidence for pre-Islamic occupation of the oases is thin apart from
a number of Neolithic hearths.103 The only notable occurrences of cairns
from the region come from the rocky Tademaït plateau to the east of the
main oasis belt. All examples are highly dispersed and may be very early in
date. Rock art is similarly scarce despite large corpora known to the north-
east at Figuig and south in the Ahaggar. Rohlfs noted an abandoned hillfort
site midway between Karsas and Tsabit, associated with an area of foggaras,
rock-cut chambers (tombs?) and a Libyan inscription but this has not been
subsequently located.104

Tabalbala
Tabalbala is a small oasis some 240 km to the east of the southern tip of the
Draa and 300 km to the west of the Tuwat. The study by Champault makes
this one of the best recorded oases in the Sahara.105 One of the languages
spoken in the oasis is Korandjé, a form of Northern Songhay, a Sahelian
language. Souag suggests that these Songhay speakers moved into the oasis
from a sedentary community (perhaps located in the Azawagh valley,
Niger) with a mixed economy c.AD 1200.106 Like Tuwat there is also
evidence of a Jewish population from a tombstone dated to AD 1322 =
5082 AM and a single dated Arabic tombstone AD 1422 = 825 AH.107 To
the east of the modern settlement there is a small hill that is covered in
rectilinear buildings, but the most remarkable features here are the c.200
foggara channels that run from the sand sea towards the escarpment.
Either these required regular replacement or the oasis sustained
a substantially larger population than it does today (around 5,000).
Champault also mentions a copper mine in Djebel Ben Tadjine, 50 km to
the south-east of Tabalbala.108
Like the Tuwat, Tabalbala does not feature in Medieval sources until the
thirteenth century at the point when both oases became more important
stopping points between the Sahel and Morocco. Souag’s thesis of the
enforced movement of Songhay speakers (with expertise in irrigated agri-
culture and copper mining) in order to supply a crucial stopping point for

103 104 105 106

Gabriel 1984. Rohlfs 2001, 202. Champault 1969. Souag 2015.
107 108
Souag 2015. Champault 1969, 15.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 269

mobile Berbers (engaged in Trans-Saharan trade) ties up many loose ends.

However, the construction of the foggaras is no trivial matter and must
relate to the much larger concentration in the Tuwat rather than a Sahelian
connection. While these may be a later technological transfer, the possibi-
lity of earlier occupation cannot be ruled out.

Conclusion

At least some of the people of the desert zones surveyed in this chapter
were described as Gaetuli and characterised as pastoralists in the
Roman literary sources.109 As we saw in the previous chapter, there is
in fact a close correlation between people called Gaetuli and oasis zones
and this may in fact help explain the widespread occurrence of people
under this nomenclature – that is, the name is more related to a way of
life (probably a mix of oasis cultivation and pastoralism) rather than an
ethnicity. In Western Algeria and Morocco the evidence for
Protohistoric origins of the oases is most compelling for those that
lay closest to the Roman frontiers. The evidence from radiocarbon
dating currently suggests this occurred (or at least accelerated greatly)
in the first millennium AD – in other words at a somewhat later date
than the examples presented from the Central Sahara and the Roman
frontier lands further east. Nonetheless, we cannot rule out
Protohistoric origins of the oases further out into the Sahara, such as
Gourara, Tuwat and Tidikelt. As noted, the presence of foggaras and of
fortified sites that bear comparison with those of Garamantian Fazzan
should give us pause for thought until more extensive radiocarbon
dating is undertaken.
The detailed work carried out in the Wadi Draa currently provides
our best evidence of the process of sedentarisation and oasis formation
in the Western Sahara in Protohistoric times. The survey in the Draa
has mainly focused to this point on the sites visible on satellite imagery
on the escarpments flanking the river, rather than investigating in
detail the oasis zone of the lower wadi terraces. There are strong
resemblances between the hillfort sites flanking the Wadi Draa and
the Early Garamantian sites like Zinkekra. A key question is whether
these fortified sites, often located an hour or more walk from the valley
floor, were accompanied by lower lying sites like the ones that emerged

109
Pliny, Natural History 5.9–10; cf. Desanges 1980, sv Gaetuli.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 270 Martin Sterry, David J. Mattingly and Youssef Bokbot

in the Garamantian case. There are practical difficulties in undertaking

fieldwork within the active oasis gardens and in and around still
inhabited traditional villages – especially while the chronology of the
handmade pottery is still being established. Nonetheless, our research
in a few places where gardens have been abandoned has revealed traces
of Protohistoric activity and our model for further testing is that there
were oasis villages on the low-lying land within and adjacent to the
cultivated garden zone. The abandonment of some hillforts and their
conversion to funerary landscapes also echoes the pattern in Libyan
Fazzan and strongly hints that there are important sites still to discover
within the Draa oasis belt.
The development of the oasis in the Draa appears to have begun some-
what later than in Fazzan, with the current radiocarbon dates suggesting
a significant upsurge in sites and activity towards the middle of the first
millennium AD. Again that is a hypothesis that will require further testing
through future excavations, which will also hopefully provide fuller detail
of the agricultural package (and dates of appearance of key crops).
The key conclusion of this chapter, however, is that the Islamic era
florescence of the Moroccan oases, the rise of urban centres and of the
Western Saharan trade routes was not the result simply of a late first-
millennium AD sedentarisation of the north-western desert. Rather, this
needs to be understood in relation to prior Protohistoric developments
that were built on spectacularly following the Islamic conquest of the
Western Maghrib.

References

Bisson, J. 1957. Le Gourara. Etude de geographie humaine. Alger: Memoires de

l’Institut de recherches sahariennes 3.
Bisson, J. 1989. The origin and evolution of foggara oases in the Algerian Sahara. In
P. Beaumont, M. Bonine and K. McLachlan (eds), Qanat, Kariz and Khattara:
Traditional Water Systems in the Middle East and North Africa. London and
Wisbech: MENAS Press, 178–209.
Bisson, J. 1999. Gourara. Encyclopédie berbère fasc 21: 3188–98.
Bokbot Y., 2005. Hydrogéologie et protohistoire du Tafilalet, quelques aspects de la
recherche archéologique en milieu présaharien. In L’eau: Source de vie à travers
les âges. Jardin des Hespérides. Rabat: Bulletin semestriel de la société marocaine
d’archéologie et de Patrimoine 1, novembre 2004 – avril 2005, 42–44.
Bokbot, Y. 2015. Protohistoire du Maroc. Encyclopédie berbère fasc. 39: 6543–61.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 271

Bokbot Y. 2019. Protohistoric and pre-Islamic funerary archaeology in the

Moroccan pre-Sahara. In Gatto et al. 2019, 315–40.
Bokbot, Y., Onrubia-Pintado, J., Rodriguez-Rodriguez, A., Rodriguez-Santana, C.
G., Velasco-Vasquez, J. and Amarir, A. 2008. Le complexe funéraire et cultuel
d’Adrar Zerzem (Anti-Atlas, Maroc): Résultats préliminaires. In Etudes
d’Antiquités Africaines. Lieux de cultes: Aires votives, temples, églises, mosquées,
Actes du IXème Colloque international sur l’histoire et l’archéologie de l’Afrique
du Nord antique et médiévale, Tripoli, 2005. Paris: Èditions du CNRS, 23–33.
Bokbot, Y., Onrubia-Pintado, J. and Salih, A. 2011. Néolithique et protohistoire
dans le bassin de l’Oued Noun (Maroc présaharien), quelques données
préliminaires. In Actes du 1er colloque de Préhistoire Maghrébine, Tamanrasset,
2007. Alger: CNRPAH, 305–22.
Bokbot, Y., Cáceres Gutíérrez, Y., Cressier, P., González Marrero, M. del C.,
Izquierdo Benito, R., de Juan Ares, J., Mabrouk, S., Onrubia Pintado, J. and
Salesse, F. 2013a. La ville natale de Sîdî Waggag: Nûl-Lamta et l’archéologie de
l’oasis d’Asrir. In Naïmi. M. (ed.), Sidi Waggag b. Zalluw al-Lamtî, aux origins du
malékisme étatique nord-ouest africain. Rabat: DTGSN, 21–64.
Bokbot, Y., Cáceres Gutíérrez, Y., Cressier, P., González Marrero, M. del C., Hervás
Herrera, M.A., Onrubia Pintado, J. 2013b. Viviendas medievales al sur del Anti-
Atlas (Marruecos). Problemas de studio y especificidades. In S. Gutiérrez Lloret
and I. Grau Mira (eds), De la estructura doméstica al espacio social. Lecturas
arqueológicas del uso social del espacio. Alicante: Publicaciones Universidad de
Alicante, 279–98.
Brito, J.C., Martínez-Freiría, F., Sierra, P., Sillero, N. and Tarroso, P. 2011. Crocodiles
in the Sahara desert: An update of distribution, habitats and population status for
conservation planning in Mauritania. PLoS ONE 6.2: e14734. https://doi.org/10
.1371/journal.pone.0014734 [last accessed 10 September 2019].
Camps, G. 1984. Rex gentium Maurorum et Romanorum. Recherches sur les
royaumes de Maurétanie des VIe et VIIe siècles. Antiquités africaines 20:
183–218.
Capot-Rey, R. 1953. Le Sahara Français (L’Afrique Blanche Francaise II). Paris:
Presses universitaires.
Capot-Rey 1956. Greniers domestiques et greniers fortifiés au Sahara. Le cas du
Gourara. Travaux de l’Institut de Recherches Sahariennes 14: 139–58.
Capot-Rey, R. and Damade, W. 1962. Irrigation et structure agraire à Tamentit
(Touat). Travaux de l’Institut de Recherches Sahariennes 21: 99–119.
Carcopino, J. 1943. Le Maroc antique. Paris: Gallimard.
Castellani, V. 1999. Civiltà dell acqua. Archeologia del territorio da Roma arcaica
alle antiche civiltà mediterrannee. Rome: Editorial Service System.
Chaintron, J.F. 1957/1958. Aoulef. Problèmes économiques et sociaux d’une oasis à
foggaras. Travaux de l’Institut de Recherches Sahariennes 16: 101–29; 17: 127–56.
Champault, D. 1969. Une oasis du Sahara nord-occidental. Tabelbala. Paris: CNRS.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 272 Martin Sterry, David J. Mattingly and Youssef Bokbot

Châtelain, L. 1944. Le Maroc des romains. Etude sur les centres antiques de la
Maurétanie occidentale. Paris: de Boccard.
Châtelain, P. 1896. En Algérie. La Kabylie et les oasis du sud. Nevers: Imp. Cloix.
Cornet, A. 1952. Essai sur l’hydrogéologie du Grand Erg Oriental et des régions
limitrophes (les foggaras). Travaux de l’Institut de Recherches Sahariennes 8: 71–122.
Daumas, E. 1845. Le Sahara algerien. Etudes geographiques, statistiques et histor-
iques sur la region au sud des etablissments francais en Algerie. Paris: Langlois et
Leclerc.
De la Chapelle, F. 1934. L’expedition de Suétonius Paulinus dans le sud-est du
Maroc. Hesperis 19: 107–24.
Desanges, J. 1962. Catalogue des tribus africaines de l’antiquité classique à l’ouest du
Nil. Dakar: University of Dakar.
Desanges, J. 1978. Recherches sur l’activité des Méditerranéens aux confins de
l’Afrique. Rome: Coll. École française de Rome 38.
Desanges, J. 1980. (Pline l’ancien) Histoire naturelle Livre V.1–46 (L’Afrique du
nord). Paris: Collection Budé.
Despois, J. and Raynal, R. 1967. Géographie de l’Afrique du Nord-Ouest. Paris:
Payot.
Dunn, R.E. 1977. Resistance in the Desert. London: Croom Helm.
Echallier, J.-C. 1972. Villages désertés et structures agraires anciennes du Touat et du
Gourara. Paris: RMG.
Echallier, J.-C. 1973. Fortéresses berbères du Gourara. Problèmes et resultants des
fouilles. Libyca 21: 293–302.
El Graoui, M., Bokbot, Y., Jungner, H. and Searight-Martinet, S. 2010. Datation
radiocarbone sur des ossements mis au jour dans un tumulus à l’Adrar Zerzem,
Oued Eç-çayad, région de Taghjijt (Sud marocain). Sahara 21: 77–80.
El-Khoumsi, W., Hammani, A., Kuper, M. and Boaziz, A. 2017. La durabilité du
système oasien face à la détérioration des ressources en eaux souterraines: Cas de
la palmeraie de Tafilalet. Revue Marocaine des sciences agronomiques et
vétérinaires 5.1: 41–51.
Euzennat, M. 1989. Le limes de Tingitane I, la frontiére méridionale. Paris: CNRS.
Gabriel, B. 1984. Zur vorzeitlichen Besiedlung Südalgeriens (Tanezrouft, Tidikelt,
Touat, Gourara). Die Erde 115: 93–109.
Gatto, M.C., Mattingly, D.J., Ray, N. and Sterry, M. (eds). 2019. Burials, Migration
and Identity in the Ancient Sahara and Beyond. Trans-Saharan Archaeology
Volume 2. Series editor D.J. Mattingly. Cambridge: Cambridge University Press.
Gaucher, G. 1948. Irrigation et mise en valeur du Tafilalet. Travaux de l’Institut de
Recherches Sahariennes 5: 95–120.
Gauthier, Y. and Gauthier, C. 2011. Des chars et des Tifinagh: Étude aréale et
corrélations. Les Cahiers de l’AARS 15: 91–118.
Gauthier, Y. and Gauthier, C. 2015. Nouvelles figurations de chars sahariens:
Technicité et positionment chronologique relativement au style de Tazina. Les
Cahiers de l’AARS 18: 5–70.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 273

Gautier, E.-F. 1905. Les oasis sahariennes. Algérie: Imp Fontana.

Gautier, E.-F. 1970. Sahara. The Great Desert (translated from the French by D.F.
Mayhew). London: Octagon.
Hamdoune, C. 2018. Ad fines Africae Romanae. Les mondes tribaux dans les
provinces maurétaniennes. Bordeaux: Scripta Antiqua.
Heywood, C. 1926. Algeria and Tunisia. Paris and London: Librarie Hachette.
Hirschberg, H.Z. 1963. The problem of the judaized Berbers. Journal of African
History 4.3: 313–39.
Homo, J.P. and Hugonot, R. 1953. Ich, oasis de montagne. Travaux de l’Institut de
Recherches Sahariennes 9: 99–124.
Jacques-Meunié, D. 1958. La nécropole de Foum le Rjam, tumuli du Maroc
présaharien. Hespéris 1958 1–2: 95–142.
Jacques-Meunié, D. 1982. Le Maroc saharien des origins à 1670, 2 vols. Paris:
Klincksieck.
Joly, F. 1948. Sur la repartition du palmier dattier en Maroc. Travaux de l’Institut
de Recherches Sahariennes 5: 203–12.
Landau, R. 1969. The Kasbahs of Southern Morocco. London: Faber and Faber.
Laporte, J.-P. 2011. Particularités de la province de Maurétanie Césarienne (Algérie
centrale et occidentale). In C. Briand-Ponsart and Y. Modéran (eds), Provinces et
identités provincials dans l’Afrique romaine. Caen: CRAHM, 111–50.
Lehuraux, A. 1943. Les origins des oasis du Tidikelt et du Bas-Touat. Travaux de
l’Institut de Recherches Sahariennes 2: 105–20.
Levtzion, N. and Hopkins, J.F.P. (eds). 2000. Corpus of Early Arabic Sources for
West African History. Princeton: Markus Wiener Publishers.
Lightfoot, D.R. and Miller, J.A. 1996. Sijilmassa: The rise and fall of a walled oasis in
Medieval Morocco. Annals of the Association of American Geographers 86.1:
78–101.
Lihoreau, M. 1993. Djorf Torba, nécropole saharienne antéislamique. Paris:
Karthala.
Lô, Capitaine 1953/1954. Les foggaras du Tidikelt. Travaux de l’Institut de
Recherches Sahariennes 10: 139–79; 11: 49–77.
Lugan, B. 2000. Histoire du Maroc des origines à nos jours. Paris: Perrin.
Lydon, G., 2009. On Trans-Saharan Trails: Islamic Law, Trade Networks and Cross-
cultural Exchange in Nineteenth-Century Western Africa. Cambridge:
Cambridge University Press.
MacDonald, K. 2011. A view from the south. Sub-Saharan evidence for contacts
between North Africa, Mauritania and the Niger 1000 BC–AD 700. In A. Dowler
and E. Galvin (eds), Money, Trade and Trade Routes in Pre-Islamic North Africa.
London: British Museum, 72–82.
Martin, A.G.P. 1908. Les oasis sahariennes (Gourara-Touat-Tidikelt). Alger: Imp.
Algérienne.
Mattingly, D.J., (ed.). 2003. The Archaeology of Fazzan. Volume 1, Synthesis.
London: Society for Libyan Studies.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 274 Martin Sterry, David J. Mattingly and Youssef Bokbot

Mattingly, D.J. (ed.). 2007. The Archaeology of Fazzan. Volume 2, Site Gazetteer,
Pottery and other Survey Finds. London: Society for Libyan Studies.
Mattingly, D.J. 2017. The Garamantes and the origins of Saharan trade. State of the
field and future agendas. In D.J. Mattingly, V. Leitch, C.N. Duckworth,
A. Cuénod, M. Sterry and F. Cole (eds), Early Saharan Trade: The Inorganic
Evidence. Cambridge: Cambridge University Press, 1–52.
Mattingly, D.J., Rushworth, A., Sterry, M. and Leitch, V. 2013. The African
Frontiers. Frontiers of the Roman Empire series. Edinburgh: Hussar Books.
Mattingly, D.J., Bokbot Y., Sterry M., Cuénod A., Fenwick C., Gatto M., Ray N.,
Rayne L., Janin K., Lamb A., Mugnai N., and Nikolaus J. 2017. Long-term
history in a Moroccan oasis zone: The Middle Draa Project 2015. Journal of
African Archaeology 15: 141–72.
Mattingly, D.J., Sterry, M., al-Haddad, M. and Bokbot, Y. 2018. Beyond the
Garamantes: The early development of Saharan oases. In L. Purdue,
J. Charbonnier and L. Khalidi (eds), Des refuges aux oasis: Vivre en milieu
aride de la Préhistoire à aujourd’hui. XXXVIIIe rencontres internationales
d’archéologie et d’histoire d’Antibes. Antibes: Éditions APDCA, 205–28.
Mattingly, D.J., Gatto, M.C., Sterry, M. and Ray, N. 2019. Burials, migration and
identity. The view from the Sahara. In Gatto et al. 2019, 1–50.
Mauny, R. 1978. Trans-Saharan contacts and the Iron Age in West Africa. In J.
D. Fage (ed.), The Cambridge History of Africa, vol. 2. Cambridge: Cambridge
University Press, 272–341.
Maxwell, G. 2004. Lords of the Atlas. The Rise and Fall of the House of Glaoua
1893–1956. London: Elland.
Messier, R.A. 1997. Sijilmasa. Five seasons of archaeological inquiry by a joint
Moroccan-American Mission. Archéologie Islamique 7: 61–92.
Messier, R.A. and Fili, A. 2011. The earliest ceramics of Sijilmasa. In P. Cressier and
E. Fentress (eds), La céramique maghrébine du haut moyen âge, VIIIe-Xe siècle:
état des recherches, problèmes et perspectives. Paris: École française de Rome,
129–46.
Messier, R. and Miller, J. 2015. The Last Civilized Place. Sijilmasa and Its Saharan
Destiny. Austin: University of Texas Press.
Monteil, V., 1951. Contribution à l’étude de la faune du Sahara occidental. Paris:
Larose.
Nesson, C., Rouvillois-Brigol, M., and Vallet, J. 1973. Oasis du sahara algérien.
Paris: Institut géographique national.
Oliel, J. 1994. Les juifs au Sahara: Le Touat au moyen age. Paris: CNRS.
Onrubia Pintado, J., Bokbot, Y., Hervás Herrera, M.A., García García, L.A.,
Marchante Ortega, A., Cáceres Gutíérrez, Y., González Marrero, M. del C.,
Juan Ares, J. de, Moreno García, M. and Rodríguez Santana, C.G. 2016.
Arqueología de Fum Asaca (Sidi Ifni-Marruecos). De probable instalación
purpuraría gétula a Torre Colonial hispano-canaria. Anuaria. Anuario de
Estudios Atlánticos 62: 1–25.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 6 Pre-Islamic Oasis Settlements in the North-Western Sahara 275

Pichler, W. 2000. The Libyco-Berber inscriptions of Foum Chenna/Morocco,

Sahara 12: 176–78.
Rebuffat, R. 1999. La frontière de la Tingitane. In C. Lepelley and X. Dupuis (eds),
Frontières et limites géographiques de l’Afrique du nord antique. Hommage à
Pierre Salama. Paris: Sorbonne, 265–93.
Remini, B. Rezoug, C. and Achour, B. 2014.The foggara of Kenadsa (Algeria).
Larhyss Journal 18: 93–105.
Reygasse, M. 1950. Monuments funéraires préislamiques de l’Afrique du Nord.
Paris: Arts et Métiers Graphiques.
Ricard, P. 1925. Le Maroc. Paris: Libraire Hachette.
Richet, E. 1920. Oued Sous Oued Draa. Paris: Figuiere.
Rinn, L. 1886. Nos frontières sahariennes. Alger: Adolphe Jourdan.
Riser, J. 1996. Dra. Encyclopédie berbère fasc 17: 2537–41.
Rohlfs, G. 2001. Voyages et explorations au Sahara. Tome I, Draa – Tafilalet – sud
oranais – Touat – Tidikelt – Rhadamès 1861–1864 (translated by J. Debetz).
Paris: Karthala.
Rosenberger, B. 1970. Tamdult, cité minière et caravanière présaharienne, IXe–
XIVe s. Hespéris-Tamuda 1970: 103–39.
Salama, P. 1981. The Sahara in classical antiquity. In G. Mokhtar (ed.), UNESCO
General History of Africa, II Ancient Civilisations of Africa. Paris: UNESCO, 513–32.
Shaw, B.D. 1987. Autonomy and tribute: Mountain and plain in Mauretania
Tingitana. Revue de l’Occident Musulman et de la Méditerranée 41–41: 66–89.
Simoneau, A. 1972a. Les prospections rupestres dans la région du Drâa (Extrême
Sud marocain), avril 1971 – avril 1972. Almogaren 3: 15–31.
Simoneau, A. 1972b. Nouvelle gravures rupestres du haut-Atlas et du Drâa.
Bulletin d’Archéologie Marocaine 8: 2–39.
Simoneau, A. 1977. Catalogue des sites rupestres du sud Marocain. Rabat: Ministère
des Affaires Culturelles.
Souag, L. 2015. Explaining Korandjé: Language contact, plantations and the trans-
Saharan trade. Journal of Pidgin and Creole Linguistics 30.2: 189–224.
Stevenson, E.L. 1991. Claudius Ptolemy, The Geography. New York: Dover.
Tayler, J. 2003. Valley of the Casbahs. A Journey across the Moroccan Sahara.
London: Little, Brown.
Vanacker, C. 1973. Géographie économique de l’Afrique du nord selon les auteurs
arabes du IXe siècle au milieu du XIIe siècle. Annales ESC 28.3: 659–80.
Voinot, L. 1909. Le Tidikelt. Etude sur la géographie, l’histoire, les moeurs du pays.
Oran: Fouque.
Wilson, A.I. 2005. Foggara irrigation, early state formation and Saharan trade: The
Garamantes of Fazzan. Schriftenreihe der Frontinus-Gesellschaft 26: 223–34.
Wilson, A.I. 2006. The spread of foggara-based irrigation in the ancient Sahara. In
D.J. Mattingly, S. McLaren, E. Savage, Y. al-Fasatwi and K. Gadgood (eds), The
Libyan Desert. Natural Resources and Cultural Heritage. London: Society for
Libyan Studies, 205–16.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 276 Martin Sterry, David J. Mattingly and Youssef Bokbot

Wilson, A.I. 2009. Foggaras in ancient North Africa or how to marry a Berber
princess. In Contrôle et distribution de l’eau dans le Maghreb antique et medieval.
Rome: Coll. École française de Rome, 19–39.
Wilson, A.I., Mattingly, D.J. and Sterry, M. Forthcoming. The diffusion of irriga-
tion technologies in the Sahara in antiquity: Settlement, trade and migration. In
C. Duckworth, A. Cuénod and D. Mattingly (eds), Mobile Technologies in the
Ancient Sahara and Beyond. Cambridge: Cambridge University Press.
Zaïnabi, A.T. (ed.). 2004. Trésors et merveilles de la vallée du Drâa. Zagora: Ed
Marsam.

https://doi.org/10.1017/9781108637978.007 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern
Sahara
martin sterry and david j. mattingly

Introduction

The southern oases of the Sahara can be split into two broad groups.1 First,
there are a series located in and around the Saharan mountain ranges.
From east to west these are: Ennedi, Tibesti, Aïr, Tassili n’Ajjer, Ahaggar
and Adrar des Ilforas. These massifs receive higher levels of rainfall than
other parts of the Sahara, resulting in seasonal rivers that feed permanent
bodies of water (small lakes or pools/gueltas) or that support a high water
table beneath the wadi beds.2 There are still populations of crocodiles living
in Ennedi and nineteenth- and early-twentieth-century reports of them
also in Ahaggar and Tibesti.3 However, these areas have traditionally been
primarily exploited by pastoralists and they have little suitable land for
oasis agriculture and highly variable rainfall. The dynamic wadi systems
may also be a factor in the poor preservation of archaeological remains
with less sturdy or older constructions either swept away or buried under
sediment.
The second group consists of oases in the intervening depressions,
including Agadez, Kawar, Djado, Kufra, Borku and a number of smaller
oases. Apart from Agadez these have almost no rainfall and are entirely
reliant on springs or lifting water from a locally restricted but elevated
water table. Further to the west, there are a number of oases in the Southern
Sahara within striking distance of Sahelian West Africa which were his-
torically important in the Medieval era. The final section of the chapter

1
Despois and Raynal 1967, 474–508 for a geographical synthesis of the western part of the Sahara.
2
Gautier 1970, 85–93 on the water sources of the mountainous enclaves. Cf. Capot-Rey 1953,
38–45 on the higher rainfall of the mountains compared to the surrounding basins, but with
annual levels still well below 100 mm.
3
Gautier 1970, 91. 277

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 278 Martin Sterry and David J. Mattingly

provides only a brief overview of these sites, as they are discussed in further
detail in the second half of this book. One of the interesting aspects of the
more southerly oases is that they played an important part in the rise of
Sub-Saharan polities (such as Kanim-Bornu, Ghana, Mali, Songhay, the
Sultanate of Agadez) and their influence on the history of Saharan trade.
The rise of the Almoravid and Almohad dynasties in the Maghrib was also
founded on deep desert contacts. Once again, a key question is whether
such Medieval actualities were prefigured by connections and develop-
ments in the Protohistoric period.
Another basic division in the material presented in this chapter
concerns a fundamental linguistic and ethnic divide between two
major groups: the Teda and the Tuareg. These peoples were instru-
mental players in the early modern caravan and slave trade and it is
highly likely that their domination of particular routes and zones of the
Sahara was of much greater antiquity. The Teda operated the main
routes between Chad and Libya and Egypt, while the Tuareg were more
engaged in the Niger Bend trade routes leading to Libya, Tunisia,
Algeria and Morocco.
Research into the southern oases has been variable and has tended to
focus either on the prehistory of the ‘Green Sahara’ (that is before
3,000 BC) or on the historical archaeology of current communities.4 The
area around Aïr and Agadez has seen the most work with a number of
major projects, while research in Chad and Mali has lagged behind albeit
with some notable and important exceptions. The Dhar Tichitt area in
south-east Mauritania for the moment provides a remarkably precocious
manifestation of nucleated settlement, whose relationship with later phases
of sedentarisation remains unclear.
In this chapter, we draw together the published data and combine it with
our direct observations of high resolution satellite imagery. It is evident
that in many of the oasis locations there is a rich and as yet poorly explored
and mostly undated archaeological heritage. For several of the regions pre-
Islamic activity is certainly attested and, while the nature of this activity
cannot be certainly linked to oasis development at this stage, it is a crucial
question that is highlighted here for future research. As in other chapters,
we have also compiled the available radiocarbon dates for the region in
Table 7.1, though detailed information on context is less readily available
for many of these.

4
Haour 2003.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 Table 7.1 Protohistoric and historic radiocarbon dates from the Southern Sahara

Region/Site Context Material Reference Date Calibrated

Ennedi
Mornou T05/88–2 Cattle skull Poz-16595 2480±30 BP 774–434 calBC
Borku/Bahr-al-Ghazal
Bahali I Gif-2895 630±90 BP calAD1224–1441
Bahali IV Gif-4201 1540±90 BP calAD 332–660
Bochiangra Gif-2611 1500±100 BP calAD 333–761
Bochiangra Gif-2612 1500±100 BP calAD 333–761
Bochiangra G.D. Paris VI 935±80 BP calAD 973–1260
Bochiangra, Koro-Toro Bone ? 1500±100 BP calAD 333–761
Bochiangra, Koro-Toro Bone ? 1015±80 BP calAD 778–1215
Bochiangra, Koro-Toro Bone ? 730±90 BP calAD 1049–1412
Bochiangra, Koro-Toro Furnace Hv-3804 1960±60 BP 108 calBC–calAD 214

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Bochiangra, Koro-Toro Furnace Charcoal with slag Hv-4694 1570±50 BP calAD 395–595
Bochiangra, Koro-Toro Furnace Charcoal with slag Hv-2751 1510±75 BP calAD 401–655
Djourab Hv-4694 1570±50 BP calAD 395–595
Kebir Bosa Gif-4202 2360±100 BP 774–205 calBC
Kebir Bosa Gif-4193 1480±100 BP calAD 348–765
Kerki Bougui (site 97) Gif-4200 1250±90 BP calAD 643–978
Kolle Charcoal with human bones Ny-83 1460±220 BP calAD 85–1012
and pottery in lake deposits
Koro Toro Hv-2751 1510±75 BP calAD 401–655
Koro Toro (Site 4) G.D. Paris 1670±80 BP calAD 143–561
Koro Toro (Site 4) Gif-4194 1580±100 BP calAD 252–649
Koro Toro (Site 4) Gif-4195 1400±100 BP calAD 421–867
Koro Toro (Site 4) Gif-4198 1230±100 BP calAD 642–995
Koro Toro (Site 4) Gif-4199 1170±90 BP calAD 678–1016
 Table 7.1 (cont.)

Region/Site Context Material Reference Date Calibrated

Krime Gif-4197 1410±100 BP calAD 413–863

Krime Gif-4196 670±100 BP calAD 1161–1440
Nemra III Gif-2613 730±90 BP calAD 1049–1412
Site 97 Gif-4200 1250±100 BP calAD 621–988
Toungour Hv-3804 1960±60 BP 108 calBC–calAD 214
Toungour Hv-2898 1655±55 BP calAD 253–539
Toungour-Salado Gif-2896 1340±100 BP calAD 434–948
Tibesti
Bardague Araye Hv-6926 4125±100 BP 2916–2463 calBC
Bardague Araye Hv-6933 3385±60 BP 1877–1528 calBC
Bardague Araye 2 Hv-6931 3900±155 BP 2872–1978 calBC
Bardague Araye 2 Hv-6929 3790±300 BP 3086–1456 calBC
Bardai Hv-4668 370±90 BP calAD 1399– . . .
Enneri Dirennao Hv-2260 2690±435 BP 2022 calBC–

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

calAD 138
Enneri Oyoroum Hv-3763 3755±335 BP 3310–1322 calBC
Enneri Oyoroum Hv-3765 2350±410 BP 1495 calBC–
calAD 530
Gabrong Hv-3709 6130±90 BP 5297–4846 calBC
Gabrong Gif-1316 1570±100 BP calAD 254–653
Gabrong Hv-2198 1440±150 BP calAD 254–940
Mount Ayando Hv-3775 690±75 BP calAD 1192–1415
Serir Tibesti Hv-2749 1840±60 BP calAD 52–335
Serir Tibesti Hv-2936 1780±60 BP calAD 89–391
Yebbigue Human bone Hv-2774 2427±225 BP 1074 calBC–calAD 24
Zouar Burial vault Human bone Gif-1183 3375±120 BP 2008–1417 calBC
Zouar Cremation burial Cremated bone Gif-1181 1330±100 BP calAD 477–967
 Zouar Gif-1318 750±100 BP calAD 1043–1405
Zoui Hv-3771 2130±40 BP 355–46 calBC
Zoui Hv-4351 2080±55 BP 350 calBC–calAD 52
Zoui Hv-4498 2065±320 BP 839 calBC–calAD 603
Zoui Hv-3773 1095±110 BP calAD 688–1155
Zoui Hv-3176 1130±55 BP calAD 772–1015
Zoui Hv-3775 690±75 BP calAD 1192–1415
Zoui Hv-4352 455±75 BP calAD 1316–1638
Zoui Hv-3774 430±35 BP calAD 1417–1619
Zoui Hv-4667 420±55 BP calAD 1412–1635
Djado
Diabessa (Djado) R-1266 420±50 BP calAD 1415–1633
Djaba Gif-3547 760±80 BP calAD 1045–1395
Djado 1 Gif-4130 90±90 BP calAD 1664– . . .
Djado 4 Gif-4131 270±90 BP calAD 1445– . . .
Termit
Do Dimmi Pa-288 2500±70 BP 795–428 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Do Dimmi DaK-145 2628±120 BP 1031–409 calBC
Do Dimmi UPS 2590±120 BP 978–404 calBC
Do Dimmi Pa-296 2580±120 BP 976–402 calBC
Do Dimmi Pa-504 2270±90 BP 736–56 calBC
Do Dimmi UPS 2065±60 BP 348 calBC–calAD 66
Do Dimmi DaK-148 1747±110 BP calAD 32–543
Gara Tchia Bo 48 a Pa-810 3260±100 BP 1861–1290 calBC
Gara Tchia Bo 48 b Pa-811 3265±100 BP 1868–1294 calBC
Tchi Guiriba 127 B Pa-669 2955±100 BP 1417–917 calBC
Tchire Ouma 146 Pa-510 3230±170 BP 1928–1055 calBC
Termit Charcoal with slag Dak-147 2924±120 BP 1414–844 calBC
Termit Charcoal with slag Dak-145 2628±120 BP 1031–409 calBC
Termit Dak-148 1747±100 BP calAD 67–536
 Table 7.1 (cont.)

Region/Site Context Material Reference Date Calibrated

Termit Ouest 8 Pa-668 2880±120 BP 1390–819 calBC

Termit ouest 95 B Pa-481 3100±100 BP 1611–1060 calBC
Tenere Desert
Tenere, No. 7 Burial at depth 1.5 m under Human bone Gif-3639 3180±100 BP 1691–1135 calBC
stone blocks 18 N 10 E
Tin Keradet C-2619 3310±160 BP
Tassili n’Ajjer
Tissoukai Tassili Gif-840 3650±130 BP 2456–1694 calBC
Tahabort Gif-2900 2560±100 BP 895–409 calBC
Tahabort Gif-3324 1110±100 BP calAD 681–1152
Tahabort Gif-2898 1050±100 BP calAD 726–1207
Tahabort Gif-3325 930±100 BP calAD 898–1273
Tahabort Gif-3258 570±90 BP calAD 1259–1609
Tamanrasset Neolithic, double burial Gif-357 3330±250 BP 2335–1000 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Tamanrasset II Mc-487 3900±100 BP 2834–2041 calBC
Tamanrasset II Gif-357 3330±250 BP 2335–1000 calBC
Ahaggar
Tin Hinan Chambered tomb ? 1480±130 BP calAD 247–856
Aïr
Aghroum Balkorene Gif-? 200±80 BP calAD 1516–..
Anisaman Gif-3522 230±80 BP calAD 1485– . . .
Arlit Faunal bone Gif-1797 2640±100 BP 1022–431 calBC
Assodé 0 635±40 BP calAD 1283–1400
Assodé 0 565±40 BP calAD 1300–1432
Assodé R-1268 360±50 BP calAD 1449–1639
Assodé 0 290±60 BP calAD 1450– . . .
Assodé 0 250±90 BP calAD 1455– . . .
 Assodé 0 245±60 BP calAD 1472– . . .
Assodé 0 260±40 BP calAD 1492– . . .
Assodé 0 205±40 BP calAD 1640– . . .
Graboun Mountain Olive tree (in situ!) UCLA-1655 180±60 BP calAD 1643– . . .
Ifanghalene Gif-2938 3410±100 BP 1966–1461 calBC
Iwelen Pa-234 3595±100 BP 2275–1687 calBC
Iwelen Pa-508 3425±120 BP 2033–1449 calBC
Iwelen Pa-133 3410±200 BP 2291–1231 calBC
Iwelen Pa-506 3160±100 BP 1664–1131 calBC
Iwelen Pa-600 3360±200 BP 2206–1130 calBC
Iwelen Pa-656 3320±200 BP 2195–1113 calBC
Iwelen Pa-683 3290±200 BP 2131–1055 calBC
Iwelen Pa-117 2935±70 BP 1381–931 calBC
Iwelen Pa-121 2940±100 BP 1411–908 calBC
Iwelen Pa-319 2930±100 BP 1408–902 calBC
Iwelen Pa-141 2795±120 BP 1374–773 calBC
Iwelen Hv-3256 2635±60 BP 926–552 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Iwelen Pa-580 2710±140 BP 1260–434 calBC
Iwelen Pa-112 2510±60 BP 799–430 calBC
Iwelen Pa-496 2520±80 BP 803–416 calBC
Iwelen Pa-884 2600±100 BP 969–412 calBC
Iwelen Pa-885 2585±110 BP 969–406 calBC
Iwelen Pa-645 2540±100 BP 891–403 calBC
Iwelen Pa-628 2530±100 BP 841–401 calBC
Iwelen Pa-655 2520±100 BP 831–401 calBC
Iwelen Pa-525 2675±200 BP 1388–386 calBC
Iwelen Pa-888 2590±150 BP 1110–386 calBC
Iwelen Pa-116 2410±80 BP 781–379 calBC
Iwelen Pa-886 2480±110 BP 832–376 calBC
Iwelen Pa-635 2415±110 BP 801–231 calBC
 Table 7.1 (cont.)

Region/Site Context Material Reference Date Calibrated

Iwelen Pa-860 2980±400 BP 2293–209 calBC

Iwelen Pa-847 2395±121 BP 800–205 calBC
Iwelen Pa-634 2305±110 BP 766–113 calBC
Iwelen Pa-146 2200±60 BP 395–107 calBC
Iwelen Pa-400 2520±250 BP 1279–43 calBC
Iwelen Pa-862 2415±200 BP 996–40 calBC
Iwelen IW68 Clothes 0 1220±60 BP calAD 669–961
Iwelen IW68 Leather 0 1165±60 BP calAD 694–991
Iwelen IW68 Carbonate 0 1175±80 BP calAD 678–995
hydroxulapatie
Iwelen IW68 Hairs 0 1145±60 BP calAD 722–1016
Iwelen IW68 Collagen 0 1160±90 BP calAD 677–1020
Iwelen S1 0 2160±50 BP 363–57 calBC
Iwelen S2 0 2680±40 BP 908–797 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Iwelen S3 0 2510±60 BP 799–430 calBC
Iwelen S4 0 2190±70 BP 394–57 calBC
Iwelen S5 0 2340±70 BP 752–206 calBC
Kori Ibine From a room in the Charcoal Dak-146 685±100 BP calAD 1058–1439
settlement
Tadeini 1 Blast furnace Charcoal Gif-3523 250±80 BP calAD 1455–present
Tadeini 2 Blast furnace Charcoal Gif-3524 350±80 BP calAD 1417–1799
Tadeliza 30–40 cm depth Gif-2594 720±90 BP calAD 1051–1417
Tadeliza Surface Gif-2593 510±90 BP calAD 1283–1631
Tezamak Gif-3525 400±80 BP calAD 1400–1660
Tezamak Gif-3526 330±80 BP calAD 1427– . . .
Wadi Mammanet ? 1280±60 BP calAD 651–884
Wadi Mammanet ? 1270±65 BP calAD 647–936
 Azawagh
Afunfun (site 162) MC-2404 2800±90 BP 1207–806 calBC
Afunfun (site 162) MC-2404 2800±90 BP 1207–806 calBC
Afunfun (site 162) MC-2406 2740±90 BP 1188–774 calBC
Afunfun (site 162) MC-2406 2740±90 BP 1188–774 calBC
Afunfun (site 162) Gif-5541 2590±90 BP 911–429 calBC
Afunfun (site 162) MC-2405 2540±90 BP 831–408 calBC
Afunfun (site 162) MC-2405 2540±90 BP 831–408 calBC
Afunfun (site 162) MC-2402 2520±90 BP 811–407 calBC
Afunfun (site 162) MC-2403 2520±90 BP 811–407 calBC
Afunfun (site 162) MC-2402 2520±90 BP 811–407 calBC
Afunfun (site 162) MC-2403 2520±90 BP 811–407 calBC
Afunfun (site 175) MC-2399 4140±90 BP 2902–2488 calBC
Afunfun (site 175) MC-2399 4140±90 BP 2902–2488 calBC
Afunfun (site 175) Gif-5175 3960±110 BP 2866–2145 calBC
Afunfun (site 175) Gif-5176 3950±110 BP 2864–2140 calBC
Afunfun (site 175) MC-2398 3920±90 BP 2835–2137 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Afunfun (site 175) MC-2398 3920±90 BP 2835–2137 calBC
Afunfun (site 175) MC-2401 3800±90 BP 2477–1977 calBC
Afunfun (site 175) MC-2401 3800±90 BP 2477–1977 calBC
Afunfun (site 175) Gif-5174 3840±110 BP 2580–1966 calBC
Afunfun (site 175) Gif-5172 3680±50 BP 2202–1931 calBC
Afunfun (site 175) Gif-5172 3680±50 BP 2202–1931 calBC
Afunfun (site 175) Gif-5177 3790±110 BP 2562–1930 calBC
Afunfun (site 175) MC-2400 3600±90 BP 2204–1696 calBC
Afunfun (site 175) MC-2400 3600±90 BP 2204–1696 calBC
Afunfun (site 175) Gif-5173 3390±110 BP 1957–1436 calBC
Afunfun (site 178) Gif-5179 2900±100 BP 1386–844 calBC
Afunfun (site 179) Gif-5178 3030±100 BP 1501–1002 calBC
Aktaouzou (site 178) Gif-5179 2900±110 BP 1391–839 calBC
 Table 7.1 (cont.)

Region/Site Context Material Reference Date Calibrated

Azelik Gif-4175 2490±90 BP 798–407 calBC

Azelik Gif-4330 2480±90 BP 795–406 calBC
Azelik Gif-? 2400±90 BP 792–239 calBC
Azelik Gif-2156 800±100 BP calAD 1023–1390
Azelik Gif-? 590±80 BP calAD 1272–1445
Azelik Gif-4174 520±80 BP calAD 1282–1619
Azelik Dak-9 353±114 BP calAD 1320– . . .
Azelik DaK-121 modern± BP 0
Azelik (site 210) Gif-5546 2510±70 BP 799–430 calBC
Azelik (site 210) Gif-4175 2490±90 BP 798–407 calBC
Azelik (site 210) Gif-4330 2480±90 BP 795–406 calBC
Azelik (site 210) Gif-? 2400±90 BP 792–239 calBC
Chin Wasararan (site 117) MC-1700 3160±95 BP 1659–1134 calBC
Chin Wasararan (site 117) MC-1702 2870±95 BP 1369–828 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Chin Wasararan (site 117) MC-1701 2795±105 BP 1257–795 calBC
Efey Washaran (site 149) MC-1703 2875±105 BP 1377–826 calBC
Ekne Wan Ataran (site 119) MC-2397 2440±90 BP 792–393 calBC
Fagochia Gif-4178 390±90 BP calAD 1324–1798
Guelele Gif-4176 680±80 BP calAD 1193–1423
In Taylalen II Gif-4170 2210±90 BP 452 calBC–calAD 1
In Taylalen II Gif-4171 2010±90 BP 351 calBC–calAD 219
In Teduq Pa-0299 3485±165 BP 2285–1431 calBC
In Teduq Pa-0231 2810±80 BP 1196–811 calBC
In Teduq Pa-1418 1890±40 BP calAD 28–230
In Tekebrin Pa-0233 2600±80 BP 923–433 calBC
In Waggeur N-2131 255±60 BP calAD 1463– . . .
In Waggeur N-2130 155±70 BP calAD 1653– . . .
Inufan Orsat 1595±80 BP calAD 256–623
 Janjari N-2129 335±60 BP calAD 1445–1659
Marandet Gif-2595 1430±100 BP calAD 392–801
Marandet Gif-2599 1420±100 BP calAD 400–860
Marandet Gif-2158 1400±100 BP calAD 421–867
Marandet Gif-2596 1360±100 BP calAD 430–891
Marandet Gif-2157 1310±100 BP calAD 551–968
Marandet Gif-2597 1310±100 BP calAD 551–968
Marandet Gif-2598 1160±100 BP calAD 661–1031
Marandet Gif-1726 1050±95 BP calAD 729–1189
Mentes Pa-1624 2740±100 BP 1215–673 calBC
Mio Site 169 MC-2395 3170±90 BP 1658–1215 calBC
Orub Gif-4173 3390±100 BP 1941–1452 calBC
Sekiret Gif-3862 3310±100 BP 1881–1407 calBC
Sekiret Gif-4177 2900±100 BP 1386–844 calBC
Sekiret Gif-5543 2700±70 BP 1016–773 calBC
Sekiret Valley Gif-5544 2470±100 BP 806–392 calBC
Takene Bawat Pa-0444 2010±200 BP 727 calBC–calAD 428

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Tamaya Mellet Pa-1366 3435±100 BP 2016–1505 calBC
Tamaya Mellet Pa-410 3060±90 BP 1505–1049 calBC
Tamaya Mellet Pa-411 3060±140 BP 1621–931 calBC
Tegef n’Agar (site 74) Gif-4172 2090±90 BP 363 calBC–calAD 69
Tuluk 211 Gif-5545 2360±70 BP 760–231 calBC
Tyeral Site 207 Gif-5542 2410±90 BP 794–265 calBC
Adrar des Ilforas
Tadmakka EKA 5 Seed (Triticum sp.) OxA-16012 961±28 BP calAD 1020–1155
Tadmakka EKA 18 Seed (Ziziphus sp.) OxA-16013 630±27 BP calAD 1288–1398
Tadmakka EKA 77 Seed (Gossypium sp.) OxA-16014 959±27 BP calAD 1021–1155
Tadmakka EKA 85 Seed (Acacia sp.) OxA-16015 960±28 BP calAD 1021–1155
Tadmakka EKA 113 Seed fragments OxA-16016 1086±28 BP calAD 894–1016
Tadmakka EKA 118 Seed fragments OxA-16017 1141±28 BP calAD 777–980
 Table 7.1 (cont.)

Region/Site Context Material Reference Date Calibrated

Tadmakka EKA 45 Pot sherd temper OxA-16150 579±30 BP calAD 1300–1419

Tadmakka EKA 114 Pot sherd temper OxA-16151 1238±30 BP calAD 685–878
Tadmakka EKB 2 Pot sherd temper OxA-16152 1032±30 BP calAD 901–1040
Tadmakka EKA 3 Pot sherd temper OxA-16768 602±26 BP calAD 1299–1406
Tadmakka EKA 87 Pot sherd temper OxA-16769 1087±27 BP calAD 894–1015
Tadmakka EKC 4 Pot sherd temper OxA-16770 650±27 BP calAD 1281–1394
Tadmakka EKA 77 Pot sherd temper OxAx-2178 19 1339±36 BP calAD 640–769
Tadmakka EKA 116 Pot sherd temper OxAx-2178 22 1245±34 BP calAD 678–876
Tilemsi Dak-3 3181±89 BP 1660–1227 calBC
Tilemsi AZ15 Gif-5306 2160±250 BP 808calBC–calAD 378
Middle Niger
Gao Ancien Gx-20623 1105±165 BP calAD 643–1247
Gao Ancien Gx-21656 1160±75 BP calAD 690–1015
Gao Ancien Gx-21657 1430±90 BP calAD 418–770

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Gao Ancien Gx-22806 1005±75 BP calAD 885–1212
Gao Saney ACGS Level 2 Beta-261360 950±60 BP calAD 990–1214
Gao Saney ACGS Level 33 Beta-261361 1170±40 BP calAD 730–972
Gao Saney GS1 Level 2 Wk-12629 1083±42 BP calAD 781–1027
Gao Saney GS1 Level 7b Wk-12630 1297±52 BP calAD 647–869
Gao Saney GS1 Level 14 Wk-12631 1227±42 BP calAD 680–890
Gao Saney GS3 Level 1 Wk-12632 1124±37 BP calAD 777–994
Gao Saney GS3 Level 5 Wk-12633 1189±54 BP calAD 690–972
Gao Saney GS3 Level 12 Wk-12634 1126±44 BP calAD 775–995
Gao Saney GS3 Level 16 Wk-12635 1246±37 BP calAD 675–879
Gao Saney Main mound Charcoal Pa-412 1000±70 BP calAD 891–1204
In Sakane AZ20 Gif-5440 2850±100 BP 1283–811 calBC
Oued Amded Gif-375 2160±150 BP 749 calBC–calAD 135
Tichet AZ25 Gif5230 2000±210 BP 727 calBC–calAD 533
 Timbuktu Gx-24763 170±40 BP calAD 1655– . . .
Timbuktu Gx-24764 240±40 BP calAD 1520– . . .
Tombouze I Beta-256795 Not given cal430–640± BP
Tombouze I Beta-256796 Not given cal120–330± BP
Mauritanian Sahara
Ma’den Ijafen Dak-1 860±108 BP calAD 971–1385
Ma’den Ijafen I-2769 785±110 BP calAD 1024–1396
Tumulus Mauny, Massif Gif-3001 480±90 BP
R’Kiz
Atlantic Coast
Amtal 64280 on bank of Sebkha Shell Gif-3259 2000±90 BP 347 calBC–calAD 226
Amtal H1 Burial in midden (date Human Bone Gif-3466 1780±100 BP calAD 16–532
considered too young)
Bguent Shell Ly-444 2700±100 BP 1187–544 calBC
Cap Timiris Shell Ly-457 2640±120 BP 1051–411 calBC
Cap Timiris Shell Gd-4270 1760±60 BP calAD 128–402
Chami (dune 1) Bone Gif-1763 2450±110 BP 817–258 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Chami (dune 1) Bone Gif-1857 2380±220 BP 996 calBC–calAD 54
Chami (dune 1) Bone Gif-1859 1520±140 BP calAD 145–774
Chami 3 km Shell Gif-2496 3220±110 BP 1756–1216 calBC
Chami (dune 7) Bone Gif-2165 6580±350 BP 6220–4780 calBC
Chami Bone Gif-2167 3310±240 BP 2281–1001 calBC
Chami Pottery temper Gif-1970 3150±140 BP 1744–1026 calBC
Chami Bone Gif-2166 2290±130 BP 766–55 calBC
Chami (Graret Zra 6) Bone Gif-2168 2450±130 BP 836–209 calBC
Chami (Graret Zra A) Shell Gif-1856 3950±80 BP 2838–2200 calBC
Chami (Graret Zra A) Pottery temper Gif-1970a 3650±140 BP 2459–1692 calBC
Chami (Graret Zra A) Charcoal Gif-2492 3500±120 BP 2141–1528 calBC
Chami (Graret Zra A) Shell Gif-2488 3450±110 BP 2112–1501 calBC
Chami (Graret Zra A) Bone Gif-1762 3100±120 BP 1625–1020 calBC
 Table 7.1 (cont.)

Region/Site Context Material Reference Date Calibrated

Chami (Graret Zra A) Bone Gif-2162 2500±100 BP 814–399 calBC

Chami (Graret Zra A) Bone Gif-2161 2170±100 BP 406 calBC–calAD 49
Chami (Graret Zra B) Shell Gif-2164 2360±100 BP 774–205 calBC
Chami (Graret Zra B) Bone Gif-2163 2100±180 BP 745 calBC–calAD 328
Chami (Graret Zra B) Bone Gif-2489 2090±120 BP 401 calBC–calAD 135
Chami (Graret Zra) Bone Gif-3060 3530±220 BP 2257–1320 calBC
Chami (Graret Zra) Ostrich eggshell Gif-2491 2960±110 BP 1437–902 calBC
Chami 10 km N Graret Zra Bone Gif-2333 4190±130 BP 3322–2457 calBC
Chami 8 km N Graret Zra bone Gif-2486 3850±120 BP 2831–1949 calBC
Chami 8 km N Graret Zra Shell Ly-346 3570±120 BP 2282–1621 calBC
Chami 8 km N Graret Zra Shell Gif-2487 3220±110 BP 1756–1216 calBC
Chami 9 km N Graret Zra Bone Gif-1765 2500±100 BP 814–399 calBC
Chami 9 km N Graret Zra Pottery temper Gif-2334 1870±240 BP 406 calBC–calAD 650

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Chedd Allal Ly-4397 2600±80 BP 923–433 calBC
Edjaila H2 Burial in dune sand Human bone Gif-3461 3020±110 BP 1502–941 calBC
Edjaila H3 Burial in dune sand Human bone Gif-3462 3310±120 BP 1911–1303 calBC
El Maharra Charcoal Dak-22 1869±115 BP 159 calBC–calAD 404
Laasailia 63293 Burial under large flat Human bone Gif-3269 3100±110 BP 1681–1265 calBC
stones next to sebkha
Mahariat H11 Burial within shell mid- Human bone Gif-3467 1080±100 BP calAD 695–1160
den (date considered too
young)
Manate Shell KI-570 3200±85 BP 1681–1265 calBC
Manate Shell KI-572 3050±65 BP 1449–1114 calBC
Manate Shell KI-571 2720±85 BP 1120–767 calBC
Manate Cardium KI-593 2100±80 BP 361 calBC–calAD 53
 Nouamghar Shell Gd-4268 3000±70 BP 1414–1031 calBC
Nouamghar Ostrea KI-565 2000±45 BP 158 calBC–calAD 85
Tenioubrar Cardium Ly-351 2660±110 BP 1111–485 calBC
Tijirit Ostrich eggshell Alger-110 1070±100 BP calAD 695–1166
Tijirit Ostrich eggshell Alger-111 1060±100 BP calAD 715–1202
Tijirit Ostrich eggshell Gif-8120 735±50 BP calAD 1189–1389
Tijirit Ostrich eggshell Alger-112 475±100 BP calAD 1297–1638
Tijirit Ostrich eggshell Alger-113 380±100 BP calAD 1319–present
Tijirit 05 Ostrich eggshell Alger-106 1580±90 BP calAD 257–645
Tijirit 05 Ostrich eggshell Gif-8122 1330±40 BP calAD 643–770
Tijirit 10B Ostrich eggshell Alger-109 2170±190 BP 767 calBC–calAD 210
Tijirit 15 Ostrich eggshell Gif-8121 3520±60 BP 2020–1692 calBC
Tijirit 15 Ostrich eggshell Alger-108 3070±110 BP 1607–1011 calBC
Tijirit 24 Ostrich eggshell Alger-107 2145±170 BP 749 calBC–calAD 227
Tijirit 73 Ostrich eggshell Alger-105 1680±100 BP calAD 127–581
Tijirit 78 Ostrich eggshell Alger-104 1560±100 BP calAD 256–657
Tijirit D11 Ostrich eggshell Alger-103 2450±140 BP 893–204 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Tintan necropole Roots Gif-2505 6070±130 BP 5312–4704 calBC
Tintan necropole Shell Orsay-85 5670±300 BP 5302–3945 calBC
Tintan necropole Shell Gif-84 5520±100 BP 4584–4062 calBC
Tintan necropole Shell Orsay-86 5320±150 BP 4455–3799 calBC
Tintan necropole Shell Orsay-87 5020±160 BP 4235–3385 calBC
Tintan necropole Shell Orsay-88 4860±160 BP 4040–3125 calBC
Tintan necropole Shell Orsay-89 4600±200 BP 3903–2780 calBC
Tintan necropole Shell Orsay-90 4570±140 BP 3633–2926 calBC
Tintan necropole Shell Orsay-91 4445±160 BP 3630–2679 calBC
Tintan necropole Shell Ly-503 4270±100 BP 3324–2576 calBC
Tintan necropole Shell Orsay-92 3970±200 BP 3012–1926 calBC
Tintan necropole Shell Orsay-93 3960±100 BP 2864–2147 calBC
Tintan necropole Pottery temper Ly-459 3930±120 BP 2864–2046 calBC
 Table 7.1 (cont.)

Region/Site Context Material Reference Date Calibrated

Tintan necropole Shell Gif-2485 3930±80 BP 2832–2147 calBC

Tintan necropole Shell Orsay-94 3850±100 BP 2576–2027 calBC
Tintan necropole Shell Orsay-95 3805±80 BP 2471–2030 calBC
Tintan necropole Shell Orsay-96 3695±80 BP 2345–1881 calBC
Tintan necropole Shell Orsay-97 3570±160 BP 2436–1527 calBC
Tintan necropole Shell Orsay-98 3550±490 BP 3339–820 calBC
Tintan necropole Shell Ly-460 3530±130 BP 2267–1529 calBC
Tintan necropole Bone Gif-2484 3240±100 BP 1751–1266 calBC
Tintan necropole Bone Ly-505 2860±170 BP 1505–592 calBC
Tintan necropole Bone Gif-1761 2470±100 BP 806–392 calBC
Tintan necropole Pottery temper Gif-1884 2460±100 BP 804–388 calBC
Akjoujt
Akjoujt Charcoal Dak-25 2776±126 BP 1378–592 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Akjoujt Charcoal Dak-29 2522±123 BP 912–384 calBC
Akjoujt Charcoal Gif-1287 2430±110 BP 806–234 calBC
Akjoujt Charcoal Gif-1285 2400±110 BP 798–210 calBC
Akjoujt Charcoal Gif-1286 2360±110 BP 779–203 calBC
Akjoujt Charcoal Gif-1284 2350±110 BP 783–197 calBC
Akjoujt Grotte aux Chauves-Souris Charcoal Gif-1823 2700±100 BP 1187–544 calBC
Akjoujt Grotte aux Chauves-Souris Charcoal Gif-1824 2500±100 BP 814–399 calBC
Akjoujt Grotte aux Chauves-Souris Charcoal Gif-1822 2460±100 BP 804–388 calBC
Akjoujt Grotte aux Chauves-Souris Charcoal Gif-1825 2460±100 BP 804–388 calBC
Akjoujt Grotte aux Chauves-Souris Dak-112 1213±110 BP calAD 639–1022
Akjoujt Grotte aux Chauves-Souris Dak-119 483±115 BP calAD 1282–1646
Lemdena Charcoal Dak-120 1760±110 BP calAD 28–537
Khatt Lemaiteg Pottery temper Ly-2501 3350±130 BP 2014–1316 calBC
 Khatt Lemaiteg Pottery temper Ly-2502 3310±200 BP 2141–1088 calBC
Khatt Lemaiteg Pottery temper Ly-2974 3040±160 BP 1640–853 calBC
Khatt Lemaiteg Pottery temper Gif-6891 2830±60 BP 1192–837 calBC
Adrar
Assabet el-Meddahia Bone Gif-5322 2590±100 BP 922–411 calBC
Assabet el-Meddahia Bone Gif-5323 2500±100 BP 814–399 calBC
Azugi Charcoal Gif-5338 1010±80 BP calAD 779–1217
Azugi Charcoal Gif-5340 950±80 BP calAD 903–1256
Azugi Charcoal Gif-5337 950±80 BP calAD 903–1256
Azugi Charcoal Gif-5336 890±80 BP calAD 1015–1272
Azugi Charcoal Gif-5339 880±80 BP calAD 1020–1271
Draa Malichigdane Human bone Gif-2551 3120±110 BP 1629–1056 calBC
Draa Malichigdane Pottery temper Gif-2550 2030±100 358 calBC–calAD 210
Foum Tizigui Pottery temper Ly-4243 2970±130 1497–856 calBC
Oum Arouaba Hippotamus bone Dak-53 2107±117 399 calBC–calAD 124
Oum Arouaba Human bone Dak-59 1689±110 calAD 86–584
Tichitt

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Akreijit ? 3850±250 BP 3009–1638 calBC
Akreijit Charcoal Dak-32 3776±120 2565–1889 calBC
Akreijit Bone Mc-427 3490±50 BP 1939–1689 calBC
Akreijit Charcoal Dak-203 3122±120 BP 1660–1050 calBC
Akreijit Charcoal I-3565 3100±105 BP 1612–1058 calBC
Akreijit Charcoal Dak-187 2975±110 BP 1447–911 calBC
Akreijit Charcoal Dak-190 2760±160 BP 1394–521 calBC
Akreijit ? Gif-6083 2430±80 BP 778–395 calBC
Akreijit Charcoal Dak-185 1663±113 BP calAD 125–615
Akreijit Aouker lake ashy layer Bone Gif-3406 730±90 BP Aouker lake ashy
layer
Baidla I Charcoal GX-1325 2780±140 BP 1390–555 calBC
Bledd Initi Charcoal GX-1326 2600±105 BP 971–411 calBC
 Table 7.1 (cont.)

Region/Site Context Material Reference Date Calibrated

El Rhimiya Hippotamus bone Gif-2884 3830±120 BP 2619–1938 calBC

Goungou A2 Fish bone GX-1421 3465±160 BP 2272–1421 calBC
Goungou A2 Charcoal I-3561 3350±110 BP 1921–1416 calBC
Goungou A5 Charcoal GX-1323 3205±95 BP 1731–1234 calBC
Goungou B3 Bone GX-1889 3425±130 BP 2121–1435 calBC
Goungou B3 Charcoal I-3563 2950±100 BP 1415–914 calBC
Goungou B3 Charcoal I-3563 2950±100 BP 1415–914 calBC
Niemilane Charcoal I-3819 2170±105 BP 411 calBC–calAD 61
Oued Chebbi Pottery temper Ly-3334 2740±160 BP 1381–484 calBC
Seyyid Ouinquil Charcoal I-3565 3100±105 BP 1612–1058 calBC
Taidart II Charcoal I-3566 2330±105 BP 770–177 calBC
Tichitt (site 17) GX-1890 3700±130 BP 2468–1763 calBC
Tichitt (site 30) Charcoal GX-1324 2700±115 BP 1195–539 calBC
Tichitt (site 30) Charcoal GX-1888 2885±140 BP 1414–808 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Tichitt (site 30) Charcoal I-3564 3205±105 BP 1741–1219 calBC
Tichitt (TIC 70) Hearth Fish bone Gif-4110 2610±110 BP 992–412 calBC
Tichitt (TIC 75) Necropolis of Tichitt village Human bone Gif-4111 1930±110 BP 201–345 calBC
Tichitt (TIC75) Lacustrine deposit Hippopotamus bone Gil-4112 2290±110 BP 756–93 calBC
Tichitt Bone Mc427 c 1540±50 BP calAD 410–615
Tichitt Bone Mc427b 1370±60 BP calAD 565–770
Tichitt Village v.39 Pa-1300 3000±120 BP 1500–919 calBC
Tichitt Village v.39 Pa-1310 2800±120 BP 1374–778 calBC
Tichitt Village v.147 Pa-1290 3060±80 BP 1500–1059 calBC
Tichitt Village v.147 Pa-1328 3590±80 BP 2195–1700 calBC
Tichitt Village v.304 Pa-1292 3360±100 BP 1898–1434 calBC
Tichitt Village v.304 Pa-1294 3085±100 BP 1606–1049 calBC
Bou Khzama II Midden feature S5 Charcoal Pa-2153 3765±35 BP 2291–2042 calBC
Bou Khzama II Pa-2437 3690±60 BP 2279–1916 calBC
 Bou Khzama II Pa-665 3330±38 BP 1730–1513 calBC
Bou Khzama II Inhumation SK1 Human bone Pa-2129 3300±100 BP 1879–1329 calBC
Bou Khzama II surface Pottery temper Pa-2170 3250±100 BP 1766–1281 calBC
Bou Khzama II Pa-653 3231±36 BP 1611–1431 calBC
Bou Khzama II Surface Pottery temper Pa-2128 2840±80 BP 1222–827 calBC
Bou Khzama II Furnace, context 1 Pottery temper GX-30818 AMS 2440±40 BP 756–407 calBC
Bou Khzama II Surface, sample 1 Pottery temper GX-28137 AMS 2340±40 BP 723–236 calBC
Bou Khzama II Pa-671 1643±34 BP calAD 268–535
Bou Khzama II Pa-666 1634±34 BP calAD 340–536
Bou Khzama II Feature S9 Charcoal Pa-2160 1530±50 BP calAD 417–625
Bou Khzama II Surface Pottery temper Pa-2127 1415±120 BP calAD 390–886
Bou Khzama II Surface Pottery temper Pa-2174 1390±100 BP calAD 426–870
Djiganayai Context 6 Pottery temper GX-29358 AMS 3550±40 BP 2016–1756 calBC
Djiganayai Surface, sample 7 Pottery temper GX-29359 AMS 3370±40 BP 1752–1533 calBC
Djiganayai Surface, sample 9 Pottery temper GX-28140 AMS 3260±40 BP 1624–1445 calBC
Djiganayai Context 2 Bos sp. tooth GX-28139 AMS 2780±30 BP 1004–844 calBC
Elb Lehnoucha Pa-661 2080±35 BP 195–1 calBC

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Ngadi Pa-667 3222±42 BP 1611–1420 calBC
Ngadi Pa-2428 3005±60 BP 1408–1056 calBC
Ngadi Pa-668 1192±34 BP calAD 713–952
Nouaoudar Pa-2438 3350±60 BP 1869–1498 calBC
Nouaoudar Pa-2436 3050±60 BP 1434–1126 calBC
Nouaoudar Pa-2435 2985±75 BP 1411–1011 calBC
Chigettomi (rkiz) Bones from under rock shel- Human bone Gif-2769 1400±90 BP calAD 340–536
ter (burial?)
Tagdaoust Dak-46 3556±142 BP 2295–1530 calBC
Tagdaoust Dak-47 2901±151 BP 1450–801 calBC
Tagdaoust Dak-194 1851±100 BP 51 calBC–calAD 401
Tagdaoust Charcoal Dak-96 1261±121 BP calAD 573–1016
Tagdaoust Charcoal Dak-63 1143±124 BP calAD 654–1154
 Table 7.1 (cont.)

Region/Site Context Material Reference Date Calibrated

Tagdaoust Charcoal Dak-195 1170±100 BP calAD 662–1024

Tagdaoust Charcoal UCLA-696 1150±80 BP calAD 689–1020
Tagdaoust Charcoal Dak-94 1066±120 BP calAD 690–1207
Tagdaoust Charcoal Dak-173 1031±100 BP calAD 773–1213
Tagdaoust Charcoal Dak-97 1003±119 BP calAD 774–1255
Tagdaoust Charcoal Dak-172 1017±100 BP calAD 776–1217
Tagdaoust Charcoal Dak-93 881±118 BP calAD 895–1383
Tagdaoust Charcoal Dak-95 831±117 BP calAD 983–1394
Tagdaoust Dak-98 765±122 BP calAD 1026–1408
Tagdaoust Dak-42 692±118 BP calAD 1045–1442
Tagdaoust Gif-265 650±120 BP calAD 1047–1476
Tagdaoust Dak-158 633±120 BP calAD 1051–1616
Tagdaoust Dak-43 595±105 BP calAD 1211–1616
Tagdaoust Dak-45 538±113 BP calAD 1261–1635

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Tagdaoust Dak-118 483±119 BP calAD 1276–1653
Tagdaoust Dak-117 276±121 BP calAD 1440– . . .
Tagdaoust Sa-219 150±100 BP calAD 1522– . . .
Tagdaoust Gif-266 Modern N/A
Tagdaoust DaK-44 Modern N/A
Tagdaoust Dak-48 Modern N/A
Kumbi Saleh Ly-1610 1400±160 BP calAD 262–982
Kumbi Saleh Ly-1792 1280±150 BP calAD 428–1026
Kumbi Saleh Ly-1616 1290±130 BP calAD 437–1018
Kumbi Saleh Ly-1613 1210±140 BP calAD 565–1151
Kumbi Saleh Dak-157 1122±125 BP calAD 665–1155
Kumbi Saleh Ly-1341 1000±150 BP calAD 695–1275
Kumbi Saleh Ly-1526 860±210 BP calAD 695–1438
Kumbi Saleh Ly-1615 980±130 BP calAD 774–1270
 Kumbi Saleh Dak-156 981±114 BP calAD 777–1264
Kumbi Saleh Ly-1612 940±120 BP calAD 778–1285
Kumbi Saleh Ly-1611 870±120 BP calAD 897–1387
Kumbi Saleh Ly-1524 550±230 BP calAD 1018– . . .
Kumbi Saleh Sa-77 740±120 BP calAD 1035–1415
Kumbi Saleh Ly-1520 590±120 BP calAD 1185–1631
Kumbi Saleh Ly-1525 440±180 BP calAD 1252– . . .
Kumbi Saleh Ly-1614 540±120 BP calAD 1257–1641
Kumbi Saleh Ly-1522 500±120 BP calAD 1273–1647
Kumbi Saleh Ly-1523 500±110 BP calAD 1283–1637

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

Kumbi Saleh Ly-1521 230±120 BP calAD 1468– . . .
Kumbi Saleh tombs P-1101 925±30 BP calAD 1026–1182
Kumbi Saleh tombs Muse-162 895±30 BP calAD 1039–1215
Kumbi Saleh tombs P-1102 890±30 BP calAD 1041–1218
Kumbi Saleh tombs Muse-160 875±30 BP calAD 1043–1241
Kumbi Saleh tombs P-1100 835±30 BP calAD 1157–1264
 298 Martin Sterry and David J. Mattingly

The Teda Lands: Kufra, Ennedi, Borku, Tibesti and Kawar

The Tibesti mountains, the Kufra oases and the Kawar oases are the
heartlands of the Tubu, a people also known as the Teda (Fig. 7.1).5 The
Teda were notoriously hardy, unwelcoming to incomers, but well-adapted
to the harsh terrain of the Tibesti massif and the oases in the surrounding
desert areas. Another related branch of the Tubu describe themselves as
Dazas, comprising today mainly semi-nomads who range between Lake
Chad and Borku on the south side of Tibesti.6 The influence of the Tubu
extended across a wide zone of desert and they controlled the key oases and
the trade routes to and from Lake Chad and northwards. The far higher
mortality rates recorded in the nineteenth century on the Tubu slave
caravans compared to the Tuareg ones further to the west can be attributed
to the limited water availability in this region, obliging the Tubu to force
the pace in their marches.7 The Teda/Tubu are linguistically distinct from
other Saharan peoples, have been little impacted on by intermarriage with
other groups (especially Arabs) and are thought to represent a long-
established Saharan people.8 They are perhaps to be equated with the
troglodyte Aethiopes described by Herodotus.9 Rock art from Tibesti
depicts warriors, some mounted on camels and horses, wearing ostrich
feather plumes and armed with large-headed metal spears, and figure of
eight and sub-rectangular shields.10
The ancestors of the Teda were the immediate southern and eastern
neighbours of the Garamantes and will have undoubtedly had profound
contacts with them. The landscapes occupied by the Teda, both ancient
and modern, are extremely harsh and challenging and a significant com-
ponent of their population has tended to be mobile and pastoralists. But,
as we shall see, the Teda territories also contain sporadic small oases and
sedentary communities and we should not underestimate the significance
of such groups in the past. The Classical sources speak of campaigns by
the Garamantes against Aethiopes, including an expedition led by a king
of the Garamantes that travelled south from Jarma for four months.11 If

5
Beltrami 2007; Chapelle 1957; Cline 1950; Nachtigal 1974; Rohlfs 2003, for studies of the Teda.
6 7
Beltrami 2007, 113. Wright 1998.
8
On their distinctiveness from Berber speaking groups in the Sahara, see Blench 2019, 439–40;
Ehret 2019, 487–93.
9
Herodotus, Histories 4.183; Beltrami 2007, 107–131, 63–86.
10
Beltrami 2007, 16–77 (this also includes useful listings for Borku, Ennedi as well as Tibesti);
Capot-Rey 1961; Huard 1953.
11
Ptolemy, Geography 1.8–11 discusses the famous expedition of Julius Maternus who
accompanied the Garamantian king as far south as the rhinoceros country of Agisymba

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 299

Figure 7.1. Eastern Sahara, main regions and sites discussed in the text.

the Garamantes were engaged, as seems likely, in enslaving Africans from

other parts of the Sahara and the Sub-Saharan regions, the Teda may well
have come to play a crucial role in procuring and transporting slaves as
they did in later eras. A hint of this is preserved in the early Arab sources.
In the ninth century, al-Ya‘qubi described the people of Kawar as respon-
sible for bringing in the slaves to Zuwila in Fazzan. Interestingly, Kawar
was the ultimate target of Uqba ibn Nafi’s great raid of AD 666–667.12

(probably Lake Chad). These ‘Aethiopians’ are distinguished in Ptolemy’s account (1.8) from
those who were directly subject to the Garamantes, which perhaps included some of the closer
Teda groups.
12
Levtzion and Hopkins 2000, 13, 22.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 300 Martin Sterry and David J. Mattingly

Kufra
There was been little research on the historic phases of the Kufra oasis and
the traditional settlements of the region are now largely obliterated as
a result of modern development relating to oil and water extraction.13
Prior to a campaign by the Ottoman rulers of Tripolitania in the early
nineteenth century, Kufra was considered a Teda oasis, but in more
modern times the population has become more mixed. Although there
are important water resources in the Kufra depression, the great distances
and difficult near waterless terrain separating this oasis and those to the
east (Dakhla 600 km), north (Awjila 575 km) and west (Tmissa 850 km)
have enhanced Kufra’s reputation as an isolated community.14
Historically, there is some evidence that Kufra has gained in importance
at times when the direct south-north route through Fazzan has been cut or
threatened, with some traffic diverting from the Borku area (c.1,000 km to
the south-west of Kufra).15 The closest links culturally appear to have been
between Kufra and the Tibesti/Ennedi area, as well as with the Teda of that
area.16 The Kufra group comprises a number of distinct oases, with
Tazurbu an outlier to the north-west (250 km from Kufra itself). Some
distinctive pre-Islamic burials have been recorded at Tazurbu, with the
corpse in a seated position, as well as an ancient castle (no longer identifi-
able in the 1990s).17 The main cluster of oases at Kufra are known as the
‘crown’ (al-Tadj), with further outlying small oases at Wadi Zighan,
Bzema, Rabiana. Although modern studies are lacking there are indica-
tions from pre-Islamic tombs and abandoned qsur of a potentially long
chronology here.18

Ennedi
The most easterly of the Saharan massifs is Ennedi. Archaeological
research remains very low in this area of northern Chad, but again there
are some strong indications of pre-Islamic activity in the form of rock art of
the horse and camel phases, funerary monuments and stone monuments of
various types.19 Work by Bailloud and Keding et al. has created a sequence

13
Aerial photographs taken in the 1930s as part of the Italian occupation of Libya provide an
important record of the oases before modern development, but they do not show evidence of
any archaeological remains.
14
See Aroca 1942 (for Waw al-Kabir, an important staging post on the route to Fazzan); Desio
1935, Scarin 1937a; 1937b; Thiry 1995.
15
Meerpohl 2013. 16 Beltrami 2007. 17 Beltrami 2007, 126. 18 Beltrami 2007, 126–29.
19
Bailloud 1969; 1997; Dalloni and Monod 1948; Keding et al. 2007.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 301

from ceramics, OSL and 14C dates, particularly from the sites known as
Shekitye T04/227, T04/234 and Mornou T05/501 and T05/88.20 The last
phase of the ceramic sequence (‘Djoki’) dates to the first millennium AD,
following a hiatus in the first millennium BC (although there is a cattle
skull from this period).21 This period is associated with numerous stone
structures and iron objects that are similar to those found in the Borku
region. This period is associated with an increase in the number of smaller
sites in the region which has been interpreted as evidence of a form of
cattle-based transhumance between Ennedi and the Wadi Howar region
and a more widespread occupation than today.
Whilst surveyed Iron Age sites still await publication, there are numer-
ous archaeological remains visible on the rock outcrops around Fada
including stone villages, enclosures and dense cairn cemeteries. Some are
undoubtedly the Medieval precursors of the modern town, but the num-
bers and association with cairns, the proximity of water sources such as the
Guelta d’Archei and the access to grazing in the highlands is suggestive of
earlier oasis settlement.
South of Ennedi in the Wadai highlands lie the impressive remains of
Wara, one-time capital of the Wadai Sultanate. The remains are considered
to date to the mid-seventeenth century and consist of a large sub-circular
wall (c.330 m in diameter) within which are a number of large enclosures,
one of which contains the mosque, palace and outhouses of the ruler.22 To
the north and south-west are many more enclosures taking the total area of
the urban-scale settlement to around 60–70 ha.

Borku
The desert region between the Ennedi and the Tibesti mountains is also
extraordinarily tough, with relatively few oases.23 The Borku region to the
south of the Tibesti contains a few scattered and small oases, such as Faya
and Fada, but crucially some important salt sources, such as Wadi Doum
and Beda in Borku and Demi to the north of Ennedi.24 The most extensive
survey of the area is by Treinen-Claustre who surveyed 127 sites and
excavated sondages at a small number of these, many of which had iron
objects from the first millennium AD.25 Archaeological remains of mud-

20
Keding et al. 2007, 36–38. 21 Site T05/88–2: Poz-16595, 2480 ± 30 BP, 774–434 calBC.
22
Lebeuf 1962.
23
See Bruce-Lockhart and Wright 2000; Richardson 1853; Rohlfs 1874/1875 for some early
accounts of the routes south of Fazzan through this region and further west towards Kawar.
24
Capot-Rey 1957; 1958; 1959; 1961, 124–30; Schiffers 1957. 25 Treinen-Claustre 1982.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 302 Martin Sterry and David J. Mattingly

brick structures including enclosures, dwellings and a monumental palace

are noted in passing by Lebeuf along the Bahr-al-Ghazal, Bornu and at
Galaka in Borku, he attributes these to the fourteenth to seventeenth
centuries or even more recent eras.26 There are numerous stone-built
enclosures and thousands of cairns across Borku, especially to the north
of Galaka.27
At the northern end of Bahr-al-Ghazal is the Koro-Toro region which is
rich in ironstone. Early iron-working sites consisting of stone structures,
iron furnaces and mounds of iron slag are associated with domestic
cattle.28 Radiocarbon dates imply a broad temporal range – samples from
the base of slag heaps tend to lie somewhere between the late fourth and
early eighth centuries AD (or perhaps a little earlier in some cases) and the
latest dates are in the tenth–fifteenth centuries AD.29

Tibesti
The fan-shaped Tibesti mountains rise to c.3,000 m and cover an area of
c.100,000 km2.30 The ranges contain several volcanic craters and the north-
eastern extremity was the main source of amazonite (a turquoise semi-
precious stone) exploited by the Garamantes.31 Settlements are sparse due
to the main water resources being dependent on irregular rain, rather than
springs. Some of the main settlements are located in the valley outlets from
the massif, others in valleys well within the massif, as at Bardai where
Nachtigal barely escaped with his life in 1850.32 It seems probable that this
has long been the pattern, to judge from the distribution of rock art.33
There are considerable numbers of burial cairns in Tibesti, of very similar
types to Garamantian Proto-Urban phase ones and again strongly hinting
at sedentary activity in pre-Islamic times.34 There are also a number of
26
Lebeuf 1963, 597.
27
Some of these cairns were visited by a team from the University of Siena – many are drum
tombs similar to those in Fazzan and some of the rouletted pottery is also similar (Gatto 2006).
28
Close 1980, 152. Despite the publication of coordinates there are no clear traces on satellite
imagery of these features.
29
Close 1980; Posnansky and McIntosh 1976, 193.
30
On the Tibesti, see, inter alia, Beltrami 2007; Desio 1942; 1950; Nachtigal 1974.
31
Beltrami 2007, 107–13; Mattingly 2003, 356–58.
32
Nachtigal 1974, 286–318, see also 379–422 for his extended discussion of the Teda and their
history.
33
Beltrami 2007; Huard 1953.
34
See Beltrami 2007, figure 2.3.1, with examples of tombs that are intrinsically similar to
Garamantian corbelled cairns (Type 1b) and drum tombs (Type 3b) in the Fazzan typology, see
Mattingly 2003, 196–200. For a full list of 13 major pre-Islamic cemeteries see Beltrami 2007,
78–92.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 303

major enclosures and ritual monuments with curving lines of stones set on
edge.35 The densest concentrations of cairns and enclosures are found
along those wadis with the most vegetation and in which settlement is
concentrated today. Two groups of settlements have been identified by the
work of Rønneseth, the earlier group with huts in the wadis and on the
slopes dates to the late first millennium BC. The second group, higher up in
the mountains dates to the eleventh–sixteenth centuries AD.36 There are
a few radiocarbon dates from Tibesti for the mid-late first millennium BC,
most are from cairn burials, although one implying elephants in the first
millennium BC is probably centuries earlier based on its recalibration
(perhaps as early as c.2000 BC).37

Kawar and Other Oases In and Close to the Bilma Sand Sea
Kawar comprises a north-south line of small oases extending down c.80 km
into the Bilma Erg from its northern edge. The most direct route from Lake
Chad to Fazzan passes along this line and Bilma at the heart of Kawar was
a crucial salt mine.38 Djado and Seguedine are important oases on the
north side of the Bilma sand sea, at the southern end of a very difficult
desert passage south from the last oasis of Fazzan over the pass of Tumu.
Djado is a collective name that is used for a group of three oases: Djado,
Djaba and Chirfa, that lies on the southern edge of a large plateau. At all
three of these sites there are the remains of still upstanding qasaba villages
made from salt-enriched mudlumps.39 According to Chappelle, the inha-
bitants of these villages were Kanuri with Tubu having erected tents
around the palm groves.40 A pair of AMS samples from Djaba returned
dates of calAD 1045–1395 and calAD 1301–1631 although it is not clear
from where these samples were taken or even which village.41 However,
there are some possible signs of earlier settlement. Also at Djaba, there is
what looks suspiciously like a mudbrick Garamantian-style qasr – a square

35
Beltrami 2007, 79–83, for 23 other stone monuments in Tibesti and Ennedi.
36
Close 1984, 10–11; Jäkel and Geyh 1982; Rønneseth 1982a; 1982b.
37
For the burials see Gabriel 1977; 1981; for dated cairns found at Yebbigue and Zoui, Close 1984,
9–10; Posnansky and McIntosh 1976, 185 (HV-2260: 2690 ± 435, 2022 calBC–calAD 138).
38
Lovejoy 1986; Retaillé 1986; Vikør 1999. See also Haour 2017; MacEachern 2019, 418–22, for
reviews of the early evolution of Trans-Saharan trade in relation to Kawar and Lake Chad.
39
Lange and Berthoud 1977, recorded a series of villages in the 1970s where they noted that
qasabah structures were still in use as date stores.
40
Chappelle 1957.
41
Gif-3547 and VRI-810, another date from Djaba and two from Djado were calibrated to post-
calAD 1443.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 304 Martin Sterry and David J. Mattingly

Figure 7.2. Sites in the Kawar Oasis.

building with square corner towers.42 To the north and east of this struc-
ture are traces of a rectilinear field-system with central wells, again similar
to Garamantian examples in eastern Fazzan. Likewise on the southern edge
of Djado there is a square fortification with a small nucleated settlement
that can be associated with an area of abandoned gardens.43 Finally,
5–10 km to the north and east of Djado and Djaba there are several
dense clusters of burial cairns on the rocky edge of the plateau.
Seguedine is a tiny oasis to the south-east of Djado and midway between
Tajirhi and Kawar that is known for the quality of its salt. There does not
appear to have ever been a large population here, but some 4 km to the west

42
Lange and Berthoud 1977 were notified in Kawar of older ruins in this area specifically
referencing a Qasr Maja.
43
Again Lange and Berthoud 1977 note a Qasr Biri and to the south-west a Qasr Garem Djawin.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 305

Figure 7.3. Satellite image of Gezabi (imagery: Google, DigitalGlobe).

of the modern settlement there is a large abandoned settlement with

a central qasaba, this looks similar to the ones at Djado and is presumably
of similar date. That the location has been much frequented in antiquity is
indicated by the cairns that sit on nearly every rocky outcrop in the vicinity.
The Kawar oases stretch underneath an escarpment from Aney in the
north to Bilma, some 80 km to the south (Fig. 7.2). By far the largest
archaeological site lies 5 km to the south-east of Aney where there are the
remains of the 20 ha Medieval town of Gezabi (Fig. 7.3). There are many
well-preserved square houses up to about 25 × 25 m and a few more
monumental buildings including a prominent mosque to the north-east
of the town. Surrounding the settlement are the remains of an extensive
area of oasis gardens (c.300 ha). Gezabi is considered by the inhabitants of
Kawar to be the oldest settlement in the oasis. Mauny associated it with al-
Qasaba (as in the capital) mentioned in al-Idrisi and a more certain
reference to a Ghazbi comes from Ibn Furtu in the fourteenth century.44

44
Levtzion and Hopkins 2000, 123; Mauny 1961, 141. Al-Idrisi also mentions a number of other
towns in a line running south: Qasr Umm Isa, Ankalus (modern Kalala?), Abzar and Talamla.
The latter two have somewhat variable geographies in other sources with Abzar (Abzan, Asben)
associated with the Songhay name for Air (but see below) and Talamla (Tamalma, Lamlama)
with Tibesti, see Levtzion and Hopkins, index entries. The relevant point is that the al-Idrisi
recognised Kawar as consisting of a linear arrangement of large settlements (similar to Fazzan).

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 306 Martin Sterry and David J. Mattingly

Agisymba in Ptolemy’s Geography is another potential, but far less con-

vincing reference.45
Just to the south of Aney there is a c.300 × 300 m settlement (Emi
Tchouma) of more square houses with a mosque. This site lies to the south-
west of a defended hilltop. The house forms are similar to those found at
Gezabi and are also presumably of Medieval date. However, the hilltop
settlement could be Protohistoric (similar ones in Libya date back to the
first millennium BC or early first millennium AD).
Along the hills to the east of the oases are a number of archaeological
sites.46 There are a further 15 villages of stone-built houses. Typically these
cluster around a promontory. There are two distinct groups based on
house morphology: a rectilinear courtyard or blockhouse type and a type
consisting of one to five circular or curved elements. There are nine villages
that only have circular-type buildings and the remaining six have clusters
of both types. In many cases it is clear that the rectilinear buildings have
been built over the circular-type buildings; in others they may be on
opposite slopes or the circular buildings are higher up than the rectilinear
buildings. Five of the six clusters of rectilinear buildings contain a mosque
(usually fairly small). At four of the sites there is also activity on the
associated hilltop with features including walls cutting off promontories
and clusters of small huts. There are a few other possible hilltop sites and
enclosures, but these are generally poorly defined. Additionally there are
a number of dense cairn cemeteries that suggest a permanent pre-Islamic
presence in several parts of the oasis. Finally, to the north of Bilma (itself
mentioned as the largest town of Kawar by Yaqut in the thirteenth
century)47 there is an undated Garamantian-style square qasr (27
× 20 m) with corner and intermediate towers and an entrance on the east
side.
Fachi is a small but important oasis south-west of Bilma on the route
through the Erg towards Agadez and the Aïr.48 To the south-west of the
main oasis there is a relict field-system close to a ruined village with large
circular-type buildings. The village is walled, constructed of mudbricks and
possibly divided into northern and southern halves. There is a sub-
rectangular qasr with square towers on the north-east corner but this is
likely Early Modern in date. There is also a large area of salt extraction

45
Ptolemy, Geography 1.8.
46
These were all noted as in ruins by 1823, Denham et al. 1985, 22–38.
47
Levtzion and Hopkins 2000, 171, 173; Yaqut also adds another town al-Bilas.
48
According to the map of Denham et al. 1985 Fachi should be associated with the town of Asben
mentioned in Medieval Arabic sources.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 307

c.7 km to the north but no visible settlement apart from the buildings
around the digging.
Although all but abandoned, Agadem is the most southerly oasis before
Lake Chad, lying c.200 km south of Bilma. In several places amongst the
hills there are small villages of small circular huts, similar (but simpler) to
those found in the Kawar and several cairn cemeteries that may be asso-
ciated to the villages. Also notable is a walled hilltop site on a small plateau
(undated) and a square enclosure (50 × 50 m).
Finally, in a remote area several hundred kilometres from the nearest
oasis, the Termit massif has evidence of iron working and burial cairns
along with copper and iron objects. A radiocarbon date placed the activity
in the first to sixth centuries AD (further dates have given early dates of
late second to early first millennium BC, but these have come under
sustained critique).49 There are no obvious signs of settlement or much
else on the massif apart from a scattering of cairns (see also the map of
Quéckon and Roset).
The Kawar oasis group (and its outliers) should be a critical focus for
future research. It potentially holds the key to understanding the relation-
ship between the Garamantes and the peoples of the Lake Chad basin to the
south. The account of Ibn Abd al-Hakam of the mid-seventh-century raid
of Uqba ibn Nafi has a ring of authenticity about its overall geographical
context, even if the historical detail is disputed.50 Uqba successively passed
through and received the surrender of the oases of the Jufra (Waddan),
Fazzan (Jarma, other ‘castles’ and Zuwila) and Kawar, following the main
coordinates of the Tripoli-Lake Chad slave route of later times. The main
qasaba of Kawar, evidently a hilltop site, initially held out against him, until
captured by a ruse. His demand of hundreds of slaves from all the rulers
who submitted to him could of course be an anachronism based on later
developments, but it seems more plausible in light of what we now know of
the Garamantes to view the raid as stimulated primarily by a desire to
establish greater Arab control of a pre-existing and lucrative trade in slaves
along this route. The Kawar oases, with their abundant salt sources, had
a valuable resource with which to trade for people, but they were also both
close enough to launch slaving expeditions and protected from pursuit by
the intervening desert terrain. Garamantian trade goods are notably absent
in the Lake Chad basin; if the wider Teda populations were responsible for
much of the slave raiding, then the Kawar oasis is perhaps the place to look

49
Poznasky and McIntosh 1976, 184; Quéckon and Roset 1974.
50
Levtzion and Hopkins 2000, 11–13.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 308 Martin Sterry and David J. Mattingly

for evidence of Garamantian material culture (items such as carnelian and

amazonite beads, glass beads, metals of distinctive alloys and textiles).51

The Tuareg Central Sahara: Tassili and Ahaggar

The Tassili n’Ajjer and Ahaggar are the two main mountain ranges in
southern Algeria with highpoints of 2,158 m and 2,908 m respectively (Fig
7.4). Historically, both have been Tuareg territory with the Tassili facing
the Tadrart Akakus across the north-south trench of Wadi Tanzzuft.52
Again, the historical emergence of the Tuareg may well have overlapped
with the Garamantian era, as their modern rangelands extend up to the
western borderlands of Fazzan. It is interesting that the massifs of the
central Sahara – Tadrart Akakus, Tassili, Immidir and Ahaggar (and Aïr to
the south) contain a great number and density of images of chariots,
especially the representations of the ‘flying gallop’ type. By way of compar-
ison, representations of chariots in the Teda lands of Tibesti and other
zones to the south and east of Fazzan are comparatively rare.53 There was
thus a distinctive cultural relationship between the western and south-
western neighbours of the Garamantes and chariot imagery. Gauthier and
Gauthier have demonstrated a strong association between the distribution
of chariot imagery and early Libyan inscriptions in the Sahara.54 Again the
contrast with the Teda area is marked, suggesting that the modern cultural
divide between Teda and Tuareg builds on something of considerable
antiquity. The rock art, the written script and the funerary monuments
of the mountain massifs to the west of Fazzan suggest that these areas had
some cultural links with the Garamantes. Although these areas were
perhaps best suited to support pastoral groups, who might have provided
beasts of burden, desert guides and military muscle for the Garamantes, we
should not neglect the possibility that some at least of the small oases of
these areas also developed early.

Tassili n’Ajjer
Although famed for its rock art (a UNESCO World Heritage Site), there is
little evidence to suggest that the Tassili ever had a substantial sedentary
51
MacEachern 2019, 417–18. 52 See Mattingly et al., Chapter 2, this volume.
53
Gauthier and Gauthier 2011, 93–95. For five examples of chariots depicted in rock art from
Teda lands, see Huard and Féval 1964, 85.
54
Gauthier and Gauthier 2011, 105–14.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 309

Figure 7.4. Mali and Niger, main regions and sites discussed in the text.

population.55 The mountain valleys are rich in pasture and can still run
with water after rain. The major oasis group is at Janet in the south-east
which comprises three centres along the Wadi Egerew. In addition to rock
art (which is of all major periods), finds from excavated burial cairns at
Chaâba Arkouya include a bronze bracelet and Roman pottery from
the second–third century AD and further tombs on the bank of the Wadi
Tin Arilane (possibly the same site) suggest that these burials are similar in
rite and material to those found in the Wadi Tanzzuft.56 However, evi-
dence of settlement is far scarcer with no oases mentioned in either
Classical or Medieval sources, and the establishment of the modern oasis
55
On the Tassili see Hachid 1998; 2000.
56
Gast and Hachid 1995; Gatto et al. 2019, 108–33; Grébénart 1988; Lhote 1971 for a description
of the Wadi Tanzzuft.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 310 Martin Sterry and David J. Mattingly

centres placed as late as the fifteenth or sixteenth centuries.57 Analysis of

satellite imagery across the Tassili shows a thin spread of small huts,
enclosures, and other monuments across the mountain range which is
consistent with a largely mobile population. On the western flank of the
Wadi Egerew, adjacent to the oasis gardens are three small nucleated
cemeteries with 20–50 probable drum tombs in each.58 There is a fourth
on the eastern flank although this has been damaged by modern develop-
ments which are encroaching up the slopes.
A smaller oasis group is found in the centre of the Tassili, in a bend of the
Wadi Iherir, which contains a seasonal river running from October to
April. Perret recorded three groups of ruins, one next to a village (Hiheras)
which he considered to be three to four centuries old and two others with
rectangular rooms which he considered likely to be Roman.59 This attribu-
tion is dubious in the extreme, but his plan of Ikakaden shows three
buildings, the most complete of which is 4.5 × 18 m, divided into five equal-
sized rooms and made from cemented stone.60 The third might still survive
(Tadjart) although there is little about the plan to support an ancient date.
As at Janet there are a few small nucleated cairn cemeteries visible on the
satellite imagery. A slightly confusing addendum is the finding of fourth-
century AD Roman coins and ceramics from Tadrart and Dider. Salama
and Wilson placed this south of Ghat, but it seems more likely that these
toponyms belong in the Wadi Iherir, giving further weight to a pre-Islamic
presence around this oasis.61 The proximity encourages comparison to the
Wadi Tanzzuft and Wadi al-Ajal in Fazzan, but given the fluvial action of
the wadi (around three days of floods a year were recorded in the mid-
twentieth century),62 oasis settlement such as is found at Fewet or Saniat
Jibril is likely to have been buried or swept away.63 However, given the
limited number of burials this is perhaps unlikely to have amounted to
more than a hamlet or two. The urban expansion of the oasis appears
directly linked to economic and political developments in the Early
Modern period. Lhote also mentioned a ruined enceinte with towers at
the head of Wadi Aitou on the track from Iherir to Dider, though we have
not been able to relocate this.64
57
Duveyrier 1864, who provides the first written account of the oasis.
58
One of these is likely the cemetery excavated by Lhote.
59
Perret 1935; 1936; cf. also Lhote 1984, 117.
60
The plan is not dissimilar to the buildings on the south-east flank of Zinkekra (ZIN003 –
Mattingly 2010, 65–74), although the construction technique of mud-mortared stone is far
more reminiscent of Medieval and Early Modern settlements in the Wadi al-Ajal.
61
Salama 1981; Wilson 2012. 62 Gast and Hachid 1995.
63
See Mattingly et al., Chapter 2, this volume. 64 Lhote 1984.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 311

Ahaggar Mountains
The Ahaggar mountains comprise one of the areas of the Sahara with the
highest archaeological potential, but it has been long neglected.65 The most
famous discovery in this area is the tomb of Tin Hinan at Abalessa on the
south-west fringe of the range.66 The impression of a Roman coin, glass-
ware, a lamp and some of the jewellery found with this burial indicates
a society that was in contact with the Roman world. Tin Hinan is regularly
described as relating to a pastoral group, but the burial site in fact lies
adjacent to a major valley with evidence of past cultivation. Furthermore,
baskets of dates and cereal grains were found with the burial.67 The burial
was actually placed within a room of a fortified building (qasr), which was
later converted into a funerary structure. It is thus equally possible that the
burial related to a leading member of an oasis/sedentary community in the
Ahaggar. Even if the burial of Tin Hinan was a late antique insertion into
a former dwelling by a pastoral group, the existence of the qasr surely
implies an earlier sedentary component of the local population. Lhote
mentions two other fortified sites (qsur) at Tit and Silet, though he sug-
gested they might be of ‘recent’ origin.68 There are numerous cairns
throughout the mountains, many of substantial size, but these do not
form clustered cemeteries. The pre-Islamic tombs of the Ahaggar include
monuments that are very similar to those of Fazzan, with some drum cairns
accompanied by proto-stelae.69
A multi-area site on a plateau has been excavated at Tahabort to the
north-east of Tamanrasset with three dates ranging from the seventh to
thirteenth centuries AD, one earlier, and one later.70 Like the Tassili, the
region is rich in rock art and has a strong prehistoric tradition. There is
a notable concentration of horse period rock art here, including numerous
chariot representations.71 The region also has a distinctive form of foggara
irrigation with channels running along wadis rather than into them.72
Little is known of the history of this region before the nineteenth century.
On the route south-west from the Ahaggar to the Adrar des Ilforas and
the Niger bend at Gao is the well of Timmissao, another suggestive site,
where Roman coins have been reported in the past as well as a chariot rock

65
Lhote 1984, 67–119 for a summary of the archaeology; Yacono 1968, for a detailed geographical
study, which makes some reference to irrigation of the wadis within the massif (148).
66
Camps 1997; Prorok 2001; Reygasse 1950, 88–106. 67 Reygasse 1950, 98.
68
Lhote 1984, 117.
69
Lhote 1984, 111 for a photo of a tomb that looks very similar to those of Garamantian Fazzan.
70
Close 1984, 13; Maître 1976. 71 Gauthier and Gauthier 2011.
72
Capot-Rey 1953, 327–29.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 312 Martin Sterry and David J. Mattingly

art scene, a possible Latin or Latino-Punic inscription and an Islamic

inscription purportedly from the eighth century.73 This again looks like
a route that was developed from an early date, though there is no modern
oasis development in the vicinity or evidence of permanent settlement in
the past.74

The Tuareg Southern Sahara: Aïr and Adrar des Ilforas

There is substantial evidence of human activity in the southern part of the

Tuareg range-lands. Multiple sites have been surveyed and funerary struc-
tures can be traced at least as far south as Tahoua (14°53’ N) making them
the most southerly group of cairns in the Sahara.

Aïr
The northern part of Aïr has lower rainfall than the south and has a very
low population density, similar to the other Saharan mountain ranges.
Burial cairns are widespread on and around the massif and are generally
thought to date between the Late Neolithic and the first millennium AD.75
Excavations at a number of sites have demonstrated the long chronology of
this form of site and samples from burials at In Teduq, Mentes, Takene
Bawat, Inufan, Tuluk, Iwelen, Mammanet and Agadez have been dated to
the period 1000 BC–AD 1000.76 In general though, the excavators have
argued that these are the graves of nomads or semi-nomads.77
The site of Iwelen in the north-eastern part of the Aïr region is the most
thoroughly investigated location and is notable for the recording of
a settlement and major rock art scenes alongside several cemetery zones,
from which more than 50 tombs have been excavated.78 Many of the tombs
have been radiocarbon dated and the funerary activity evidently spanned
a long period from the second millennium BC to the late first
millennium AD. On the other hand, the main use of the settlement

73
Capot-Rey 1953, 187; Mauny 1956, 252; Salama 1981, 514–15. Pichler 1997 for a possible
Latino-Punic inscription with Libyan undertones.
74
There are also a very large collection of crescent cairns on the eastern side of the massif of
presumably Neolithic date.
75
Beltrami 1987. 76 Haour 2003; Paris 1996; Paris et al. 1986.
77
Though note Dupuy 2010, argues that this was ‘un pastoralisme peu mobile’ on account of the
emphasis on horses in the rock art, which are not suited to wide-ranging nomadism in the
Sahara.
78
Paris 1990; Roset 1988; 2007.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 313

apparently fell in the mid-first millennium BC.79 The settlement and rock
art lies in a bend in the Wadi Iwelen, which is known for seasonal floods,
with the cemeteries mainly on the rocky plateau above the Wadi. The
settlement comprises an area with traces of stone footings, many saddle
querns and abundant pottery. Although some lithics were found here, the
site is also notable for the occurrence of copper alloy items, including
spearheads. The rock art includes images of biconical human figures,
associated with chariots and horses, wearing elaborate headdresses and
carrying spears. The finds from the burials included leather shrouds,
patterned textiles, copper alloy and bone bracelets, handmade pottery,
saddle querns and grinding stones. Although Iwelen is located in a very
forbidding environment today, with no sedentary settlement in the vici-
nity, the seasonal floods in the wadi could have supported some cultivation,
especially if rainfall was marginally higher or more predictable in the past.
In the south of the massif, there are a number of investigated settlements
of Medieval or more recent date. Agadez has been the primary urban centre
in the Aïr region since the fifteenth century and part of a sequence of
capitals (in order Assodé, Tadeliza, Tin Chamane, Agadez) culminating in
the Tuareg Sultanate of Agadez. From Agadez, there are routes that run
north towards Tamanrasset and the Ahaggar (c.650 km), east (with
a difficult crossing of the Ténéré sand sea) to Bilma and the Kawar oases
(c.750 km) and south-west towards Niamey and the Niger (600 km). Its
foundation is normally placed in the eleventh or fifteenth century with the
most reliable early description that of Leo Africanus who implied that it
consisted of ‘Barbary’ (that is, Berber) houses. Little remains of the
Medieval town apart from the mosque, the earliest phase of which is
dated to the fifteenth century.80 The most northerly of the capitals of Aïr
is Assodé, the ruins of which encompass an area of 60 ha (although the
large compound houses imply a much lower population density than
towns in the northern Sahara). It is positioned high in a basin on the
wetter, west side of the massif with a few satellite settlements of similar
construction type nearby. Excavations in six locations have recovered
material that dates no earlier than the late thirteenth/fourteenth
century AD, although as elsewhere in the massif, cairns of ‘Protohistoric’
type are widespread. A little to the south, closer to the still inhabited Timia
region, there are some small hilltop enclosures with possible huts within
them and to the north-east there are many villages of stone-built structures
79
Paris 1990, 67–71.
80
Cressier and Bernus 1984. The site was recently added to the UNESCO World Heritage List
which provides the most up-to-date survey of the old town.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 314 Martin Sterry and David J. Mattingly

of different types to those found at Assodé, some have nearby Islamic

cemeteries. Together, these point to a more complex settlement history
than has previously been put forward and raises the possibility of earlier
agriculture in this region. On the southern edge of the Aïr massif are
Anisamane and Tadeliza, 31 and 20 km north-west of Agadez respectively
and as with Assodé, both are considered to be former capitals of the region.
These have been examined by Lhote with Anisamane dated to the fifteenth
century or later and the citadel of Tadeliza to the eleventh–fifteenth
century.81 A further Medieval settlement is known at Kori Ihine on the
eastern edge of Aïr.82 There are around 20 small rectangular houses (3–4
× 2.5 m) with thick stone walls and a high concentration of pottery and
lithics (including a red-burnished pitcher with black painted designs).
Satellite prospection in this area suggests a diffuse settlement with many
sub-circular enclosures and many drum tombs in the surrounding hills.
Earlier settlements are somewhat harder to come by. The tail end of
occupation at a number of Neolithic sites (for example, Arlit) has yielded
a few first millennium BC radiocarbon dates, but these do not indicate
anything more than seasonal occupations in the Aïr massif. Given the
nature of habitation remains (or lack thereof) from Marandet it is difficult
to draw hard conclusions about whether there was or was not permanent
settlement in the first millennium BC/AD, but certainly there were suitable
conditions.

The Azawagh Region

The Azawagh valley is a dry wadi system that runs from the Aïr mountains
westwards towards the Niger river. Although a running river has not
existed since the Neolithic, the lowlands have some vegetation – indicating
sporadic run-off. To the west of Agadez is Azelik the assumed location of
the mines and town of Takadda.83 Here archaeological remains stretch
over several kilometres including settlement agglomerations, copper
mining and cemeteries. The most important part of the site is Azelik wan
Birni where there is an urban-scale site of some tens of hectares sur-
rounded by Islamic burials.84 Excavations and radiocarbon dates suggest
that the main period of occupation was the fourteenth to sixteenth

81
Calvocoressi and David 1979, 20; Lhote 1973; Posnansky and McIntosh 1976, 183.
82
Roset 1977. Dak-146, 685±100 BP, calAD 1058–1439.
83
Bernus and Gouletquer 1976; Lhote 1972; see also Nixon, Chapter 17, this volume.
84
See Bernus and Gouletquer 1976, 27–31; regrettably much of the site has been disturbed or
damaged since its initial exploration.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 315

centuries, but tumuli just to the south imply some earlier occupation of the
area. A few kilometres to the west are the salt mines and adjoining settle-
ment of Tegidda-n-tesemt, on the basis of ceramics this appears to be
a successor to rather than contemporary with Azelik. Between Agadez and
Azelik six Medieval settlements have been identified, with three phases of
settlement that show variation in trade routes, copper mining and
demography.85 From Azelik to the area of Agadez there are a number of
sites with evidence of copper metallurgy and whose dating is the subject of
some argument. As this review is concerned with evidence for settlement,
we shall not revisit this debate except to note that despite the many sites,
there is little direct evidence of Protohistoric habitation or structures. They
do however imply the presence of metalworking communities in the
general region from at least the late first millennium BC and into the mid-
second millennium AD.
Further west again, a sondage in the so-called ‘western structure’ in the
small settlement of In Teduq produced two early dates of material (ninth to
third century BC and seventh to tenth century AD) although it is difficult
from the excavation to distinguish whether this related to an earlier settle-
ment or the continual use of the place as a convenient stopping point.86
The rocky outcrops to the south and west of the site have a dense covering
of cairns of various types.

Marandet
In recent years there has been some important work at the site of Marandet,
which lies at a convenient break in the topography to the south of Agadez.
An area of c.60 ha is covered with scatters containing c.50,000 clay crucibles
used for copper working.87 Geophysical survey and excavation have shown
that at least some of the crucibles are from pits whose primary use appears
to have been for dumping domestic waste.88 These pits have been radio-
carbon dated between the sixth–ninth century AD and along with evidence
for copper working they contained finds of imported pottery from North
Africa, carnelian and amazonite beads, glass objects and date stones. These
finds reopen the debate about pre-Islamic Saharan trade and by extension
of settlements and oases in this region as well.89 By the ninth century the
site was appearing in Arabic sources as a key node between Gao and
85
Bernus and Cressier 1991; 1999, 370; Haour 2003.
86
Bernus and Cressier 1999, 288–90. 2450±150 and 1210±60.
87 88
Grébénart 1985; Lhote 1972; Magnavita 2013; Mauny 1961. Magnavita 2013.
89
Magnavita 2013; 2017.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 316 Martin Sterry and David J. Mattingly

Zuwila. However, in the twelfth century al-Idrisi described it as populous,

but with little merchandise or industry.90 It is also helpful to note that
similar pits in the modern village were dug for house construction and this
may be the only archaeological remains of the structures of the settlement.
These features are not visible from the surface and show the limits of what
can be viewed from satellite imagery.

Adrar des Ilforas

The rock art of the Adrar des Ilforas range shares many similarities with the
material from Aïr, including the presence of horses, chariots and metal
weapons.91 However, less is known of the contemporary cemeteries and
there is a lack of certain settlement evidence. Recent research at the early
Islamic town of Tadmakka has not recorded any evidence of pre-Islamic
activity there.92 There are no major oases in the Adrar des Ilforas, but the
wadis flowing out of the massif do receive sporadic floodwaters after rains.
As with all the massifs, cairns are widespread and do concentrate around
those wadis with more vegetation, particularly on the western side of the
massif. There are also occasional abandoned stone buildings of unknown
date, but not too dissimilar to Tadmakka in form. The exception to this
sparse evidence is the region immediately around Kidal. The hills here have
many stone structures on them whose date and function is unknown.

Taghaza and Tawdenni

The salt mining centres of Taghaza and Tawdenni lie in the Western
Sahara c.700 km north of Timbuktu and played a significant role in the
Medieval trading networks of that region.93 The former is dated from
the thirteenth–sixteenth centuries, after which point it was replaced by
the latter site, where mining still continues. At Taghaza there are two
settlements on either side of the sabkha, possibly these are of different
date, but both have remains of mosques.94 At Tawdenni, to the north of the
prison are the ruins of Ksar de Smida which covers around 2 ha and has
large cemeteries to the south-east. As at Bilma, it would be interesting to
investigate the possibility that salt extraction began earlier.

90
Levtzion and Hopkins 2000, 46, 125. 91 Dupuy 2010.
92
Nixon 2009; 2013; 2017; Chapter 17, this volume. 93 Lydon 2009, 89–90; McDougall 2006.
94
Mauny 1961; Monod 1938.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 317

Figure 7.5. Mauritania sites discussed in the text.

A little further south on the route to the Niger is Arawan, there is little
information about this site and today it consists of a handful of houses in
between the sand dunes with a dozen or so palms. There are traces of some
structures to the south, but it is unclear what the antiquity of this place is.

Middle Niger ‘Oases’

The Niger Bend

This area is covered later in this volume by the chapters of McIntosh,
MacDonald and Nixon. Occupation from the early Medieval period has
been convincingly demonstrated by modern excavation at Timbuktu and

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 318 Martin Sterry and David J. Mattingly

Gao.95 However, it remains an open question whether there may be sites

close to these important centres with earlier antecedents. In the case of
Timbuktu, a plausible case can be advanced for Tombouze 1.96 There was
certainly no shortage of advanced pre-Islamic urban-scale settlements
around the Niger Bend area as a whole and there are increasingly signs
that some at least of these were in contact with the Mediterranean world,
albeit probably indirectly.97

Dhar Tichitt, Nema, Walata and Tagant

The Tichitt culture is described by MacDonald elsewhere in this volume.98

There are numerous settlements located along the escarpments that have
seen several campaigns of survey and excavation. The settlements date
mainly to c.1600–400 BC with evidence of precocious cultivation of cereals,
metallurgy and a four-tier settlement hierarchy. There are also many tombs
and rock art of a Saharan tradition. The settlements are easily detectable
from satellite imagery and a thorough survey of remains would likely
identify hundreds of archaeological sites. Of particular interest is the area
of the Tagant to the west where there is very varied morphology including
both circular and rectilinear buildings that suggest multiple chronological
phases (as in southern Morocco). Ould Khattar has identified a late phase
of the Tichitt culture in this region and the presence of water (and
crocodiles) in its wadis makes continuous occupation more likely, but
not certain.99 A tomb with a funerary chapel identified by Monod at
Foum el Hadjar is potentially of first millennium AD date.100

Trading Centres in the South-West Sahara

In the South-western Sahara there are a number of major trading centres of

Medieval date, contemporary with the towns of Sijilmasa, Zagora and
Tamdult in Morocco (Fig. 7.5). The best known are Tagdaoust/
Awdaghast and Kumbi Saleh both of which have had several campaigns

95
Cissé 2010; 2017; Insoll 1996; 2000; Nixon 2009; 2013.
96
MacDonald 2011; Chapter 13, this volume; Park 2010.
97 98
McIntosh, Chapter 14, this volume. MacDonald, Chapter 13, this volume.
99
Ould Khattar 1995.
100
Chapelle and Monod 1937, 518 (one of a number of examples of ‘tombes à chapelle’ in
Mauretania).

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 319

of excavations that date the early phases broadly to calAD 660–1020 and
calAD 430–1020, respectively.101 Azugi, Wadan, Chinguetti, Rachid,
Tijkja, Kasr el Barka, Tichitt and Walata are other Mauritanian oases
with substantial settlements of Islamic date in the Western Sahara that
might all repay investigation for earlier antecedents – again if not at the
known Islamic centres, then in the vicinity.102

Azugi
Although mentioned by both al-Bakri and al-Idrisi as a strategic stop on
routes to the south, little is known of this site and it is not mentioned after
the twelfth century.103 It apparently existed as a town before falling under
the control of the Almoravids in the eleventh century. It is located in the
centre of the Adrar area of Mauritania, close to modern Atar. When the site
was investigated in the 1950s there were remains of a fortress (100 × 70–
80 m with 20 bastions) located to the south of the palmerie. Around 1 ha of
ruined buildings were located 100 m to the north-west.104 Excavations
outside the fortress in the 1980s identified two phases of rectangular stone
buildings with the earliest (at a depth of 5 m) producing dates of calAD
779–1217 and calAD 1020–1271 (similar dates also come from deposits
under the walls of the fortress).105 Three radiocarbon dates from the
fortress walls themselves indicate that this was much later, probably con-
structed in the fifteenth–seventeenth centuries.106 Around 7 km to the
north-west of the main oasis there are the remains of a c.2 ha hilltop
settlement of enclosures and small round huts of unknown date. A loose
scatter of cairns and walls are also found on the escarpments to the east and
west of the oasis.

Chinguetti
Moving to the east, Chinguetti is a much better preserved town with
a famous dry-stone mosque and major Medieval manuscript libraries
reflecting its history as a gathering place for pilgrims.107 Little archaeological
101
On Tagdaoust see Devisse 1983; Polet 1985; Robert et al. 1970; Robert-Chaleix 1989; Vanacker
1979. On Kumbi Saleh see Berthier 1997; Capel et al. 2015. For a full discussion of both see
Nixon, Chapter 17, this volume.
102
See Robert 1982 for a study of mosques from several of these sites.
103
Levtzion and Hopkins 2000, 73, 127–28. 104 Mauny 1955.
105
Saison 1981; Gif-5337, Gif-5340, Gif-5336, Gif-5338 – 880 ± 80, Gif-5339 – 1010 ± 80.
106
Gif-5331, Gif-5332, Gif-5333. Also Gif-5334 – 200 ± 80 BP.
107
Mauny 1955; Ould Mohamed Yahya and Rebstock 1997; Stewart et al. 1990.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 320 Martin Sterry and David J. Mattingly

investigation has taken place at the site, but the town dates back to the
twelfth to fifteenth centuries.

Wadan
The most easterly town in Adrar is Wadan. Like Chinguetti this is probably
of Medieval origin, although the earliest references to the toponym are in
Portuguese accounts from the fifteenth century.108 A mosque of presumed
Medieval date, part of an enceinte and many other buildings are preserved
on the west side of the town.
There is potential for early settlement along the valleys of Adrar and
indeed some cairn cemeteries to the north of Atar and along the south-
eastern edge may be indicative of such activity north of Atar, but to date no
habitation sites have been reported.

Ksar el Barka
Between Adrar and the Dhars, Ksar el Barka is a town of c.10 ha reputedly
founded in 1690. The ruins today are partly obscured by windblown sand
but were visited by Monod.

Walata
Walata is situated in the Dhars of southern Mauritania at a nodal point on
the edge of the Sahara. The earliest settlement is of the Tichitt tradition (see
above, p. 317) from c.1000–200 BC of similar type to that described by
MacDonald.109 From the thirteenth century AD a town started to develop
with links to trade from the salt mines of northern Mali and the site’s name
was changed from Biru (Mande) to Iwalaten (Berber).110 It appeared rather
late in Arabic sources in the fourteenth century where it was described as
the next stopping point south after Tabalbala (see above, p. 278) and was
visited by Ibn Battuta.111

Other towns in the Dhars

Within the area of the Dhars are a number of towns that have seen little
archaeological attention: Rachid, Tijika and Tichitt (at the foot of the
108
Mauny 1955.
109
Munson 1980; see MacDonald, Chapter 13, this volume for discussion of Tichitt.
110
Cleaveland 2002. 111 Levtzion and Hopkins 2000, 276, 282–86.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 321

escarpment of the eponymous Dhar). They all have stone-built rectilinear

architecture in varying states of preservation and are most likely Medieval
or later in date.112

Akjoujt

The area of Akjoujt is an area of relief c.200 km to the east of the Atlantic
coast and at a latitude that has at points been at the border of the Sahel.
There is minimal water today, although there is an aquifer at the town of
Akjoujt. The region is well known for its mineral resources and there is
evidence of copper mining and processing sites, the best-known of which is
the Grotte aux Chauvres-souris which dates to the mid-first
millennium BC.113 Further evidence of copper working comes from the
sites of Lemdana and Touizigt where metalworking slags have been found
along with weapons, tools and objects of personal adornment. Although
there is a large cemetery of drum-tombs in Akjoujt at Lembetet el Kbir, no
contemporary settlements have been identified. There is also very little
evidence of activity in this region between the late first millennium BC
until the Medieval period when activity restarted in the Grotte aux
Chauvres-souris. It has been suggested that this mining activity reflects
the periods which were wetter and in which the edge of the Sahel lay further
to the north.114 Finally, there are a number of Roman finds from southern
Mauritania, including coins from Resseremt (near Akjoujt) and
Tamkartkart (near Tichitt) and a bronze fibula from Akjoujt.115

The Atlantic Coast

Between the Draa (in Morocco) and Adrar there are no major oases and few
wells. However, there are some distinctive sites that suggest a degree of
sedentarisation occurred here. Along the coast a number of sites have been
identified that have their origins in the Neolithic, but continued to be used
into the first millennium BC.116 These have little or no significant architec-
tural remains, but artefact scatters stretch over more than 1 ha in area.
Shellfish, fish and marine mammals were of primary importance to the
diet of these communities, particularly given the aridity of the region
112
Monod 1948. 113 Lambert 1986; Vernet 1993.
114
See McIntosh Chapter 14, this volume. 115 Mauny 1956; 1978; Salama 1981, 514–15.
116
Close 1984.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 322 Martin Sterry and David J. Mattingly

(although shell middens do not occur until south of Nouackchott). Further

south in Mauritania the sites are found along the edges of coastal lagoons.
Cemeteries were part of, or close by many of these sites, and follow Saharan
styles of burial including both earlier (for example, antenna tomb) and later
(chapel and drum tomb) types.117 Important sites include Tarfaya in
Morocco; Amtal, Wadi Tifarti and Mahariat in Western Sahara; and
Chami, Tintan and a cluster of sites around Nouakchott in Mauritania.
Vernet has studied the Mauritanian coast in more detail and identified
a later phase of shellfish harvesters: the Tin Mahham culture 400 BC–AD
300.118

Conclusions

The southern Sahara is a region of few major oases, though some of the
salt-producing centres have been historically very important on account of
that vital resource of Saharan trade. The relative paucity of water sources,
apart from sporadic and unpredictable rainfall watering pasture lands in
the wadis flowing out of the massifs, has favoured adventitious exploitation
of these desert landscapes by pastoral groups. In recent times, the southern
Sahara has been dominated by two major groups, the Teda in the Libyan
Sahara and the Tuareg between Fazzan and the Niger. The archaeological
evidence reviewed above suggests that this broad division is of long rele-
vance. The burial of Tin Hinan was celebrated by the Ahaggar Tuareg as
their founding princess long before Prorok dug her up in the 1920s. The
proto-Teda and proto-Tuareg of the Garamantian era of the areas surveyed
above were thus most likely predominantly pastoralists and occasionally
bellicose. But the evidence of potential sedentary farming elements should
be recognised and future research needs to investigate this possibility in the
environmental niches where cultivation is possible.
A final point to make about the Tuareg (Berber) and Teda peoples of the
southern Sahara is that they were clearly operating in the early Medieval
period from centres within the desert north of the Sahelian communities of
the Chad Basin and the Niger. The Kawar oases played a key role not only
in supplying Chad with salt, but also were a gathering point for slaves to be
traded north. It is possible that some of the centres that were established
north of the Niger bend (Tadmakka is a case in point) served a similar
117
Vernet 1993. There are a large number of radiocarbon dates from these cemeteries that
similarly confirm their long chronology from the fourth–first millennium BC.
118
Vernet and Tous 2004.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 323

purpose in relation to developing trade there. Such sites were located far
enough into the desert to be difficult to reach from the Sahelian lands, but
close enough to facilitate exchanges and predation by slave-trading
societies.

References

Aroca, A. 1942. Ual el Chebir: L’oasi della redenzione. Milano: Edizioni Alpe.
Bailloud, G. 1969. L’évolution des styles céramiques en Ennedi (République du Tchad).
In Mémoires I, Actes du Premier Colloque International d’Archéologie Africaine. Fort
Lamy: Institut National Tchadien pour les Sciences Humaines, 31–45.
Bailloud, G., 1997. Art rupestre en Ennedi. Saint-Maur: Éditions Sépia.
Beltrami, V. 1987. Repertorio preistorico-archeologico del territorio dell’Aïr ed aree
limitrofe. Rome: Istituto Italo-Africano.
Beltrami, V. 2007. Il Sahara Centro-Orientale Dalla Preistoria Ai Tempi Dei
Nomadi Tubu: The Central-Oriental Sahara from Prehistory to the Times of the
Nomadic Tubus. Oxford: Archaeopress.
Bernus, E. and Cressier, P. 1991. La région d’In Gall-Tegidda n Tesemt (Niger)-
programme archéologique d’urgence, 1977–1981. Azelik-Takadda et l’implantation
sédentaire médiévale. Niamey: Institut de Recherches en Sciences Humaines.
Bernus, E. and Cressier, P. 1999. In Teduq du moyen-âge à l’époque actuelle. (Vallée
de l’Azawagh (Sahara du Niger) vol 2). St Maur: Editions Sepia.
Bernus, S. and Gouletquer, P., 1976. Du cuivre au sel. Recherches ethno-
archéologiques sur la région d’Azelik (campagnes 1973–1975). Journal des
africanistes 46.1: 7–68.
Berthier, S., 1997. Recherches archéologiques sur la capitale de l’empire de Ghana:
Étude d’un secteur d’habitat à Koumbi Saleh, Mauritanie. Campagnes II-III-IV-V
(1975–1976)-(1980–1981) (Cambridge Monographs in African Archaeology 41/
BAR International Series 680). Oxford: Archaeopress.
Blench, R. 2019. The linguistic prehistory of the Sahara. In Gatto et al. 2019,
431–63.
Bruce-Lockhart, J. and Wright, J. 2000. Difficult and Dangerous Roads. Hugh
Clapperton’s Travels in Sahara and Fezzan 1822–1825. London: Sickle Moon Books.
Calvocoressi, D. and David, N. 1979. A new survey of radiocarbon and thermo-
luminescence dates for West Africa. Journal of African History 20.1: 1–29.
Camps, G. 1997. Tin Hinan et sa légende. A propos du tumulus princier d’Abalessa
(Ahaggar, Algérie). Bulletin archéologique du comité des travaux historiques et
scientifiques ns 24: 173–95.
Capel, C., Zazzo, A., Saliège, J-F. and Polet, J. 2015. The end of a hundred-year-old
archaeological riddle: First dating of the columns tomb of Kumbi Saleh
(Mauritania). Radiocarbon 57.1: 65–75.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 324 Martin Sterry and David J. Mattingly

Capot-Rey, R. 1953. Le Sahara Français (L’Afrique Blanche Francaise II). Paris:

Presses universitaires.
Capot-Rey, R. 1957. Introduction à une géographie humaine du Borkou. Travaux
de l’Institut de Recherches Sahariennes 16: 41–72.
Capot-Rey, R. 1958. Decouvertes archeologiques au Borkou. Travaux de l’Institut
de Recherches Sahariennes 17: 203–5.
Capot-Rey, R. 1959. Le sel et le commerce du sel au Borkou-Ennedi-Tibesti.
Travaux de l’Institut de Recherches Sahariennes 18: 187–93.
Capot-Rey, R. 1961. Borkou et Ounianya. Etude de géographie régionale. Alger: IRS.
Chapelle, J. 1957. Black Nomads of the Sahara (translated by F. Schütze). Paris:
Librairie Plon.
Chapelle, J. and Monod, T. 1937. Notes sur la grande sepulture d’El Mreïti. Bulletin
du Comité des études historiques de l’A.O.F. 20: 507–21.
Cissé, M. 2010. Archaeological Investigations of Early Trade and Urbanism at Gao
Saney (Mali). Unpublished PhD thesis, Rice University.
Cissé, M. 2017. The Trans-Saharan trade connection with Gao during the first
millennium AD. In D.J. Mattingly, V. Leitch, C.N. Duckworth, A. Cuénod,
M. Sterry and F. Cole (eds), Trade in the Ancient Sahara and Beyond.
Cambridge: Cambridge University Press, 101–130.
Cleaveland, T. 2002. Becoming Walata: A History of Saharan Social Formation and
Transformation. Portsmouth, N.H.: Heinemann.
Cline, W. 1950. The Teda of Tibesti, Borkou and Kawar. Mwenasha: General Series
in Anthropology.
Close, A.E. 1980. Current research and recent radiocarbon dates from northern
Africa. Journal of African History 21.2: 145–67.
Close, A.E., 1984. Current research and recent radiocarbon dates from northern
Africa, II. Journal of African History 25.1: 1–24.
Cressier, P. and Bernus, S. 1984. La grande mosquée d’Agadez. Journal des africa-
nistes 54.1: 5–40.
Dalloni, M. and Monod, T. 1948. Mission scientifique du Fezzan (1944–1945). VI
Géologie et préhistoire (Fezzan méridionale, Kaouar et Tibesti). Alger: Impr.
Imbert.
Denham, D., Clapperton, H. and Oudney, W. 1985. Narratives of Travels and
Discoveries in North and Central Africa in the years 1822, 1823, and 1824.
London: Darf Publishers (reprint).
Desio, A. 1935. Missione scientifica della Reale Accademia d’Italia a Cufra (1931–
32). Vols I–III. Rome: Reale Accademia d’Italia.
Desio, A. 1942. Sahara Italiano. Il Tibesti nord-orientale. Rome: università la
Sapienza.
Desio, A. 1950. Le vie della sete. Milan: Hoepli.
Despois, J. and Raynal, R. 1967. Géographie de l’Afrique du Nord-Ouest. Paris:
Payot.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 325

Devisse, J. (ed.). 1983. Tegdaoust III: Recherches sur Aoudaghost. Campagnes 1960/
65, enquête generals. Paris: Editions Recherches sur les Civilisation.
Dupuy, C. 2010. Les apports archéologiques des gravures rupestres de l’Aïr (Niger)
et de l’Adrar des Iforas (Mali). Les Nouvelles d’archéologie 120–121: 29–37.
Duveyrier, H. 1864. Les Touareg du nord. Exploration du Sahara. Paris: Challamel.
Ehret, C. 2019. Berber peoples in the Sahara and North Africa. Liguistic historical
proposals. In Gatto et al. 2019, 464–94.
Gabriel, B. 1977. Zum ökologischen Wandel im Neolithikum der östlichen
Zentralsahara. Berliner Geographische. Abhandlungen 27: 1–94.
Gabriel, B. 1981. Die oestliche Sahara in Holozaen: Klima, Landschaft und
Kulturen. In C. Roubet, H.L. Hugot and G. Souville (eds), Mélanges offerts au
Doyen Lionel Balout. Paris: A.D.P.F., 195–211.
Gast, M. and Hachid, M. 1995. Djanet. Encyclopédie berbère fasc 16: 2379–90.
Gatto, M.C. 2006. The local pottery. In M. Liverani (ed.), Aghram Nadarif.
A Garamantian Citadel in the Wadi Tannezzuft. Firenze: All’Insegna del
Giglio, 201–40.
Gatto, M., Mattingly, D.J., Ray, N. and Sterry, M. (eds). 2019. Burials, Migration
and Identity in the Ancient Sahara and Beyond. Trans-Saharan Archaeology
Volume II. Series editor D.J. Mattingly. Cambridge: Cambridge University Press
and The Society for Libyan Studies.
Gautier, E.-F. 1970. Sahara. The Great Desert (translated by D.F. Mayhew).
London: Octagon.
Gauthier, Y. and Gauthier, C. 2011. Des chars et des Tifinagh: étude aréale et
corrélations. Cahiers de l’association des amis de l’art rupestre saharien 15:
91–118.
Grébénart, D. 1985. Le néolithique final et les débuts de la métallurgie. Niamey: Inst.
de recherches en sciences humaines.
Grébénart, D. 1988. Les origines de la métallurgie en Afrique occidentale. Paris:
Errance.
Hachid, M. 1998. Le Tassili des Ajjer. Aux sources de l’Afrique. Paris: Ed Paris-
Méditerranée.
Hachid, M. 2000. Les premiers berbères. Entre Méditerannée, Tassili et Nil. Aix-en-
Provence: Edisud.
Haour, A.C. 2003. One hundred years of archaeology in Niger. Journal of World
Prehistory 17.2: 181–234.
Haour, A.C. 2017. What made Islamic trade distinctive as compared to pre-Islamic
trade? In Mattingly et al. 2017, 80–100.
Huard, P. 1953. Gravures rupestres de la lisiere nord-occidentale de Tibesti.
Travaux de l’Institut de Recherches Sahariennes 10: 75–106.
Huard, P. and Féval, J.-C. 1964. Figurations rupestres des confins Algero-Niger-
Tchadiens. Travaux de l’Institut de Recherches Sahariennes 23: 61–94.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 326 Martin Sterry and David J. Mattingly

Insoll, T. 1996. Islam, Archaeology and History. Gao Region (Mali) Ca.AD 900–
1250. Cambridge Monographs in African Archaeology 39. BAR S647. Oxford:
Tempus Reparatum.
Insoll, T. 2000. The origins of Timbuktu. Antiquity 74: 495–96.
Jäkel, D. and Geyh, M.A. 1982. 14C-Daten aus dem Gebiet der Sahara. Berliner
Geographische Abhandlungen 32: 143–65.
Keding, B., Lenssen-Erz, T. and Pastoors, A. 2007. Pictures and pots from pastor-
alists. Investigations into the prehistory of the Ennedi highlands in NE Chad.
Sahara 18; 23–46.
Lambert. N. 1986. Akjoujt (Axamuk). In Encyclopédie berbère 3, Ahaggar – Ali ben
Ghaniya, 417–19.
Lange, D. and Berthoud, S. 1977. Al-Qasaba et d’autres villes de la route centrale du
Sahara. Paideuma 23: 19–40.
Lebeuf, J.P. 1962. Archéologie tchadienne: Les Sao de Cameroun et du Tchad. Paris:
Hermann.
Lebeuf, J.P. 1963. Prehistory, proto-history and history in Chad. Journal of the
Historical Society of Nigeria 2.4: 593–601.
Levtzion, N. and Hopkins, J.F.P. (eds). 2000. Corpus of Early Arabic Sources for
West African History. Princeton: Markus Wiener.
Lhote, H. 1971. Ronde-Bosse néolithique du Tassili représentant un Bovidé. Objets
et Mondes 11.2: 227–35.
Lhote, H. 1972. Recherches sur Takedda, ville décrite par Ibn Battouta et située en
Aïr. Bulletin de l’Institut Français d’Afrique Noire 34: 429–47.
Lhote, H. 1973. Découverte des ruines de Tadeliza: ancienne résidence des sultans
de l’Aïr. Notes africaines 137: 9–16.
Lhote, H. 1984. Le Hoggar. Espace et temps. Paris: Armand Colin.
Lovejoy, P. 1986. Salt of the Desert Sun. History of Salt Production and Trade in the
Central Sudan. Cambridge: Cambridge University Press.
Lydon, G. 2009. On Trans-Saharan Trails: Islamic Law, Trade Networks and Cross-
cultural Exchange in Nineteenth-century Western Africa. Cambridge: Cambridge
University Press.
MacDonald, K.C. 2011. A view from the south: Sub-Saharan evidence for contacts
between North Africa, Mauritania and the Niger, 1000 BC – AD 700. In
A. Dowler and E.R. Galvin (eds), Money, Trade and Trade Routes in Pre-
Islamic North Africa. London: British Museum Press, 72–82.
MacEachern, S. 2019. Burial practices, settlement and regional connections around
the southern Lake Chad basin 1500 BC–AD 1500. In Gatto et al. 2019, 399–428.
Magnavita, S. 2013. Initial encounters: Seeking traces of ancient trade connections
between West Africa and the wider world. Afriques: débats, méthodes et terrains
d’histoire 04. Available at: http://afriques.revues.org/1145 [last accessed
12 September 2019].
Magnavita, S. 2017. Track and trace. Archeometric approaches to the study of early
Trans-Saharan trade. In Mattingly et al. 2017, 393–413.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 327

Maître, J.-P. 1976. Inventaire archéologique du sud-est nigérien. Niamey: Institut de

Recherches en Sciences Humaines.
Mattingly, D.J. (ed.). 2003. The Archaeology of Fazzan. Volume 1, Synthesis.
London: Society for Libyan Studies, Department of Antiquities.
Mattingly, D.J. (ed.). 2010. The Archaeology of Fazzan. Volume 3, Excavations
Carried out by C. M. Daniels. London: Society for Libyan Studies, Department of
Antiquities.
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017. Trade in the Ancient Sahara and Beyond. Cambridge: Cambridge
University Press.
Mauny, R. 1955. Notes d’histoire et d’archéologie sur Azougui, Chinguetti et
Ouadane. Bulletin de l’Institut Français d’Afrique Noire (B) 17: 142–62.
Mauny, R. 1956. Monnaies antiques trouvées en Afrique au sud du limes romain.
Libyca 4: 249–61.
Mauny, R. 1961. Tableau géographique d’ouest africain au moyen âge.
Dakar: Mémoires de l’I.F.A.N.
Mauny, R. 1978. Trans-Saharan contacts and the Iron Age in West Africa. In
J.D. Fage (ed.), The Cambridge History of Africa, Volume 2. Cambridge:
Cambridge University Press, 272–341.
McDougall, E.A. 2006. Snapshots from the Sahara: Salt the essence of being. In
D. Mattingly, S. McLaren, E. Savage, Y. al-Fasatwi and K. Gadgood (eds), The
Libyan Desert: Natural Resources and Cultural Heritage. London: Society for
Libyan Studies, 295–303.
Meerpohl, M. 2013. Footprints in the sand: Recent long distance camel trade in the
Libyan Desert (north-east Chad/southeast Libya). In F. Förster and H. Reimer
(eds), Desert Road Archaeology in Ancient Egypt and Beyond. Köln: H. Barth
Inst., 167–91.
Monod, T. 1938. Teghaza, la ville en sel gemme (Sahara occidental). La Nature
3025: 289–96.
Monod, T. 1948. Sur quelques constructions anciennes du Sahara occidental.
Bulletin de la Société de Géographie et d’Archéologie d’Oran 71: 1–30.
Munson, P.J. 1980. Archaeology and the prehistoric origins of the Ghana Empire.
Journal of African History 21: 457–66.
Nachtigal, G. 1974. Sahara and Sudan, vol I, Tripoli and Fezzan, Tibesti or Tu
(translated by A.G.B. and H.J. Fisher). London: C. Martin.
Nixon, S. 2009. Excavating Essouk-Tadmakka (Mali): New archaeological investi-
gations of early Islamic trans-Saharan trade. Azania: Archaeological Research in
Africa 44.2: 217–55.
Nixon, S. 2013. Tadmekka: Archéologie d’une ville caravanière des premiers temps
du commerce transsaharien. Afriques 4. Available at: http://afriques.revues.org
/1237 [last accessed 12 September 2019].
Nixon, S. (ed.). 2017. Essouk-Tadmekka: An Early Islamic Trans-Saharan Market
Town. Leiden: Brill.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 328 Martin Sterry and David J. Mattingly

Ould Khattar, M. 1995. La fin des temps préhistoriques dans le sud-est Mauritanien.
Unpublished PhD dissertation, Université de Paris-I, Pantheon-Sorbonne.
Ould Mohamed Yahya, A. and Rebstock, U. 1997. Handlist of Manuscripts in
Shinqit and Wadan. London: Al-Furqan Islamic Heritage Foundation.
Park, D.P. 2010. Prehistoric Timbuktu and its hinterland. Antiquity 84: 1076–88.
Paris, F. 1990. Les sépultures monumentales d’Iwelen (Niger). Journal des
Africanistes 60.1: 47–75.
Paris, F. 1996. Les sépultures du Sahara nigérian, du Néoloithique à l’Islamisation.
Paris: ORSTOM.
Paris, F., Roset, J.-P. and Saliège, J.F. 1986. Une sépulture musulmane ancienne
dans l’Aïr septentrional (Niger). Comptes-Rendus de l’Académie des Sciences de
Paris 303 (Série III.12): 513–18.
Perret, R. 1935. Observations géographiques faites au cours d’un voyage au Tassili
des Ajjers. Bulletin de l’Association de géographes français 12.89: 96–100.
Perret, R. 1936. Recherches archéologiques et ethnographiques au Tassili des Ajjers
(Sahara central). Les gravures rupestres de l’Oued Djaret, la population et les
ruines d’Iherir. Journal de la Société des Africanistes 6.1: 41–64.
Pichler, W. 1997. A Latin inscription at Ti-m Missaou (Algeria). Sahara 9: 150–51.
Polet, J. 1985. Tegdaoust IV: Fouille d’un quartier de Tegdaoust. Paris: Editions
Recherche sur les Civilisations.
Posnansky, M. and McIntosh, R. 1976. New radiocarbon dates for northern and
western Africa. The Journal of African History 17.2:161–95.
Prorok, B.K. de. 2001. In Quest of Lost Worlds: Five Archaeological Expeditions
1925–1934. Santa Barbara: Narrative Press (reprint of 1935 book).
Quéchon, G. and Roset, J.-P. 1974. Prospection archéologique du massif de Termit
(Niger). Cahiers ORSTOM, Série Sciences Humaines 11.1: 85–104.
Retaillé, D. 1986. Le Kawar, problème géographique. Cahiers géographiques de
Rouen 26: 37–60.
Reygasse, M. 1950. Monuments funéraires préislamiques de l’Afrique du Nord.
Paris: Arts et Métiers Graphiques.
Richardson, J. 1853. Narrative of a Mission to Central Africa, 2 vols. London:
Chapman and Hall.
Robert, D., Robert, S., Devisse, J. 1970. Tegdaoust I: Recheches sur Aoudaghost.
Paris: Arts et Métiers Graphiques.
Robert, S. 1982. Rapport sur les mosquées historiques des villes anciennes de
Oualata, Tichitt et Chinguetti. Nouakchott: IMRS.
Robert-Chaleix, D. 1989. Tegdaoust V: Une concession médiévale à Tegdaoust:
implantation, évolution d’une unité d’habitation. Paris: Editions Recherche sur
les Civilisations.
Rohlfs, G. 1874/1875. Quer durch Afrika: Reise vom Mittelmeer nach dem
Tschadsee und zum Golf von Guinea, 2 vols. Leipzig: F.A. Brockhaus Verlag.
Rohlfs, G. 2003. Voyages et explorations au Sahara. Tome V, Koufra – les oasis de
Djofra et de Djalo 1878–1879 (translated by J. Debetz). Paris: Karthala.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 7 Pre-Islamic Oasis Settlements in the Southern Sahara 329

Rønneseth, O. 1982a. Verlassene Siedlungen im Tibestigebirge. Berliner

Geographische Abhandlungen 32: 27–62.
Rønneseth, O. 1982b. Gräber im nordwestlichen Tibesti (Tschad). Munich: Zabern.
Roset, J.-P. 1977. Un site à céramique peinte dans l’Aïr oriental (Niger). Cahiers Office
de la Recherche Scientifique et Technique Outre-Mer, ser Sci Hum 14: 337–46.
Roset, J.-P. 1988. Iwelen – an archaeological site of the chariot period in Northern
Aïr, Niger. In Libya Antiqua. Paris: UNESCO, 13–46.
Roset, J.-P. 2007. La culture d’Iwelen et les débuts de la métallurgie du cuivre dans
l’Aïr, au Niger. In J. Guilaine (ed.), Le chalcolithique et la construction des
inégalités. II, Proche et Moyen-Orient, Amérique, Afrique. Paris: Errance, 107–36.
Saison, B. 1981. Azugi: Archéologie at historie en Adrar mauritanien. Recherche,
pédagogie et culture 11.55: 66–74.
Salama, P. 1981. The Sahara in classical antiquity. In G. Mokhtar (ed.), UNESCO
General History of Africa, II Ancient Civilisations of Africa. Paris: UNESCO, 513–32.
Scarin, E. 1937a. Le oasi cirenaiche del 29˚ parallelo. Firenze: Sansoni.
Scarin, E. 1937b. Descrizione delle oasi e gruppi di oasi. In Il Sahara Italiano,
I. Fezzan e oasi di Gat. Rome: Società Italiana arto grafiche, 603–44.
Schiffers, H. 1957. Le Borkou et ses inhabitants. Travaux de l’Institut de Recherches
Sahariennes 15: 65–88.
Stewart, C.C., Salim, S.A. and Yahya, A.O. 1990. Catalogue of Arabic manuscripts
at the Institut Mauritanien de Recherche Scientifique. Islam et sociétés au Sud du
Sahara 2.1: 179–84.
Thiry, J. 1995. Le Sahara libyen dans l’Afrique du nord médiévale. Leuven: Peeters.
Treinen-Claustre, F. 1982. Sahara et Sahel à l’âge du fer. Borkou, Tchad. Mémoires
de la Société des Africanistes. Paris: Musée de l’Homme.
Vanacker, C. 1979. Tegdaoust II. Recherches sur Aoudaghost. Fouille d’un Quartier
Artisinal. Nouakchott: Institut Mauritanien de la Recherche Scientifique.
Vernet, R. 1993. Préhistoire de la Mauritanie. Nouakchott: Centre Culturel
Français A. de Saint Exupéry – Sépia.
Vernet, R. and Tous, P. 2004. Les amas coquilliers de Mauritanie occidentale et leur
contexte paléoenvironnemental (VIIe–IIe millénaires BP). Préhistoires
méditerranéennes 13: 55–69.
Vikør, K. 1999. Oasis of Salt. History of Kawar: Saharan Centre of Salt Production.
Bergen: Centre for Middle Eastern and Islamic Studies.
Wilson, A.I. 2012. Saharan trade: Short-, medium- and long-distance trade net-
works in the Roman period. Azania: Archaeological Research in Africa 47.4:
409–49.
Wright, J. 1998. Murzuk and the Saharan slave trade in the 19th century. Libyan
Studies 29: 89–96.
Yacono, D. 1968. L’Ahaggar. Travaux de l’Institut de Recherches Sahariennes 27,
fasc. 1–2.

https://doi.org/10.1017/9781108637978.008 Published online by Cambridge University Press

 8 Discussion
Sedentarisation and Urbanisation in the Sahara

martin sterry and david j. mattingly

Introduction

This chapter is intended as an analytical counterpoint to the descriptive

presentation of the evidence of pre-Islamic activity at Saharan oases and as
an engagement with some of the more theoretical introduction to the
volume.1 Here we make the case for a dynamic Saharan history with
frequent episodes of sedentarisation and urbanisation in the past. The
related issue of state formation will be considered as the conclusion of
the book, with a discussion of how we might define states and state
formation in the Saharan milieu and the contribution that oasis centres
have made to early desert states, like the Garamantes.2 We do not seek to
deny the importance of pastoral lifeways, or the social complexity of such
mobile societies, but we argue that the pastoral and mobile dimension of
the Trans-Saharan world has been over-emphasised to the neglect of its
sedentary communities. Several factors play into the choices made by
communities as to how to live: social and political circumstance, environ-
mental affordances, regional dynamics and trade. The fact that pastoral
groups have tended to be more dominant in later history does not exclude
the possibility that in some earlier ages sedentary and oasis populations
have had greater social and political weight. Trans-Saharan history we
suggest has been characterised by a continually changing power balance
between pastoralists and sedentarists (as is also evident at times in the
recent past).
In this chapter we shall test two hypotheses, developed from the pre-
ceding chapters. These are that:

1 2
330 See Chapter 1 and Chapters 2–7 above. Sterry and Mattingly, Chapter 19, this volume.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 331

• sedentary oasis communities were established much earlier and across

much broader areas than has conventionally been recognised, totalling
many hundreds of settlements, and following different chronological
trajectories in different zones (broadly earlier in the Eastern Sahara
compared to the Central and Western Sahara);3
• sites that we can recognise as urban in character, albeit with some
distinctive Saharan aspects, existed by the late first millennium BC/
early first millennium AD in several Saharan regions, but always in
small numbers in relation to the totality of oases;
Our discussion here will first explore these factors with regard to the
Garamantes of Fazzan, as this is the oasis group with the largest con-
centration of evidence for the early spread of sedentary oasis cultivators.
We present the factors that persuade us that this society was urban and
set out some ideas about the characteristics of Saharan urbanisation,
reflecting on how this may differ from Mediterranean or Sub-Saharan
urbanism. We then broaden our analysis to consider the evidence from
the wider Sahara.
At the outset, we should be clear that we do not view these elements in an
evolutionary or strictly linear relationship with each other. However, it is
clear from the archaeological evidence from Fazzan that Garamantian
urban centres (and their putative state) only developed after oasis agricul-
ture and several other new technologies were well established.4 On this
basis the first question we would pose is how and on what basis were the
first permanent settlements associated with oasis cultivation formed?
A related issue concerns the extent to which sedentary cultivators and
semi-sedentarised pastoral groups were distinct from one another. To put
it another way, was there also a process towards reduced mobility in
pastoral communities?

Sedentarisation

Sedentarisation is a deceptive paradigm, as for many scholars, it is the

successful end of an evolutionary process away from pastoralism and
mobility. The fallacies of such a mind-set in the context of the Sahara are
well established.5 Faced with the environmental challenges thrown up by
the desert climate and landscapes, it is far from certain that communities
3
See in particular the discussion of chronology in Chapters 2, 3, 5 and 6 above.
4
See Duckworth et al. forthcoming. 5 Scheele 2010; 2012; 2017.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 332 Martin Sterry and David J. Mattingly

would favour a life as sedentary agriculturalists over the dynamic possibi-

lities of adaptive pastoral strategies. Pastoral groups often seem to have
been more appropriately scaled for resilience, while the labour needs of
oasis communities are disproportionately high.
In many oases, irrigation schemes depend on the maintenance of dras-
tically unequal labour relations, involving slaves or other notionally free
but highly dependent groups.6 An interesting phenomenon of the early
modern Sahara has been the leading role played by far-flung pastoral
groups like the Tuareg in maintaining dependency relations over oasis
cultivators. The historical depth of such relationships is uncertain and
a further variable for future work to consider is how the balance of
power between different types of pastoralist and different sedentary groups
within the varied parts of the Sahara have developed and changed over
time.
Sedentarisation is not a one way street and decline in water sources or
diminished access to dependent labour could easily spark a move back to
more pastoral-based lifestyles. While some oases have grown over time,
many others have shrunk dramatically, especially impacted by changes to
the groundwater availability. As we shall see, a recurrent theme of this
volume is the cyclical nature of development and decline in Saharan oases,
rather than a linear evolutionary trajectory.
The processes of sedentarisation and its end results were far from uni-
form across time and space within the Trans-Saharan zone. Equally it was
not an either/or choice between pastoral and agricultural lifestyles. Mixed
strategies are apparent at Early Garamantian sites like Zinkekra and the
earliest cultivation in some areas may have reflected a shift to fixed base
transhumance linked to locations where scratch cultivation could exploit
seasonally available water. Cereal cultivation may have preceded the arrival
of the date palm in some regions, with the spread of the latter being a much
clearer marker of fully sedentary farming communities.7 Sedentarisation
also evidently took many different forms as is evident from a comparison
between the run-off farming of the Libyan pre-desert and the oases of the
Garamantes.8 In this volume our focus is primarily on the emergence
of intensive oasis agriculture, rather than on more extensive floodwater
farming regimes. That is not to deny the importance of the latter, but rather

6
For example, the population of the Tabalbala oasis, Sterry et al., Chapter 6, this volume.
7
It is interesting that thus far our fieldwork in Wadi Draa (southern Morocco) has identified
Protohistoric cerealiculture, but no certain evidence for the date palm, though this may be an
artefact of the small scale till now of our trial sondages at sites identified.
8
Mattingly 2004.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 333

to highlight the overall significance of oases in underpinning the Trans-

Saharan networks that we suggest supported the development of urban
centres and states. It is also apparent that sedentarisation was reversible (or
not necessarily sustainable) and that in many areas of the Sahara, the
balance has tipped back at some point towards pastoral exploitation of
the landscape.

Chronology of Sedentarisation and Oasis Formation

A key aim of this volume is to review and reassess the chronology of oasis
agriculture in the Sahara. There are signs that the processes described for
Fazzan were replicated in other parts of the Sahara, but not in any uniform
manner. However, there are hints in the available dating evidence of
a progressive intensification of activity from east to west.
The earliest evidence of Saharan sedentarisation comes from opposite
sides of the Sahara: in the Dakhla oasis of Egypt during the Old Kingdom
(2649–2150 BC) and Second Intermediate Period (1650–1550 BC),9 and in
Mauritania with the Tichitt tradition of escarpment settlements and millet
cultivation around the Dhars Tichitt, Walata and Nema dating to the mid-
second millennium BC.10
As might be expected the first Saharan oases appear to have developed
in the Western Desert of Egypt, an area strongly integrated into Nilotic
socio-economic and political systems. However, although oasis agriculture
and urban sites appeared from the third millennium BC, it was not until
the Hellenistic and Roman periods that there was expansion into all parts
of the Siwa, Bahariya, Farafra, Dakhla and Kharga oases.11
Among the oases of Tripolitania, southern Tunisia and eastern Algeria,
we have little knowledge of the first millennium BC, although hilltop and
escarpment sites are located in or near many of them. However, as we saw
in Chapters 5 and 6, the Roman period saw the growth of urban centres
around Roman forts (e.g. Bu Nijim, Negrine, Gemellae), in oasis groups
such as the Nefzaoua and Jarid and the expansion of irrigation-based
agriculture on the southern flank of the Aures mountains, as at Badias
and in the Libyan pre-desert. Beyond the Roman frontier the settlements
are essentially undated, though given the scale of oasis development in the
frontier zone, we suggest our default assumption should now be that the

9
Boozer, Chapter 4, this volume.
10
MacDonald et al. 2009; MacDonald, Chapter 13, this volume.
11
As discussed by Mattingly et al., Chapter 3, and Boozer, Chapter 4, this volume. See also Gill
2015.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 334 Martin Sterry and David J. Mattingly

major spring-fed oases just beyond the frontier (for example, Wadi Rihr
and Tuggurt) were in all probability already active by this date.
Thus, despite a number of lacunae, and a particular lack of evidence for
the second millennium BC, it is clear that by the late first millennium BC
there had been a substantial change in oases of the Northern, Eastern and
Central Sahara with a large growth in the numbers of settlements (oases in
Fazzan, al-Jufra, Western Desert, Roman frontier zone in Algeria and
Tunisia) that were largely sustained by oasis agriculture (that is, date
palms, fruits, cereals, chickens and in some cases pigs). Technological
transfers must have been key to facilitating this transition with the adop-
tion of Near Eastern/Mediterranean crops, transport animals, irrigation
technologies and iron-working affecting both the make-up and rhythms of
communities, but also their connectivity with one other. Other factors may
also have been prominent. Although the major drying of the Sahara took
place millennia before sites such as Zinkekra, the evidence from the
Tanzzuft implies that many surface water sources were still shrinking
because of aridification until at least the first millennium AD.12 This may
have encouraged pastoral communities to develop more permanent (and
better defended) sites at the locations with the most reliable water sources.
The locations with the most reliable or easily accessible water sources will
also have been those to which pioneer agriculturalists were drawn.
Logically, either these oasis cultivators will have been in competition
with pre-existing pastoral groups or they must have involved groups of
the latter.

Sedentary Settlements
What were the earliest forms of sedentary settlements? In the Wadi
al-Ajal there are 12 hilltop sites of varying morphology that seem to
represent the earliest sedentary settlement in the region. Of these sites,
only Zinkekra has seen any serious excavations, producing evidence of
oasis agriculture and stone buildings in the tenth to seventh centuries BC
(the first stages of the Early Garamantian phase).13 A second site, Tinda
(near Ubari), has also yielded evidence of oasis agriculture dated to the
latter centuries BC.14
Similar sites have also been identified from satellite imagery in other
Saharan oases (for example, al-Jufra, Kawar, Aïr) and in northern
12
Cremaschi 2006. 13 Mattingly 2003, 136–42; 2010, 19–119.
14
Mattingly 2007, 294–95 for AMS dates on barley, pearl millet and sorghum spanning 390
calBC–calAD 20. See also Mattingly 2010, 78–82 for radiocarbon dates from Zinkekra.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 335

pre-desert areas such as the Libyan Valleys, the Aures Mountains in

Algeria, the Anti-Atlas Mountains and the Draa Valley in Morocco.15
However, it could be argued that such hillforts are not in themselves
indicative of a shift to agriculture. Some Saharan communities appear to
have constructed permanent or at least semi-permanent settlements whilst
maintaining pastoralist lifeways.16 With this in mind we might put Saharan
settlements on a continuum, with sites such as Afunfun and Gobero at one
end (semi-permanent),17 Dakhlet el Atrous in the middle (some mobile
pastoralism, some agriculture),18 and Zinkekra at the other end (sedentary
oasis agriculture alongside localised livestock raising, although this does
not preclude the continuation of mobile parts of the population). What is
most notable about the Zinkekra and Tinda evidence is that the develop-
ment of agricultural lifeways preceded major architectural and cultural
change at these sites. The initial buildings at Zinkekra were oval one- or
two-room structures, associated with enclosures, with abundant traces
of the presence of livestock alongside the clear signature of adoption of
a developed package of agricultural crops (dates, wheat, barley, grapes,
etc.). Although the later buildings at Zinkekra in the latter centuries BC
had more complex and extensive rectangular layouts, at the Tinda escarp-
ment settlement, the curvilinear architectural form seems to have persisted
to the end of the site’s occupation.19 The material culture from Zinkekra
(pottery and lithics) also suggests a strong continuity with pre-existing
Pastoral communities of this part of the Sahara.20
This Early Garamantian phase of settlement in Fazzan appears to be
characterised by the adoption of new lifeways, but overall these commu-
nities remained small, and may have been only a minority of the overall
Saharan populations in the first millennium BC. However, in the last
few centuries BC (the Proto-Urban Garamantian phase) mudbrick settle-
ments started to appear in greater numbers in the centre of the oasis
depressions.21 Initially these may have remained fairly close to the fortified
hilltop sites (some of which were still in use), but they gradually spread
15
Fentress 1979, 30–39; Gsell 1972, 232–45; Mattingly 1995, 41–49. For the Libyan Valleys,
Barker et al. 1996a, 116–18 and individual entries in 1996b; for the Draa, see now Mattingly
et al. 2017b.
16
Garcea 2013. 17 Garcea 2013. 18 MacDonald, Chapter 13, this volume.
19
Mattingly 2010, 78–84 (Zinkekra) and 85–102 (Tinda).
20
See especially the finds assemblages illustrated in Mattingly 2010, 27–62.
21
A particularly clear example is the Taqallit area, where the pioneer phase of oasis cultivation,
marked by the first valley-centre villages, the construction of foggara irrigation systems and the
arrival in the region of sub-Saharan cereals like sorghum and pearl millet (that allowed more
intensive summer cropping) can be comprehensively dated to the last centuries BC by multiple
forms of evidence, see Mattingly et al. 2010.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 336 Martin Sterry and David J. Mattingly

more widely in the landscape supported by the digging of foggaras and

wells. It is around this time that we start seeing the first material evidence
of regular trade links across the Sahara, with Punic imports to Fazzan
common by the second century BC. Sedentary populations continued to
increase as evidenced by the greater numbers of settlements and the
appearance of nucleated cairn cemeteries. However, the real boom appears
to have taken place in the first–sixth centuries AD (the Classic and Late
Garamantian phases – also the period in which ceramic imports to Fazzan
peaked). During this time we see the intensification of the settlement
system and its expansion into new areas with little evidence for prior
occupation. The Murzuq region is the prime example with no certain
settlement from the first millennium BC, but over one hundred substantial
settlements supported by well and foggara irrigation by the fourth
century AD.
A recent article by Fentress and Wilson has postulated a Late Antique
period of Berber ‘migration’ in the Sahara and Maghrib.22 How does this
fit with the evidence presented in this volume? The overall picture for
the development of mudbrick fortified structures within oasis villages
highlights Late Antiquity as a time of significant change, with some oases
exhibiting signs of declining populations, as in the Western Desert or the
North African limes, but others showing an expansion in oasis settlements,
as in the Murzuq area in the Central Sahara or the Wadi Draa in Morocco.
The dating of the expansion of fortified oasis settlement across these
regions is broadly consistent with the Fentress and Wilson model, but it
is not clear if these phenomena are related or if the changes in settlement
systems are exceptional or symptomatic of oasis living (where water
resources are always finite).

The Implications of Garamantian Sedentarisation and Oasis

Settlement
There are several important lessons to be learned from the recent work
on the Garamantes. First, intensive ground survey work has revealed
numerous Garamantian settlements to have existed in places where no
evidence could be adduced previously. Absence of evidence in other areas
of the Sahara is clearly not evidence of absence.23 It is apparent that
successive phases of oasis development and exploitation can obliterate
22
Fentress and Wilson 2016; see also Fentress 2019 for a development of the argument based on
linguistic evidence.
23
Sterry and Mattingly 2011.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 337

many of the physical traces of preceding sites, but intensive survey work is
capable of detecting traces. For example, archaeological survey along
a systematic grid of lines criss-crossing the Jarma oasis in advance of
oil exploration seismic survey discovered pottery scatters and traces
of mudbrick structures relating to numerous previously unsuspected
Garamantian sites.24 This is the most likely explanation for the paucity of
Garamantian settlements recorded to date in the al-Barkat oasis, where
Aghram Nadharif is the only securely identified settlement, closely adja-
cent to a cemetery on the east side of the oasis. There are a further 11
ancient cemeteries that surround the al-Barkat oasis on all sides and the
existence of a similar number of separate settlements seems a logical
conclusion.25 The lesser visibility of unfortified sites in comparison to
fortified sites should also give us pause for thought where our survey is
based on remote sensing and major standing monuments. Settlement
phases based on unfortified and dispersed sites may well be drastically
underrepresented in the recorded archaeology.
Second, our AMS dating programme of recent years has completely
overturned initial assumptions about the dating of sites based on their
plan or state of preservation.26 It is now clear that many of the classic
settlement forms believed to characterise Medieval villages and fortified
sites (qsur) in the Sahara can be closely paralleled in Garamantian Fazzan
or the Protohistoric period of the Draa.27 This does not mean that we can
make a clear association between similar site types in the Murzuq area and
the Tuwat oases – indeed it is clear that some of these forms have been very
long lived. But neither is it any longer acceptable to assume a default
Medieval date for all the sites in the Tuwat (and elsewhere), especially
those that are not associated with mosques and Islamic cemeteries. As our
survey above demonstrates, in many areas of the Sahara there are undated
hillforts and castle-like fortified sites, sometimes close to funerary zones
of pre-Islamic type burial monuments.28 Each site needs to be assessed
on its merits and there is an urgent need for an extension of scientific
dating methods to these types of sites to establish more secure regional
chronologies.
Third, it should also be recognised that the centres of power in Saharan
oasis zones have moved around over time, rather than always remaining

24
Lahr et al. 2009. 25 Liverani 2006, 2, 25–27.
26
Sterry and Mattingly 2013; Sterry et al. 2012.
27
See for example the work of Echallier 1972; 1973; Chekhab-Abudaya 2016 seems to start with
the assumption that these types of site must be Medieval in date.
28
For our parallel work on the burial monuments, see Gatto et al. 2019.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 338 Martin Sterry and David J. Mattingly

fixed on specific sites. Until more extensive work has been done at a range
of sites, we should be wary of drawing definitive historical conclusions for
an oasis group on the basis of excavation at only a single site.
The expansion of settlement in Saharan oases has regularly been
considered to have gone hand-in-hand with Islamic-period Trans-
Saharan trade, especially at oases such as Tuwat and Kawar. The
connection between oasis settlement and trade is further explored
below, but whilst trade may bring in investment and food security for
supporting small settlement development, this may only have been at
local rather than Trans-Saharan levels. The population at agricultural
sites may have grown with the increasing food resources, but also, to the
north and south of the Sahara, there were likely booming populations.
We may therefore consider increased connectivity and migration as
a factor. It is worth noting that our evidence would suggest that the
earliest agricultural settlements pre-dated visible evidence of trade in
commodities.29 While we believe the existence of trade, or at least the
possibility of trade, was an important factor in the investment in oasis
development, there was no simple linear progression from establishing
oases to creating flows of trade. The distribution of settlements in
Fazzan and other large oasis chains cannot be seen as a string of trading
posts. Rather, in each case they need to be evaluated firstly as local
systems of settlement and only secondly as parts of interconnected
networks. The development of oases was not some sort of diffusionist
tidal wave that spread evenly and predictably across the Sahara.
The substantial hydrological differences between many neighbouring
locations in the Sahara have posed significant barriers to a neat and
predictable diffusion of oasis farming technology (see Figs 1.4–1.5). The
Medieval Trans-Saharan network certainly built on an embryonic net-
work established earlier, but the particular historical circumstances in
each phase emphasised different nodes and routes – sparking further
investment and intensification of activity around these key sites and
routes.

Factors Affecting Oasis Development

What were the mechanisms by which new settlements were created? The
earliest development should have favoured the areas with the most
29
Though we should be cautious here as much Saharan trade at all periods has been in ‘invisible’
commodities, see Mattingly et al. 2017a. Moreover, there is clear evidence for the transference
of technologies along the Saharan trade routes from an early date, see Liverani, forthcoming.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 339

abundant and most easily accessible water. For ease of exploitation we can
distinguish between, on the one hand, oases fed by surficial waters (rivers
like the Nile or Draa, flash floods in desert wadis, artesian and other natural
springs) and, on the other, oases dependent on artificial means of accessing
sub-surface water (foggaras, man-made artesian wells, other wells, palm-
eries tapping into a high water table).30 These different scenarios have
varied potential and problems for oasis farmers to overcome.31 The natu-
rally occurring water sources in the Sahara will have been the most obvious
ones to find and potentially the easiest for population groups to start to
exploit. These were locations that will have already been exploited by
Saharan pastoral groups, so their development probably involved partner-
ship between those people and incomers with knowledge of new agricul-
tural technologies. The scale of exploitation of the naturally occurring
water sources within the Sahara was always limited by and proportional
to the volume of water available. The effective use of a spring could be
improved by constructing efficient tanks at the spring head and canals to
conduct the water to individual gardens, but there would always be a limit
to the absolute area that can be irrigated with a given volume of water.
Beyond a certain point the further expansion of oases can only be obtained
by the introduction of additional technologies to augment the available
water.
In accordance with this idea, many of the oasis locations surveyed in the
earlier chapters appear to have embraced a range of different technologies
for exploiting water. For example, the Wadi al-Ajal area may originally
have benefitted from a number of perennial springs that encouraged
agricultural development at a few select locations. However, the intensifi-
cation of agriculture in the valley after 400 BC was almost certainly due to
the introduction of foggara technology, allowing the dramatic augmenta-
tion of the supply of running water. At a later date the foggara irrigation
was replaced by irrigation using only well technology – perhaps due to the
exhaustion of the aquifer that had fed the foggara channels.
In some instances, it is possible that technological barriers inhibited the
initial growth of some oasis areas. This is particularly the case with oases
where the natural artesian springs were few or not productive enough to
support a large population until augmented by additional man-made
artesian wells – as seems to have been the case in the Wadi ash-Shati or
Wadi Righ.
30
For a fuller discussion of the importance of hydrology, see the chapter by Wilson et al.
forthcoming.
31
See also Sterry et al. 2019.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 340 Martin Sterry and David J. Mattingly

Another issue concerns the transference of oasis farming technology

beyond an area where it became established. This was not always a case of
the application of the same technical competencies at a neighbouring
location. As we have seen, the earliest settlements tended to rely on knowl-
edge of springs and other surface sources of water, whereas subsequent
phases of expansion often involved the application of more labour-
intensive methods of accessing ground water. Furthermore, the movement
of sedentary oasis farming often took place over large distances and often
into zones with different and more difficult hydrological conditions. This
implies both prior knowledge and investment in order to develop success-
ful settlements and their associated infrastructure (one wonders how many
failed enterprises there were). In thinking about these, we should not be
drawn to analogies of island-hopping ocean exploration, but instead reflect
that there must have been a good societal knowledge of water sources
among the Saharan pastoral groups that had used them for millennia.32
The crucial aspect is to what extent this knowledge of springs, gueltas,
wadis with vegetation supported by mountain run-off, lakes/playas, shal-
low groundwater and other hydrological sources could be used as a basis
for irrigation schemes. Overall, there are significant indications of mobility
and migration within the populations responsible for the first oases.
A number of important technological innovations were brought in
developed form from outside the Sahara: a package of crops and some
new domesticated animals from the direction of the Nile/Western Desert,
balance wells and foggaras can both be found at earlier dates in the same
area, several Sub-Saharan crops were added later.
The model put forward by Hardesty (itself derived from Kirch) is
perhaps appropriate in this case.33 In this model, the first colonisers
bring with them standardised knowledge of the environment (where the
water is), technology (how to reach it) and social relations (who they are
and how to organise themselves). Once established we might then expect
the diversification of these categories with a better understanding of the
water sources, how to access it and changing social relations within and
between groups. Finally in the third stage we might expect the proliferation
of the most effective strategies.
32
Rockman 2003 draws attention to the difference between landscape learning with and without
resident populations. In the case of the Sahara, we should consider there to be resident
populations although the degree to which pastoral and sedentary societies were integrated or
friendly to one another is difficult to assess. The Roman and early Arabic sources seem to make
little distinction between sedentary and mobile groups perhaps suggesting a high degree of
integration, but they are far from conclusive.
33
Hardesty 2003, 83; Kirch 1980.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 341

We want to emphasise the point about technological choices being

driven by pragmatic and empirical knowledge, not by evolutionary schema
from simplest to most complex. Two different examples of ancient foggara
irrigation at Ghadamis and Murzuq illustrate this point. At Ghadamis, the
limitations of the main spring has for much of its history constrained the
absolute development of the size of the oasis and its population. At various
times (not closely dated) attempts have been made to create additional
irrigated gardens around the periphery of the oasis through the labour-
intensive construction of foggaras and wells. By contrast, in the Murzuq
region the earliest dated settlements, comprising nucleated settlements
similar to those in the Wadi al-Ajal, were located close to playas and
used a form of foggara technology to irrigate a system of fields. But the
relative unsuitability of this landscape for foggara irrigation constrained
that development. By the fourth–fifth centuries AD, a wider range of
settlement forms and sizes had been built, but the later settlements fell
into two more easily defined categories – the qasr and the walled village
each supported by their own system of well-based gardens. In this case, the
more appropriate technology was actually the simplest form of balance
well, facilitated by the existence of a high water table. On the other hand, it
is equally apparent that the existence of technology (or knowledge of it) is
not an adequate explanation for the spread of oases. The expansion of oasis
agriculture in almost every scenario involves significant labour inputs and
without the availability and investment of that human capital, the enter-
prise cannot advance.
To summarise our model for the development of oasis and sedentary
settlement in the Sahara, we have suggested that the first settlements sat
at one end of a spectrum of permanent bases used by pastoral groups.
Once these appeared, and in conjunction with a number of converging
technologies, intensive oasis agriculture could be developed (or imported),
probably through social partnerships involving both established pastoral
communities and some migrant groups. For fully fledged oases to emerge,
external contacts were important for accessing crops and animals domes-
ticated elsewhere and the supporting technologies. Finally, this accumu-
lated knowledge then formed the basis for developing new areas and new
oases in the Sahara.
In our descriptive chapters on early Saharan oasis development, we have
identified a broad trend from precocious oasis agriculture and permanent
settlements in the Nile Valley and the Western Desert in the third–second
millennia BC, to the emergence of Central Saharan oases by the early first
millennium BC, to increasing evidence for oases in the Northern Sahara

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 342 Martin Sterry and David J. Mattingly

Figure 8.1. Spread of urbanisation across the Trans-Saharan region (circles indicate
urban or other significant population agglomerations, smaller oases are not shown).

within and just outside the Roman frontier zone by the early centuries AD,
to possible penetration of oasis settlements into most of the suitable
hydrological niches of the Southern and Western Sahara by the mid-first
millennium AD (Fig. 8.1). The dating schema is increasingly weak as one
moves further from the Roman frontier zone and further west and south,
where most of the investigated sites have been of Medieval date and in part
influenced by the descriptions in the Arabic sources.
The implications of this new model for conventional understanding
of the socio-economic status of the indigenous peoples of the Roman
frontier zone and beyond (Gaetuli and Garamantes) are profound.34
The conventional narrative has portrayed Roman military advances
in the late first century BC and first century AD as essentially engaging
with pastoral peoples (though with the Garamantes now recognised as
a possible exception).35 The orthodox interpretation has presented Roman
34
Lassère 2015, 21–45, acknowledges the fact that not all indigenous peoples of the Maghrib were
pastoralists on the eve of Roman conquest, but limits his recognition of sedentary living and
proto-urban development to the areas far from the desert. Cf. Mattingly 2016, for
a deconstruction of the commonly assumed Roman responsibility for the urban and
agricultural boom of the early centuries AD.
35
Even in recent summaries, where the old emphasis on ‘nomads’ is more nuanced, the presence
and significance of sedentary oasis dwellers remains a bit elusive: Guédon 2018; Leveau
2018; Reddé 2014, 121–34.; cf. Mattingly et al. 2013a.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 343

pacification of the pastoral peoples of the desert margins as the essential

starting point for their sedentarisation in a progressive evolutionary model.
The new evidence presented above changes the paradigm and our starting
assumptions. For one thing, it is clear that the start-up date of many oases
in the Eastern and Central Sahara preceded military contact with Rome
and we have argued that it is more plausible that the Roman forts of the
Roman frontier zone of Libya, southern Tunisia and Algeria were estab-
lished alongside already existing oasis communities rather than the other
way around. For instance, the interactions between Rome and the various
peoples described as Gaetuli in the sources were thus more complex than
containment of transhumance lifestyles; they also involved large sedentary
oasis populations by the early first millennium AD. This is a new starting
point for scholarly debate and it impacts the entire northern fringes of the
Sahara from Egypt to Morocco. Oases were, of course, not restricted to the
area of the Roman frontier and a priority in future research must be to
investigate in greater detail the early history of oases in regions that
remained well outside the Roman frontier zone.

Urbanisation

In the introduction to this volume, we reviewed theoretical debate on

ancient urbanism, expressing doubts about the ready application of check-
list models developed in relation to Near Eastern or Mediterranean socie-
ties to the African context. That is not to say that Saharan sites cannot
conform to external models related to urban function or settlement hier-
archies, but rather to acknowledge that the differences in application (and
the relevant lack of excavation of the Saharan sites) can confound expecta-
tions. A key issue for us in thinking about Saharan towns is the need to look
beyond traditional urban models of ancient Egypt, Mesopotamia or the
Mediterranean. Although those well-known urban traditions may have
had some influence on what emerged in the Sahara, through proximity
and contacts, there are other possible explanations for the emergence of
Saharan urban centres.36 Given the evidence for Sub-Saharan urbanisation
in the first millennium BC, we cannot exclude the possibility of influence
from that direction.37 More importantly, if oasis settlements became
36
Mattingly and MacDonald 2013; Mattingly and Sterry 2013.
37
MacDonald 2013; Sinclair 2013 – we might note that his survey of African urbanisation by
region excludes the Sahara from consideration! Cf. Mattingly and MacDonald 2013, for an
attempt to reintegrate the Sahara into such discussions.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 344 Martin Sterry and David J. Mattingly

widespread before towns and state formation occurred, we should expect

a proto-urban phase of development reflecting internal characteristics of
desert societies. In other words, the idea of the town and its physical
manifestation in the Sahara may have reflected both the wider Trans-
Saharan world and the particularities of oasis communities. A further
point to consider with regard to Saharan urbanisation is the relationship
between the overall spread of oasis settlements (as modelled in Fig. 8.1) and
the eventual emergence of certain settlements with urban characteristics.
If we accept the proposition that Saharan towns may have had some
distinctive characteristics that differentiate them from other contemporary
urban forms, the next question is to ask what are the defining character-
istics of desert urbanisation and what makes them stand out as something
distinctive in the settlement system?
A useful starting test of ‘Saharan towns’ is the extent to which they meet
the basic parameters of multiple models of towns. How far can we identify
them as:
• centres or agglomerations of size and population density;38
• foci of distinctive functions and craft specialisations;39
• nodes of economic networks;40
• organising hubs of farming hinterlands or zones of raw material
exploitation;41
• places of social complexity and authority;42
• transformational settlements reorganising town–country interactions?43
Within the overall picture of sedentary oasis settlements, a case can
certainly be made for the special status of certain sites in terms of the
bulleted characteristics, of which the Garamantian capital at Jarma is the
prime example. Although small by Mediterranean standards at c.10 ha, it
was at the larger end of Garamantian settlements, evidently with distinctive
features and functions, its architecture reflecting unusual social complexity
and authority, its material culture demonstrating the existence of atypical
access to distant markets. For a defined period it was clearly a nodal point
in Saharan trade, as well as a focal point in the exploitation of regional
resources, including things like metals and semi-precious stones, as well as
agricultural production and labour.
We have argued that it was in the Classic and Late Garamantian phases
that the evidence for urban sites and a Garamantian polity is clearest. Our

38
Marzano 2011. 39 McIntosh, R. 2005; 2015. 40 Wilson 2011. 41 Cowgill 2004.
42
Smith, A.T. 2003; Smith, M. 2003; cf. McIntosh, S. 1999. 43 Yoffee 2005, 61–62.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 345

excavations at the Garamantian capital at Jarma suggests that monumental

stone buildings first appeared at the very end of the Proto-Urban phase in
the first century BC, but that the most exceptional development at the site
occurred in the first–fourth centuries AD.44
The modern debate about how to define towns, and the evidence
thresholds laid down by different models, complicate things to some
degree. Some modern theorists ignore the testimony of contemporary
witnesses. While Greco-Roman sources were notoriously biased against
‘barbarian’ societies, often presenting them in stereotypical terms as
underdeveloped, there are references that should give us pause for
thought in seemingly pointing to higher development. For example, it
is commonplace for these sources to refer to some significant settlements
beyond the Roman frontiers as ‘towns’ (oppidum, urbs, polis). We should
be cautious here as it is certainly conceivable that these terms for upper
echelon settlements in the Mediterranean region were being inappropri-
ately applied to non-urban sites in regions for which first-hand accounts
were lacking or inadequately detailed. On the other hand, there are
indications that distinctions were being reflected in the choice of
vocabulary. In the section of Ptolemy’s Geography relating to North
Africa, he does discriminate between sites that he describes as ‘towns’
(poleis) and others he refers to as ‘villages’ (komai).45 The description of
sites in the Libyan interior, essentially the Sahara, contains reference to
numerous ‘towns’ (poleis) and three sites that he highlights by the term
metropolis, Garama, Nigira and Gira.46 We can add to this the fact that
a number of oasis settlements in Roman frontier zones were explicitly
assigned urban rank by the Romans, with an administrative function (as
is described below in more detail). This suggests that however Saharan
towns may have deviated from Mediterranean norms, they were still
recognisable as urban equivalents and adaptable within Roman admin-
istrative structures when required. In some cases these Saharan towns
were explicitly identified as the principal place of a prominent people,
such as the Garamantes. By the early first century AD the Romans were
dealing with Garamantian ‘kings’ (another term not normally invoked
for petty chieftains), implying the existence of not only an urban society
but a genuine polity.47 In the past our assumptions about the relatively
unsophisticated nature of ancient Saharan societies and doubts about
44
Mattingly 2013.
45
See for instance, Ptolemy, Geography 4.5, where Awjila is indicated as a village, while Ammon
(Siwa) is identified as a town.
46
Ptolemy, Geography 4.6. 47 Sterry and Mattingly, Chapter 19, this volume.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 346 Martin Sterry and David J. Mattingly

the interconnectedness of the Roman provinces with the Saharan

hinterland, led to a default reading of such source material as
being ill-informed or fanciful. The archaeological evidence from the
Garamantian case study in particular, but with support from the
persuasive evidence for more widespread oasis development presented
in this book, suggests that we need to recalibrate our starting
assumptions. The implications are that within many of the areas of
Protohistoric oasis settlement that we have postulated, one or more
primate centres emerged with characteristics and functions that differ-
entiated them from other settlements. Visitors from the Mediterranean
world or the Sub-Saharan zone would have readily seen these centres as
equivalents to urban sites outside the Sahara.

What Was Distinctive about Saharan Towns?

What our survey of Protohistoric Saharan oasis settlement demonstrates is
that there were many villages, but relatively few sites of demonstrable
urban scale and status. Nonetheless the urban sites we can identify (or
postulate) played a significant role in these desert societies. Understanding
the urbanisation process in the desert requires careful consideration of the
particularities of these exceptional sites. If we start by reflecting on the
main groups of towns identified on Figure 8.1, some regional specificities
emerge or, to put it more bluntly, a distinct heterogeneity. The oases that
were closest to the Mediterranean and Nile were much more likely to
reflect some influences from the established civilisations of those regions.
The towns in the oases of the Western Desert are distinctive as a group
from those of the rest of the Sahara. In layout, architecture and cultural
reference points they look towards the Nile Valley and Egypt and were
organised following historical urban categories of Pharaonic, Hellenistic
and Roman governance. Most of the oasis depressions contained at least
one town, with externally recognised authority in local government: for
Kharga, these are Hibis and Kysis; for Dakhla, Kellis, Mothis, and
Trimithis; for Bahariya, al-Bawiti; for Siwa, Ammon. These main centres
were marked out by their size, by monumental architecture and by the
richness of associated cemeteries. Many of these Egyptian oases towns
show a degree of planning around a major temple complex, as at Kellis
and Trimithis in the Dakhla oasis.48 In addition to these few urban centres,
there were dozens of smaller villages.

48
See Boozer, Chapter 4, this volume.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 347

Within the province of Africa, a number of pre-Roman oasis centres

were formally recognised as Roman towns – most notably Tacapae (Gabes)
and Capsa (Gafsa), though there are strong reasons to suggest that these
deviated in important respects from the towns of northern Tunisia. Some
other pre-existing indigenous centres further out in the frontier zone
appear to have been recognised or promoted to be civitas towns, municipia
and even coloniae: in the Nefzaoua, Turris Tamalleni, in the Jarid, Thiges,
Nepta, Thusuros. However, these appear to have been much smaller and
less monumentalised than the cities in Northern Tunisia. In some cases,
rather than a single coherent urban unit, as Mediterranean models might
lead us to expect, it seems possible that they were in fact fragmented into
multiple village-scale complexes within a single oasis, as in the Nefzaoua,
Jarid and at Gabes.
The few oasis sites in the Roman frontier zone of Algeria and Tunisia
that were accorded urban status under Rome have quite a varied morphol-
ogy. Gemellae and Nigrenses Maiores are two examples that we know were
promoted to municipium rank, but which seem essentially to have been
walled garrison settlements around long-term Roman forts situated out-
side the oasis proper. There were quite a number of such sites along the
African frontier and these are the locations where stone architecture
appears to be most commonly encountered.
These oasis towns of the Western Desert and the Roman frontier zones
were to some extent administrative towns of more distant political autho-
rities – serving as centres for local tax collection, law and governance.
Further out into the Sahara, towns were both rarer and more specialised.
The clearest indication that there were African or Saharan urban prin-
ciples at play, alongside Punic or Roman ones, within the oases is the
existence of settlements of urban character beyond the frontiers. The urban
centres of the Garamantes appear to be rather different in form and scale to
the towns of Egypt, with a prominent emphasis on fortified complexes at
their core.49
Ancient oasis towns were in general characterised by a relatively small
area and population, but with a significantly enhanced role for craft
production and trade specialists when compared to other oasis settlements.
They were served by a hinterland of intensive irrigated cultivation that
needed to be large enough to support the settlement, but practical difficul-
ties of walking far in Saharan summer heat favoured the development of
satellite villages over expansion of the urban core. Trade rather than

49
Mattingly and Sterry 2013; Mattingly et al. 2013b, 530–34.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 348 Martin Sterry and David J. Mattingly

agriculture created the additional surplus required to support the growth of

the town, although water and agricultural supplies were needed to support
caravans. Saharan oasis towns have thus attracted and facilitated trade and
their wellbeing was intimately tied to the flow of commerce. Islamic
Ghadamis, Ghat, Murzuq and Zuwila can be seen to have operated in
this manner. When Trans-Saharan trade routes were disrupted, the cities
stagnated and shrank in population.50 Similarly, in the Tuwat oases, when
French colonial forces stopped all movement of trade goods, shortages and
malnutrition quickly followed.51 Therefore some external stimulus was
required for an oasis city to develop successfully.52
New capital investment could come from another already developed
oasis city, but could also have been drawn from outside the desert region.
In the Islamic era, for example, states to the north and south of Fazzan
deliberately developed new urban centres, as at Traghan and Murzuq.
Similarly, Roman period expansion of the Dakhla oasis has been attributed
to direct imperial interest.53 A key conclusion of this chapter is that the
same socio-economic factors for urban growth and decline also applied to
pre-Islamic Saharan society. It follows that the wellbeing of these desert
towns could be just as profoundly affected by political and economic shifts
as by environmental changes.
Yoffee has described cities as ‘not things or essences but points of entry
into . . . what changed in the urban revolution’.54 Any examination of ancient
urbanisation in the Sahara and its environs should thus investigate
what changed. Another question that needs to be addressed is how isolated
were the different urbanised zones from one another? We are certainly
not advocating a return to diffusionist models of towns and states in relation
to these entities in Saharan and Sub-Saharan Africa. However, we do think
that towns and urban networks were an important element in the emergence
of extensive patterns of Trans-Saharan connections and that this was happen-
ing across a wide space as early as the second half of the first millennium BC.
In the wider Saharan context, preliminary observations and reports
suggest that there are remains of urban character in a number of other
oasis zones, though by no means all.55 The picture is not always clear;

50
Eldblom 1968, 305–14; Mattingly et al. 2015. 51 Scheele 2010.
52
Some operated at an agricultural deficit, as proposed for Tadmakka by Nixon 2009; 2017.
53
Mills 1993, 194–95. 54 Yoffee 2009, 282.
55
See Chapters 2–7, this volume. An example of an area that may not have developed towns until
a later date, despite having a number of Garamantian era settlements is the Ghat oasis, see
Liverani 2006, 457–62; Mori 2013. The presence of a notably richer cemetery by Ghat compared
to the other known cemeteries is a possible indication that there was in fact a Garamantian-era
antecedent to the Medieval and Early Modern town of Ghat.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 349

excavations at Tadmakka and Timbuktu have suggested that there is little

evidence at those locations of Trans-Saharan trade before the late first
millennium AD. Yet as we have discussed, there is no reason to assume
that Protohistoric centres would underlie those of Medieval date – centres
in one period have frequently been replaced by new nodal sites in
a succeeding era.56 More research is urgently needed to investigate the
possibility of pre-Islamic activity at oases further afield: Djado and Bilma in
Niger, Tuwat and Tamanrasset in Algeria; Wadi Ziz and Wadi Draa in
southern Morocco.57
In Fazzan, we have suggested that 4 ha (and perhaps c.400–800
inhabitants) represents the upper limit for a purely agricultural settle-
ment that was reliant on wells or foggaras for irrigation. There are also
centres of c.7.5–10 ha and above and we have argued that these belong
to an urban element in the pre-modern settlement hierarchy.58 These
figures are significantly below what we encounter in many other parts
of the world, but we must recognise that the harsh desert conditions
constrain the emergence of larger nucleated centres. Fletcher provides
a possible explanation when he links maximum settlement area
with communication range and traffic capacity.59 We suggest that
hyper-intensive farming in gardens requiring hours of irrigation
every day increases traffic (in Fletcher’s terms), whilst the low avail-
ability of (and restricted access to) water and high summer tempera-
tures effectively lowers communication range far below that seen in
temperate agricultural regimes. It is no coincidence that many oases
comprise multiple small villages spread through the palm groves,
reducing travel distance between dwelling and gardens. As demon-
strated in Chapter 5, some of the oases in the Tunisian Gabes, Jarid
and Nefzaoua oases were recognised as ‘proper’ towns by Rome, but on
the ground seem to have comprised a series of closely spaced villages,
with no one element being significantly larger or more developed
architecturally. Some oasis towns may thus be seen as distributed or
composite centres, comprising multiple village-scale settlements. The
scatter of centres at Waddan is a prime archaeological example. But
sites in the Central Sahara like Jarma, Qasr ash-Sharraba, and Zuwila
demonstrate that the alternative model of a single primate centre was
also present.

56
Insoll 1998; Nixon 2009.
57
For other recent reviews of urbanisation in Africa, see MacDonald 2013; Mattingly and
MacDonald 2013; McIntosh 2015; Sinclair 2013.
58
Mattingly and Sterry 2013. 59 Fletcher 1995, 84.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 350 Martin Sterry and David J. Mattingly

How to Build (and Sustain) a Saharan Town?

Urban settlements of Fazzan were not the consequence of the region
having particularly high potential for its fertility or privileged access to
natural water sources. There were lots of oases, but relatively few of them
were spring-fed and many have historically been dependent on well
irrigation.60 The reality seems to be that substantial outside investment
had to be made to create urban centres that had functions beyond farming.
The most likely reason behind this investment is undoubtedly in securing
trade routes or exercising political authority.
A paradox of Saharan oases is that they were vulnerable economically if
they were solely dependent on irrigated agriculture. The capital-intensive
costs of irrigated farming have constrained the self-sufficiency of most oases
through time. Trade is a crucial adjunct to agriculture, even if it is mainly
routed through a few nodal settlements, which have acquired higher eco-
nomic and political status as a consequence.61 It has been hypothesised that
oasis societies are non-viable without parallel development of trade. We
would add the additional argument that successful oasis agriculture coupled
with long-range trade was also liable to result in urbanisation. It can equally
be argued that towns were unlikely to have developed in the Sahara on the
basis of oasis agriculture alone and their sustainability was closely linked to
the vagaries of Saharan trade. The Trans-Saharan networks of interconnec-
tions changed over time, with new nodes emerging periodically, often at least
in part as a result of external events and changes.62 The fortunes of Jarma
changed dramatically when trade routes shifted away from that town in the
Islamic era.63 Where we see evidence of intensive growth in oasis agriculture,
it is likely that there would also be a related primate centre either in the oasis
or closely connected to it. The Saharan town is thus another example of
transformative urban forms.64
So far we have mainly focused on the possible identification of
Protohistoric origins of some Saharan urban settlements, though noting
also the importance of Islamic foundations later, especially in the Western
and Southern Sahara. Several chapters in the next part of the book will
assess in more detail the specific character of the caravan cities associated
with Islamic era Trans-Saharan trade. A case could be made, perhaps, for
Islamic innovation in the spread of Saharan urbanisation. However, at this
point we would simply restate a key point of this section of the book: the
existence of pre-Islamic origins of large-scale and atypical oasis settlements
60
Scarin 1934. 61 See Scheele 2010, 284–88. 62 Lydon 2009.
63
Mattingly et al. 2013b, 512–38. 64 Cf. Renfrew 2008.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 351

at locations that went on to be Islamic era centres (for example, Jarma,

Zuwila, Ghadamis, etc.) introduces a major difficulty for that argument.
Even if the Medieval Saharan towns had some unique features, it is hard to
deny the existence of pre-existing urban or proto-urban traditions. Though
these remain ill-defined archaeologically in many cases, they need to be
taken account of in relation to the Medieval developments.

References

Barker, G., Gilbertson, D., Jones, B. and Mattingly, D.J. 1996a. Farming the Desert. The
UNESCO Libyan Valleys Archaeological Survey. Volume 1, Synthesis. (Principal
editor, G. Barker). Paris/London: UNESCO, Society for Libyan Studies.
Barker, G., Gilbertson, D., Jones, B. and Mattingly, D.J. 1996b. Farming the Desert.
The UNESCO Libyan Valleys Archaeological Survey. Volume 2, Gazetteer and
Pottery. (Principal editor, D.J. Mattingly). Paris/London: UNESCO, Society for
Libyan Studies.
Bowman, A. and Wilson, A. (eds). 2011. Settlement, Urbanization, and Population.
Oxford: Oxford University Press.
Chekhab-Abudaya, M. 2016. Le qsar, type d’implantation humaine au Sahara:
˙
Architecture du Sud Algérien. Oxford: Archaeopress.
Cowgill, G.L. 2004. Origins and development of urbanism: Archaeological
perspectives. Annual Review of Anthropology 33: 525–49.
Cremaschi, M. 2006. The Barkat oasis in the changing landscape of Wadi
Tannezzuft during the Holocene. In Liverani 2006, 13–24.
Duckworth, C., Cuénod, A., and Mattingly, D.J. (eds). Forthcoming. Mobile
Technologies in the Ancient Sahara and Beyond. Trans-Saharan Archaeology
Volume 4. Series editor D.J. Mattingly. Cambridge: Cambridge University Press
and The Society for Libyan Studies.
Echallier, J.-C. 1972. Villages désertés et structures agraires anciennes du Touat et du
Gourara. Paris: RMG.
Echallier, J.-C. 1973. Fortéresses berbères du Gourara. Problèmes et resultants des
fouilles. Libyca 21: 293–302.
Eldblom, L. 1968. Structure foncière, organisation et structure sociale. Une étude sur
la vie socio-économique dans les trois oasis libyennes de Ghat, Mourzouk et
particulièrement Ghadamès. Lund: Uniskol.
Fentress, E.W.B. 1979. Numidia and the Roman army. Social, military and
economic aspects of the frontier zone. Oxford: BAR S53.
Fentress, E. 2019. The archaeological and genetic correlates of Amazigh linguistics.
In Gatto et al. 2019, 495–24.
Fentress, E. and Wilson, A. 2016. The Saharan Berber diaspora and the southern
frontiers of Byzantine North Africa. In S.T. Stevens and J.P. Conant (eds), North

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 352 Martin Sterry and David J. Mattingly

Africa under Byzantium and Early Islam, Dumbarton Oaks Byzantine Symposia
and Colloquia 7. Washington, DC: Dumbarton Oaks Research Library and
Collection, 41–63.
Fletcher, R. 1995. The Limits of Settlement Growth. A Theoretical Outline.
Cambridge: Cambridge University Press.
Garcea, E.A.A. (ed.). 2013. Gobero: The No-Return Frontier: Archaeology and
Landscape at the Saharo-Sahelian Borderland. Frankfurt: Africa Magna Verlag 9.
Gatto, M.C., Mattingly, D.J., Ray, N. and Sterry, M. (eds). 2019. Burials, Migration
and Identity in the Ancient Sahara and Beyond. Trans-Saharan Archaeology
Volume 2. Series editor D.J. Mattingly. Cambridge: Cambridge University Press.
Gill, J.C.R. 2015. Dakhleh Oasis and the Western Desert in Egypt under the
Ptolemies. Oxford: Oxbow Books.
Gsell, S. 1972. Histoire ancienne de l’Afrique du nord. Tome V Les royaume indigènes,
orientation sociale, politique et économique. Osnabrück: Otto Zeller Verlag.
Guédon, S. 2018. La frontière romaine de l’Africa sous le Haut-Empire. Madrid:
Casa de Velazquez.
Hardesty, D.L. 2003. Mining rushes and landscape learning. In Rockman and
Steele 2003, 81–95.
Insoll, T. 1998. Archaeological research in Timbuktu, Mali. Antiquity 72: 413–17.
Kirch, P.V. 1980. The archaeological study of adaptation: Theoretical and metho-
dological issues. In M. Schiffer (ed.), Advances in Archaeological Method and
Theory 3. New York: Academic Press, 101–56.
Lahr, M.M., Foley, R.F., Mattingly, D.J. and Le Quesne, C. 2009. Area 131, Jarma,
Fazzan: Archaeological Mitigation of Seismic Acquisition 2006–2008 – Final
Report. Unpublished Consultancy report for Occidental Libya Oil and Gas BV
and Libyan Department of Antiquities.
Lassère, J.-M. 2015. Africa, quasi Roma (256 av. J.C. – 711 apr. J.C.). Paris: CNRS
editions.
Leveau, P. 2018. Climat, sociétés et environment aux marges sahariennes du Maghreb:
Une approche historiographique. In S. Guédon (ed.), La frontière méridionale du
Maghreb. Approches croisées (Antiquité-Moyen Âge). Bordeaux: Ausonius, 19–106.
Liverani, M. (ed.). 2006. Aghram Nadharif. The Barkat Oasis (Sha‘abiya of Ghat,
Libyan Sahara) in Garamantian Times. Firenze: All’Insegna del Giglio.
Liverani, M. Forthcoming. Technological innovation transfer through the hyper-
arid belt. In Duckworth et al., Forthcoming.
Lydon, G. 2009. On Trans-Saharan Trails: Islamic Law, Trade Networks and Cross-
cultural Exchange in Nineteenth-Century Western Africa. Cambridge: Cambridge
University Press.
MacDonald, K. 2013. Complex societies, urbanism and trade in the western Sahel.
In Mitchell and Lane 2013, 829–44.
MacDonald, K.C., Vernet, R., Martinon-Torres, M., and Fuller, D.Q. 2009. Dhar
Nema: From early agriculture to metallurgy in southeastern Mauritania. Azania:
Archaeological Research in Africa 44.2: 280.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 353

McIntosh, R. 2005. Ancient Middle Niger: Urbanism and the Self-Organizing

Landscape. Cambridge: Cambridge University Press.
McIntosh, R. 2015. Different cities: Jenne-jeno and African urbanism. In N. Yoffee
(ed.), The Cambridge World History: Volume 3, Early Cities in Comparative
Perspective, 4000 BCE–1200 CE. Cambridge: Cambridge University Press, 364–80.
McIntosh, S. 1999. Beyond Chiefdoms: Pathways to Complexity in Africa.
Cambridge: Cambridge University Press.
Marcus, J. and Sabloff, J.A. (eds). 2008. The Ancient City: New Perspectives on
Urbanism in the Old and New World. Santa Fe: School for Advanced Research
Press.
Marzano, A. 2011. Rank-size analysis and the Roman cities of the Iberian Peninsula
and Britain: Some considerations. In Bowman and Wilson 2011, 196–228.
Mattingly, D.J. 1995. Tripolitania. London: Batsford.
Mattingly, D.J. 2003. The Archaeology of Fazzan, Volume 1 Synthesis. London:
Society for Libyan Studies.
Mattingly, D.J. 2004. Surveying the desert: From the Libyan Valleys to Saharan
oases. In M. Iacovou (ed.), Archaeological Field Survey in Cyprus. Past History,
Future Potentials. Proceedings of a Conference Held at the University of Cyprus,
1–2 December 2000. London: BSA monograph 11, 163–76.
Mattingly, D.J. 2007. The Archaeology of Fazzan, Volume 2. Site Gazetteer, Pottery
and Other Finds. London: Society for Libyan Studies.
Mattingly, D.J. 2010. The Archaeology of Fazzan. Volume 3, Excavations Carried
out by C.M. Daniels. London: Society for Libyan Studies.
Mattingly, D.J. 2013. The Archaeology of Fazzan, Volume 4. Survey and Excavations
at Old Jarma (Ancient Garama) Carried out by C.M. Daniels (1962–69) and the
Fazzan Project (1997–2001). London: Society for Libyan Studies.
Mattingly, D.J. 2016. Who shaped Africa? The origins of agriculture and urbanism
in Maghreb and Sahara. In N. Mugnai, J. Nikolaus and N. Ray (eds), De Africa
Romaque. Merging Cultures across North Africa. London: Society for Libyan
Studies, 11–25.
Mattingly, D.J. and MacDonald, K. 2013. Early cities: Africa. In P. Clark (ed.), The
Oxford Handbook of Cities in World History. Oxford: Oxford University Press,
66–82.
Mattingly, D.J. and Sterry, M. 2013. The first towns in the Central Sahara.
Antiquity 87.366: 503–18.
Mattingly, D.J., al-Aghab, S., Ahmed, A., Moussa, F., Sterry, M and Wilson, A.I.
2010. DMP X: Survey and landscape conservation issues around the Taqallit
headland. Libyan Studies 41: 105–32.
Mattingly, D.J., Rushworth, A., Sterry, M. and Leitch, V. 2013a. The African
Frontiers. Frontiers of the Roman Empire series, general editor, D. Breeze.
Edinburgh: Hussar.
Mattingly, D.J., Sterry, M. and Thomas, D. 2013b. Jarma in its Saharan context: An
urban biography. In Mattingly 2013, 505–44.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 354 Martin Sterry and David J. Mattingly

Mattingly, D.J., Sterry, M. and Edwards, D.N. 2015. The origins and development
of Zuwīla, Libyan Sahara: An archaeological and historical overview. Azania:
Archaeological Research in Africa 50.1: 27–75.
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017a. Trade in the Ancient Sahara and Beyond. Trans-Saharan
Archaeology Volume 1. Series editor D.J. Mattingly. Cambridge: Cambridge
University Press and The Society for Libyan Studies.
Mattingly, D.J., Bokbot, Y., Sterry, M., Cuénod, A., Fenwick, C., Gatto, M.C.,
Ray, N., Rayne, L., Janin, K., Lamb, A., Mugnai, N. and Nikolaus, J. 2017b. Long-
term history in a Moroccan oasis zone: The Middle Draa Project 2015. Journal of
African Archaeology 15: 141–72.
Mills, A.J. 1993. The Dakhleh oasis colombarium farmhouse. Bulletin de la Societé
archéologique d’Alexandrie 45: 192–8.
Mitchell, P. and Lane, P. (eds). 2013. The Oxford Handbook of African Archaeology.
Oxford: Oxford University Press.
Mori, L. (ed.). 2013. Life and Death of a Rural Village in Garamantian Times. The
Archaeological Research in the Oasis of Fewet (Libyan Sahara), AZA Monographs
6. Firenze: All’Insegna del Giglio.
Nixon, S. 2009. Excavating Essouk-Tadmakka (Mali): New archaeological investi-
gations of early Islamic trans-Saharan trade. Azania: Archaeological Research in
Africa 44.2: 217–55.
Nixon, S. (ed.). 2017. Essouk-Tadmekka: An Early Islamic Trans-Saharan Market
Town. Leiden: Brill.
Reddé, M. 2014. Les frontières de l’empire romain (1er siècle avant J.C. – 5e siècle
après J.C.). Lacapelle-Marival: Archéologie Nouvelle.
Renfrew, C. 2008. The city through time and space. In Marcus and Sabloff 2008,
29–52.
Rockman, M. 2003. Knowledge and learning in the archaeology of colonization. In
Rockman and Steele 2003, 3–24.
Rockman, M. and Steele, J. (eds). 2003. Colonization of Unfamiliar Landscapes: The
Archaeology of Adaptation. London: Routledge.
Scarin, E. 1934. Le oasi del Fezzan. Bologna: Nicola Zanichelli.
Scheele, J. 2010. Traders, saints, and irrigation: Reflections on Saharan connectivity.
The Journal of African History 51.3: 281–300.
Scheele, J. 2012. Smugglers and Saints of the Sahara: Regional Connectivity in the
Twentieth Century. Cambridge: Cambridge University Press.
Scheele, J. 2017. The need for nomads: Camel-herding, raiding, Saharan trade and
settlement. In Mattingly et al. 2017a, 55–79.
Sinclair, P. 2013. The archaeology of African urbanism. In Mitchell and Lane 2013,
689–702.
Smith, A.T. 2003. The Political Landscape: Constellations of Authority in Early
Complex Polities. Los Angeles: University of California Press.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

 8 Discussion: Sedentarisation and Urbanisation in the Sahara 355

Smith, M. (ed.). 2003. The Social Construction of Ancient Cities. Washington, DC:
Smithsonian Books.
Sterry, M. and Mattingly, D.J. 2011. DMP XIII: Reconnaissance survey of archae-
ological sites in the Murzuq area. Libyan Studies 42: 89–116.
Sterry, M., and Mattingly, D.J. 2013. Desert Migrations Project XVII: Further AMS
dates for historic settlements from Fazzan, south-west Libya. Libyan Studies 44:
127–40.
Sterry, M., Mattingly, D.J. and Higham, T. 2012. Desert Migrations Project XVI:
Radiocarbon dates from the Murzuq region, Southern Libya. Libyan Studies 43:
137–47.
Sterry, M., Mattingly, D.J. and Wilson, A.I. 2019. Foggaras and the Garamantes:
Hydraulic landscapes in the Central Sahara. In S. Rost (ed.), Irrigation in Early
States: New Directions. Chicago: The Oriental Institute of the University of
Chicago.
Wilson, A.I. 2011. City sizes and urbanization in the Roman Empire. In Bowman
and Wilson 2011, 161–95.
Wilson, A.I., Mattingly, D.J. and Sterry, M. Forthcoming. The diffusion of irriga-
tion technologies in the Sahara in antiquity: Settlement, trade and migration. In
Duckworth et al. Forthcoming.
Yoffee, N. 2005. Myths of the Archaic State: Evolution of the Earliest Cities, States
and Civilizations. Cambridge: Cambridge University Press.
Yoffee, N. 2009. Making ancient cities plausible. Reviews in Anthropology 38:
264–89.

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press

https://doi.org/10.1017/9781108637978.009 Published online by Cambridge University Press
 part iii

Neighbours and Comparanda

Published online by Cambridge University Press

Published online by Cambridge University Press
 18 Urbanisation, Inequality and Political Authority
in the Sahara
judith scheele

Introduction

Recent archaeological research has shown that Garamantian civilisation was

much more extensive than previously thought, allowing for surprisingly high
population density, agricultural exploitation and urbanisation. This, then,
poses the question of how to make sense of the available material, and how to
best describe the Garamantian polity and its evolution. Several aspects of the
archaeological and historical record are at issue here: the nature of settle-
ments and their interrelation, their connections with the surrounding poli-
tical and economic landscape, their ecology and economic base, and the
nature and organisation of their polity. Most recent publications have inter-
preted these features as necessary components of one coherent political
complex, conveniently labelled as an ‘early state’. Thus, Wilson describes
the Garamantes as a ‘substantial agriculturally-based trading state’.1 For
Mattingly and Sterry, Garamantian settlements ‘give clear evidence of state-
level activity in the north-eastern Sahara’; Mattingly postulates the ‘existence
of an incipient state in Fazzan from the first millennium BC’; and Mattingly,
Sterry and Thomas speak of a ‘literate, state-level society’.2 By the early
centuries AD, Mattingly claims elsewhere, this was moreover a territorial
state, ‘covering c. 250,000 km2 of the Libyan Sahara’.3 For Liverani, the
matter leaves even less room for doubt:
We can confidently postulate the existence of a Garamantian complex chiefdom in
mid-first millennium BC; while a true or proper Garamantian state or kingdom
(well attested in Roman times) could have begun slightly later, with the first
urbanization in Germa.4

1
Wilson 2012, 409.
2
Mattingly 2003, 355; Mattingly and Sterry 2013, 517; Mattingly et al. 2013a, 517.
3
Mattingly and MacDonald 2013, 72. 4 Liverani 2000, 511. 667

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 668 Judith Scheele

Elsewhere, and perhaps more cautiously, the term ‘Garamantian kingdom’

is used, with references to classical authors that speak of a ‘Garamantian
king’. In all these cases, the development of the Garamantian polity is
mapped on an evolutionary scale: ‘the rise of the Garamantes bears on
a long-term process involving the transformation of pastoral communities
of the central Sahara into increasingly unequal and territorial agricultural
societies, eventually coalescing as a slave-using kingdom’.5
This paper attempts to unpick the knot here implied, of urbanisation,
sedentarism, socio-economic inequality, labour exploitation, territoriality,
and central state power, with reference to more recent and hence better
documented Saharan history and ethnography. Such evidence cannot of
course describe social and economic developments in the Fazzan of two
millennia ago, and the aim of this chapter is in no way to suggest stasis, or
even structural continuity. More recent evidence can, however, alert us to
our own, forcibly parochial assumptions, and to problematic shortcuts that
link phenomena that are a priori separate and independent – such as the
existence of social inequality and urbanisation, and the development of
territorial states – by showing how these, in the Sahara at least, did not
always necessarily imply each other. Much of this might simply be
a question of the terminology chosen (‘state’, versus ‘complex polity’, for
instance), but words matter, inasmuch as they bear assumptions that in
turn might shut off potentially fruitful avenues of inquiry, pre-empting
important questions and preventing others from being asked.

Urbanisation

The archaeological evidence leaves no doubt that ancient Garama is best

understood as a town:
Garama was a site of unusual size and complexity within the Garamantian king-
dom, and much of its monumental architecture is unique at present. There are
ample indications that this was a place of high political and religious authority and
an unusually well-connected economic focus. Garama seems urban by these
criteria, though Saharan urbanism clearly differed in important respects from
Mediterranean (or indeed sub-Saharan) urbanism.6

For Mattingly and Sterry, this degree of urbanisation and traces of complex
settlement hierarchies provide ‘key evidence of . . . complexity’:

5 6
Mattingly et al. 2013a, 510. Mattingly et al. 2013b, 290.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 669

The large urban centres of Fazzan were not the consequence of particular fertile
locales, but places where substantial outside investment had been made to create
a centre that had functions beyond farming. The most likely reasons behind this
investment are undoubtedly control of trade routes or exercising of political
authority.7

This impression is furthered by the existence of ‘public buildings’ and ‘the

large-scale erection of forts and urban defences’.8 Indeed, according to
archaeological markers – the existence of a town wall, a qasaba, two
mosques, and several saints’ tombs – Jarma continued to be ‘urban’ until
the end of the seventeenth century.9 But can we conclude from this that
Garama was also necessarily the centre of a state, or indeed that it func-
tioned within a settlement hierarchy that would pull its hinterland together
in a hierarchical fashion?
In Fazzan as in the Sahara more generally, the existence of urban centres
can only be justified through functions and sources of income that go
beyond local resources: trade and ‘the exercise of political authority’ are
indeed obvious candidates here. Yet while there is ample evidence for the
existence of trade, and indeed for the decline of Garama/Jarma after trade-
routes shifted,10 the notion of ‘political authority’ is more difficult to
capture. As much as we can assume that a town like Garama, distinctive
from everything that surrounded it, exerted some kind of dominance over
its ‘hinterland’, and might have been constructed with external, state-based
models in mind, the term ‘political authority’ can cover many realities
apart from centralised state administration. And indeed, factors that are
usually seen to indicate the presence of states – records of taxes, adminis-
tration, or coined money – are conspicuously absent from the findings at
Garama, although the necessary techniques and raw materials were locally
available.11 Similar caution is warranted when analysing the existence of
monumental ‘public’ buildings in Garama. These indeed seem to have been
impressive; but too little is known of their function and use to be able to say
with certainty that these were really ‘public’, rather than privately con-
structed temples. Indeed, Mattingly et al. themselves use the term ‘public’
in inverted commas, and note that ‘we need to keep an open mind about
domestic/public attribution of our fragmentary remains’.12 And although
the notion of the state necessarily implies some idea of a public sphere or
good, the converse is not true: ‘public’ buildings can be erected and
7
Mattingly and Sterry 2013, 506, 515; see also Mattingly et al., Chapter 2, this volume.
8
Mattingly et al. 2013b, 292; Mattingly and MacDonald 2013, 73.
9
Mattingly et al. 2013a, 523. 10 Mattingly et al. 2013a, 522.
11
Mattingly et al. 2013a, 515, 517. 12 Mattingly et al. 2013b, 295.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 670 Judith Scheele

maintained by other forms of collective authority, such as local councils, or

they can be provided by leading families or individuals. Susan McIntosh
thus calls for caution in the interpretation of the archaeological record:
In societies where political action and coordination is achieved through assemblies,
councils, and other forms of horizontally arrayed or democratic structures, and not
through conspicuous vertical control hierarchies, we have relatively little idea of
how we might recognise the material manifestations of such an organisation.13

In the Sahara of written history, meanwhile, there are many more examples
of municipal councils and local initiatives than of state projects. Despois
noted that most qsur in the Fazzan were built by individual families whose
names were still known in the 1940s.14 Local archival records and French
colonial sources attest the importance of municipal self-government
throughout the area, in the eighteenth and nineteenth centuries at least.15
Before then, jama‘at (local assemblies or tribal councils) were seen as
a defining feature of the bilad al-siba, the country beyond the reach of
central government – which covered all of the contemporary Sahara and
much of the Maghrib.16 These councils were routinely in charge of the
construction and maintenance of complex irrigation systems, fortresses
and defensive walls, markets, mosques and shrines.17 Lest these settlements
are seen as too small to qualify as ‘urban’, there are plenty of examples
elsewhere in the Sahara of longstanding trading towns that, at a later date,
grew to impressive size and clearly deserved to be qualified as ‘towns’, and
that prospered in the absence or margins of states, such as, for instance,
Timbuktu, Walata, Ghadamis, or the cities of the Mzab.18 For the four-
teenth century, Nixon thus speaks of urban centres that were strongly
associated not with states but with ‘strong nomadic groups’.19 These
urban centres were usually based on independent internal forms of gov-
ernment, through councils or large dominant families, with little internal
taxation, a clear preoccupation with trade and mobility, little territorial
appetite, and no independent state institutions.20 In the Sahara perhaps

13
McIntosh 1999, 22. 14 Despois 1946, 101. 15 Scheele 2010.
16
Touati 1993, 98; Voguet 2009.
17
Irrigation: Bédoucha 1987; Grandguillaume 1978; Powers 2002, 95–140; fortresses: Montagne
1930; Nehlil 1916; markets: Aspinion 1954; mosques and shrines: Scheele 2012, 48.
18
See Cleaveland 2002; Eldblom 1968; Masqueray 1886; Milliot and Giacobetti 1930; Saad 1983.
19
Nixon, Chapter 17, this volume.
20
These towns can perhaps be described as ‘city-states’, but they show little in the way of a public
sphere or independent institutions, and indeed leave one to wonder what might be gained by
such a label, apart from a (possibly mistaken) impression of familiarity.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 671

even more than elsewhere, the development of cities and that of states
hence needs to be treated separately.21

Settlement Hierarchies

Arguments derived from settlement hierarchies pose similar problems.

Mattingly et al. identify a number of distinct supporting settlements near
ancient Garama, in ‘areas of higher potential for exploiting agriculture and
perhaps more importantly trade’, that would have provided supplies for
Garama, and in turn received services, perhaps of an administrative kind.22
In the Sahara, however, classic models of settlement hierarchies need to be
qualified. Oases, of whatever size and even those that relied on a similar
technological infrastructure as the Garamantian Fazzan, rarely produced
a reliable surplus of any kind, and barely managed to reproduce a labour
force sufficient to maintain their existence. Indeed, an argument can be
made for their inherent outside dependency, whatever their size (as the
cost of irrigation does not diminish proportionally) – a point that
Mattingly and Sterry fully recognise elsewhere.23 Such fundamental sub-
sistence problems cannot simply be solved through political fiat or
increased exploitation of local populations or neighbouring villages, as
surplus can only be produced through imports – of labour and goods –
mostly from areas beyond the Sahara proper, or that rely on different forms
of production (such as neighbouring pastoral economies).24 Saharan
economies thus never evolved within a closed system, making regional
settlement hierarchies necessarily porous on all levels, and shifting pro-
blems of political control from internal exploitation to external relations,
mobility and transport. Concepts of ‘dispersed hinterlands’, developed by
Horden and Purcell might thus be more apposite here.25
In the Sahara of recorded history, production for export was thus
crucial, but limited in kind. It mostly consisted in dates, while some

21
Compare also Cowgill 2004, 526; McIntosh, R. 2005. 22 Mattingly et al. 2013a, 533.
23
Mattingly and Sterry 2013, 516; and Chapter 8, this volume.
24
This makes the Sahara fundamentally different from the model of state inception proposed by
Sanmartí et al., Chapter 11, this volume (drawing on Godelier 1980 and Johnson and Earle
2000), who explain the development of hierarchies through the need for increased productivity
and technological change.
25
‘If the term hinterland is to be retained, then the presumption should be that what is in question
is fragmented, not a compact domain that can be mathematically modelled and limned on
a map . . . The dispersed, changeable, hinterland has, we suggest, been the Mediterranean norm’
(Horden and Purcell 2000, 112, also 115–22).

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 672 Judith Scheele

oases could draw on nearby salt-mines (salt-mines that had no agricultural

base near them were mostly worked by anybody who could organise
transport).26 This means that there was little incentive to develop exchange
between neighbouring qsur. Most qsur had no markets: this was as true of
the early twentieth-century Fazzan as of southern Algeria:
There is no market in the Tidikelt, nor is there a special site [for trade]. Business is
conducted at the arrival of caravans who are offered hospitality in different qsur
and carry out their exchanges while staying with their hosts. . . . Exchanges are
carried out behind locked doors. Rich qasr-dwellers receive caravaneers into their
homes, lodge them and are in charge of selling their goods.27

Trade thus connected places far afield, not neighbouring settlements; and
until colonial times, it seems that most Saharan oases groups had no
‘central place’, neither in economic nor in political terms. Economic
regional integration was not a matter of exchange or hierarchical supply
towards an ‘urban’ centre, as even the smallest qsur were not self-sufficient
in all they needed and hence relied on exchange (and therefore had in
a sense ‘urban’ economies).28 Rather, it was a matter of overlapping
property rights. Individual owners invested in several qsur, neighbouring
and distant, mostly, it seems to have spread risk.29
As a result, hierarchies, or relations of mutual dependency, tended to be
established with places much further afield, within the Sahara or on its
boundaries. Hence, in 1919, the Tuwat as a whole had close economic
relations with ‘the Doui Menia and Ouled Djérid (today near the border
with Morocco), with the Ahaggar Tuareg (in the far south) and caravaneers
from El Goléa and the Hauts Plateaux (in north and central Algeria)’.30
Similarly, the ‘hinterland’ of the oases of Fazzan in the early twentieth
century included not only the pastures of northern Arabic-speaking pas-
toralists, but also the Ajjer mountains, whose inhabitants ‘absolutely
needed’ Fazzani grain and dates, while themselves being indispensable to
Fazzani trade and thus survival.31 Indeed, the droughts that hit the Sahel in
the 1970s and 1980s were catastrophic not because of their unusual sever-
ity, but because southern Saharan economies had been cut off from their
usual Sahelian or even Sudanese outlets.32

26
See for example, Clauzel 1960; Gouletquer and Kleinmann 1975; Vikør 1982.
27
Lethielleux 1948, 70; and ‘Rapports annuels, annexe du Tidikelt’, 1908 and 1911, Archives
nationales d’outre mer (ANOM), box 23H102.
28
The French geographer Capot-Rey (1953, 239) thus remained uncertain whether qsur are best
qualified as villages or towns, ‘since their rural and urban characteristics are so closely mixed’.
29
Scheele 2012, 42–43. 30 ‘Rapport annuel, annexe du Touat’, 1919, ANOM 22H50.
31
Despois 1946, 191. 32 Baier and Lovejoy 1975.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 673

As a result perhaps, internal Saharan classifications of settlements tend not

to speak in terms of ranks, with villages, towns and cities, but rather in terms of
different kinds of settlement, some used primarily by nomadic populations,
others by agriculturalists, some which allow for permanent settlement and
production, and others acting as storehouses or defensive structures in times of
need.33 As Mattingly and Sterry observe, traces of settlement do not necessarily
imply a fully sedentary lifestyle, but should rather be understood to lie on
a continuum of practices.34 In the western and central Sahara, the most
common term used to refer to oases is thus ‘qasr’ (plural qsur) or ‘fortress’,
indicating an emphasis on safe storage and defence corresponding to regional
rather than local economic strategies.35 In southern Morocco, collective store-
houses (hisn, agadir or tighremt) count among the oldest settlements, and have
given rise to a lively municipal and customary law tradition.36 These had few
permanent residents, and clearly relied on a widely dispersed nomadic hinter-
land, which could shift over time, as access to agadir was mediated through
property rights rather than membership in social groups. Despois noted the
architectural similarities between the qsur in Fazzan that he visited in the 1940s,
and the Moroccan collective storehouses – a similarity that was, decades later,
endorsed by Thiry, who warned that ‘the beauty of the Moroccan tighremt
should not make us forget the similar plan, and apparently also the similar use,
that they share with the castles (qsur) in the Fazzan’.37
Settlement hierarchies as they appear in the archaeological record might
thus not just rank more or less complex manifestations of the same thing,
but rather describe different kinds of constructions. As Fenwick notes,
drawing on the much better documented example of Umayyad North
Africa, the function of towns cannot merely be understood through their
location or appearance. They need to be seen as part of larger networks and
connections: archaeologically, a town might still look the same although
structurally, it has fundamentally changed, say from an agro-town on the
apex of its own regional hierarchy (with a classic productive ‘hinterland’) to
a relay-post in a much larger network.38 This structural diversity that might
not be apparent in the physical appearance of towns makes the use of the

33
For a description of the early-twentieth-century Fazzan in these terms, see Despois 1946, 101.
34
Mattingly and Sterry, Chapters 1 and 8, this volume.
35
Hence, Ibn Khaldun describing Berber pastoralists in the fourteenth century says that ‘in their
homelands [among which he cites the Algerian Tuwat] they have built qsur and strongholds and
have made gardens of palms and grapes and other fruits’ (in Hopkins and Levtzion 1981, 339);
five centuries later, Auguste Geoffroy (1887) writes of the qasr of Aïn Madhi as built by and for
nomads: ‘settlement was only developed with regards to the seasonal needs of the nomads’.
36
See for example, Montagne 1930. 37 Despois 1946, 60; Thiry 1995, 92.
38
Fenwick 2013, 15; forthcoming.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 674 Judith Scheele

term ‘settlement hierarchy’ potentially difficult, especially in an environ-

ment that is as ecologically fragile and complex as the Sahara: as Taylor et al.
note, ‘to show the existence of a hierarchy, power relations between levels
must be demonstrated’.39

Spiritual Centres

Other types of centrality can be identified in Saharan contexts. From the

eleventh century onwards, the most visible and perhaps also the most
common in the Sahara is that established by religious settlements, or zawaya
(singular zawiya). These are settlements founded by saints, connected gen-
ealogically and spiritually to the wider Islamic world. They tend to be centred
on the founding saint’s tomb, which continues to be a pole of baraka and an
attraction for pilgrims (and thus of outside goods, funds and labour), and
often but not always administered by his descendents, spiritual or otherwise.
Some of these zawaya belong to larger Sufi turuq, and some set up colonies
further afield, often in areas where the original zawiya has economic inter-
ests; many doubled up as more or less temporary markets or fairs, under the
saint’s protection. The more successful among them accumulated consider-
able property holdings, again situated in a widely dispersed hinterland. They
thereby bound various sedentary and nomadic populations together in
complex networks of overlapping ownership, protection and donations,
providing initial funding for agricultural and commercial projects, and
redistributing surplus. The Awlad Sidi Shaykh, for instance, originally from
north-west Algeria, invested in houses and storage places in the Gourara and
Tuwat in southern Algeria; several zawaya had long been set up in their name
throughout the greater Tuwat, owning large gardens as ahbas (endowments)
and worked by slaves or haratin (freed slaves) given to the zawiya. These
zawaya acted as markets and travel lodges, and sent their revenues back to
the mother zawiya in Abiodh Sidi Cheikh, their original fief.40 In southern
Morocco, the zawiya of Tamgrout described by Abdullah Hammoudi and
David Gutelius owned 10 per cent of all irrigated land in the Draa valley, as
well as 35,000 palm trees throughout the area; meanwhile, the zawiya of Iligh,
also in southern Morocco, invested heavily in irrigation in its immediate
vicinity and beyond, as well as controlling and financing regional and trans-
regional trade.41 Further east, the Sanusiyya spread from central and
39
Taylor et al. 2010, 2806.
40
Simon, ‘Notices sur le Tidikelt’, 20 June, 1900, ANOM 22H50; see also Scheele 2012, 45–46, 62–70.
41
Gutelius 2002; Hammoudi 1980; Pascon 1980.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 675

southern Libya south, setting up agricultural colonies in north-eastern Niger

and northern Chad, and providing eastern Trans-Saharan trade routes with
a much needed infrastructure.42
The flow of goods – tribute, rents or gifts – from colonies and clients to the
‘mother-zawaya’ described its own kind of regional hierarchy. Yet these
zawaya were usually established far afield, and no compact hinterland
emerged; moreover, zawaya were inherently colonies, inverting images of
settlement hierarchies constructed from the bottom up. Secondary places
were established from the core outwards: a zawiya did not earn its prestige
from the sway it held over its daughter establishments, but rather, the latter
prospered owing to the former’s baraka (and founder’s investment, and ability
to centralise funds and labour). Hierarchy more generally was imagined on
a family model: saints established original settlements, while their ‘sons’
travelled on and set up dependencies, thereby diffusing their father’s baraka
throughout the region. As segmentary imaginings more widely, such patterns
were limited in range, and eventually lead to fission: no independent state
apparatus ever emerged. Classic political anthropology of the Maghrib pos-
tulates that zawaya were essentially set up in the interstices of tribal structures,
acting as mediators and arbitrators, and deriving their influence from their
lack of political power which was vested in the surrounding, acephalous
tribes.43 This is historically inaccurate, as zawaya accumulated independent
influence, as economic agents, repositories and arbitrators of Islamic legality,
but also as mediators with coastal states.44 Nonetheless, zawaya never devel-
oped into states themselves, and indeed lost their influence if they became too
closely associated with state endeavours.45 Economic and spiritual power, in
other words, was not converted into political dominance, but rather used to
play the locally influential off against each other. Saints acted as arbitrators
and mediators, not as territorial sovereigns.46
Hence, the Sanusiyya, although it was commonly decried as a rival
‘colonial state’ by French military officers,47 seems to have had some
success in northern Chad precisely because it did not act as a state in the
42
Triaud 1995. 43 See for example, Evans-Pritchard 1949; Gellner 1969.
44
Hammoudi 1974; 1980. 45 Ennaji and Pascon 1988, 17.
46
Edwards, Chapter 9, this volume, similarly stresses the importance of spiritual authority in the
development of cities in early Meroe; see also Crone 1986.
47
Hence, according to the French colonial officer Boujol in 1939, in Chad, the Sanusiyya acted like
a ‘genuine state, I would even say a colonial state . . . [Their] zawaya were almost exactly like
colonial outposts, around which local populations crystalised. Through a regular courier service,
they constantly kept in touch with their capital at Kufra, and the muqaddamin who managed
them were genuine administrators. Taxes were paid to them, forced labour carried out,
agriculture and trade developed, and caravans found in them safe relay posts and protection
against plunderers’ (Boujol, ‘La Senoussya au Tchad’, 7 June, 1939, ANOM 10 APOM 401).

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 676 Judith Scheele

European sense, aiming at territorial sovereignty. Rather, it attempted to

control strategic points, by setting up trade relays and agricultural colonies
with labour they had brought with them. Their focus was on control of
(some) people, rather than territory, arbitration rather than force,
a reliance on gifts rather than taxes, and flexibility instead of centralised
control.48 Its internal administrative structure was loose, based primarily
on family and spiritual ties, with different zawaya vying for resources, and
little control at all exerted over their nomadic auxiliaries.49 This was so
because the Sanusiyya intrinsically relied on existing infrastructures and
alliances, without which they could not have survived; and they adopted
and partially restructured existing regional relations. Instead of super-
imposing an alien political construct on local resources that could not
finance it, the Sanusiyya infiltrated existing patterns and structures, tem-
porarily pulling them together into a larger whole. Hence, perhaps, their
visible fragility (they were easily defeated by the French, and ostensibly left
few traces when they collapsed), coupled with an underlying long-term
resistance (the patterns of ownership they established in northern Chad for
instance are still visible, as are the trade networks they established or rather
co-opted and expanded). The zawiya of Iligh in southern Morocco
similarly

attempted to play, among the tribes, the role of a state, but without a state’s
prerogatives or legitimacy. It seems that Iligh never raised taxes . . . nor did it
intervene militarily among the tribes in other cases than those that immediately
concerned it. Iligh carried out reconciliations, acted as guarantors for securities
and received fines for breach of tribal peace agreements . . . without getting
involved in relations among tribes as long as these did not threaten them.50

A different type of internal Saharan colonisation is exemplified by a more

regionally and historically specific type of Saharan settlement, the zariba
(plural zara’ib). Most contemporary towns in the eastern Sahara and
Sahel originated as zara’ib – that is, temporary camps set up by slave
raiders on their way south.51 Many of these turned into more permanent
settlements, through the agricultural settlement of slaves around them.52

48
As Evans-Pritchard (1945, 187) noted: ‘Within this vast theocratic empire there was little
centralised planning or control. New lodges were founded often by the initiative of influential
families and the head of the order might even not hear of a new foundation till after it had been
completed. In the more distant parts the lodges were autonomous or came under the control of
a local mother lodge.’
49
Djian 1996 [1915–1916], 123. 50 Ennaji and Pascon 1988, 27–28.
51
Johnson 1989, 80; see also O’Fahey 1973, 34. 52 Johnson 1989, 78.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 677

These slave raiders were initially either sent by the Egyptian state, or by
private companies that supplied the Egyptian armies with slave soldiers,
but they sparked imitations through the eastern and central Sahel, often
by slave soldiers themselves.53 They thus constituted another form of
Saharan settlement, again bound up with external logics of trade, produc-
tion and conquest that were closely connected to state aspirations, but
that never aimed at establishing state control in the region itself.54
Although they might thus look at first sight similar to the slave towns
identified by MacDonald and Camara in Segou, they were clearly per-
ipheral to the state, and often constructed at great distance from areas
under state control – as they could effectively only function beyond the
state’s boundaries.55 Indeed, Douglas Johnson claims that one of
the reasons why the British conquest of Sudan was so slow was because
the British needed a periphery beyond their direct control, ideally
a periphery torn apart by warfare, whence they could draw supplies for
their own slave army (which they had taken over directly from the
Egyptians).56 Similar arguments for the maintenance of a necessary
slaving frontier could be made for regions further west, although here
this has not resulted in the development of permanent settlements of the
zariba-type.57
There can, then, be no doubt about the outside dependency of
Saharan settlements, or indeed the fact that they were often bound up
with projects of regional control. But these projects of regional control
might be spiritual and commercial, or based on the hunting and
gathering of slaves, as much as on stable institutionalised political
rule by a state apparatus independent of local social structures
(which, again, taking into account the limited resources of the Sahara
proper, would have been difficult to sustain). Nor did they necessarily
aim at a monopoly of force, but rather inscribed themselves into
a polycentric political world – a world they attempted to manipulate,
not to dominate. In a sense, they developed a particular form of
centrality, parasitic on economic developments elsewhere, beyond the
boundaries of the Sahara proper, and that favoured extraversion over
territorial consolidation.58

53
Cordell 2003, 34; Johnson 1989, 73.
54
Note, however, possible parallels here with the garrison towns, identified by Fenwick
forthcoming, in Umayyad North Africa.
55
MacDonald and Camara 2012, 178. 56 Johnson 1989, 81. 57 Testart 1998, 61.
58
On extraversion, see Bayart 2000; on the utility of the concept in archaeological research, see
Edwards 2004; Mattingly et al. 2015, 58; Stahl 2014.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 678 Judith Scheele

Inequality, Slavery and the Exploitation of Labour

In much of the literature, the stark inequality that characterised

Garamantian society – as is apparent from evidence both for the use of
slave labour and for internal socio-economic distinctions – is seen as an
additional proof of the existence of a centralised state. Hence, Mattingly
concludes that monumental burials, expressing status distinctions, indicate
social hierarchies of a kind generally associated with a state.59 Mattingly
and MacDonald, meanwhile, cite as one of the main proofs for the exis-
tence of a centralised Garamantian state ‘the simple consideration of the
amount of work required just to construct the irrigation systems’.60 This,
however, assumes that inequality, labour exploitation and especially slav-
ery are necessarily related to the existence of states.61 Conversely, the
absence of traces of state centralisation would indicate equality, or ‘heter-
archy’, and presumably the absence of slave labour.62 Saharan history
belies both assumptions.63 While states are rare (although not unheard
of) in the Saharan historical record, slavery has long been at the heart of
Saharan modes of production, and has probably preceded and certainly
outlasted particular political formations. Again, beyond the historical
evidence, a structural argument can be made for this: if the only way in
which Saharan oases can be made to yield a surplus is by starving its
workforce, the regular importation of slaves appears a fundamental
necessity.64 At a much later period, the French colonial sources leave no
doubt that oasis populations were unable to reproduce themselves, and
chronic demographic decline – only offset by mostly forced immigration –

59
Mattingly 2003, 87; but see McIntosh 1999, 14 for a different possible explanation.
60
Mattingly and MacDonald 2013, 75; see also Mattingly et al. 2013a, 510.
61
In a similar vein, MacDonald and Camara (2012, 171, 174) link the observation of a generally
recognised need for labour with the close connection established in Segou (and other Sahelian
states) between states, warfare and slave-raiding: for a similar argument, see Azevado 1998;
Reyna 1990; Roberts 1987. All of these latter cases, however, inscribe themselves into
a particular historical context: that is, Sahelian and Sudanic Africa while it was rapidly exposed
to international slave trade, and suffering from a high degree of internal warfare. See also
Sanmartí et al., Chapter 11, this volume, who state that the formation of states ‘normally
accompanies’ hierarchical organisation of human societies.
62
McIntosh, R. 2005; McIntosh, S. 1999.
63
More generally, there is no reason to assume that slavery – or any other form of labour
exploitation – was ever the monopoly of states: ‘Slavery is a wild card whose use depends on
social strategies, defined elsewhere. The existence of slavery does not define a type of society,
but particular societies define types of slavery’ (Testart 1998, 41).
64
As Meillassoux (1986) famously argued, slavery is a means of obtaining labour without having
to meet the cost of the biological and social reproduction of people; it can thus operate at
a permanent disequilibrium.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 679

is a recurrent theme in historical descriptions of virtually all settled

Saharan societies.65 As Edwards notes in the very different context of
early Meroe, cities were often ‘consumers of people’, at a scale that we
now find difficult to imagine.66 They were also unpleasant places, and
people did not necessarily choose to live there unless coerced.
However they were structured politically, Saharan societies historically
relied heavily on servile and slave labour, for herding, guarding the flocks,
and salt mining, but even more so for irrigated horticulture, religious
settlement, and, especially in the eastern Sahara, military endeavours.67
Slaves were also crucial to local and regional reproductive strategies, lead-
ing to complex gradations of hierarchy and status within families.68 In the
Sahara as elsewhere, slaves were first and foremost strangers who could
claim no prior ties with their captors or buyers; their servile status was
defined by their lack of connectivity, and hence mobility.69 This means that
once a slave was put to work in an oasis and deprived of independent
means of transport, no state apparatus was needed to keep him or her
there; connections made between low status and immobility were self-
fulfilling.70 Conversely, enslavement was not only a way of incorporating
much needed outside labour into Saharan economies, but also a guarantee
of internal flexibility, crucial in an environment where crises are always to
be expected: status hierarchy ‘functioned primarily to assure first access to
resources in times of scarcity’.71 Slaves and clients could be set free to fend
for themselves when resources were scarce: it is thus not the state-like
aspect of the system, but rather its unbounded and decentralised nature,
coupled with structural uncertainty and the need for risk management,
that explains the persistence of slave labour and other forms of
dependency.72

65
Faced with demographic catastrophe after the abolition of the slave trade, the French colonial
administration thus proposed to forcibly settle French West African soldiers in the Saharan
oases, with little success: see ‘Note sur la question noire en Algérie’; ‘Lettre du Gouverneur
Général de l’Algérie au Général commandant le 19e corps de l’Armée’, 23 April 1913; and
‘Rapport du Capitaine Vincent, chef d’annexe de Beni Abbès, sur l’installation des troupes
noires en Algérie’, 3 March 1919, all ANOM 3H13.
66
Edwards, Chapter 9, this volume.
67
See Bonte 1998; Gutelius 2002; Johnson 1989; Lovejoy 1986; and McDougall 1986, respectively.
68
McDougall 1998. 69 Meillassoux 1986; Rossi 2009; Testart 1998.
70
This makes the Saharan case very different from regions with more abundant natural resources:
see for example, Cooper (1980) on the great difficulties experienced by British colonial officers
in setting former slaves to work on plantations, as the latter tended to run off into the forest
where they could easily appropriate land to grow their own food, far from any form of coercion.
In the Sahara, this was just not a possibility.
71
Baier and Lovejoy 1975, 561. 72 Baier and Lovejoy 1977.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 680 Judith Scheele

Slavery, moreover, only describes the bottom rung of a complex system of

gradation that admits no centre and no boundaries. The classic account of
such hierarchies is given by Stewart on Mauritania, who provides
a description of a complex system of gradations between full slavery and
‘nobility’, on the one hand, and Islamic excellence on the other, with tribute-
paying free men and freedman somewhere in between.73 Although the dis-
tribution of wealth did not necessarily follow such status-distinctions – as
those at the ‘top’ were defined by their boundless generosity, but also by their
unlimited access to other people’s property – real differences in socio-
economic position were clearly very much part of Saharan life for most of
its history, in ‘tribal’ as much as in urban settings.74 An early twentieth-
century manuscript from Timbuktu gives us an internal view of these dis-
tinctions: all people of the Sahara and the Sahel are ranked here according to
their genealogical excellence, which can be mapped into Islamic history; these
differences matter, inasmuch as they describe relations of tribute-paying and
protection, many of which still persist today (and explain, for instance,
political alignments in the current northern Malian civil war).75 This allows
for much internal differentiation; meanwhile, centrality and value here is not
given by the existence of a state-apparatus, but rather by the Islamic revelation
that reasons in terms of spiritual rather than political power.76 Indeed, con-
temporary states and their armies are often seen by those on the bottom of
a scale as a way of breaking out of such status hierarchies; and we can surmise
that this was also the case of historic states, in those cases where they had
a direct impact on Saharan realities.77 Social complexity, in other words, much
as exploitation, might thrive in the margins of states as much as at their core.78
Although there can be no doubt that these Saharan status hierarchies did
not derive from state influence, they at times led to the formation of state-
like structures, in particular on the edges of the Sahara. Hence, Pierre Bonte
describes the emergence of the Mauritanian emirates from the seventeenth

73
Stewart 1973.
74
There is a tendency, in much of the literature, to associate tribal systems with equality; this only
holds true if we believe local ideologies, and discount all those – in many cases a majority – who
cannot pretend to full tribal status, and hence ‘drop out’ of such reckonings altogether. For an
account of the longevity of Saharan inequalities in one such setting, see Hall 2011.
75
Muhammad Mahmūd wuld Shaykh, Kitab al-turjaman fī ta’rīkh al-sahara wa al-sudan wa bilad
˙ ˙
tinbuktu wa shinjit wa Arawan wa nubadh fi ta’rikh al-zaman fi jami‘a al-buldan, further
discussed in Scheele 2012, 139–42.
76
Corresponding status categories can be found among Tuareg, see e.g. Casajus 1990; Clauzel
1962; and, in a somewhat attenuated form, among Tubu-speakers, Brandily 1988.
77
Klein 2005, 832; Schmitz 2009.
78
For a different period, Fenwick 2013, 18 similarly notes clear indications of socio-economic
inequalities within North African tribal societies.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 681

century as a result of internal stratification among the ‘nobles’ brought

about by marriage strategies.79 Yet the ‘states’ thus developed had certainly
no territorial reality, nor much administrative power (or indeed aspira-
tion): the amirs received tribute, and acted as supreme arbiters, but they
never developed anything that resembled an administrative state apparatus
that could become independent of the regional tribal structure: ‘the
Moorish emirate reveal themselves as original tribal political
formations’.80 They are not tribes turned into states, but rather ‘tribal
states’ (or rather state-like constructs) that were momentarily superim-
posed on a tribal structure without profoundly altering it.81 Elsewhere in
the Sahara, ‘sovereigns’ similarly turn out to be primarily arbitrators – as is
the case of the ‘sultan’ of Agadez, who famously had to be of slave
descent.82 The scope of political action of the Tuareg amenokal was
similarly limited, unless he acted in times of conflict.83 Such ‘sultans’
might look the part to outsiders, and even act it in external relations, but
they had in fact little influence on the ground, and certainly did not develop
an independent centralised state administration, or even territorial ambi-
tions beyond the control of strategic trade routes and settlements.84

Centralisation and Territoriality

The notion of pre-colonial African states is a contested one, to the point

where some say that we should stop using the term altogether.85 Hence,
Skalník warns us that ‘what students of the African past labelled as ‘state’
has, on the whole, little to do with the state as it developed both as a concept
and as a social reality within the orbit of western civilisation’. Worse,
by labelling e.g. a certain kind of African political organisation a ‘state’ one does
disservice to African history and the Africans themselves because not only is
African originality forced into a Eurocentric straight-jacket but worst of all this

79
Bonte 2008. 80 Bonte 2008, 519.
81
‘A tribal state is a state superimposed on a society that is designed to cope without it and which may
accordingly revert to statelessness at any time. It is only when the autonomous and self-sufficient
nature of the building blocks has been undermined that we have a state as opposed to a tribe’
(Crone 1986).
82
Claudot-Hawad 1998, 23; Urvoy 1934. 83 Claudot-Hawad 1990; Keenan 1977.
84
As Fentress 2006, 13 notes for the North African coast and its immediate hinterland, Berber
Hellenistic princes or kings might have looked such to their ‘civilised’ neighbours and
overbearing allies, while back home, they acted like rotating chieftains within a segmentary
world (see also Brett and Fentress 1996, 33).
85
Lonsdale 1981, 139.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 682 Judith Scheele

bias denies the possibility that, in Africa, viable alternatives to the state, as it
evolved in the West, may have been developed.86

Instead, as Lonsdale notes, ‘the most distinctively African contribution to

human history could be said to have been precisely the civilised art of living
fairly peaceably together not in states’.87 But even those scholars who are less
sceptical mostly agree that it is difficult to apply Western European territorial
models of statehood to Sub-Saharan Africa. Most of Sub-Saharan Africa has
historically been marked by an abundance of land and a shortage of labour,
which means that notions of power, wealth and influence were mediated
through people rather than land, while agricultural surplus was exceptional
and in any case difficult to transport and store.88 As a result, Sub-Saharan
polities, however coercive they might have been, rarely had defined bound-
aries, but rather a core and an ill-defined periphery. This also means that
historically, much of Sub-Saharan Africa was best characterised as a ‘frontier
society’, making it possible for people to try their luck elsewhere, and thereby
limiting state coercion, or alternatively, basing it on continual violence.89 As
a result, state revenue was generally derived less from taxation than the
control of trade and booty, in other words, raids in the state’s peripheries.90
Similar features seem to emerge from the archaeological record in
Fazzan, where Mattingly and MacDonald point to ‘the disparity of material
culture between the core and the periphery of Garamantian territory’.91 Yet
in the Sahara, where permanent agriculture is limited to a few nodal points,
and most land is used extensively by inherently mobile people (whether
pastoralists or farmers), notions of territory, let alone of territorial states,
have to be treated with even more caution.92 Political formations were
inherently mobile, to the point where Grémont, in his study of the histori-
cally dominant Tuareg group north of the Niger Bend, the Iwellimiden,

86
Skalník 1983, 25, 26. 87 Lonsdale 1981, 139. 88 Coquery-Vidrovitch 1975.
89
Kopytoff 1987.
90
Edwards 1998; Edwards, Chapter 9, this volume, applied these arguments to his analysis of
Meroe on the Upper Nile. Citing historical and contemporary Sudanic examples, he concludes
that in Meroe also, ‘the control of land did not form a primary power base. The major
determinant of production remained the availability of labour rather than land, and political
power associated with it was derived from the control of people rather than territory’. As
a result, ‘the most important state revenues were derived elsewhere, from non-agricultural
exotic resources’ (Edwards 1998, 178, 183).
91
Mattingly and MacDonald 2013, 73.
92
In any case, in light of the increased evidence for early urbanisation and perhaps even state
formation in Sub-Saharan Africa, there is no good reason to assume that northern political
models would have had more impact on the Garamantes than southern ones. As MacDonald
notes in this volume, ‘at a broad, syncretic level, I prefer to see the Trans-Sahara as a sphere of
interaction in trade and ideas – much like the Mediterranean. These ideas go both ways’.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 683

quickly gave up all attempts to describe their political structure in terms of

territory, as such questions simply made no sense to his interlocutors.93
This high degree of mobility was also true of earlier dominant polities in
the area.94 More generally, although there is no shortage in Saharan history
of the emergence and decline of dominant groups, these generally did not
attempt to establish territorial states, but rather vied with each other for the
control over people, routes and strategic way-points: they established ‘net-
work territories’, as Pliez writes of the Sanusiyya.95 In the Sahara, local
production mostly did not even allow for self-sufficiency, making the
extraction of regular taxes difficult, while distances were such that any
practical continued projection of power was severely limited.96 Tribute,
raiding and sporadic taxation of mobility thus appear as the dominant
forms of ‘governance’; ‘power lies in movement’, as Retaillé summarises,
and internal hierarchies are perhaps best expressed as gradations of mobi-
lity and immobility.97 The inherent mobility of dominant political groups
was exacerbated by the fact that strategically important points were them-
selves rarely fixed: in the Sahara, ‘connections produce places, and not vice
versa; places that matter are crossroads. Centrality moves across the desert
according to the flow of social networks’.98
This corresponds to Saharan ethnography and history more widely:
historically, there simply has been no point in attempting to control vast
stretches of desert, nor has it ever been profitable. What mattered were
routes and crossroads, and certain groups of people, and these were best
controlled – or rather, taxed – either sporadically, or through proxy.99
From the internal point of view, questions of power and coercion were
relegated to the domestic sphere of slavery and tribute extraction, whereas
‘politics’ was generally couched in terms of opposition, arbitration, hospi-
tality and protection.100 As Fenwick notes, even in contexts where the
existence of a central state cannot be doubted, such as Umayyad and post-
Umayyad North Africa, we need to be careful not to impose our own
93
Grémont 2012, 133; see also Grémont 2010.
94 95
See, for example, McIntosh 2008, 360; Retaillé and Walther 2013, 607. Pliez 2006, 692.
96
Witness, for instance, the repeated failure of nineteenth- and twentieth-century imperial
powers to ‘govern’ the Sahara in any meaningful way, until they could rely on motorised
transport and aviation; and witness also contemporary difficulties to militarily control most of
the Sahara, despite foreign intervention.
97
Retaillé 1998, 77; Rossi 2009; and see the discussion of slavery above.
98
Retaillé and Walther 2013, 608.
99
For notions of ‘wealth in people’ expressed in qualitative rather than quantitative terms, see
Guyer 1993, 246.
100
As Bonte 1998, 124 notes with regards to the Mauritanian Adrar, status conflicts were mostly
fought out in the economic rather than the political sphere.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 684 Judith Scheele

contemporary assumptions of the purpose and functioning of states. She

hence describes the Umayyad empire as a ‘low-cost imperial endeavour’
leading to a ‘node-and-corridor’ strategy that meant that many areas
remained beyond state control.101 This, of course, only becomes
a ‘problem’ if we assume a territorial approach, whose historical validity
is often questionable.102
Although we cannot know at this point whether similar reflections
applied to the Garamantes, the limited narrative historical evidence for
the existence of a Garamantian state tends to emphasise control over trade
routes and mobility rather than territory. The Garamantian ‘kings’ of
Greco-Roman writers were never portrayed as sedentary governors of
a territorial state, but rather as inherently mobile: setting out on a raid, in
Ptolemy, or coming in aid of the ‘brigand’ Tacfarinas, in Tacitus.103 This,
of course, reflects a Mediterranean sedentary perspective, describing
a potentially dangerous ‘other’, whose mobility was a military problem,
but also a necessary asset for trade; and clearly it cannot, on its own, sum up
Garamantian civilisation.104 Yet we should not dismiss the reference to
mobility and raiding altogether: not only does it fit with a ‘nodes-and-
corridor’ model, or with the idea of a ‘network-territory’, but the predatory
rather than territorial nature of many pre-colonial states, in particular in
Sub-Saharan Africa, should be kept in mind.105 In the Sahara as much as in
the Sahel, raiding was not an accidental attribute of states, but rather at the
core of their development, a means of ‘importing’ crucial resources, in
a context where state raids in the periphery were probably more productive
than taxation in core areas.106 As Edwards puts it, with regards to one of
the examples that Mattingly and Sterry cite as a potential inspiration to
Garamantian state formation, ‘the ability of Meroitic kings to directly
control subsistence production was likely to be limited to a relatively
small core region, while more diffuse forms of power could be exerted
over often more mobile populations’.107 Control of mobility and exchange
101
Fenwick 2013, 33; Forthcoming.
102
Compare, Retaillé 1993; Trousset 1984. This might also go some way towards solving the
puzzle posed by Sanmartí et al., Chapter 11, this volume, of the huge territorial extension of
what they label early North African states, and their resulting oddity within the Mediterranean
context.
103
Ptolemy, Geography 1.8.4, and Tacitus, Annals 4.23. 104 Mattingly 2003, 86.
105
Azevedo 1998; Diamond 1996; MacDonald and Camara 2012; Reyna 1990; Roberts 1987. The
argument for the inherently predatory nature of early states was first made by Tilly 1985; for
contemporary analyses that link warfare, crime and state formation, see for example, Charlton
and May 1989.
106
Compare Edwards 1998, 181.
107
Edwards, Chapter 9, this volume; Sterry and Mattingly, Chapter 8, this volume.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 685

systems remained thus key to state power, while territorial concerns were
secondary. Whether such political formations are best described as ‘states’
or not is a separate question, but we certainly need to take the label ‘state’ as
a starting rather than the end-point of analysis.

Conclusion

Although the ‘Garamantian state’ has by now become a convenient short-

hand for the complex polity that developed in Fazzan from the fifth
century BC onwards, we have to be wary of terms that assume both too
much – that is, in this case, an independent state administration, centralisa-
tion and territoriality – and too little, by potentially masking local and
regional originality. Whatever the Garamantian ‘kings’ of the Greco-
Roman writers were, and however they interacted with the Garamantian
‘tribes’ attested in the same sources, it is important not to stop at outside
descriptions, but to attempt to understand how Garamantian civilisation was
held together, whether this implied any kind of political centralisation, and
what this meant on the ground. Similarly, while Garamantian towns might
have exercised political authority in one form or another, and while labour
exploitation clearly was at the heart of Garamantian – as of all Saharan –
settlement, it is more difficult to grasp the nature of this authority. In order
to do this, we need to keep in mind the rarity of historically attested state
formations in the Sahara, the inherent difficulty of territorial control in
desert areas, and the longevity of other political traditions. Also, the
Saharan context provides ecological givens that, far from determining poli-
tical structure or even social organisation, nonetheless indicate the limits of
what could be done, even by the most ambitious of leaders. Assertions of vast
territorial control exercised by a Garamantian ‘kingdom’ hence have to be
taken with a pinch of salt, as this control was forcibly very different in kind
and degree from what our current assumptions and experiences might lead
us to believe. Similarly, questions of urban status and of socio-economic
inequality and labour exploitation should be analytically divorced from
assumptions about statehood, unless there is clear proof that they really
were related. Exploitation can take many forms, and central territorial state
administration is only one of them, and not necessarily historically the most
common – or, indeed, the most efficient. The problem with labelling the
Garamantian polity a ‘state’ is thus not so much that it might be wrong, but
rather that it stops us from asking certain types of questions, by creating
a false impression of recognition.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 686 Judith Scheele

References

Aspinion, R. 1954. Un llouh du Sous: règlement coutumier du Souq el Jemaa des

Ida ou Gnidif. Hespéris 41.3–4: 395–409.
Azevedo, M. 1998. Roots of Violence: A History of War in Chad. Amsterdam:
Horden and Breach.
Baier, S. and Lovejoy, P. 1975. The desert-side economy of the Central Sudan.
International Journal of African Historical Studies 8.4: 551–81.
Baier, S. and Lovejoy, P. 1977. The Tuareg of the Central Sudan: Gradations in
servility at the desert’s edge (Niger and Nigeria). In I. Kopytoff and S. Miers
(eds), Slavery in Africa: Historical and Anthropological Perspectives. Madison:
University of Wisconsin Press, 291–311.
Bayart, J.-F. 2000. Africa in the world: A history of extraversion. African Affairs 99:
217–67.
Bédoucha, G. 1987. L’eau, l’amie du puissant. Une communauté oasienne du sud
tunisien. Paris: Éditions des archives contemporaines.
Bonte, P. 1998. Esclaves ou cousins. Évolution du statut servile dans la société
mauritanienne. In B. Schlemmer (ed.), Terrains et engagements de Claude
Meillassoux. Paris: Karthala, 157–82.
Bonte, P. 2008. L’émirat de l’Adrar mauritanien: Harīm, compétition et protection
dans une société tribale saharienne. Paris: Karthala.
Brett, M. and Fentress, E. 1996. The Berbers. Oxford: Blackwell.
Brandily, M. 1988. Les inégalités dans la société du Tibesti. In C. Baroin (ed.), Gens
du roc et du sable. Les Toubous. Hommage à Charles et Marguerite Le Cœur.
Paris: Éditions du CNRS, 37–72.
Capot-Rey, R. 1953. Le Sahara français. Paris: Presses universitaires de
France.
Casajus, D. 1990. Islam et noblesse chez les Touaregs. L’Homme 115: 7–30.
Charlton, R. and May, R. 1989. Warlords and militarism in Chad. Review of African
Political Economy 16.45–46: 12–25.
Claudot-Hawad, H. 1990. Honneur et politique: Les choix stratégiques des
Touaregs pendant la colonisation française. Revue du monde musulman et de
la Méditerranée 57: 11–47.
Claudot-Hawad, H. 1998. La hiérarchie des savoirs et des pouvoirs dans la
société touarègue précoloniale et la recomposition des rôles socio-politiques
pendant la guerre anti-coloniale et après la défaite. Nomadic Peoples 2.1–2:
17–38.
Clauzel, J. 1960. L’exploitation des salines de Taoudeni. Algiers: Institut des
recherches sahariennes.
Clauzel, J. 1962. Les hiérarchies sociales en pays touareg. Travaux de l’institut de
recherches sahariennes 31: 120–75.
Cleaveland, T. 2002. Becoming Walata: A History of Saharan Social Formation and
Transformation. Portsmouth: Heinemann.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 687

Cooper, F. 1980. From Slaves to Squatters: Plantation Agriculture in Zanzibar and

Coastal Kenya. New Haven: Yale University Press.
Coquery-Vidrovitch, C. 1975. Research on an African mode of production.
Critique of Anthropology 2.4–5: 38–71.
Cordell, D. 2003. The myth of inevitability and invincibility: Resistance to slavers
and the slave trade in Central Africa, 1850–1910. In S.A. Diouf (ed.), Fighting the
Slave Trade. West African Strategies. Athens: Ohio University Press, 31–49.
Cowgill, G.L. 2004. Origins and development of urbanism: Archaeological
perspectives. Annual Review of Anthropology 33: 525–49.
Crone, P. 1986. The tribe and the state. In J.A. Hall (ed.), States in History. Oxford:
Blackwell, 48–77.
Despois, J.J. 1946. Mission scientifique au Fezzân (1944–1945). Tome 3: Géographie
humaine. Algiers: Institut de recherches sahariennes.
Diamond, S. 1996. Dahomey: The development of a proto-state. Dialectical
Anthropology 21.2: 121–216.
Djian, G. 1996. Le Tchad et sa conquête, 1900–1914 (reprint of 1915–1916 edition).
Paris: L’Harmattan.
Edwards, D. 1998. Meroe and the Sudanic kingdoms. Journal of African History
39.2: 175–93.
Edwards, D. 2004. The Nubian Past: An Archaeology of the Sudan. London:
Routledge.
Eldblom, L. 1968. Structure foncière, organisation et structure sociale. Une étude sur
la vie socio-économique dans les trois oasis libyennes de Ghat, Mourzouk et
particulièrement Ghadamès. Lund: Uniskol.
Ennaji, M. and Pascon, P. 1988. Le makhzen and the Sous al-aqsa: La
correspondance politique de la maison d’Illigh (1821–1894). Paris: Éditions
du CNRS.
Evans-Pritchard, E.E. 1945. The distribution of Sanusi lodges. Africa 15.4:
183–87.
Evans-Pritchard, E.E. 1949. The Sanusi of Cyrenaica. Oxford: Clarendon Press.
Fentress, E. 2006. Romanizing the Berbers. Past and Present 190: 3–34.
Fenwick, C. 2013. From Africa to Ifrîqiya: Settlement and society in early medieval
North Africa (650–800). Al-Masaq 25: 9–33.
Fenwick, C. Forthcoming. Archaeology, empire and the Arab conquest of North
Africa. Past and Present.
Gellner, E. 1969. Saints of the Atlas. London: Weidenfeld and Nicolson.
Geoffroy, A. 1887. Arabes pasteurs nomades de la tribu des Larbas. Les Ouvriers des
deux mondes 1.8: 409–64.
Godelier, M. 1980. L’État: Les processus de sa formation, la diversité de ses formes
et de ses bases. Revue internationale des sciences sociales 32.4: 657–71.
Gouletquer, P.L. and Kleinmann, D. 1975. Structure sociale et commerce du sel
dans l’économie touarègue. Revue de l’occident musulman et de la Méditerranée
22: 131–39.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 688 Judith Scheele

Grandguillaume, G. 1978. De la coutume à la loi: Droit de l’eau et statut des

communautés locales dans le Touat précolonial. Peuples méditerranéens 2:
119–33.
Grémont, C. 2010. Les Touaregs Iwellemmedan (1647–1896). Un ensemble politique
de la boucle du Niger. Paris: Karthala.
Grémont, C. 2012. Villages and crossroads: Changing territorialities among the
Tuareg of Northern Mali in the twentieth century. In J. McDougall and J. Scheele
(eds), Saharan Frontiers: Space and Mobility in Northwest Africa. Bloomington:
Indiana University Press, 131–45.
Gutelius, D. 2002. ‘The path is easy and the benefits large’: The Nasiriyya, social
networks and economic change in Morocco, 1640–1830. Journal of African History
43: 27–49.
Guyer, J. 1993. Wealth in people and self-realization in equatorial Africa. Man
(n. s.) 28.2: 243–65.
Hall, B. 2011. A History of Race in Muslim West Africa, 1600–1960. Cambridge:
Cambridge University Press.
Hammoudi, A. 1974. Segmentarité, stratification sociale, pouvoir politique et
sainteté. Hespéris – Tamuda 16: 147–80.
Hammoudi, A. 1980. Sainteté, pouvoir et société: Tamgrout aux XVIIe et XVIIIe
siècles. Annales Économies, Sociétés, Civilisations 35.3–4: 615–41.
Hopkins, J.F.P. and Levtzion, N. 1981. Corpus of Early Arabic Sources for West
African History. Cambridge: Cambridge University Press.
Horden, P. and Purcell, N. 2000. The Corrupting Sea. Oxford: Blackwell.
Johnson, A.W. and Earle, T. 2000. The Evolution of Human Societies From Foraging
Group to Agrarian State. Stanford: Stanford University Press.
Johnson, D. 1989. The structure of a legacy: Military slavery in northeast Africa.
Ethnohistory 36.1: 72–88.
Keenan, J. 1977. Power and wealth are cousins: Descent, class and marital
strategies among the Kel Ahaggar (Tuareg – Sahara). Part 1. Africa 47.3:
242–52.
Klein, M. 2005. The concept of honour and the persistence of servility in the
Western Sudan. Cahiers d’études africaines 45.179–80: 831–51.
Kopytoff, I. 1987. The internal African frontier: The making of African political
culture. In I. Kopytoff (ed.), The African Frontier: The Reproduction of Traditional
African Societies. Bloomington: Indiana University Press, 3–84.
Lethielleux, J. 1948. Le Fezzan, ses jardins, ses palmiers: notes d’ethnographie et
d’histoire. Tunis: Imprimerie Bascone et Muscat.
Liverani, M. 2000. The Libyan caravan road in Herodotus IV.181–185. Journal of
the Economic and Social History of the Orient 43.4: 496–520.
Lonsdale, J. 1981. States and social processes in Africa: A historiographical survey.
African Studies Review 24.2–3: 139–225.
Lovejoy, P. 1986. Salt of the Desert Sun: A History of Salt Production and Trade in
Central Sudan. Cambridge: Cambridge University Press.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 689

MacDonald, K. and Camara, S. 2012. Segou, slavery and Sifinso. In J. Cameron

Monroe and A. Ogundiram (eds), Power and Landscape in Atlantic West Africa.
Cambridge: Cambridge University Press, 169–90.
McDougall, E.A. 1986. The economies of Islam in the Southern Sahara: The rise of
the Kunta clan. Asian and African Studies 20: 45–60.
McDougall, E.A. 1998. A sense of self: The life of Fatma Barka. Canadian Journal of
African Studies 32.2: 285–315.
McIntosh, R.J. 2005. Ancient Middle Niger. Urbanism and Self-Organising
Landscape. Cambridge: Cambridge University Press.
McIntosh, S. 1999. Pathways to complexity: An African perspective. In S. McIntosh
(ed.), Beyond Chiefdoms: Pathways to Complexity in Africa. Cambridge:
Cambridge University Press, 1–30.
McIntosh, S. 2008. Reconceptualizing early Ghana. Canadian Journal of African
Studies 42.2–3: 347–73.
Masqueray, E. 1886. Formation des cités chez les populations sédentaires de l’Algérie:
Kabyles du Djurdjura, Chaouïa de l’Aourâs, Beni Mezâb. Paris: E. Leroux.
Mattingly, D. (ed.). 2003. The Archaeology of Fazzan. Volume 1. London: The
Society for Libyan Studies.
Mattingly, D. (ed.). 2013. The Archaeology of Fazzan. Volume 4. London: The
Society for Libyan Studies.
Mattingly, D. and MacDonald, K. 2013. Africa. In P. Clark (ed.), The Oxford
Handbook of Cities in World History. Oxford: Oxford University Press, 66–82.
Mattingly, D. and Sterry, M. 2013. The first towns in the central Sahara. Antiquity
87: 503–18.
Mattingly, D., Sterry, M. and Edwards, D.N. 2015. The origins and development of
Zuwîla, Libyan Sahara: An archaeological and historical overview of an ancient oasis
town and caravan centre. Azania: Archaeological Research in Africa 50.1: 27–75.
Mattingly, D., Sterry, M. and Thomas, D.C. 2013a. Jarma in its Saharan context: An
urban biography. In Mattingly 2013, 505–44.
Mattingly, D., Thomas, D.C., Sterry, M. and Preston, J. 2013b. Public and domestic
architecture of Garama. In Mattingly 2013, 287–98.
Meillassoux, C. 1986. Anthropologie de l’esclavage. Le ventre de fer et d’argent. Paris:
Presses universitaires de France.
Milliot, L. and Giacobetti, A. 1930. Recueil des délibérations des djema‘a du Mzab.
Revue des études islamiques 4: 171–230.
Montagne, R. 1930. Un magasin collectif de l’Anti-Atlas, l’agadir des Ikounka. Paris:
Larose.
Nehlil. 1916. L’azref des tribus et qsour berbères du Haut Guir. Archives Berbères
1.1: 77–89; 2: 88–103, 109–34.
O’Fahey, R. 1973. Slavery and the slave trade in Dâr Fûr. Journal of African History
14.1: 29–43.
Pascon, P. 1980. Le commerce de la maison d’Ilîgh d’après le register comptable de
Husayn b. Hachem. Annales Économies, Sociétés, Civilisations 35: 700–29.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 690 Judith Scheele

Pliez, O. 2006. Nomades d’hier, nomades d’aujourd’hui. Les migrants africains

réactivent-ils les territoires nomades au Sahara? Annales de Géographie 652:
688–707.
Powers, D. 2002. Law, Society and Culture in the Maghrib: 1300–1500. Cambridge:
Cambridge University Press.
Retaillé, D. 1993. Afrique: Le besoin de parler autrement qu’en surface. Espaces
Temps 51–2: 52–62.
Retaillé, D. 1998. L’espace nomade. Revue de géographie de Lyon 73.1: 71–81.
Retaillé, D. and O. Walther. 2013. L’actualité sahélo-saharienne au Mali: Une
invitation à penser l’espace mobile. Annales de Géographie 694: 595–618.
Reyna, S.P. 1990. Wars without End. The Political Economy of a Precolonial State.
Hanover: University Press of New England.
Roberts, R.L. 1987. Warriors, Merchants and Slaves: The State and the Economy in
the Middle Niger Valley, 1700–1914. Stanford: Stanford University Press.
Rossi, B. (ed.). 2009. Reconfiguring Slavery: West African Trajectories. Liverpool:
Liverpool University Press.
Rossi, B. 2009. Slavery and migration: social and physical mobility in Ader (Niger).
In Rossi 2009, 182–203.
Saad, E. 1983. Social History of Timbuktu: The Role of Muslim Scholars and
Notables 1400–1900. Cambridge: Cambridge University Press.
Scheele, J. 2010. Councils without customs, qadis without states: Property and
community in the Algerian Touat. Islamic Law and Society 17.3: 350–74.
Scheele, J. 2012. Smugglers and Saints of the Sahara: Regional Connectivity in the
Twentieth Century. Cambridge: Cambridge University Press.
Schmitz, J. 2009. Islamic patronage and republican emancipation: The slaves of the
Almaami in the Senegal River Valley. In Rossi 2009, 85–115.
Skalník, P. 1983. Questioning the concept of the state in indigenous Africa. Social
Dynamics 9: 11–28.
Stahl, A.B. 2014. Africa in the world: (Re)centering African history through
archaeology. Journal of Anthropological Research 70: 5–33.
Stewart, C. 1973. Islam and Social Order in Mauritania. Oxford: Clarendon
Press.
Taylor, P., Hoyler, M. and Verbruggen, R. 2010. External urban relational process:
Introducing central flow theory to central place theory. Urban Studies 47.13:
2803–18.
Testart, A. 1998. L’esclavage comme institution. L’Homme 38.145: 31–69.
Thiry, J. 1995. Le Sahara libyen dans l’Afrique du nord médiévale. Leuven: Peeters
Press.
Tilly, Charles. 1985. War making and state making as organized crime. In
P.B. Evans, D. Rueschemeyer and T. Skocpol (eds), Bringing the State Back In.
Cambridge: Cambridge University Press, 169–91.
Touati, H. 1993. La loi et l’écriture. Fiqh, ‘urf et société au Maghreb d’après les
Ajwiba d’Ibn Nâsir (m. 1085/1674). Annales islamologiques 27: 93–108.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

 18 Urbanisation, Inequality and Political Authority in the Sahara 691

Triaud, J.-L. 1995. La légende noire de la Sanûsiyya. Une confrérie musulmane

saharienne sous le regard français (1840–1930). Paris: MSH.
Trousset, P. 1984. L’idée de la frontière au Sahara et les données archéologiques. In
P.-R. Baduel (ed.), Enjeux sahariens. Paris: Éditions du CNRS, 47–78.
Urvoy, Y. 1934. Chroniques d’Agadez. Journal de la société des africanistes 4.2:
145–77.
Vikør, K. 1982. The desert-side salt trade of Kawar. African Economic History 11:
115–44.
Voguet, E. 2009. Islamisation de ‘l’intérieur du Maghreb’: Les fuqahâ’ et les
communautés rurales. Revue du monde musulman et de la Méditerranée 126:
141–52.
Wilson, A. 2012. Saharan trade in the Roman period: Short-, medium- and
long-distance trade networks. Azania: Archaeological Research in Africa 47.4:
409–49.

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press

https://doi.org/10.1017/9781108637978.019 Published online by Cambridge University Press
 9 Early States and Urban Forms in the Middle Nile
david n. edwards

Introduction

Studies of the political organisation of the Meroitic state (c.350 BC–AD

300) have often underlined the importance of an explicit recognition of its
Sudanic context, and place within longer-term traditions of Sudanic poli-
tical developments.1 Its study may also benefit from a more explicit
engagement with African political anthropology, along with a greater
concern with political practices, as more recently envisaged by Smith.2 In
turn, it may also be argued that many features of early Sudanic polities of
the Middle Nile (for example, Kerma) differ in some fundamental ways
from their northern neighbours of the Egyptian Lower Nile, or indeed the
Mediterranean. Meroe shares a number of distinctive characteristics of
more recent Sudanic kingdoms; some are emphasised in this chapter.
A number of challenges to rulership are readily identifiable in the Sahelian-
Sudanic landscapes of the Middle Nile, and many of these may be familiar
to those working in other contexts across the continent. It can be sug-
gested, for example, that the ability of the Meroitic kings to directly control
subsistence production was probably limited to a relatively small core
region, while more diffuse forms of power could be exerted beyond,
extending also to more mobile populations. Control of exchange systems
and certain forms of production may have remained more royally focused;
forms of ritual suzerainty may also have provided an important power
source (potentially with a long history). It should be emphasised that urban
settlements remained exceptional, as they have in the longer-term history
of the region. As such their presence invites close scrutiny when and where
it occurred. The extent to which urban forms were relevant to the

1 2
Edwards 1996; 1998. Smith 2011. 359

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 360 David N. Edwards

development of the Meroitic kingdom, or indeed other early states of the

Middle Nile also merits further exploration. Considering other possible
foci of early states in this region, the often fundamental differences from
the Egyptian state in the Lower Nile may be reiterated, notwithstanding
well-recognised cultural linkages. Looking in other directions, demon-
strable links with the Ethiopian highlands remain limited. Over the long
term, however, the distinctiveness of Ethiopian cultural trajectories must
also be recognised.
In recent years much new archaeological data has been acquired
relevant to state development in the Middle Nile, relating not just to
the Meroitic state but its Napatan predecessor, which briefly ruled Egypt
in the seventh century BC, as the Twenty-fifth Dynasty,3 and indeed the
Bronze Age Kerma kingdom, centred at old Kerma, of the third–second
millennia BC.4 Drawing on such material, this paper will revisit
some general issues concerning our conceptualisation of the Meroitic
kingdom, its sources of social and political power, and their material
manifestations. Some suggestions will also be offered on some of the
more noteworthy features of its settlement landscapes, and especially its
‘urban’ centres, which seem likely to be closely linked to the state
and disappeared with it in the third–fourth centuries AD. Rather than
considering these ‘settlement patterns’ as politically neutral elements it
will be suggested that they should be understood as deeply imbued with
political intent. How such sites were created and what purposes they may
have served will be considered; not least as centres of production, and in
turn central to the creation of a Meroitic imperial culture.5 Comparisons
can be made with examples from the Meroitic period of the features
of later kingdoms such as the Sultanate of Sinnar.6 There are also
many points of comparison in some of the more recent state-generated
landscapes of West African states,7 suggesting that there may be much
here for fruitful discussion of the features of states in and around the
Sahara.
In the Middle Nile Basin, as across most of the continent,8 the early
history of urbanism remains poorly understood. Settlement has remained
relatively dispersed, with a non-urban character into recent times,9
3
Pope 2014. 4 Bonnet and Valbelle 2014.
5
In view of the huge scale of the Meroitic state, its reach extending from the Gezira region south
of Khartoum to the Egyptian frontier in the north (>1,500 km of the Nile valley and its
hinterlands), its characterisation as ‘imperial’ is perhaps justified, not least in its likely
incorporation of many and varied populations.
6
Spaulding 1985. 7 For example, MacDonald and Camara 2012.
8
For example, Anderson and Rathbone 2000. 9 Lebon 1965.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 361

although with some possible early exceptions – potentially the earliest

urban centres in Sub-Saharan Africa. Historically, while few in number
and perhaps more enigmatic than we commonly allow, these are likely to
have had a disproportionate significance as foci of political and economic
activity. Archaeology has begun to explore a number of these, ranging from
Bronze Age Kerma to the Medieval metropolises of Dongola and Soba,
near the confluence of the Niles.10 Such work has revealed parts of
what may be quite extensive townscapes, and some of our most detailed
information for early urban sites anywhere on the continent.
However, if there do seem to have been some remarkably early urban
developments in the region, our understanding of their character remains
uncertain. As recent research in West Africa has suggested,11 urban devel-
opments may take many forms, although defining the ‘urban’ continues
to present many challenges, beyond the formal definition of any large
nucleated settlement as a ‘town’. In the Middle Nile, what we call urban
may well have been highly variable in both form and function(s). If we are
indeed finding early urban forms here, or in other parts of Sub-Saharan (or
Saharan) Africa, their particular interest would seem to lie in understand-
ing the roles they served. Many may have been centres of religious and
ritual power, albeit often closely linked to political authority. Others may
have developed as centres of trade and economic activity. In historical
times, as is also the case in West Africa, such trading towns may have been
deliberately set apart from political centres. Their histories may be very
sensitive to many external factors, not least those that may determine
shifting caravan routes. Those that were political foci were commonly
also centres of consumption and display, focused on royal palaces.
Contemporary accounts, such as Nachtigal’s of more recent royal centres
(El-Fasher in north Darfur), may be instructive in highlighting the
structuring principles underlying the creation of such urban centres.12
Unlike some other parts of Sub-Saharan Africa,13 fortified towns are
not commonly encountered in the Middle Nile. As is also more widely
encountered, towns and their inhabitants may have been, or have become
culturally distinct from, their surrounding countryside. They may have
been the homes of immigrants and traders, of elites speaking foreign
languages, but also places of moral danger of many kinds, not least in the
genesis of new kinds of urban populations, with new identities. In more
recent histories of the Sudanese Middle Nile, the ‘Sudani’ identity is
10
On Dongola – Jakobielski and Scholz 2001; Soba – Welsby 1998; Welsby and Daniels 1991.
11
McIntosh and McIntosh 1993. 12 Nachtigal 1971, 259 and following.
13
See for example Magnavita, Chapter 15, this volume.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 362 David N. Edwards

strongly associated with urban living. In such terms urban centres have the
potential to play formative roles in the creation of new identities, as well as
being places of hardship and vulnerability.14
Their histories were also dynamic. In the Middle Nile, few of the early
urban experiments established deep roots and most disappeared without
trace; few modern Sudanese towns can claim an urban ancestry of more
than a few centuries. Most originated in initiatives of the Turco-Egyptian
government in the nineteenth century and the zariba (fortified camps) of
traders and slavers,15 which were further developed by the Anglo-Egyptian
Condominium government of the twentieth century. The most prominent
contribution was of course the foundation and development of Khartoum
in the 1830s, which acquired a population of perhaps 30,000 by the mid-
nineteenth century. This population was formed around, and to
a considerable degree serviced the new government and a substantial
military garrison – sometimes in the order of 10,000 troops. With many
and varied contemporary accounts we may draw on, one persistent and
recurrent theme which we might discuss concerns the very real challenges
of densely populated urban living in tropical environments, with condi-
tions being particularly difficult in the rainy season. At Khartoum, disease
(typhus) outbreaks soon decimated the population in the 1860s; malaria
was also a major and recurrent problem. Interesting insights can be gained
in the early reports of the Wellcome Research Laboratories in Khartoum
(1908) which were greatly concerned with sanitation and the threat of
malaria, in attempts to ‘improve’ the Khartoum environment. While such
issues need not be pursued here, we should not underestimate the chal-
lenges of sustaining urban populations.16 With their many and varied
functions, we may consider that our urban centres were ‘consumers of
people’. This raises interesting questions about the composition of their
populations, which may return us to more fundamental questions con-
cerning the sources and exercise of state power in the Meroitic world.
Notwithstanding the absence of many forms of data that we would wish
to have, some of the more pertinent questions we might ask in the Middle
Nile, and indeed elsewhere, might include: ‘Why would one live in a town?’
or indeed ‘Would one choose to live in a town?’ Whatever the perceived
merits of urbanitas amongst some ancient populations in some parts of the
Mediterranean, these were not necessarily self-evident and universal

14
See Leonardi 2013, 156. 15 Lane and Johnson 2009.
16
We seem, however, to remain poorly equipped to establish even the most basic demographic
parameters for our research, building on early work in this field (for example, Fyfe and
McMaster 1981).

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 363

truths. It need not be something that should be aspired to or, as we have

been reminded in Mediterranean contexts, ‘an accolade to be awarded’.17
Within the broad geographical region we are dealing with there may have
been strong reasons for not living in towns, while many of those who did
live in towns might have been viewed as undesirable for varied reasons. In
the Meroitic Middle Nile, it may not overstate the case that urban spaces
were where the aspirations of an imperialistic state confronted the rather
different values, desires, and indeed the objections of its subjects.18

Urban Forms before Meroe

Kerma
While this chapter will focus mainly on the Meroitic period of the later first
millennium BC to the early first millennium AD, it may be useful to
contextualise this to some extent in relation to earlier manifestations of
possible urban forms in the Middle Nile. Much of this work has only been
recently published. The site of Kerma is of some interest as, over more than
a millennium, it developed as a major settlement agglomeration (Fig. 9.1),
apparently associated with what grew to be a massive necropolis. This was set
within a regional landscape of dispersed farmsteads and a rich environment
of braided river channels that annually flooded a wide alluvial plain. Over the
long term, such a site clearly has particular significance in potentially estab-
lishing a model for urban living in the Middle Nile. However, it has some
curious features. For example, it can be noted that it did not necessarily
originate as a specifically political centre. It would seem its centre, from an
early date, was a temple, manifested in its latest forms as a massive mudbrick
monument (Fig. 9.2).19
While it seems reasonably clear that by the mid-second millennium BC
forms of kingship with a strong military capability had developed at
Kerma, and indeed were to pose significant challenges to Pharaonic
Egypt, it is far less certain that Kerma had necessarily been the seat of
kings in earlier centuries. It was not obviously dominated by great palaces,
while the rare external historical sources hint at the existence of a number

17
Osborne 2005, 7.
18
For example, Scott 1998. Such issues are not unrelated to current debates about African
urbanisation and their perceived benefits, or otherwise (Potts 2012).
19
Its survival across 4,000 years, until today, also makes it one of the most remarkable ancient
monuments of Sub-Saharan Africa.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 364 David N. Edwards

Figure 9.1. Plan of Kerma ‘town’ c.2300–1450 BC (after Bonnet and Valbelle 2014).

of Kushite polities in earlier centuries.20 The manifestations of what we

perceive as royal power in the late phases of the Classic Kerma period
(c.1750–1450 BC), are also suggestive of quite novel forms of kingship

20
Edwards 2004, 78–79.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 365

Figure 9.2. Massive mudbrick temple, c.2000 BC – the ‘Defuffa’ – in the centre
of Kerma.

developing in that period, associated with what may have been a new warrior
elite.21 On present evidence, the settlement at Kerma is also perhaps most
remarkable for its uniqueness. If this was an early experiment in urban living,
it did not catch on. However, in relation to African traditions of kingship in
which ritual and the religious are often highly developed,22 it may not be
surprising to find such kingship – if such it was – developing at a religious
centre. Its focal role may in turn provide a basis for the longer-term devel-
opment of shared ‘Kerma’ cultural forms over large areas of what is now
northern Sudan. It is worth bearing in mind that researchers have yet to
move beyond potentially rather teleological approaches to the history of
Kerma, or indeed to attempt to disentangle Kerma the place, the culture and
potentially complex and dynamic political structures.

Temple-Towns of the New Kingdom Occupation

Following the conquest of Kerma c.1450 BC by the revived and expansive
New Kingdom Egyptian state, the old townsite seems to have been
destroyed. The locale seems however to have been maintained as a
significant centre through the construction of a number of temples and
associated buildings just to the north, on a site previously occupied by
a series of Kerma/Kushite monuments. Research has focused mainly on the
21
Hafsaas-Tsakos 2013. The most striking are perhaps the massive royal tumulus burials, which
incorporated large numbers of human sacrifices, as well as numerous later secondary burials
(O’Connor 2013).
22
For example, Mair 1977, Chapter 2.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 366 David N. Edwards

series of temples built by a succession of New Kingdom pharaohs. While

these temples were apparently enclosed, whether there was ever an
attempt to establish a more permanent Egyptian settlement at the site
remains less certain.23 Clearer manifestations of the Egyptian colonial
presence and their own distinctive urban forms (commonly described as
temple-towns) are found further north, beyond the Third Cataract. This
may well have been the southern limit of Egyptian vice-regal adminis-
tration during the New Kingdom occupation of Nubia (c.1450–1150 BC).
Notwithstanding their urban label, it may be useful to draw attention
to some of their specialised characteristics. In addition to housing
a number of Egyptian officials, there is a growing body of evidence that
they also had direct links to gold-mining enterprises in their vicinity.
As such, it seems likely that they also housed the workforce for these
enterprises, in regimented buildings, along with their workshops.24
In the case of Sesibi (Fig. 9.3), the domestic (as opposed to temple)
component extended to c.2 ha.25 Amara West,26 a new foundation
established several generations later is also suggestive, with an initial
formal layout including large magazines/storehouses, but which was then
reconstructed with new residential layouts. It seems likely that these
modifications reflect the agency of its inhabitants finding more appro-
priate ways of living in what may have been quite trying circumstances.
Already by that period, climatic deterioration was probably becoming
a significant factor, with increasingly mobile Saharan sands infiltrating
and ultimately overwhelming the settlement.
The small scale of these temple-towns suggests they had relatively
modest populations, but including both Egyptian and Nubian material
elements, suggestive of potentially complex colonial interactions and
entanglements. Their role as seats of colonial governance seem clear,
while also having more specific productive functions, notably in the higher
level management of Nubian gold extraction. What still remains uncertain
is how they sat in the wider landscape, and the extent to which they
interacted with or drew in surrounding populations. An enduring issue

23
The status of other Egyptian temples, for example at Kawa some 55 km to the south, and at the
sacred mountain of Jebel Barkal, also remains unclear. We see perhaps the appropriation of
elements of sacred landscape (Barkal) as well as the imposition of an alien constructed
landscape framed around Egyptian temples.
24
Comparisons may perhaps be drawn with dense low-status housing in the Amarna ‘workmen’s
village’ – Kemp 1987.
25
Such a size might be compared with a Roman auxiliary fort. The scale is suggestive of
populations in the 100s rather than 1,000s.
26
Spencer 2012, Spencer et al. 2012.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 367

Figure 9.3. Plan of colonial Egyptian ‘temple-town’ of Sesibi, combining substantial

temple complex and residential quarter.

of the New Kingdom colonial period in Nubia is the apparent decline of

overall population levels over several centuries of Egyptian rule.27 Since the
Egyptian presence was focused on the (increasingly) arid northern parts of
Nubia, this may not be unconnected with ongoing desiccation, as the rains
retreated southwards.28 The nature of the Egyptian colonial presence may
also be a factor that contributed to this rural depopulation. For example,
it seems likely that population dispersal rather than nucleation would
offer better opportunities in such increasingly challenging environmental
conditions. It may be noted that these northern regions only saw a significant
revival of rural settlement during the first millennium AD, which was made
possible by the introduction of the new irrigation technology of the saqia
waterwheel (which represented an advance in the capability of water-lifting
equipment over the balance well or Archimedes screw).

27
Säve-Söderbergh and Troy 1991, 10–13.
28
One result was a growing threat from increasingly mobile sand on the west bank of the Nile,
known to have posed a significant threat in some areas. It may also be noted that it remains far
from clear as to what extent date cultivation (an essential component of later Nubian farming
communities) was established in this area during this period.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 368 David N. Edwards

The Napatan Period

With the revival of a Kushite state in the early first millennium BC, new
political and religious centres emerged in the Twenty-fifth Dynasty and
Napatan period. Their urban components still remained poorly defined,
although in the period of Egyptian rule, the Kushite kings would have been
exposed to what may have been very different Egyptian forms of living. We
may differentiate between royal centres such as Sanam (a palace-town?)
and already ancient religious foci such as the Holy Mountain of Barkal,
across the river, where a series of royal palaces were added to a temple
complex. Recent work at Sanam has clarified the nature of its vast store-
rooms (the ‘Treasury’, 267 × 68 m) and associated structures in what is
presumed to have been the primary royal centre of the eighth–sixth
centuries BC.29 Extraordinary finds point to it serving as a major centre
for the accumulation of raw materials, local and exotically rare, consum-
ables such as imported wines, as well as the production of an array of
prestigious as well as more mundane artefacts. The massive scale of this
royal ‘Treasury’ cannot be over-emphasised. A religious dimension to this
evidence of material accumulation and manufacture has recently been
highlighted.30 Pope’s largely textual studies emphasise how Napatan
governance remained focused on a series of six (?) palaces, each perhaps
serving distinct territories.31 As he also notes, there appears to be an
absence of an obviously hierarchical articulation of major officials, which
may in turn be suggestive of the character of royal claims on these
territories. Ongoing excavations in and around the large settlement of
this period at Kawa may throw further light on such centres. Here, hitherto
unsuspected pyramid burials of an impressive size have been identified.32
Kawa itself seems to have been an extended unenclosed linear river-bank
site exploiting a hinterland that was watered by seasonally flooding braided
river channels.
On current evidence, the urban centres we can identify in the earlier first
millennium BC seem likely to have been based on a framework of palace
centres, at that time still focused on the Dongola Reach. While some
Napatan presence evidently extended much further south, it remains far
from certain as to whether this was framed around major settlements. No
Napatan towns seem to have been fortified, although a large fortress-like
structure at Qala Abu Ahmed has recently been explored c.110 km from
the Nile along the Wadi Howar, a major route running towards Darfur.33
29
Vincentelli 2007; 2011. 30 Pope 2014. 31
Pope 2014, 150. 32
Welsby 2014.
33
Fiedler and Jesse 2012; Jesse 2013.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 369

Finds there of Greek lekythoi (a narrow jug used for storing oil) and glass
vessels raise interesting new questions about the possible role of this site
in trade between the Nile and the Darfur region during this period.
North of the Dongola Reach, on the river route to Egypt, some more
ancient sites seem to have been reoccupied as way-stations. That in this
period the state was contending with substantial and mobile pastoral
populations in the hinterlands away from the Nile should be envisaged.
The presence of another fortified site (not dissimilar to that at Qala Abu
Ahmed) at Fura Wells, a major water point in the central Baiyuda Desert,
could perhaps be evidence for an early (Napatan) state presence reaching
out into this hinterland. Control of permanent water supplies of course
remains one key strategy for imposing state control over more mobile
potential subjects.

The Meroitic Kingdom

The later centuries of the first millennium BC saw a major restructuring of

the Kushite state, traditionally associated with a shift in royal cemeteries
southwards to Meroe. This initiated a distinct Meroitic kingdom in the
Middle Nile which persisted to the third–fourth centuries AD (Fig. 9.4).
Notwithstanding longstanding debates about the significance of this
perceived southwards shift, several features distinguish this period that
bring into focus something of the structure of what became a massive
kingdom (perhaps empire?) whose reach extended from the southern
margins of Egypt to somewhere south of the confluence of the Niles.
Within this huge region, territorial cores emerged and what are generally
conceived of as Meroitic urban centres seem to have played a crucial role.
These centres were focused on a relatively small region of the east Nile
bank with a hinterland extending to the south-east, towards an area of
greater rainfall and the grazing lands it supported. Bounded by the seasonal
Atbara river on its north-east side, this may be envisaged as a core territory
of perhaps 10,000–15,000 km2, marked by the presence of a number of
mudbrick and stone-built urban centres as well as other monuments. This
may be contrasted with territories further north (the core territory of the
Napatan state in the early part of the millennium) in which permanent
settlements were confined to the river. The large hinterland of what is now
the Baiyuda desert seems to have lacked a state presence beyond a handful
of stations along the main routes traversing it and at water points. The
almost continuous spreads of burial cairns along the west bank of the Nile

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 370 David N. Edwards

Figure 9.4. General map of central Meroitic territories.

would seem to relate to predominantly pastoral populations grounded in

the Baiyuda, but whose relationship to the Meroitic state would seem to
have been very different from those with populations of the Western
Butana.
We have little hard evidence concerning the core Meroitic region prior
to the fourth–third centuries BC. In the first millennium BC this region
seems likely to have received significantly greater and more reliable rainfall
than in more recent centuries. Under such circumstances the non-riverine
areas will have had great potential both for rain-fed cultivation and collect-
ing, as well as grazing. Notwithstanding the considerable potential for
riverside farming, it would perhaps be premature to assume that this was

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 371

necessarily the more preferable or productive form of exploitation. The

risks inherent in the unpredictable Nile floods were not insignificant, while
the relatively late development of domesticated tropical crops in the region
may reflect deeper-seated pastoral preferences. However we might expect
that populations choose to exploit both riverine areas as well as the interior.
We can be reasonably sure that prior to the Meroitic period, such exploita-
tion was not based around large nucleated settlements. Archaeological
traces of earlier settlement in the region that was to become the Meroitic
heartlands remain, to date, elusive, notwithstanding some recent detailed
survey work close to Meroe.34
In this context, as in many other parts of the continent where we may
also confidently assume relatively low population densities combined with
relatively abundant productive land, it seems likely that the scarcity of
people and a need to establish control of people, was of crucial importance
to Meroitic state building. This was one of several ways in which state
formation in the Middle Nile followed a quite different course from the
Egyptian Lower Nile.35 Here we might also emphasise the lack of environ-
mental constraints which may circumscribe populations, a crucial and
fundamental difference from both the Egyptian Lower Nile, and Saharan
oases. As in other savannah regions ‘wealth and power in men rather than
acres’ was crucial, while ‘those who exercised authority were people-
owners rather than landowners’.36
Such imperatives may in turn be related both to Meroitic perceptions of
identity as well as throwing light on key aspects of rulership. Such percep-
tions may be traced back through external historical references to the
‘Noba’ (Nubians), though this term has traditionally been interpreted as
an ethnic designation. However, recent work suggests that some rather
different understandings of the ‘Noba’ are possible, which relate to funda-
mental principles of the Meroitic state. Briefly stated, the ‘Nob’ appear in
Meroitic texts as defeated and subjugated peoples; they may be represented
as bound prisoners (one, identified as ‘a king of the Noba’), as well as in lists
of slaughtered enemies.37 However, rather than as a specific ethnic identi-
fier, the term in fact seems to serve as a generic and pejorative identifier,
identifying slaves. If we may seek a hypothetical etymology for Meroitic
‘Nob’ (slaves/Nubians) this may lie in a term meaning something like ‘men
of the land’, suggesting a derogatory term for agriculturalists, who were in
turn potential slaves. In this context, the significance of such an

34 35 36 37
Wolf 2015. Edwards 1996, 14–15. Hopkins 1973, 26. Rilly 2008, 216–17.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 372 David N. Edwards

understanding is that the term is an effective reflection of practical power

relations, rather than ethnic designations.38
If this was a reality of the Meroitic period, this sits well within the longer-
term history of the region, where just such designations are a familiar
presence within the traditions of later kingdoms of the Middle Nile, where
the idioms of slavery and enslavement have remained central to the
structuring of social and political relations.39 The ‘Nuba’ of the medieval
Arab texts, and indeed more recent usages maintain almost exactly these
connotations of the Meroitic usage. The work of Spaulding and Kapteijns
in particular,40 has argued for the persistence of social structures ‘com-
posed almost exclusively of a landowning elite, called in later times the
hukkam (those who rule)’, and subject commoners. Communities of sub-
jects might be obliged to ‘buy peace from the government with precious
goods, to seize neighbours as captives at tax times, or to endure govern-
ment reprisals in the form of slave-collecting raids’.41 When discussing
notions of a ‘Sudanic serfdom’ we are reminded of the narrowness of the
line dividing uncooperative subjects from slavery; the threat of enslave-
ment remained the essential inducement to ‘free’ subjects to continue to
serve their masters. Enslavement was the fate visited on whoever failed in
their obligations to those who ruled; the terminology of slaves is in fact the
natural idiom in which power relations would be expressed. I suggest that
such a perspective can help us understand the relationship of the Meroitic
state to all its subjects. Where this may be most obvious is in dealing with
its more peripheral subjects and the familiar pattern of extracting resources
from the peripheries. Such a relationship may also be relevant in its heart-
lands, where a direct state presence was established and constructed
around a framework of urban centres.

Meroitic Urbanism?
Meroe
The origins of Meroe itself still remain obscure, although the location may
have been occupied early in the first millennium BC, if not before.42 It
should be noted that there is no textual reference to a major centre at this
site from the period of the Twenty-fifth Dynasty, though a Napatan
38
Early in the first millennium BCE we find donations of ‘landworkers’ to temple establishments
recorded in (Egyptian language) royal texts – Pope 2014, 120.
39
Kopytoff and Miers 1977. 40 Kapteijns and Spaulding 2005; Spaulding and Kapteijns 2002.
41
Spaulding 2006; Spaulding and Kapteijns 2002, 46–47.
42
See Pope 2014 for a recent overview.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 373

presence in the region is evident in the following centuries to the

mid-first millennium BC.43 Jeremy Pope has characterised this as
‘a gradual annexation of the Meroe region’.44 The process by which
this region was integrated into the larger state may also be seen as
fundamentally transformative. The urban centres would seem likely
to have been fundamental to such transformations, creating new state-
generated landscapes.45
At Meroe itself, the beginning of the Meroitic period (c.350 BC) does
seem to have witnessed major building episodes. Perhaps the most promi-
nent of these was the construction of a massive rectilinear enclosure, whose
heavy masonry walls surrounded an area of c.8 ha (Fig. 9.5). Its construc-
tion seems to mark a new development in defining a nucleated core to the
settlement. Traditionally referred to as the ‘Royal City’, it certainly
included several palatial buildings, and notwithstanding the caution
that must be exercised in assuming the existence of a single capital,
Meroe would seem marked out as a centre with some unique qualities.
Its construction may have taken place at about the same time as that of
a major new Amun temple against its east side, which was extended over
the next few centuries.46 A series of smaller shrines were later built to the
east of its entrance, facing on to a processional way, oriented like most
temples on the winter solstice, replacing earlier domestic structures in
that area.
The more general character of Meroe the town remains unclear. To the
north-east of the enclosure was a large settlement mound covering c.10 ha,
with a second large mound (3–4 ha) to the south. Outlying features include
a temple complex (the ‘Isis temple’) to the north, as well as an area of
pottery kilns and a large temple complex with a hafir (known to earlier
generations as the ‘Sun Temple’) to the south-east of the town. Only small
areas of the northern settlement have been excavated, revealing deep
stratified deposits, up to 10 m thick; as noted above, these potentially
date back to the early first millennium BC. In the absence of widespread
open-area excavations, we as yet have very little idea about the character of
such areas as only small parts of buildings have been exposed. It still
remains to be demonstrated that the mounds represent the remains of
nucleated domestic settlement that we would conventionally expect for an
urban area. Parts of the north mound were certainly used for other
industrial activities, as workshops were found there for smelting and

43
Pope 2014, 27–31; Wolf 2015. 44 Pope 2014, 33.
45
These were ‘political landscapes’, in the sense explored by Smith 2003. 46
Török 1997.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 374 David N. Edwards

Figure 9.5. Meroe ‘royal city’ and environs.

forging iron. Cotton textile manufacture (evidenced in abundant spindle

whorls) is also likely to have been important. Cloth is likely to have been
a very significant manufacture, both as a status marker and perhaps
a medium of exchange,47 as in later periods.48
47
Edwards 2004, 170.
48
One of the identities of elites in later periods may be as the ‘clothes-wearing people’ – Spaulding
1985, 78–83.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 375

Such productive activities also required considerable resource inputs.

This must be considered as part of wider questions concerning how
urban living/activities relate to a much wider hinterland and interlinking
supply networks. Urban-rural linkages and interdependence are perhaps
central to any model we can suggest for Meroe and other Meroitic urban
centres. Meroitic ironworking is likely to be of particular importance
here, requiring a range of key raw materials (for example, ores and
charcoal), in turn requiring significant labour inputs. This iron industry
has attracted considerable attention since the first excavations at
Meroe,49 and is now the subject of renewed interest.50 The scale of
production appears to have been considerable,51 demanding significant
inputs of various resources, with the supply of wood/charcoal in parti-
cular probably generating significant environmental impacts,52 although
recent work suggests that the volume of production may have been more
modest than once supposed.53
Many uncertainties still surround the nature of the urban centre at
Meroe. Was it primarily a religious centre, or a palace-city inhabited by
royal households and their attendants? To what extent did it attract
a broader population base with its own independent dynamics? The size
of its population remains a matter of speculation, although it is worth
bearing in mind that the total area of the site is rather less than that of
the two modern villages (a few hundred households) which adjoin the
archaeological site.

Other Meroitic Urban Centres

Meroe was clearly not an isolated phenomenon. Several similar large sites,
often combining the presence of palatial structures with temples, are
known to lie both to the south and north (Fig. 9.6).54 Some of these are
again of considerable size, covering c.10–15 ha, including both official,
religious and domestic quarters. Most are known mainly from finds of
architectural elements and none have as yet been investigated in any detail.
However, survey and test-excavation at a number of these have in recent
years begun to provide better definition of their character.
Wad ben Naqa is the southernmost yet identified, c.80 km upriver from
Meroe. The main palace, measuring 61 m square, would seem to have been
constructed by Queen Amanishakheto, currently dated to the late first
49
For example, Garstang et al. 1911, 55; Shinnie 1985; Shinnie and Kense 1982.
50
Humphris 2014; Rehren 2001. 51 Haaland 1985; Tylecote 1982. 52 Haaland 1985.
53
Humphris 2014, 128. 54 Baud 2011, 211–24; Lenoble 2008.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 376 David N. Edwards

Figure 9.6. Examples of regularly designed ‘palace’ structures, with examples from
Meroe, Muweis and Wad ben Naqa in the Meroitic riverine heartlands, as well as from
the religious centre of Jebel Barkal (after Maillot 2014).

century BC, or early first century AD. Here, as in other examples, the
surviving ground floor provided both storerooms and a casemate founda-
tion for a raised upper storey, now lost apart from architectural fragments.
How long this survived remains unclear but elements seem to have been
reused in the construction of a temple and adjoining storerooms in
the second–third centuries.55 The purpose of a large and, to-date, unique

55
Onderka and Vrtal 2014, 147.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 377

circular brick building remains unknown. Early speculations suggested it

might be a granary, but a religious function has also been suggested.56
Indirect evidence for the role of this, and other urban sites, in the
accumulation, storage and redistribution of goods is found in a range
of mud-sealings. Comparisons may be drawn with other archaic adminis-
trative systems, operating without more complex written accounting
systems,57 albeit coexisting with some use of written accounts (preserved
in ostraca). The context and character of such sealings may also support
a case that it was palaces, rather than temples which played central roles in
economic administration.58
Muweis, c.30 km from Wad ben Naqa had at its core another palatial
building of very similar dimensions and plan to that at Wad ben Naqa,
a temple complex, and two associated settlement areas, at least one with
industrial areas.59 While early results suggest a history spanning several
centuries, major building episodes are identifiable at the start of the first
century AD and abandonment of the palatial building by perhaps the
mid-third century, soon followed by robbing for reuse of building
materials.60 Finds of mud-sealings again indicate some of the adminis-
trative features of the palace.61 Another temple and palace complex at El
Hassa-Damboya displays a similar combination of a substantial temple,
with one or more palatial structures set within a substantially larger
settlement.62 Preliminary investigations here beyond the temple have
raised the possibility of a rather larger settlement of more ephemeral
post-built structures around the more formal brick buildings. This is, as
yet, a still poorly documented aspect of Meroitic settlements.
While there are few smaller-scale settlements yet known, there are some
sites with both temples and substantial brick-built architecture (although
less impressive than palatial buildings described above). One such complex
is located within 10 km of Meroe near the mouth of the Wadi Hawad
drainage at Abu Erteila-Awlib. Initial excavations suggest the presence of
a temple, ‘official’ structures and a hafir water reservoir.63 This is very
reminiscent of the arrangement of the so-called ‘Sun Temple’, ‘Priest
House’ and hafir just south-east of Meroe. The locations of a number of
other centres may be suggested, potentially as far north as Dangeil, the site
of a large royal temple, although, as yet with no associated palace or urban
centre.

56
Onderka and Vrtal 2014, 153–55. 57 Vincentelli 1993, 41. 58 Edwards 1996, 26.
59
Baud 2008; 2014; Maillot 2014; 2015; 2016. 60 Baud 2014, 771. 61 Baud 2011, 244–45.
62
Lenoble and Rondot 2003. 63 Borkowski and Paner 2005.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 378 David N. Edwards

Figure 9.7. Map of planned enclosed settlement and later ‘suburbs’ to

the south, with pottery kilns and iron-working slag heaps (after Wolf
et al. 2014, Fig.2).

Hamadab – A Planned Settlement

Another rather different kind of settlement is found closer to Meroe at
Hamadab located only c.3 km south of the ‘royal city’ (Fig. 9.7). Some unusual
features provide further suggestive data relating to its foundation. With

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 379

a number of distinctive features, it is our most convincing example of an urban

space. It was clearly planned and was enclosed (I will avoid the term fortified),
c.110 m square, with a small temple on a central axis.64 The provision of a small
(20 m square) tower-house in the south-east corner of the site is also an
otherwise unique feature, similar in construction to, but much less substantial
than, the large palaces encountered elsewhere. The enclosed and planned
settlement appears to date to the first century BC. In the early centuries AD
this planned structure was substantially reshaped, however, with the abandon-
ment (and robbing) of the enclosure wall and outward spread into substantial
‘suburbs’. As such it may be differentiated on a number of counts from other
known urban centres of the Meroitic heartlands, lacking both a large palace
core, major temples, but also provided with a substantial, but soon superseded
enclosure wall. Like other urban centres the site was engaged in ceramic and
textile production, with direct archaeobotanical evidence for cotton as well as
finds of spindle whorls and loomweights.65 Long recognised as the site of
major iron-working, recent work suggests that this may largely be a feature of
the later, or indeed post-Meroitic period, raising a number of interesting issues
concerning its relationship with Meroe and its iron-workers.66 Its specific
origins are perhaps suggested in two Meroitic stele of Queen Amanirenas
and Akinidad associated with the central temple. Rilly’s restudy of these notes
explicit reference to ‘Tameya prisoners’ and may well relate to military cam-
paigns, and indeed the acquisition of ‘non-African’ captives. While we must
await a full presentation of these texts, many features of the site seem suggestive
of an enclosed settlement (walls to keep people in as much as out?) including
a range of royal craft-workers (see further discussion below).

The Western Butana

Unlike most of the centres spread along the Nile, two other major centres
are known within the hinterland of Western Butana, an area that received
considerable seasonal rainfall and that had good agricultural and grazing
potential along major drainages. It is likely, as is the case in more recent
periods, that these drainages were actively managed with forms of water-
harvesting techniques. However, most Meroitic settlement within this
region seemingly relied on more ephemeral and mobile architectural
forms, appropriate to seasonal cultivation and grazing. Permanent centres
with mudbrick and stone architecture are much rarer. The two main
centres, Mussawarat and Naqa, take very different forms.

64 65 66
Wolf et al. 2014a; 2014b. Fuller 2014, 173. Humphris 2014.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 380 David N. Edwards

Figure 9.8. Musawwarat es Sofra – royal pilgrimage centre – a periodic royal centre?

Mussawarat, known to the Meroites as Aborepe, seems primarily to

have served as a religious pilgrimage centre, notwithstanding a number
of other, more or less imaginative, interpretations offered in the past
(Fig. 9.8).67 Within the ‘Great Enclosure’, there was a complex of
interconnected temples, set in courtyards spread over 5.5 ha. At least
some of the courtyards were irrigated gardens, laid out with trees,
brought-in from the Nile; the pots of Nile silt in which they were
planted were readily identifiable during excavation.68 Indications of
larger public spaces within the complex, the nature of its monuments,

67
These include that the site might have been used for training elephants (Shinnie 1967) and that
the complex may have been some form of ‘animal garden’ or hunting palace (Török 1997).
68
Wolf 1999.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 381

the many thousands of graffiti69 and the large natural arena which
surrounds the site would be consistent with its use as a periodic
(pilgrimage?) centre. While the Meroitic monarchs were undoubtedly
closely linked to the site, and may well have played a central role in
whatever festivals or other events were enacted there, its very special
status seems clear. A group of ancillary buildings, the ‘Small
Enclosure’, had a more domestic character, with cooking and store-
rooms, probably relating to the priests and other staff servicing the
main complex.70 A small pottery workshop was also located beside the
north wall of the enclosure.71
By contrast, a quite substantial settlement grew up around a series of
small temples located on a major wadi c.35 km from the Nile at Naqa,
ancient Tolkte (Fig. 9.9). Occupying a very different savannah landscape
than the riverine settlements, this site was located at the confluence of
two large wadis. Best known for a series of temples, the larger settlement
complex would seem to include several palatial structures and ancillary
buildings. Naqa may also perhaps be the focus of a larger complex of
smaller settlements, marked by a series of small temples; one such site
lies c.5 km away on the west side of the wadi, at Nasb es Sami. It is set
within large areas suitable for rain-fed agriculture, with managed graz-
ing on its margins. This will in turn have required careful management
of arable and pastoral exploitation, and of perennial water supplies, the
latter likely to have been important in managing (and controlling) more
mobile and pastoral parts of the population. Their reorientation towards
arable production seems probable. The many wells and the hafirs (water
reservoirs) in the region, some marked with Meroitic monuments will
have formed part of this wider system of population management.
Overall, within this region we might identify the larger productive
hinterland which fed in to a series of regional urban centres on the
Nile, including Meroe itself. Naqa appears unique as the largest settle-
ment of the interior with major investment in brick and stone
architecture.
The particular importance of the Western Butana hinterland would
seem to lie in its role as a core productive territory, capable of
supporting both arable and pastoral farming, on a larger scale, this
may well be the only such area so directly controlled, and controllable,
by the Meroitic state.72 Only here is a state presence clearly manifested
in the form of temples, often associated with hafirs and (a small

69 70 71 72
MGA 2015. Fitzenreiter et al. 1999. Edwards 1999. Edwards 1996, 25–26.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 382 David N. Edwards

Figure 9.9. Hinterland settlement of Naqa – palace complexes and temples.

number) of more permanent settlements.73 The investment in water

supplies in turn seems likely to have also provided further opportu-
nities for state controls, not least in the extraction of taxes/tribute.74
The southern margins of this area may well be marked by the sugges-
tive iconography of an otherwise unique Meroitic stele at Jebel Qeili,
a point of transition beyond which lay the open plains of the Butana
proper, which were perhaps beyond any direct state control (Fig. 9.10).
Notwithstanding long-standing suggestions of an earlier Napatan

73 74
Näser 2011. Hinkel 1991.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 383

Figure 9.10. Jebel Qeili inscription.

exploitation of the Western Butana, this state presence manifested

in monumental structures (and perhaps a religious colonisation) may
well be a phenomenon of the Meroitic period of the later first
millennium BC.75 That this reflects significant and fundamental
changes in state organisation also seems likely. As in other contexts,
the productions of territory are likely to have had profound practical
implications for subject populations.76

The Roles of Meroitic Towns

While it has become usual to represent these temple-palace centres as
towns it is already possible to recognise considerable variety in their

75 76
Näser 2011. Smith 2011, 423–24.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 384 David N. Edwards

Figure 9.11. Examples of throne dais, emphasising subjugation – mainly fragmentary

finds from Meroitic palace centres (based on Baud 2014).

forms and scale. The royal presence at many of these centres is emphasised
in temple foundations of royal cults, as well as the larger palaces, although
they are mostly associated with only a small number of particularly active
monarchs in the later first century BC–first century AD. The royal pre-
sence is also very evident in other associated material such as throne daises
(Fig. 9.11), which have been found at several of them. Representations of
the subjugation of bound captives were a common feature of these royal
thrones.77 Such major riverine centres may be part of a wider network of
nodal points of different kinds, including key watering points (wells, hafirs)
and religious centres.
In general terms, what we currently characterise as urban centres seem
likely to have performed crucial roles in the exercise of royal control of the
heartlands of Meroe, forming its core productive territories. The nature of
these centres, with their strong royal associations, emphasised in such
displays of rulership, suggest they may be perceived very much as key
nodes in a state-generated landscape, not dissimilar to examples known
from elsewhere in Sub-Saharan Africa in more recent centuries.78 Such
77 78
Baud 2014, 775–78, figure 8. For example, MacDonald and Camara 2012, 187.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 385

comparisons might also suggest that we should consider the extent to

which redeployments of population may have been required, and that
the creation of these urban spaces may have involved coercion, to varying
degrees. As we have seen, representations of the coercive reach of royal
power are much in evidence and we should not discount that these may be
expressing fundamental aspects of royal rule over local audiences (and not
just external threats). While the Jebel Qeili stele may project Meroitic royal
power outwards into the open grazing lands of the Butana, the royal
thrones at each of the riverine urban centres provide more localised
reminders of the obligations of subjects, and how they were originally
imposed. The lavish decoration and scale of the palaces which lay at the
heart of many of these centres also contributed to their role in ‘aesthetics of
awe’; an aspect of sovereignty encountered in other African contexts, and
beyond.79
As has been suggested, these centres show strong associations with
pottery and textile manufacture and ironworking and other related spe-
cialist skills (such as faience manufacture). Hamadab in its original
enclosed form, dominated by a tower-house (at least as first conceived)
and with its possible associations with captives, is particularly marked out.
The ceramic workshops widely encountered at urban centres may also be
linked to the requirements of managing and processing grain crops, not
least in the form of grain beers, a likely dietary staple. One specific feature
of the pottery workshops of the Meroitic heartlands is their production of
distinctive and apparently quite standardised beer-jar forms. To date, their
distribution also suggests their use/circulation within a quite circum-
scribed zone. Abundant on settlement sites these also became increasingly
prominent in burial contexts through the early centuries AD. Interestingly,
by the later Meroitic centuries large quantities of grain-beer were deployed
in funerary contexts,80 at least within the Meroitic ‘core’, marking a very
major shift in practice, one which survives beyond the Meroitic period.81 If
related to status displays, this could in turn be linked to the control of
agricultural resources.82 Such a shift may well have been one outcome of
the increased state-level management of the Western Butana. In such ways
we are reminded that urban and rural histories cannot be studied
independently.
The social significance of textiles and their production has already been
suggested. Cotton growing and processing, spinning and weaving may be

79 80 81
Monroe 2010; Smith 2011, 424–25. Lenoble 1994. Edwards 2011.
82
See also Arthur 2003, 523.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 386 David N. Edwards

linked to sites such as Hamadab, and other urban centres. Whether or

not cotton growing developed as a specialised crop remains to be seen,
but this is one possibility we might consider, brought on by societal
changes wrought by the Meroitic state, of the kind envisaged by Brite
and Marston.83 In later periods, cotton-growers continued to have specific
obligations to their lords.84 The practical demands of cotton-growing
also suggest some significant level of management was required, perhaps
distinct from generalised grain-based agriculture practised along the wadis
of the interior, or following the Nile flood.85
On current evidence a regional system of royal palace towns may have
been sustained for at least a few centuries, but it may also have changed
significantly within the Meroitic period. We may suppose that once these
palaces were being quarried for building materials, the original system of
control which had been constructed around these urban centres was
already unravelling. If, as is suggested these palaces and their towns played
central roles in economic administration, the decay and abandonment of
the palaces may be very tangible demonstrations of both economic and
political restructurings underway by the third century AD, if not before. It
is possible that these urban sites may have persisted as nodal points in local
economies, although control may already have been passing out of the
hands of the central monarchy in the face of political fragmentation,
presumed to be a central feature of the ‘end of Meroe’. While our evidence
remains slight, it seems likely that the changing fate of these urban centres
had significant wider impacts.
Notwithstanding the conventional identification of Meroitic towns in
more areas of Middle and Lower Nubia, on the route to Egypt, urban
terminology may perhaps be misleading, at a number of levels. In terms of
scale, few if any could match the substantial size of contemporary Egyptian
villages.86 In terms of function, most known settlements have features that
mark them off from what we might expect of rural (agricultural) settle-
ments, perhaps, suggesting more urban roles. However, when examined
together at a landscape scale, it is apparent that if they were towns then they
lacked any rural hinterlands.87 That they were Meroitic ‘colonies’ may not
be an inappropriate characterisation; their unusual features reflecting their
origins and purpose. Where we can date Meroitic activity in the region,
patterns emerge which are suggestive of phases of site foundation, as well as
83
Brite and Marston 2013, 51; cf Griffith and Crowfoot 1934. 84 Spaulding 1985, 81.
85
Fuller 2014. 86 See Boozer, Chapter 4, this volume.
87
The fullest study of the known extent of Meroitic settlement in the north remains Edwards
1996.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 387

abandonment and reoccupation. Wider cultural patterns are suggestive of

the gradual emergence of a distinct regional culture within the Meroitic
north, increasingly distinct from its Sudanic origins, also demonstrating
developing cultural links with Roman Egypt.
The proximity and accessibility of Roman Egypt is manifest in many
features of northern Meroitic culture, especially between the Second
and First Cataracts. Notwithstanding the presence of long-established
religious/pilgrimage centres such as Qasr Ibrim, it remains difficult to
identify Meroitic urban centres that are distinct from the series of small
nucleated settlements that traverse Middle and Lower Nubia. By the fourth
century AD, with the gradual appearance of the saqia waterwheel, we
may however begin to see the transformation of this northern form of
settlement built around a new type of agricultural system, which was to
later sustain the early medieval kingdom of Nobadia. In the same period,
the accessibility of Egypt was similarly evidenced in many aspects of the
political and religious transformation of northern Nubia.88
While we know a great deal about well-preserved Meroitic archaeology
in the north, the Nubian ‘corridor’ across the Sahara is less studied. There is
good reason to believe that Kushite settlements in the north through the
first millennium BC and beyond were deliberately established to manage
what were in effect Trans-Saharan communications.

Urban Centres and New Identities?

A final issue to explore concerns the wider role that Meroitic urban centres
may have played in the creation of both the material world of the Meroitic
state and new forms of social identities. That many of the most distinctive
forms of Meroitic material culture were created in these urban centres
seems increasingly likely. It is also probable that access to most forms
of imported materials was controlled by such urban centres (with the
exception of the northern frontier zone in Lower Nubia).
As new forms of living in these landscapes, these centres may also have
had wider social impacts.89 In addition to craftworkers, other elements of
extended royal retinues and households may also have settled in new
palace-centred settlements. The fundamental importance of such royal

88
Edwards 2014b.
89
As elsewhere in the continent urban centres may also have played other important roles, not
least in language development (Coquery-Vidrovitch 2005, xxix); the development of written
standardised Meroitic, and indeed in medieval ‘Old Nubian’ may perhaps be linked particularly
with their urban centres.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 388 David N. Edwards

dependents as a primary source of power should be reiterated; such ‘wealth

in people’ still being central to more recent constructions of power in
the region.90 Whether or not the Meroitic royal households possessed
a significant military component remains unknown, but this also seems
likely. Bearing in mind the very long traditions of slave troops in the
region,91 it is possible that this was already an institutionalised practice
in the Meroitic period, although this may again represent one extreme in
a continuum of dependency amongst royal retinues. In later periods, slave
troops were settled in garrison communities and engaged in cultivation to
support themselves; within a more general pattern of settlement wherein
the proportion of slaves/captives increased as a factor of proximity to the
capital.92 While ‘slaughtering the men [and] enslaving the women’ was
a recurrent feature of Kushite royal activities,93 enslavement was likely
an important component of a range of strategies deployed to extend
retinues.94 It may also be suggested that these strategies provided the
population base for the urban centres of the Meroitic core. Similar techni-
ques of spatial production may be widely encountered, as for example
suggested in Parker’s study of Assyrian imperial construction.95 This is
perhaps a hypothesis that could be further explored in investigations of the
socio-political character of the Meroitic urban settlements currently being
excavated. It is a reasonable premise that these settlements did not arise
organically as voluntary aggregations.
The afterlife and legacies of such urban communities also invite further
investigation. Since not all these urban centres were necessarily the same
(Hamadab, for example, being rather different on a number of counts),
varied later histories should perhaps be expected in the aftermath of
political changes. While some sites may well have been abandoned, others
may have been more resilient, finding new purposes. One interesting
outcome of recent work at Hamadab has been to demonstrate the persis-
tence of ironworking at the site, apparently well into the post-Meroitic
period. As suggested in some preliminary work this may well be linked
with changes in the (royal) control of ironworking.96 Such a persistence
raises a number of interesting questions, not least concerning the nature of
the community established in the enclosed settlement. It is clear that
ironworkers had long-existed at Meroe, although their status relationship
(free or servile) to the Meroitic crown remains unknown. But equally, like

90
Leonardi 2013, 27; Simonse 1992, 233. 91 Johnson 1989. 92 Spaulding 1985, 208–9.
93
Pope 2014, 98–108; Rilly and De Voogt 2012. 94 Simonse 1992. 95 Parker 2003.
96
Humphris 2014.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 389

potters, their crafts were transformed and developed during the Meroitic
period.
In terms of the legacies of such Meroitic urban centres, one further
outcome that may be suggested is that within communities such as
Hamadab we might find one explanation for how craft-based castes take
form. While very familiar in various forms across much of Sahelian-
Sudanic Africa, how these might originate has remained a puzzle.97 In
the Meroitic case we might suggest the central role for the state in the
generation of craft-based communities and also identities in such urban
centres. Their likely central role in technological development may also be
highlighted. In relation to ceramic manufacture, the first century BC seems
to have been a period of significant transformations of already very long-
established potting traditions, with the re-appearance of wheel-made
mass-produced coarsewares. It is apparent that these were able to integrate
both long-established craft practices and technologies with new ones, for
example in new pottery kiln technologies, within an emerging new ‘imper-
ial’ culture.98 We might also consider that other craft groups may also have
undergone similar transformations, within royally controlled contexts.
Meroitic towns provided these contexts.
Like other forms of identity, these urban/craft identities would have
been dynamic. A political collapse of Meroitic kingship would have opened
up new trajectories for such groups, already distinct from other population
elements. The extent to which urban centres created their own ‘internal
frontier zone’ around them, as recently explored in Leonardi’s work,99
might be usefully explored in relation to these much earlier urban
experiments. Equally, the implications of their disappearance must also
be considered. While we have little hard evidence for the exact nature of
the social transformations which accompanied the disintegration of the
Meroitic state, it is possible that this was a period of pastoral resurgence
that unpicked the structures which encouraged or supported more inten-
sive agriculture in the Meroitic heartland. More specifically however we see
an abandonment of the urban forms which had been such distinctive
material manifestations of the state. Under such circumstances, the options
open to groups with craft-based identities, the product of several centuries
of enclosed, or at least controlled living, would seem to be very different.
What did the post-Meroitic future hold for royal potters, or indeed royal
ironworkers if they chose to or were required to ‘go feral’? Possessed of
valuable skills, but quite possibly also lacking claims to land or owning

97 98 99
Haaland and Haaland 2008; Tamari 2012. Edwards 2014a. Leonardi 2013, 217.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 390 David N. Edwards

herds, or indeed social ties to those who did, their possible futures may
have been more constrained than many. However, their skills would
continue to represent an important power source that was desirable to
control.100 Their subsequent histories doubtless developed and took on
new forms in the new regional centres of power that emerged by the sixth
century in what was to become Nubia.
Notwithstanding many significant gaps in our knowledge of Meroitic
urbanism, the many-layered political implications of the very existence of
these centres should not be overlooked. A consideration of the necessary
conditions for their creation, persistence, and practical purposes takes us
close to many of the fundamental bases of the Meroitic state and enforce-
ment of its sovereignty. Echoing Smith, we may here recognise material
manifestations of authority and subjection at many different scales.101
Within political landscapes we encounter re-orderings of space and terri-
tory; while within urban spaces, the built environment may itself represent
a technology of order and control.

References

Anderson, D. and Rathbone, R. (eds). 2000. Africa’s Urban Past. Oxford: James
Currey.
Anderson, J. and Welsby, D. (eds). 2014. The Fourth Cataract and Beyond. Proceedings
of the 12th International Conference for Nubian Studies. Leuven: Peeters.
Arthur, J. 2003. Brewing beer: Status, wealth and ceramic use alteration among the
Gamo of south-western Ethiopia. World Archaeology 34.3: 516–28.
Baud, M. 2008. The Meroitic royal city of Muweis: First steps into an urban
settlement of riverine Upper Nubia. Sudan & Nubia 12: 52–63.
Baud, M. (ed.). 2011. Meroe, un empire sur le Nil. Paris: Louvre.
Baud, M. 2014. Downtown Muweis – a progress report (2007–2014). In Anderson
and Welsby 2014, 763–82.
Bonnet, C. and Valbelle, D. 2014. La ville de Kerma – Une capitale nubienne au sud
de l’Egypte. Lausanne: Favre.
Borkowski, Z. and Paner, H. 2005. The Awlib Temple Complex: Report on the
2001 and 2003 Excavation Seasons. Gdansk Archaeological Museum African
Reports 3: 47–60.
Brite, E.B. and Marston, J.M. 2013. Environmental change, agricultural innovation,
and the spread of cotton agriculture in the Old World. Journal of Anthropological
Archaeology 32: 39–53.

100 101
Compare Simonse 1992, 237–38. Smith 2011, 423–26.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 391

Coquery-Vidrovitch, C. 2005. African urban spaces. History and culture. In S.

J. Salm and T. Falola (eds), African Urban Spaces in Historical Perspective.
Rochester, N.Y.: University of Rochester Press, xv–xl.
Edwards, D.N. 1996. The Archaeology of the Meroitic State: New Perspectives on Its
Social and Political Organization. Oxford: Archaeopress.
Edwards, D.N. 1998. Meroe and the Sudanic kingdoms. Journal of African History
39.2: 175–93.
Edwards, D.N. 1999. A Meroitic Pottery Workshop at Musawwarat es Sufra.
(Meroitica 17.2). Wiesbaden: Harrassowitz.
Edwards, D.N. 2004. The Nubian Past: An Archaeology of the Sudan. London:
Routledge.
Edwards, D.N. 2011. From Meroe to ‘Nubia’ – exploring culture change without
the ‘Noba’. In V. Rondot, F. Alpi and F. Villeneuve (eds), La Pioche et La Plume
(Hommages Archéologiques à Patrice Lenoble). Paris: PUPS, 501–14.
Edwards, D.N. 2014a. Early Meroitic pottery and the creation of an early imperial
culture? In A. Lohwasser and P. Wolf (eds), Ein Forscherleben zwischen den
Welten. Zum 80. Geburtstag Steffen Wenig, Berlin: SAG, 51–63.
Edwards, D.N. 2014b. Creating Christian Nubia: Processes and events on the
Egyptian frontier. In J. Dijkstra and G. Fisher (eds), Inside and Out. Interactions
between Rome and the Peoples on the Arabian and Egyptian Frontiers [Late
Antique History and Religion 8]. Leuven: Peeters, 407–31.
Fiedler, M. and Jesse, F. 2012. Griechische Keramik aus der Festung Gala Abu
Ahmed im Nordsudan. Archäologischer Anzeiger 2011–12: 67–80.
Fitzenreiter, M, Seiler, A. and Gerullat, I. 1999. Musawwarat es Sufra III. Die Kleine
Anlage. Wiesbaden: Harrassowitz.
Fuller, D. 2014. Agricultural innovation and state collapse in Meroitic Nubia. The
impact of the Savannah package. In C.J. Stevens, S. Nixon, M.A. Murray and D.
Q. Fuller (eds), The Archaeology of African Plant Use. Walnut Creek: Left Coast
Press, 165–77.
Fyfe, C. and McMaster, D. (eds). 1981. African Historical Demography:
Proceedings of a Seminar held in the Centre of African Studies, University of
Edinburgh, 24th and 25th April 1981. Vol. 2. Edinburgh: Centre of African
Studies.
Garstang, J., Sayce, A.H. and Griffith, F.L. 1911. Meroe. The City of the Ethiopians.
Oxford: Clarendon Press.
Griffith, F.L. and Crowfoot, G.M. 1934. On the early use of cotton in the Nile
Valley. Journal of Egyptian Archaeology 20: 5–12.
Haaland, R. 1985. Iron producing, its socio-cultural context and ecological impli-
cations. In R. Haaland and P.L. Shinnie (eds), African Iron Working – Ancient
and Traditional. Oslo: Norwegian University Press, 50–72.
Haaland, R. and Haaland, G. 2008. Craft specialization, caste identities and poli-
tical centralization. The use of anthropological perspectives in reconstructing
archaic forms of economic organization. In K. Chilidis, J. Lund and C. Prescott

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 392 David N. Edwards

(eds), Facets of Archaeology. Essays in Honour of Lotte Hedeager on her 60th

Birthday, Oslo Archaeological Series 10. Oslo: Unipub, 155–65.
Hinkel, M. 1991. Hafire im antiken Sudan. Zeitschrift für Ägyptische Sprache 118.1:
32–48.
Hafsaas-Tsakos, H. 2013. Edges of bronze and expressions of masculinity: The
emergence of a warrior class at Kerma in Sudan. Antiquity 87: 79–91.
Hopkins, A.G. 1973. An Economic History of West Africa. London: Longman.
Humphris, J. 2014. Post-Meroitic iron production: Initial results and interpretations.
Sudan & Nubia 18: 121–29.
Jakobielski, S. and Scholz, P. (eds). 2001. Dongola-Studien. Warsaw: ZAS.
Jesse, F. 2013. The Gala Abu Ahmed fortress in Lower Wadi Howar (Northern
Sudan). In F. Jesse and C. Vogel (eds), The Power of Walls – Fortifications in
Ancient Northeastern Africa. Cologne: Heinrich-Barth-Institut, 321–52.
Johnson, D.H. 1989. The structure of a legacy: Military slavery in Northeast Africa.
Ethnohistory 36.1: 72–88.
Kapteijns, L. and Spaulding, J. 2005. The conceptualisation of land tenure in
the precolonial Sudan: Evidence and interpretation. In D. Crummey (ed.),
Land, Literacy and the State in Sudanic Africa. Trenton, N.J.: Red Sea Press,
21–41.
Kemp, B.J. 1987. The Amarna workmen’s village in retrospect. Journal of Egyptian
Archaeology 73: 21–50.
Kopytoff, I. and Miers, S. 1977. African ‘slavery’ as an institution of marginality. In
S. Miers and I. Kopytoff (eds), Slavery in Africa. Madison: University of
Wisconsin Press, 3–81.
Lane, P. and Johnson, D. 2009. The archaeology and history of slavery in South
Sudan in the nineteenth century. In A. Peacock (ed.), The Frontiers of the
Ottoman World. Oxford: British Academy, 509–37.
Lebon, J.H.B. 1965. Landuse in Sudan. Bude: Geographical Publications.
Lenoble, P. 1994. Du Méroïtique au Postméroïtique dans la Région méridionale du
Royaume de Méroé. Recherches sur la période de transition. Unpublished PhD
thesis, Paris: Sorbonne.
Lenoble, P. 2008. Une carte des derniers siècles de Méroé. Sites préchrétiens autour
de l’ancienne capital entre Wad ben Naqa et Gabati. Kush 19: 59–65.
Lenoble, P. and Rondot, V. 2003. À la redécouverte d’El-Hassa. Temple à Amon,
palais royal et ville de l’empire méroitique. Cahiers de recherches de l’Institut de
papyrologie et d’égyptologie de Lille 23: 101–15.
Leonardi, C. 2013. Dealing with Government in South Sudan. Woodbridge: James
Currey.
MacDonald, K. and Camara, S. 2012. Segou, slavery and Sifinso. In J.C. Monroe
and A. Ogundiran (eds), Power and Landscape in Atlantic West Africa:
Archaeological Perspectives. Cambridge: Cambridge University Press, 169–90.
McIntosh, S. and McIntosh, R. 1993. Cities without citadels: Understanding urban
origins along the middle Niger. In T. Shaw, P. Sinclair, B. Andah and A. Okpoko

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 393

(eds), The Archaeology of Africa: Food, Metals and Towns. London: Routledge,
622–41.
Maillot, M. 2014. The palace of Muweis in the Shendi Reach: A comparative
approach. In Anderson and Welsby 2014, 783–95.
Maillot, M. 2015. The Meroitic palace and royal city. Sudan & Nubia 19: 80–87.
Maillot, M. 2016. Palais et grandes demeures du royaume de Méroe. Paris: PUPS.
Mair, L. 1977. African Kingdoms. Oxford: Oxford University Press.
Monroe, J.C. 2010. Power by design: Architecture and politics in precolonial
Dahomey. Journal of Social Archaeology 10: 367–97.
MGA 2015. Musawwarat Graffiti Archive. Exploring Pictures in Place. Available
at: http://musawwaratgraffiti.mpiwg-berlin.mpg.de/index_html [last accessed
12 September 2019].
Näser, C. 2011. Early Musawwarat. In Rondot et al. 2011, 317–38.
Nachtigal, G. 1971. Sahara and Sudan (translated by A.G.B. Fisher and H.J. Fisher).
London: Hurst.
O’Connor, D. 2013. Kerma in Nubia, the last mystery: the political and social
dynamics of an early Nilotic state. In R.B. Koehl (ed.), Amilla: The Quest for
Excellence. Studies Presented to Guenter Kopcke in Celebration of His 75th
Birthday. Philadelphia: INSTAP Academic Press, 189–205.
Onderka, P. and Vrtal, V. 2014. Nubia. A Land on the Crossroads of Cultures. Wad
Ben Naga 2014. Prague: National Museum.
Osborne, R. 2005. Urban sprawl: What is urbanization and why does it matter? In
R. Osborne and B. Cunliffe (eds), Mediterranean Urbanization 800–600 BC.
Oxford: British Academy, 1–16.
Parker, B. 2003. Archaeological manifestations of empire: Assyria’s imprint on
southeastern Anatolia. American Journal of Archaeology 107: 525–57.
Pope, J.W. 2014. The Double Kingdom Under Taharqo. Leiden: Brill.
Potts, D. 2012. Challenging the myths of urban dynamics in Sub-Saharan Africa:
The evidence from Nigeria. World Development 40.7: 1382–93.
Rehren, T. 2001. Meroe, iron and Africa. Antike Sudan 12: 102–9.
Rilly, C. 2008. Enemy brothers, kinship and relationship between Meroites and
Nubians (Noba). In W. Godlewski and A. Lajtar (eds), Between the Cataracts.
Warsaw: Warsaw University Press, 211–25.
Rilly, C. and DeVoogt, A. 2012. The Meroitic Language and Writing System.
Cambridge: Cambridge University Press.
Rondot, V., Alpi, F. and Villeneuve, F. (eds). 2011. La pioche et la plume: Autour du
Soudan, du Liban et de la Jordanie: hommages archéologiques à Patrice Lenoble.
Paris: PUPS.
Säve-Söderbergh, T. and Troy, L. 1991. New Kingdom Pharaonic Sites (The Finds
and Sites (Volume 5:2). Uppsala: Scandinavian Joint Expedition to Sudanese
Nubia.
Scott, J.C. 1998. Seeing Like A State: How Certain Schemes to Improve the Human
Condition Have Failed. New Haven: Yale University Press.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 394 David N. Edwards

Shinnie, P.L. 1967. Meroe: A Civilization of the Sudan. London: Thames and
Hudson.
Shinnie, P.L. 1985. Iron working at Meroe. In R. Haaland and P.L. Shinnie (eds),
African Iron Working – Ancient and Traditional. Oslo: Norwegian University
Press, 28–35.
Shinnie, P.L. and Kense, F. 1982. Meroitic Iron Working. In N. Millet and A.
L. Kelley (eds), Meroitic Studies (Meroitica 6). Berlin: Akademie-Verlag,
17–28.
Simonse, S. 1992. Kings of Disaster: Dualism, Centralism, and the Scapegoat King in
Southeastern Sudan. Leiden: Brill.
Smith, A.T. 2003. The Political Landscape: Constellations of Authority in Early
Complex Polities. Berkeley: University of California Press.
Smith, A.T. 2011. Archaeologies of sovereignty. Annual Review of Anthropology 40:
415–32.
Spaulding, J. 1985. The Heroic Age in Sinnār. East Lansing: Michigan State
University.
Spaulding, J. 2006. Pastoralism, slavery, commerce, culture and the fate of the
Nubians of Northern and Central Kordofan under Dar Fur Rule, c.1750 –
c.1850. International Journal of African Historical Studies 39.3: 393–412.
Spaulding, J. and Kapteijns, L. 2002. Land tenure and the state in the precolonial
Sudan. Northeast African Studies 19.1: 33–66.
Spencer, N. 2012. Insights into life in occupied Kush during the New Kingdom:
new research at Amara West. Antike Sudan 23: 21–28.
Spencer, N., Woodward, J. and Macklin, M. 2012. Re-assessing the abandonment
of Amara West: the impact of a changing Nile? Sudan & Nubia 16: 37–43.
Tamari, T. 2012. De l’apparition et de l’expansion des groups de spécialistes endo-
games en Afrique: Essai d’explication. In C. Roboin-Brunner and B. Martinelli
(eds), Métallurgie du fer et Sociétés africaines. Oxford: Archaeopress, 5–31.
Török, L. 1997. The Kingdom of Kush. Leiden, Brill.
Tylecote, R.F. 1982. Metal working at Meroe, Sudan. In N. Millet and A.L. Kelley
(eds), Meroitic Studies. (Meroitica 6). Berlin: Akademie-Verlag, 29–42.
Vincentelli, I. 1993. A discharge of clay sealings from the Natakamani palace. Kush
16: 116–41.
Vincentelli, I. 2007. Some clay sealings from Sanam Abu Dom. Cahiers de
recherches de l’Institut de papyrologie et d’égyptologie de Lille 26: 371–78.
Vincentelli, I. 2011. The treasury and other buildings at Sanam. In Rondot et al.
2011, 269–82.
Welsby, D. 1998. Soba II. London: British Institute in Eastern Africa.
Welsby, D. 2014. The Qatar-Sudan Archaeological project in the Northern
Dongola Reach. Sudan & Nubia 18: 47–68.
Welsby, D. and Daniels, C.M. 1991. Soba: Archaeological Research at a Medieval
Capital on the Blue Nile. London: British Institute in Eastern Africa.
Wolf, P. 1999. Recent fieldwork at Musawwarat es Sufra, Sudan & Nubia 1: 20–9.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 9 Early States and Urban Forms in the Middle Nile 395

Wolf, P. 2015. The Qatar-Sudan Archaeological project – the Meroitic town of

Hamadab and palaeo-environment in the Meroe region. Sudan & Nubia 19:
115–31.
Wolf, P., Nowotnik, U. and Hof, C. 2014a. The Meroitic urban town of Hamadab
in 2010. In Anderson and Welsby 2014, 719–37.
Wolf, P., Nowotnik, U. and Woss, F. 2014b. Meroitic Hamadab – a century after its
discovery. Sudan & Nubia 18: 104–20.
Wolf, P. 2015. The Qatar-Sudan Archaeological Project – The Meroitic town of
Hamadab and the palaeo-environment of the Meroe region. Sudan & Nubia 19:
115–31.

https://doi.org/10.1017/9781108637978.010 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa
Greek, Punic and Roman Models

andrew i. wilson

Introduction and Historical Background

This chapter seeks to identify the distinctive characteristics of Greek, Punic

and Roman urbanism in North Africa, and to explore similarities and
differences between them. It presents an overview of urban morphology,
infrastructure (streets, water supply), architectural characteristics (materi-
als and aesthetics), and the common range of public buildings and types of
domestic housing found in the various cultures; it also explores the extent
to which we can reconstruct the use of public space and the character of
urban life from inscriptions and the evidence of the statue habit in Roman
towns. Questions of size and population, and economic roles, will also be
considered. The main aim is to provide a succinct summary of funda-
mental information to enable comparison and contrast with other chapters
in the collection which look at indigenous state formation and urbanism in
the Maghrib and the Sahara.
The first cities along the North African coast were founded as
Phoenician trading centres. The traditional date for the foundation of the
earliest of these, Utica, equates to 1101 BC, although in fact no archae-
ological remains earlier than the eighth century BC have been discovered.
Likewise, although Carthage was said to have been founded in 814 BC, the
earliest archaeological material is about a hundred years later.
The Greek colonisation of Cyrenaica (Eastern Libya) began in the later
seventh century BC, with the founding of Cyrene, traditionally in c.631 BC,
by settlers from Thera under their leader Battos, who became the city’s first
king.1 Its harbour, Apollonia, some 14 km distant, must have developed
gradually in the wake of Cyrene’s foundation. Barca (modern al-Marj),

1
396 Herodotos, Histories 4, 150–59.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 397

inland on the middle terrace of the Cyrenaican plateau, and the coastal city
of Taucheira (Tocra), were founded after Cyrene in the later seventh
century; the port of Barca was distant from it, and by the third
century BC developed into a city in its own right, Ptolemais. Together
with Euesperides (modern Benghazi), the last of the cities to be founded,
perhaps in the early sixth century BC, these five cities formed the
Cyrenaican Pentapolis.2
In the Archaic and Classical period down to c.440 BC, Cyrene was ruled
by the Battiad dynasty, descendants of King Battos, alternately named
Battos and Arkesilas. Dynastic struggles eventually undid them, and the
last king, Arkesilas IV, was murdered at Euesperides. Thereafter Cyrene
appears to have had a more democratic or republican form of government,
but the end of royal control saw Cyrene pitted against shifting alliances of
the other Cyrenaican cities, sometimes Cyrene with Euesperides against
Barca and Taucheira; at other times Euesperides allied with Barca. A decree
(late fourth- or possibly even early third-century BC) from Euesperides
informs us that the city had at that period a boule (council), with gerontes
(elders), and ephors (magistrates) involved in the city’s governance.3 But
Cyrenaica was also affected by external factors, notably in the Late Classical
and Hellenistic periods when the cities were subject at times to control by
Persian and Ptolemaic rulers in Egypt.
Alexander’s conquest of Egypt (332 BC) brought an end to Persian rule
there and to Persian hegemony over Cyrenaica. Following his death,
Cyrenaica came under Ptolemaic control, and it was in the reign of
Ptolemy II that the port of Barca took the name Ptolemais, probably
being laid out on an ambitious grid plan in this period, while Taucheira
took the name Arsinoë after Ptolemy’s wife.
In the Maghrib, inland from the coastal belt of Phoenician/Punic
emporia, we hear in ancient authors principally of two cities in
the second century BC, both called Cirta: Constantine in Algeria and El
Kef in Tunisia, the latter more commonly known as Sicca Veneria. This for
long encouraged the view that Numidian society knew little in the way of
urban life; but more recent archaeological work is now changing that
picture. Excavations at Althiburos in inland Tunisia show settled occupa-
tion, with an agricultural component, from the ninth century BC (see
below).4

2
Though the subsequent further development of Apollonia and the foundation of a further city in
the second century AD, Hadrianopolis, eventually meant the region supported seven major cities.
3
Fraser 1951.
4
See Sanmartí et al., Chapter 11, this volume on Althiburos and Numidia more generally.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 398 Andrew I. Wilson

Rome acquired territory in North Africa progressively: in 146 BC fol-

lowing the final defeat of Carthage in the Third Punic War; Cyrenaica in
96 BC, Numidia in 46 BC, Mauretania in the AD 40s, with subsequent
expansion into the desert margins. Under Roman rule, North Africa – at
least from the Lesser Syrtes westward – became one of the most densely
urbanised regions of the empire, comparable with Italy and coastal Asia
Minor.5 This was in part due to a development of the pre-Roman urban
base along the littoral, but it is clear also that the dense network of
Numidian settlements away from the coast were developed, expanded
and monumentalised as cities.6

Punic Cities

Many early Phoenician emporia were founded on promontories or projecting

capes which offered alternative anchorages either side, sheltered from different
winds; many of these are identifiable by the common toponym Rus- (‘cape’,
‘headland’): Rusicade, Rusaddir, Rusucurru, Rusazu, Rusguniae.
Carthage, which became the pre-eminent of these Punic cities, was
founded on a broad promontory with marine inlets to the north and
south, later supplemented by artificial harbour works.7 Overlooking
these, the Byrsa hill formed the nucleus and defended citadel. The ancient
artificial harbour – the Cothon – is represented today by two lagoons north
of the bay of el-Kram. The Punic harbour had two parts: an outer part
(referred to as the rectangular harbour but actually originally a very elon-
gated hexagon) for merchant shipping, and an interior, circular, basin with
a central island, reserved for warships.
The harbours had communication with each other, and a common entrance from
the sea seventy feet wide, which could be closed with iron chains. The first port was
for merchant vessels, and here were collected all kinds of ships’ tackle. Within
the second port was an island, and great quays were set at intervals round both the
harbour and the island. These embankments were full of shipyards which had
capacity for 220 vessels. In addition to them were magazines for their tackle and
furniture. Two Ionic columns stood in front of each dock, giving the appearance of
a continuous portico to both the harbour and the island. On the island was built the
admiral’s house, from which the trumpeter gave signals, the herald delivered
orders, and the admiral himself oversaw everything.8

5
See now Hanson 2016, for a study of the urban geography of the Roman world.
6
Mattingly 2016. 7 For a history of Carthage, see Lancel 1995.
8
Appian, Punic Wars 14.95 (Loeb translation).

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 399

Excavations have broadly confirmed Appian’s description and show that

the ports of Carthage were impressive technical achievements: c.120,000 m3
of earth were removed to dig the merchant harbour basin, and another
115,000 m3 for the circular harbour.9
Well-appointed Punic houses of the fourth century BC were originally
supplied with wells, but during the third century most, if not all, of the
houses underwent a change in their water supply. This involved the con-
struction of cisterns with capacities between 5.5 and 14.3 m3, which collected
rain run-off from roofs and courtyards by means of terracotta downspouts.
They were of the classic Punic ‘cigar-shaped’ type – deep, long and narrow
with rounded ends, their narrowness resulting from the use of stone slabs for
the roof, which could not be made very long without cracking (Fig. 10.1).
The construction of such cisterns within existing houses involved consider-
able disruption, necessitating the demolition of some walls to get the cover
slabs in, and since ordinarily water from wells was preferable to water from
cisterns, there must have been some compelling reason to make the switch.

Figure 10.1. Carthage – Punic houses on the Byrsa hill, destroyed in the Third Punic War,
146 BC. The houses are constructed in opus africanum, with large orthostats and smaller
masonry infill between them. A cistern, long and narrow with rounded ends, is visible in the
room in the right foreground where some of its massive cover slabs have been removed. The
concrete rubble masonry to the right and in the background belongs to the foundations of
Roman structures built over the ruins of the Punic city (Photo: A. Wilson).

9
Hurst 1993a; 1993b.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 400 Andrew I. Wilson

The change must have been a necessary response to changing conditions in

water supply that rendered the wells unusable, either a lowering of the local
water table due to declining rainfall, or to increased abstraction by a growing
population, or perhaps a contamination of the water table by sewage and
effluent.10
The Tophet to the south of Carthage and west of the harbours was an
area where children (up to four years old) were sacrificed and buried.
Sometimes animals would be substituted in place of children, but human
sacrifice continued through to the end of the Punic Wars (if not beyond).
Squat, gray funeral stelae marked the location of cinerary urns containing
the ashes of infants, small children and animals. Some stelae are inscribed
with texts saying that the burial is an offering to the gods; many are
inscribed with the triangular symbol of the goddess Tanit. One third-
century BC stone even bears a depiction of a priest carrying a small
child. From this evidence, most (but not all) scholars conclude that the
Tophet was a burial ground for infants sacrificed to the gods Tanit and Baal
Hammon.11
The siege of Carthage in the third Punic War lasted more than three
years – Scipio Aemilianus eventually took command and imposed
a blockade on the port, aiming to starve the city into submission.
Carthage finally fell to Scipio’s forces in 146 BC – it took six days of
urban fighting, house-to-house, for the Roman forces fully to take the
city, which was sacked and burnt.
The Punic town whose archaeology is most completely known to us is
the site of Kerkouane on Cap Bon, chiefly because it was destroyed in the
mid-third century BC and not subsequently overbuilt. The site was occu-
pied from the sixth century BC, and two major phases of destruction have
been associated with the invasion of Agathocles in 310 BC, and the First
Punic War in 253 BC. A rampart enclosed an area of about 8 ha with
a perimeter street around the inside of the rampart, and within this the
bulk of the site was laid out on an irregular grid plan, with large public
spaces at some street intersections. A large courtyard building near the
centre of the town has been identified as a sanctuary of Semitic type. The
masonry houses were equipped with bathrooms with a hip bath, water-
proofed with cocciopesto lining, and an open channel drain through the
entrance corridor out to the street (Fig. 10.2). Evidence for the production
of salted fish and murex dye shows the importance of marine resources to

10 11
Wilson 1998. Xella et al. 2013.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 401

Figure 10.2. Kerkouane – interior of Punic house with basin and bath tub (Photo:
A. Wilson).

the city’s economy; glassmaking debris and a cluster of pottery kilns

provide further evidence of craft activities.12
Meninx, on the island of Jerba, was in existence by 253 BC and was
sacked by the Romans in 217 BC. The Punic site covered at least some 19 ha
and was engaged in purple dye production. The other main town on Jerba,
Burgu (probably ancient Thoar or Phoar), reached its maximum size
(c.20 ha) in the second century BC; it is known largely as a surface scatter
of pottery, with traces of structures possibly suggesting a centripetal
arrangement of streets rather than an orthogonal grid, and a Punic tower
tomb on its south-eastern edge. A steep scarp around the perimeter of the
site may suggest that it was originally walled (in mudbrick?).13

Greek Colonies

Much of what is visible at many of the Cyrenaican cities belongs to the

Roman and late Roman periods, but Cyrene, Tocra and Euesperides in
particular provide good information about the Archaic, Classical and
Hellenistic phases (Fig. 10.3).

12 13
Fantar 1984–1986; 1998. Fentress 2009.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 402 Andrew I. Wilson

Figure 10.3. Plan of the Greek city of Euesperides (Benghazi). The original Archaic
nucleus of the city was on the low mound occupied in the twentieth century by the
cemetery of Sidi Abeid; in the Classical period (by the fourth century BC) the city had
been extended considerably to the south, where the grid plan of rectangular city blocks,
enclosed by a defensive wall, could be traced from air photographs. The extent of
pottery scatters suggests that suburban activity extended beyond the limits of the wall
circuit (A. Wilson).

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 403

Figure 10.4. Cyrene – Temple of Zeus (sixth century BC), built with massive columns in
the Doric order (Photo: A. Wilson).

Cyrene was founded on the edge of the upper Cyrenaican terrace, where
a copious spring, the Fountain of Apollo, emerges from the limestone
karst. Next to this developed a sanctuary of Apollo, with a monumental
temple in the Doric style that gradually accreted a number of smaller
temples and shrines around it. Above the spring and overlooking the
sanctuary of Apollo was a fortified citadel or acropolis; to the south, on
flatter ground, lay the agora or main market place, and civic centre. A large
temple of Zeus lay in the north-east part of the city (Fig. 10.4), and there
was a third temple of the Archaic period, dedicated perhaps to Demeter,
just outside the city walls to the south.14 This temple lay within its own
walled sanctuary area, which included a theatre for ritual dramatic perfor-
mances, and a monumental propylon. A smaller sanctuary of Demeter with
a set of shrines, excavated in the 1970s along the Wadi Belgadir immedi-
ately to the north,15 is probably an outlier of the more recently discovered
main sanctuary. The public architecture of Archaic and Classical Cyrene
was built in monumental limestone ashlars, with colonnaded temples. Less
is known about the domestic housing of the period, largely overbuilt by
Roman phases, and in any case most excavation has concentrated on the
monumental public structures. Outside the gates, the roads leading out of
town were lined with necropoleis (it being forbidden to bury the dead
within the city limits), and their expansion and development over time can
14 15
Luni 2001; 2006; cf. Kane and White 2007. White 1984; 1993.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 404 Andrew I. Wilson

clearly be traced: in the northern necropolis, rock-cut tombs of the Archaic

period were carved into the rock face above the road leading to Apollonia
and the coast just outside the sanctuary of Apollo, with later rock-cut tomb
complexes and sarcophagi of the Classical period increasingly further out.
In the southern necropolis, increasing competition in funerary display led
to spectacularly ornamented columnar façade tombs in the Hellenistic
period.16
The earliest phases of Cyrene’s port, Apollonia, are largely buried below
Roman and Byzantine layers, but parts of an Archaic/Classical sanctuary
with statuettes and ritual offerings have been found on the highest point of
the site. The harbour and ship-sheds go back to the Classical/Hellenistic
periods.
Little is known of the archaeology of Barca, founded after 565 BC by the
brothers of king Arkesilas II, but its port, Ptolemais (Tolmeita) has yielded
Archaic pottery from the shore near the modern lighthouse, and the
theatre, cut into the hillside overlooking the site, is probably of Classical
date. A Hellenistic wall circuit can be traced, though largely dismantled in
the late Roman period, the west gate towers surviving as a fortified block-
house. The city was laid out on a regular grid plan with city blocks as
elongated rectangles. In the western necropolis, situated in an area of
quarries that provided building stone for the city, an imposing tower
tomb, with several layers of loculi, must represent the mausoleum of one
of the city’s leading families in the Hellenistic period.
At Taucheira, on the coast further to the west, excavations in the 1960s
found a deposit of Archaic pottery from the late seventh century BC,
interpreted as a part of a deposit within a sanctuary, perhaps of Demeter.
The pottery indicates trade links with Corinth and the East Greek islands
and cities of the Asia Minor coast.17 The approximate limits of the early
city can be gauged from the traces of the city wall, of mudbrick on a stone
rubble foundation, visible in the seaward cliff where it has been exposed by
coastal erosion,18 and the locations of some stone quarries which were
originally extramural, later incorporated within the much expanded ashlar
defensive wall circuit laid out in the Hellenistic period (though later
rebuilt).19
At Euesperides, the most westerly of the cities of the Pentapolis, whose
site now lies within the suburbs of Benghazi at Sidi Abeid, rather more of
the Greek levels have been investigated because the city was abandoned in

16
Cherstich 2008. 17 Boardman and Hayes 1966. 18
Bennett et al. 2004.
19
Smith and Crow 1998.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 405

the mid-third century BC and not rebuilt. The earliest phases belong to the
early sixth century BC, c.580 BC, suggesting that Euesperides was founded
after the other Cyrenaican colonies (its location, on the less fertile Benghazi
plain, is consistent with this, the most favourable territory having been
colonised first). The city was surrounded by a defensive wall of mudbrick
on a rubble foundation, as at Taucheria, later widened and fronted with
a facing of mudbrick; part of a projecting tower on ashlar foundations was
also discovered. Outside the wall a quarry ditch provided extra defence,
and into the sides of the quarry tombs were subsequently cut in the late
fourth century BC.20
Within the walls, two phases of Archaic-period housing have been
excavated in different parts of the site: the houses were of mudbrick on
rubble dwarf wall foundations, with beaten earth floors; an amphora or jar
in the corner of many rooms acted as a storage receptacle. The streets were
unpaved and without drains or sewers; refuse was thrown out into the
street, and trodden in; periodically they were resurfaced with a layer of clay
to even out the worst potholes. Structures of the later fourth and third
centuries BC are better preserved, and seem to show better built walls; by
the third century BC some of the houses had floors in pebble mosaic, mixed
pebble and tessera technique, or even some purely in tesserae. Most of the
mosaic floors were fairly plain, but one had a wave-crest border, and some
indications of figural compositions, including a pair of dolphins, have been
inferred from very damaged fragments from a phase of the early third
century BC. Water supply came mainly from wells, with some houses
provided with rock-cut cisterns towards the end of the city’s life.
The city appears to have suffered an earthquake between 262 and
250 BC, and though rebuilt thereafter, its final abandonment no later
than 250 BC is shown by the abrupt cessation of coin finds. The deliberate
filling of wells, abundant finds of slingshot, and a Hellenistic epigram by
Callimachus apparently referring to a siege of the city by Queen Berenice,
combine to suggest that the city was sacked during the civil wars in
Cyrenaica that followed the death of King Magas in 250. The population
was transferred to the Sidi Khrebish area of modern Benghazi where a new
city was founded, called Berenice after the victor of these wars.21
Of the Cyrenaican cities, Cyrene and Barca in particular enjoyed access
to a very fertile agricultural hinterland, abundant not only in grain, vines
and olives (these latter two probably introduced to the region by the
Greeks), but also, to the south, the medicinal and aphrodisiac plant

20 21
Buzaian and Lloyd 1996; Lloyd et al. 1995; 1998. Wilson 2005.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 406 Andrew I. Wilson

silphium, exported for very high prices. The excavations at Euesperides

have also highlighted the role of the exploitation of marine resources in the
economy of the coastal cities; at Euesperides large deposits of crushed
Murex trunculus attest the production of purple dye, and imply that
some kind of textile, probably wool, was dyed with it. This in turn hints
at the raising of sheep on the thin soils, poorly suited to agriculture, of the
Benghazi plain.22
Pottery from Cyrene and Tocra shows the expected trade connections
with the Aegean in the Archaic and Classical periods, but only the fine
pottery has been studied and published, and not in any quantified form.
More detailed study and quantification of the material from Euesperides
shows that in addition to the usual links with the Greek world (although
there is no identified material from the Black Sea region), from the fourth
century BC trading links with Punic North Africa begin to intensify, and
that by the third century BC, 30 per cent of the cooking wares were
imported from outside Cyrenaica, and half of these imports were Punic.
Punic amphorae accounted for about 5 per cent of the total amphorae from
the site.23

Numidian Urbanisation

As noted already, recent excavations have revealed permanent settlement

at the inland site of Althiburos in Tunisia by the ninth century BC.24
Certainly, one cannot assume that Althiburos was a town at this early
date, and only a relatively small extent of these earliest levels has been
excavated, but it is surely important that there was continuous occupation
from the ninth century BC through the Roman period, and that what had
by the Roman period become a town with a full complement of monu-
mental architecture had developed out of a Numidian village. Although it
may be difficult to state at what point Althiburos developed urban char-
acteristics, comparative evidence from other sites shows that we should
certainly not assume that it did not happen before the Roman period. The
last three centuries BC saw an increasing Mediterranean influence of
Numidian architecture: at Bulla Regia, for example, excavations below
the Baths of Julia Memmia revealed ashlar foundations of a monumental
stone building, dating to the late second or early first century BC,
22
Wilson 2005; 2013. 23 Göransson 2004; Swift 2006; Wilson 2005; 2013.
24
See Sanmartí et al., Chapter 11 and Bokbot, Chapter 12 for further reflections on the
contributions of indigenous North African populations to urbanism in the Maghrib.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 407

suggesting public urban architecture already at this period.25 At Chemtou

there is a Numidian monument below the Roman forum, and an elaborate
altar on the top of the hill above the marble quarries, showing strong
Hellenistic influence in its carved motifs.
A major problem for our knowledge of Numidian urbanism is the fact
that most if not all Numidian settlements continued to be occupied, and
indeed were expanded and rebuilt, in the Roman period, and relatively
little excavation has taken place below Roman levels in these towns. The
physical evidence for the architecture, layout and appearance of these
Numidian settlements is thus scant. But the fact that the vast majority of
toponyms of Roman cities in North Africa have Berber roots is a clear
indication that the massive urbanisation of the Roman period was not
imposed on an empty landscape, but implanted in pre-existing settle-
ments, which were expanded and (perhaps rather gradually) monumen-
talised. Mattingly has observed that many of these sites with Berber
toponyms, Althiburos included, are sited on spurs with steep drops either
side of a central ridge, and that this appears to have been a preferred
situation.26
Sallust, in his account of the Jugurthine War (112–106 BC), refers to
a handful of Numidian towns and sheds a little light on their nature. He
mentions ‘. . . a town of the Numidians called Vaga, the most frequented
emporium of the entire kingdom, where many men of Italic race traded
and made their homes’.27 He also calls Zama ‘a large city’ and mentions its
walls,28and that it was ‘situated in an open plain and fortified rather by art
than by nature’;29 his reference also to the gates of Sicca shows that it too
was a walled city.30 Thala, too, is described as ‘a large and wealthy town in
which the greater part of his [Jugurtha’s] treasure was kept, and his
children were being brought up in grand style’.31 It was clearly walled,
and with a palace. Indeed, nearly all the cities Sallust mentions are said to
be large and wealthy; this may be mere rhetoric, and in any case we have no
way of knowing what that meant in absolute terms. But the presence of
Roman traders at Vaga, and at Cirta in 113 BC (where their massacre by
Jugurtha was a casus belli or a cause of war for Rome),32 is important, as it
shows these towns hosting Roman and Italian merchant communities
engaged in trade, certainly in grain and apparently also in other goods, in
the late second century BC.

25
Broise and Thébert 1993, 204–5. 26 Mattingly 2016, 15. 27 Sallust, Jugurthine War 47.
28
Sallust, Jugurthine War 56. 29 Sallust, Jugurthine War 57. 30 Sallust, Jugurthine War 56.
31
Sallust, Jugurthine War 75–76. 32 Sallust, Jugurthine War 21.2, 26–27.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 408 Andrew I. Wilson

Roman Urbanism

Roman Carthage
Roman urbanisation in Africa consisted only in part of deliberate acts of
colonial foundation, and most of those long after the territory first fell
under Rome’s control with the final defeat of Carthage in 146 BC. There
was a short-lived and abortive attempt to re-found Carthage as a colony for
settlers from Italy in 133 BC, as part of C. Gracchus’s programme of land
reforms, but it clearly did not prosper, and archaeological traces of it are
scant.33 Julius Caesar founded a series of veteran colonies around Cap Bon
in Africa Proconsularis (modern Tunisia);34 and Augustus founded addi-
tional colonies across a wider region of North Africa to settle more veterans
after the end of the civil wars.35 The successful refoundation of Carthage, as
Colonia Concordia Iulia Karthago, belongs to this period; Caesar had
certainly drawn up plans for a colony at Carthage, although it remains
unclear whether it was actually begun before his assassination, or whether
it was carried out after Caesar’s death by Octavian/Augustus. The 3,000
colonists sent from Italy were too few to fill out the city’s plan, and must
have been supplemented by locals as well.
As refounded in the late first century BC, the city was laid out on
a regular grid plan, rigidly imposed on the undulating topography so that
where they ascended to the forum, on the summit of the Byrsa hill, the
streets had to turn into flights of steps (Fig. 10.5). The main streets,
the cardo and decumanus maximus, met at right angles in the forum, and
the city blocks were laid out as rectangular insulae twice as long as they
were wide. The city grid takes its alignment from the coastline between the
Bordj Djedid hill and the harbours, following part of the Punic street plan,
but diverges from other streets. The harbours were put back into service,
and by the fourth century AD the island in the middle of the circular
harbour housed a facility where shipments of olive oil received perhaps as
tax in kind (the canon olei?) were weighed and checked for quality.36 Most
of the Roman houses excavated in Carthage are elite residences, peristyle
houses with reception rooms lavishly decorated with mosaics.

33
Saumagne 1928.
34
Clupea (Kelibia), Curubis (Korbous), Carpi (Henchir Mraïssa), Neapolis (Nabeul).
35
Thuburbo Minus (Tebourba), Hippo Diarrhytus (Bizerte), Maxula (Rades), Caspis,
Thuburnica (Tebournuc), Simitthus (Chemtou), Sicca Veneria (El Kef), and Assuras in what is
now Tunisia; Igilgili (Jijel), Saldae (Bejaia), Rusucurru, Rusazu, Rusguniae, Aquae Calidae,
Zuccabar, Gunugu and Cartenna in what is now Algeria.
36
Peña 1998.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 409

Figure 10.5. Plan of Roman Carthage showing the main public buildings and known
elements of water-supply infrastructure. The theatre and odeon are on the Odeon Hill;
the circus is towards the south-west edge of the city, and the amphitheatre is the oval
structure between this and the La Malga cisterns. Numbered features: 1) Large La Malga
cisterns. 2) Small La Malga cisterns. 3) Castellum excavated by Vernaz. 4) Castellum
excavated by Ellis. 5) Aqueduct branch traced by Vernaz under decumanus III N. 6)
Bordj Djedid cisterns. 7) Antonine Baths. 8) Dar Saniat cisterns. 9) Turris Aquaria
(Wilson 1998, 66, Fig. 1).

Little is known of the buildings of Augustan Carthage, although the

German excavations down by the seashore have documented the re-use of
Punic domestic cisterns in houses of the Augustan period, and the employ-
ment of rubble from the Punic destruction layers. Most of the public
architecture now surviving belongs to the first century AD or later, and

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 410 Andrew I. Wilson

in particular to a large-scale rebuilding of the urban centre in the Antonine

period after a fire around AD 150.37 The Byrsa hill was monumentalised to
transform the former Punic citadel into a lavish forum complex, the top of
the hill being levelled to create a massive platform extended out across the
natural slope of the hill by a dumped terrace fill, to create a platform
30,000 m2, three times the size of the Augustan forum at Rome. This was
the commercial, religious and civic heart of the city, with a monumental
paved forum plaza surrounded by a basilica, library and temples including
a Capitolium (to Jupiter, Juno and Minerva). This latter is known only
from inscriptions that mention it, and probably lies beneath the Cathedral
of St Louis, built on the highest point of the site in the late nineteenth
century AD.
Of the theatre only a part of the cavea, backed against the side of the ‘Hill
of Juno’, to the north of the Byrsa, is still intact. The Augustan city probably
had a theatre from its foundation (in book four of the Aeneid, Virgil has
Dido building a theatre in a scene which has been taken to reflect
Augustus’s refoundation of the city at the time Virgil was writing), but
the structure visible today is of the second century AD, badly reconstructed
in the early twentieth century with the seating rake at the wrong angle.
The odeon at Carthage is one of the largest known from the Roman world,
built in c.AD 200 over earlier houses; it has been suggested that its construc-
tion may have been an imperial initiative, linked with Septimius Severus
granting permission to Carthage to hold the Pythian Games in 203.38 The
circus, too, is the longest known outside Rome, with a total length of
570–580 m, and an arena at least 496 m long; with 27 rows of seats, the
total capacity is estimated at 60,000–63,000 spectators. The amphitheatre,
whose capacity is estimated at 30,000, dates probably from the early first
century AD with a rebuild and expansion in the mid-second century AD.39 It
was built of sandstone from Cap Bon for its annular and radial walls, with
limestone used for the exterior façade, and the characteristically Italian
technique of opus reticulatum used for facing the cavea vaults.
As in most Mediterranean towns in the ancient world, rainwater collection
cisterns were a standard part of domestic water supply, and besides the reuse of
some Punic cisterns, new cisterns were also constructed in concrete with
barrel-vaulted roofs, a technological advance enabling the creation of larger
domestic cisterns than had been possible in the Punic period. But in addition,
several aqueducts were built; ultimately, Carthage came to have four or five of
them. The earliest probably tapped the sources closest to the city, while the

37 38 39
Thomas 2007, 146. Wells 2005. Bomgardner 1989, 145–46.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 411

later ones came from further away. In the Julio-Claudian period, and possibly
as early as the reign of Augustus, a massive set of reservoir cisterns on the
north-western outskirts of the city was constructed, with 16 barrel-vaulted
chambers holding a total of 50,000 m3 of water. The complex was clearly
aqueduct-fed, probably from a spring at Sidi Bou Said. In the mid-second
century, probably under Hadrian, a long aqueduct was built from springs at
Zaghouan, 98 km away, and this was extended under Septimius Severus to
additional springs at Aïn Djouker, bringing the total length to 132 km. This
lengthy and massive project was probably the last in the series of Carthage’s
aqueducts; other shorter systems, which probably fall between the Julio-
Claudian period and Hadrian, include a subterranean aqueduct in the plain
of La Soukra (serving at least in part for irrigation), an underground aqueduct
near La Malga, and a channel feeding reservoir cisterns at Dar Saniat to the
north of the city proper.40 Recent excavations at La Malga have also exposed
part of the arcade of another aqueduct running alongside the aqueduct from
Zaghouan, at a slightly higher level and feeding reservoir cisterns and baths
beside the modern Phoenix restaurant.
This multitude of aqueducts reflects the demands of a populous city in
a hot climate, demands that were both essential and ornamental. The
aqueducts supplemented domestic cisterns, and the richest houses were
directly connected, via lead or terracotta pipes, to the aqueduct network.
Yet these elite houses tended to use their piped water principally for
display, feeding ornamental fountains in peristyle courtyards. On the
other hand, the public fountains, while monumental and spectacular,
decorated with coloured marble veneer and with bronze and marble
statues, also served as an important source of water for those houses
without piped water, and as the primary source for those without wells
or cisterns; slaves would fetch water from the public fountains. The large
reservoir cisterns at La Malga enabled regulation of the supply of at least
one of the aqueducts, enabling for example night-time supply to be stored
to augment the distribution through the network during the day.41
The aqueduct engineering projects also supported a typically Roman
culture of public bathing, and the construction of the long aqueduct from
Zaghouan enabled the subsequent development of the massive Antonine
baths, a huge imperial bath complex built under Antoninus Pius (con-
struction began in the late 140s or 150s and the building was dedicated in
161–162) on a site overlooking the sea.42 The most extensive example of

40
Wilson 1998. 41 Wilson 1998.
42
For the date, CIL 8.12513 and AE 1949, 27; Wilson 1997, 242–43.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 412 Andrew I. Wilson

Roman baths in North Africa, they were once the largest in the Roman
world, with a central pool the size of a present-day Olympic swimming
pool; the whole complex was lavishly equipped with mosaics and imported
marbles.

Figure 10.6. Plan of Roman Timgad, showing fulleries (shaded) and other workshops.
The Trajanic core of the original city is clearly visible, with the forum near the centre
and the theatre below this; the Capitolium is in the south-west, with the Market of
Sertius and a possible cloth market above it. At the top centre are the large Northern
Baths (Wilson 2001a, 279, Fig. 12.08).

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 413

City Planning
Like Carthage, those sites which were colonial foundations, or which saw the
implantation of veteran colonies on pre-existing sites, tended to be planned on
a grid layout. At Utica, an extensive grid plan layout has been traced from
excavation, air photography, and recently geophysics, with rectangular insulae
similar to Carthage, though without an evident cardo maximus and decuma-
nus maximus; its date, however, remains to be established and a pre-Roman
origin cannot yet be excluded.43 The most celebrated example of a grid-
planned town in North Africa is Timgad (ancient Thamugadi), founded as
a colony under Trajan in AD 100, and laid out on a square grid with the forum
at the centre at the junction of the decumanus maximus and cardo maximus
(Fig. 10.6). Because the forum lies at the dead-centre of the plan, the southern
part of the cardo maximus has to make a dog-leg around it. The plan is similar
in many respects to that of legionary fortresses, with the forum in the place of
the fort’s headquarters building or principia. However, for all that Timgad is
often taken as a textbook example of Roman town planning, such regularity is
actually extremely rare. Indeed, even at Timgad, the subsequent development
of the city, growing from its initial size of just under 10 ha to some 47.5 ha,
lacks the order of the original foundation. The wall that originally surrounded
the Trajanic colony was demolished in the later second century and the town
expanded over it, the original gates marked by monumental arches (Fig. 10.7).
These expansions chiefly took the form of ribbon development following the
roads out of town, in stark contrast to the regularity of the original grid; and the
third-century Capitolium temple complex is located outside the original col-
ony, on a completely different alignment.
Timgad in fact shows the reverse of the much more common pattern of
urban development in North Africa, which generally comprises an early
nucleus (Punic or Numidian) of irregular streets, with later outgrowths of
regularly planned expansion, as at Sabratha, where the early core around
the forum contrasts with the regularly planned area (second century AD)
to the north of the theatre, or Volubilis, where the north-east quarter is
a more regular and spaciously planned later development. At Lepcis
Magna, successive phases of expansion are traceable through different
alignments of sections of the city plan.
Cuicul (modern Djemila), a Nervan colony founded in AD 96, shortly
before Timgad, shows a nearly regular plan adapted slightly to the topogra-
phy of the narrow ridge on which it sits, although here too later development
along the ridge takes a less regular form, determined as much by the
43
Ben Jerbania et al. 2015; Hay et al. 2010; Kallala et al. 2010; Lézine 1968.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 414 Andrew I. Wilson

Figure 10.7. Timgad – the view along the decumanus maximus towards the so-called
‘Arch of Trajan’, in fact a late second-century honorific arch built on the site of the
original west gate when the wall circuit was torn down as the city expanded. The
decumanus maximus (the main east-west street) and the cardo maximus (the main
north-south street, seen departing to the left) are both colonnaded and paved in blue
lias, while the minor streets (right) are paved in local limestone. Note how the paving
slabs are set at 45 degrees to the line of the road so that the ruts caused by wheeled traffic
do not enlarge the joints between them (Photo: A. Wilson).

constraints of topography as by the main street axis along the ridge. Cuicul,
like Thamugadi, is a Berber toponym, suggesting a pre-Roman settlement
here, and it is noteworthy that like so many of the other Berber settlements it
occupies a site on a spur delimited by ravines either side.
The areas of Roman cities in North Africa range from a few hectares up
to some 450+ ha for Lepcis Magna (Table 10.1). We have no direct evidence
for the size of their population, which can only be roughly estimated from
the areas, by applying approximate population densities. Evidence from
sites where housing has been extensively excavated suggest that at Sabratha
a range of 165–414 inhabitants/ha is plausible, giving a total population of
between 5,730 and 14,330, while Trajanic Timgad seems to have had c.3,550
inhabitants in 9.96 ha at a density of c.357/ha; the city’s later expansion may
have taken the population up to perhaps between 8,000 and 14,000 but at
a rather lower density per hectare.44 But in addition to these better known,
larger cities, there were numerous smaller cities, many developed from the

44
Wilson 2011a, 175–76.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 415

Table 10.1 Areas of selected large and

medium-sized cities in Roman North
Africa, c.second century AD

City Area (ha)

Lepcis Magna 452

Iol Caesarea 318
Ptolemais 217
Thelepte 180
Hadrumetum 155
Cyrene 123
Uthina 120
Utica 95
Ammaedara 61
Hippo Regius 60
Theveste 56
Tipasa 55
Ksar el-Guellal 53
Thamugadi 50
Tocra 41
Volubilis 40
Thapsus 39
Sufetula 38
Leptiminus 38
Sabratha 35
Cillium 31
Bulla Regia 31
Thugga 25
Thuburbo Maius 25

Numidian villages. These number two or three hundred, and by the early
third century AD had acquired some at least of the categories of public
buildings discussed below, even if their populations were never more than
a thousand or two. It is the near-ubiquity of these small but monumentally
impressive settlements that is one of the most striking features of North
African urbanism in the Roman period.

Architecture and Buildings

No two Roman cities are exactly alike, but they share a common architectural
language and the same types of public buildings. Usually a rectangular space,
the forum was the religious, political, and often also the commercial heart of

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 416 Andrew I. Wilson

the city. From the first century AD onwards fora were regularly paved, and in
several towns (e.g. Lepcis Magna, Hippo Regius) bronze letters set into the
paving record the generosity of the benefactor who paid for the work. One side
of the forum was frequently bounded by a basilica or law court, and overlooked
by one or more temples, dedicated to the protective deities of the city or to
Roma and Augustus (indeed, the forum at Lepcis Magna has three temples to
Roma and Augustus, to Hercules, and to Liber Pater), or to the Capitoline
Triad of Jupiter, Juno and Minerva (e.g. Thuburbo Maius). Contrary to what is
commonly assumed, however, Capitolia were not by any means a universal
phenomenon, and even where they do exist many post-date the foundation or
development of a city (for example, Cuicul, or Thugga). The Capitolium at
Timgad was built well outside the original colony and far from its forum
(Fig 10.6).45 Temples to other deities might be found at other points through-
out the city, or even on its outskirts: temples of Mercury Sobrius, for example,
seem to have been located on the edge of town on major routes, and may have
served also as the location of periodic markets or fairs.46 Specialised market
buildings or macella supplemented the commercial function of the forum; they
took the form of a square, or more usually, rectangular enclosure with stalls
around the edges and a circular or polygonal kiosk in the centre (the macellum
at Lepcis is unusual in having two central kiosks). The permanent fixtures of
the stalls consisted of little more than a stone counter between columns, and
a space behind for the stall holder to stand. In addition, a macellum usually had
a set of standard weights and measures for dry or liquid goods, set in a stone
table called a mensa ponderaria, that enabled city magistrates to check trading
standards. Roman-period inscriptions in Punic from Lepcis Magna record
public amenities (such as benches) set up by magistrates from the fines
imposed on market traders. Macella may have served in particular for the
sale of meat or fish, although a stone engraved with standard length measure-
ments from the macellum at Lepcis indicate that the goods sold included cloth,
and one of the stalls in the macellum at Timgad was selling children’s toys at
the time of the city’s destruction in the late fifth century AD. A noticeable
characteristic of Roman city development in North Africa as elsewhere was the
imbrication of shops and elite residences: many larger houses, and especially
those on the most frequented thoroughfares, had shops built into their street
frontages, which might be staffed by slaves, operated through freedmen, or
rented to other individuals.
A distinguishing feature of Roman cities, in North Africa as elsewhere,
was their dependence on long-distance aqueducts that tapped water

45 46
Quinn and Wilson 2013. Fentress 2007.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 417

Figure 10.8. The Severan nymphaeum (monumental fountain) at Lepcis Magna. A two-
storey columnar façade of coloured marble columns framed niches which originally
housed statues of gods and the imperial family (Photo: A. Wilson).

sources at some distance from the city. There has been some debate over
how necessary they really were, usually couched as a false opposition
between aqueducts as a utility and aqueducts as a luxury.47 In reality they
were simultaneously both: they celebrated a lavish command of water
resources through monumental fountains and in the thermae or baths,
sumptuous palaces for public bathing, but at the same time these fountains
(Fig. 10.8) provided the basic water supply for the majority of the urban
population, and supplemented with better and fresher supplies the water
stored in domestic cisterns. The fact that aqueducts, cisterns and wells, and
occasionally local springs, coexisted within the same settlement is not an
argument that aqueducts were redundant. The Romans were clearly alive
to the different qualities of water provided by different sources, and how
they might best be suited to different purposes. It is apparent that aque-
ducts both allowed a larger urban population to be sustained than would
have been possible otherwise, and that they facilitated a peculiarly Roman
lifestyle of public bathing and lavish public and private display of water in
ornamental fountains.48 In addition to the public baths, aqueducts enabled
the construction of public latrines flushed by (usually continuous) running

47
Leveau 1987; Leveau and Paillet 1976; 1983; Shaw 1984; 1991.
48
Wilson 1997; 1998; 2001b; 2012.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 418 Andrew I. Wilson

water; toilets were often included in public bath complexes (although

sometimes accessible directly from the street, without having to use the
baths), or adjacent to the forum, as at Timgad.49 Most Roman cities of any
size acquired one or more aqueducts, often in the course of the second
century AD; we have already seen that Carthage had several, as did for
example Thysdrus (two),50 Volubilis, Cirta, Timgad, and Berenice (three
each), and Meninx (four). Towns without an aqueduct are rare, and
perhaps the only certain example is Tiddis, a small settlement in
Numidia perched on a hill to which it would have been well-nigh impos-
sible to bring an aqueduct. Even here, substantial water supply works were
undertaken in AD 251 to create a water-collecting area just below the
summit of the hill that fed large reservoir cisterns supplying a small set
of public baths, a rare example in the empire of such an amenity fed
entirely by rainfall collection.
Entertainment buildings were usually (though not always) to be found
towards the edges of towns where space was less at a premium than in the
centre, although Timgad, with its theatre behind the forum, provides an
exception. Of the main types of building dedicated to entertainment,
theatres were much the most common, reflecting both their lesser con-
struction cost compared to other entertainment buildings, and, perhaps
more importantly, the lower cost of putting on theatrical performances
compared with gladiatorial shows or circus games. Amphitheatres were
less common than theatres, although the larger cities have them and over
38 are known from Africa Proconsularis alone.51 They were used for
displays of gladiatorial combat and for wild beast hunts, by professional
troupes of animal-hunters such as the Telegenii, who are recorded on
several inscriptions and mosaics.52 Several amphitheatres had tunnels
and cells below the arena from which animals could be winched up in
cages and released through trapdoors into the arena.53 Circuses were even
less common, and (with one exception) restricted to the larger cities:
Carthage, Iol Caesarea, Thysdrus, Lepcis Magna – and, oddly, Thugga,
a small town of perhaps 5,000 inhabitants. In general, entertainment
buildings broadly reflect a hierarchy of urban rank and significance,
suggesting different catchment areas for the different categories of enter-
tainment provided.

49
Wilson 2000; 2011b.
50
Cintas 1956, pace Shaw 1991, 81 who erroneously believed that it had none.
51
Lachaux 1979, 16, 156–57.
52
See, for example, the mosaic of Henchir Smirat: Beschaouch 1966.
53
See, for example, Carthage, Thysdrus (El Jem).

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 419

Figure 10.9. Aerial view of Sabratha, looking west. The theatre is clearly visible on the
left, with several blocks of housing excavated between it and the sea. The second major
area of excavation, in the upper centre of the picture, includes the forum and several
temple complexes, with the Seaward Baths by the shore, set off-axis to the main grid
(Wilson 2001a, 108, Fig. 1).

Outside the cities of the living stretched the cities of the dead, necropo-
leis lining the main routes out of town, with tombs large and small jostling
for the attention of the passer-by, in a competition of commemoration.
Roman cities in general exhibit an extraordinary high level of monu-
mental overhead – a vast mass of imposing public buildings considering
the level of population (Fig. 10.9).54 These buildings projected messages
about the fruits of empire, the stability and permanence of Roman rule, the
scale of imperial power, and the wealth of the local elites who had paid for
many of these construction projects. Cities in North Africa displayed all
these features, and the impact of their buildings was enhanced by the
abundance of local good quality stone, enabling many to be built in good
limestone or sandstone ashlar, in contrast to many domestic houses whose
superstructure was built in mudbrick or pisé (rammed earth). In particular,
it is remarkable how relatively small places put up monumental arches,

54
Wilson 2011a, 177.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 420 Andrew I. Wilson

structures with no practical utility, but which served to project an image of

prosperity and of loyalty to the empire (something apparently especially
frequent in the Severan period).55 Sometimes this was part of a campaign
to petition the emperor for an upgrade in civic status.
The monumentality of public buildings was further underscored by the
Roman practice of inscribing them prominently with the names of their
builders. Members of the local elite in many cases would promise to
construct a particular building if they were elected to office, spending
above and beyond the stipulated fee or summa honoraria that was levied
on magistrates, priests and other holders of municipal office. The built
landscape became an arena where competition between wealthy families
was played out, with the building inscriptions often specifying the sums
expended. This practice of public benefaction, or euergetism, had
Hellenistic roots, but became especially widespread under Rome: it har-
nessed to useful ends the politically competitive instincts of local elites,
who with incorporation into the Roman empire had become small fish in
a very large pond, and at the same time it devolved the costs of urban
development from the state onto rich local families.
Imperial involvement becomes more common in building projects of
the second century AD. Phrases like ex indulgentia (as at Verecunda in
Numidia, or occurring as the legend with the symbolic depiction of the
Carthage aqueduct on a Severan coin issue) may imply state aid to a town,
perhaps in the form of a remission of taxes. The words ex auctoritate
imperatoris or imperator . . . fieri iussit (‘by imperial authority’ or ‘the
emperor commanded it be done’) show that emperors, generally through
the intermediaries of their procurators or legates, either approved con-
struction or even commanded works to be carried out which were paid
for by the towns.56 The expense might be lightened by the use of military
labour, as with the aqueduct of Saldae.57 Cities financed many projects
from public money (pecunia publica), and individual euergetism became
very common, often in connection with the fees paid (summa honoraria)
for holding public office, and more rarely as a testamentary benefaction.
In return for their buildings, benefactors received the honour of a statue
erected at civic expense in a prominent public space – often the forum, or
the theatre – and this ‘statue habit’ caused cities progressively to accu-
mulate a sculpted population of stone and bronze honorands.

55 56 57
Wilson 2007b. Jouffroy 1986, 233. ILS 5795.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 421

The Economies of Roman Cities

The wealth of Roman Africa came principally from agriculture, and espe-
cially from its fertile fields of grain and its olive groves; it exported grain
and olive oil in large quantities, but there were also appreciable exports of
wine; and of salted fish and derived products (garum, liquamen, and allec,
fish sauces and paste). The wealth of the cities was founded on this rich
productive base, and it is clear that North African agriculture was suffi-
ciently productive to support a considerable population of city dwellers
who were not engaged in primary agriculture. But the cities too produced
goods and services: textiles, dyes, metalwork and craft goods. The coastal
cities also frequently acted as the bottling points for the olive oil and wine
exported from the interior of the province. This is not the place to reprise
the lengthy debate over whether the ancient city was a consumer city or
a producer city, a debate whose very terms are largely false because it
assumes that a city plus its territory formed a cellular self-sufficient unit,
ignoring the plentiful evidence for long-distance trade between cities.
Some cities do indeed seem to have acted as agrovilles, residential and
administrative centres a large part of whose population worked the sur-
rounding land – Volubilis, Madauros and site KS 022 in the Kasserine
Survey area seem to have played this role, on the evidence of the numerous
oil presses they contain.58 But the full extent, and indeed the nature, of
urban production remains difficult to grasp, largely because of evidential
problems connected with the size of production units, and the preservation
(or lack of it) in the archaeological record of diagnostic traces of produc-
tion, whether infrastructure or equipment, or the waste or residues from
production processes.59
Most production units in the ancient world were relatively small,
although in fact some large factories, and especially fish-salting factories,
do indeed survive in urban contexts (e.g. Lixus in Morocco). Establishing
that there was, or that there was not, significant production of a particular
item, or even of urban production in general, requires the investigation of
large swathes of an urban landscape in order to prove the existence – or the
absence – of numerous small production units.60 But in those cases where
excavation has been carried out extensively across large areas of Roman
cities – for example, Timgad between c.1880 and 1920, or Sabratha between
1923 and 1942 – the quality of excavation has been poor and finds record-
ing very limited; in most cases the key evidence of production debris will
have gone unnoticed or unrecorded. And yet in fact in both these cases
58 59 60
Hitchner 1990. Wilson 2002b. Wilson 2002b.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 422 Andrew I. Wilson

plentiful evidence was uncovered (some of it not recognised at the time) of

productive activities that required particular built infrastructure – prob-
able fulleries at Timgad (Fig. 10.6), and fish-salting vats at Sabratha.61 At
Sabratha, too, the evidence of crushed murex shells re-used as aggregate in
cement floors indicates the production of purple dye, and indeed the
dyeing of some textile fibres (probably wool) with it, somewhere at the
site. By contrast, modern excavation techniques usually cover too small an
area to establish whether or not there are multiple production units –
although with the aid of magnetometry and surface survey and pottery
collection, for example, clusters of kilns have been found at some sites:
coarseware kilns on the outskirts of Utica, and amphora and coarseware
kilns at Leptiminus and at Sullecthum (Salakta).62
Geophysical survey and surface sherd collection have been employed at
some sites to investigate the question of the extent and scale of some
production industries – at least, those that leave readily detectable produc-
tion waste. Besides the examples just cited, numerous kilns are indicated at
Ptolemais in Cyrenaica by kiln wasters and overfired pottery scattered
across parts of the site. Purple dye production is indicated by crushed
murex shell reused as aggregate in structural mortars at Lepcis Magna, and
large spreads of crushed murex shell at Berenice (Benghazi) and, most
spectacularly, at Meninx on Jerba, where a dump of crushed murex at least
500 × 300 m, and over 2.5 m deep, can still be seen, attesting the production
of purple dye on a massive scale and over a long period of time.63 Many of
the amphora sherds associated with the dump are from first- to third-
century AD forms, but the mention of an imperial dyeworks at the site in
the Notitia Dignitata, shows that production continued into the late
Roman period.64 While such non-invasive surface investigation techniques
potentially allow city-wide examination of the question of urban produc-
tion, they provide very little information on chronology, and of course are
limited to those activities which produce either clear geophysical signals, or
durable and recognisable production waste. Nonetheless, they do show
a mixture of activities at those sites at which they have been conducted:
pottery production and fish-salting at Utica; fish-salting, metalworking (of
imported iron ore), coarseware and amphora production (for fish
61
Timgad: Wilson 2001a. Sabratha: Wilson 1999; 2007a.
62
Utica: Ben Jerbania et al. 2015; Fentress et al. 2013. Leptiminus: Mattingly et al. 2011, 224–52.
Sullecthum: Peacock et al. 1989, 192–94.
63
Wilson 2002a; 2002b.
64
Notitia Dignitata, occidentis 11.70; Wilson 2002a. Excavation of parts of two buildings with
evidence for murex dye production shows that this activity continued into the fourth and fifth
centuries: Aït Kaci et al. 2009, 9–20, 215–16.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 423

products, wine and olive oil) at Leptiminus, fish-salting and amphora

production at Sullecthum; fish-salting, olive oil production and glasswork-
ing at Sabratha.65

Urban Systems and Networks

Cities controlled an area of territory around them, from which they drew
rents on land and farms owned directly by the city, and levied local taxes.
The territory of self-governing cities might include smaller dependent
towns without self-governing status. Control of territory was important:
Oea (Tripoli) and Lepcis Magna went to war with each other in AD 68–69
over a boundary dispute. The Roman road system networked cities to their
nearest neighbours and through them to larger and more important towns,
and ultimately the large administrative centres like Carthage.
The juridical statuses of Roman urban settlements, in North Africa as
elsewhere, do not map neatly onto differences in size or even in
monumentality.66 While the provincial capital (Carthage for Africa
Proconsularis, Sitifis for Mauretania Sitifensis, Iol Caesarea for
Mauretania Caesariensis, Tingis for Mauretania Tingitana; and, later,
Cirta for Numidia) was usually one of the largest cities of its province,
coloniae were not always larger than municipia, and some vici (small
towns) rivalled towns with a higher status. Political links can be traced
upwards from small towns to the provincial capital and even to Rome, as at
Dougga, some of whose magistrates also held political office at Carthage,
and one was adlected into the jury courts at Rome. Similarly, one of the
richest councillors at Timgad, Sertius, was a priest of the cult of the city of
Rome. Entertainers, theatrical players, gladiatorial troupes and venatores
(professional animal hunters for amphitheatre shows) will have travelled
their own circuits between towns to give performances; and just as the
catchment areas for different kinds of entertainment varied according to
which towns had circuses, amphitheatres, or theatres, so too would the
entertainment circuits travelled by these performers.
Many of the most prosperous and largest cities appear to have been
coastal ports. More systematic research would be needed to prove this as
a rule, but it is a strong impression. Their role as export points connected to
a wider Mediterranean network of markets was apparently a source of their
wealth. Amphora production at many of the large coastal cities shows that
they served as bottling points for olive oil and even wine, which was

65 66
Wilson 1999. Cf. Hanson 2016, 81–87.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 424 Andrew I. Wilson

produced inland and then exported through these cities together with the
salted fish products from seasonal catches of migratory species caught off
their coasts. Cooking wares produced at these coastal sites seem to have
piggy-backed on the export flows of amphora-borne products to reach
a wide distribution of overseas coastal markets around the western
Mediterranean, and to a lesser extent the eastern Mediterranean too.
The cities with their fora, macella, and streets lined with fixed shops or
tabernae forming a cluster of fixed retail outlets constituted permanent,
not merely periodic, markets. As such they must have had a different
relationship with their surrounding villages than had been the case in
a world of periodic markets; although there were in addition some periodic
markets (or nundinae) at towns – or on their outskirts – and at rural sites
too.67
Inland cities served as local markets for a rural economy based on
agriculture and livestock with a smaller radius of action. Sites in the
Tunisian Tell and Sahel within some 200 km of the coast seem to have
exported their oil and grain to coastal sites. Along with the agricultural
products went the shiny red table pottery (African Red Slip Ware), which
was exported around the Mediterranean from AD 90 onwards. By contrast,
cooking wares produced at sites away from the coast were not exported and
had only a limited distribution. Further west and further away from the
coast (in what is modern Algeria), the inland Numidian table pottery
industries had only a limited distribution, over a radius of up to c.35 km.68
As with the Numidian period, the larger towns of the interior were in the
major grain-growing plains.69 Using the example of Sétif, an inland town
in Numidia, Fentress identifies two economies: a regionally self-sufficient
economy in which the town played the part of a regional market for surplus
crops, largely wheat, and for locally made goods in leather, pottery and
cloth; and a second economy based on the surrounding imperial estates
and the grain market. Tax grain collected by the State and the rent in kind
from tenants (coloni) on imperial estates were collected in horrea (gran-
aries) and transported to coastal ports from where it was exported to Rome.
It was not until the fourth century that the two economies merged, when
the military response to Moorish raids and incursions saw a growing body
of administrative and military personnel stationed in the town, with cash
income for the city derived both from these administrative salaries and the
opportunities for profiting from supplying the army.70
67
Fentress 2007; Shaw 1981. 68 Fentress 2013, 332.
69
Fentress 2006, 21, speaking of the Numidian towns in the fourth–second centuries BC.
70
Fentress 1990.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 425

For the cities in the immediate vicinity of large troop concentrations the
supply of the army will have been an important feature of their economy.
The small town of Diana Veteranorum was no doubt involved in the
supply of the legionary fortress at Lambaesis until the legio III Augusta
was disbanded in the mid-third century; thereafter, Fentress suggests that it
supplied units stationed south of the Aurès mountain range.71 Timgad was
presumably engaged in supplying the legionary base at Lambaesis, 19 km to
the west, with food over and above the local market. A large number of
fulleries (22) and a cloth market (mid-fourth century) suggest that by the
late Roman period the town had become a major centre for textile produc-
tion, playing a coordinating role in turning the wool from the livestock
economy of the surrounding countryside into textiles. Thelepte in the
Tunisian steppe may have played a similar role, as Hitchner suggests that
enclosures on its outskirts may have been yards for livestock.72
Unsurprisingly, the density of cities decreases towards the southern
margins of the Roman African provinces, as more arid environments
make it harder (though not impossible) to support large nucleated popula-
tions. As Chapter 5 above has demonstrated, in the frontier zone, the major
settlements were at oasis centres, and these often became the sites of
frontier forts.73 But while we tend to think of these as oasis forts, with
military vici (settlements) around them, in all probability some kind of oasis
settlement existed before the forts (this was certainly the case at Cidamus/
Ghadamis). The case of al-Qurayyat al-Gharbia, where the Severan fort
built in the early third century was abandoned c.275–280 and reoccupied in
the fourth and early fifth centuries, but with occupation continuing until
c.540 possibly under a Libyan chieftain, and also at times (whether con-
tinuous or not) during the Medieval period, shows that the viability of the
site was not entirely dependent on the presence of a Roman military unit.74
Rather, the oasis marked the northern end of the longest waterless stretch
on the Saharan route down to Fazzan. Little is known in any detail of the
nature and morphology of these oasis settlements, although the 18-ha
settlement around the fort of Bu Nijim shows that some could be sizeable
(Fig. 10.10). The vici at Bu Nijim and al-Qurayyat al-Gharbia had temples
and shrines, and a bathhouse, although no other kinds of public buildings
are known. Ostraca from the fort of Bu Nijim suggest that here the oasis
centre acted as a major terminus for camel caravans coming up from the
Sahara via the Jufra oasis group further south.75

71
Fentress 2013. 72 Hitchner 1994, 39–40. 73 Mattingly et al., Chapter 5, this volume.
74
Mackensen 2012; Mackensen et al. 2010, 379–85. 75 Marichal 1992.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 426 Andrew I. Wilson

Figure 10.10. Satellite image of the Roman fort of Bu Nijim and its surrounding vicus or
settlement, before encroachment by modern intensive olive culture between 2006 and
2012 (imagery: Google, DigitalGlobe, 24 March 2006).

Late Antique Urbanism

Late antiquity, and especially the Vandal and Byzantine periods in North
Africa, saw profound transformations in the appearance and roles of cities.76
From the mid-third century onwards there was a marked decline in euerget-
ism and the construction of public buildings. A resurgence of activity in the
360s and 370s, especially in Numidia, is to a considerable extent accounted
for by repairs (although there was new construction too), but whether this
was prompted by a period of intense seismic activity or reflects a bout of
fourth-century prosperity remains an open question. New construction
effort was indeed channelled into churches, but it is hard to see this before
the fifth century, and even then this activity was on a much smaller aggregate
scale than the building activity of the second century. The argument that by
the third century most cities had all the public buildings they needed is not
wholly persuasive (more temples, to other deities, could always be built), and
does not explain the decay or change in use of some buildings that already
existed. In cities with more than one set of large public baths we find that in

76
See in particular Leone 2007. Sears 2007 places the accent on continuity rather than decline, and
argues that cities were ‘thriving’ into the fifth century, a view achieved largely by ignoring
contrary evidence.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 427

Figure 10.11. Sufetula (Sbeïtla): Late antique building with two oil presses, attached to
the church of Saints Gervasius and Tryphon (out of the picture, to the right), built
across and entirely blocking a Roman street (Photo: A. Wilson).

late antiquity (by the later fourth century) these had been renamed ‘Summer
Baths’ and ‘Winter Baths’, the ‘Winter Baths’ always being the smaller and
easier to heat.77 Civic finances no longer ran to heating two sets of baths, and
the second-century urban lifestyle could no longer be funded. Conversely,
though, we see some of the richer private houses enlarging in the third and
fourth centuries, sometimes with lavish apsidal or tri-apsidal reception halls.
The reasons for these changes were multiple and not entirely evident, but
clearly involved fiscal pressure. They included: the military and economic
crises of the mid- to late third century, whose effect on the imperial finances
had repercussions on local taxation; reduced autonomy of town councils
and a consequent reluctance of local elites to undertake the euergetic
burdens of municipal office, preferring to expend their wealth on their
own dwellings; depopulation caused by the ‘Plague of Cyprian’ (AD
251–270),78 and later, the Justinianic Plague of the 540s; and changing
religious practices. In the space left by the retreat of town councils, the
church increasingly insinuated itself into the expanding administrative
vacuum. At Lepcis Magna a fifth-century church and baptistery were built
77
For example, at Madauros, Sufetula, and Thuburbo Maius: Nielsen 1992, 138–40. This
phenomenon is also seen in the Eastern Mediterranean, as for example at Aphrodisias: Wilson
2016, 193.
78
Harper 2015.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 428 Andrew I. Wilson

Figure 10.12. Sabratha – late antique graves (in the foreground) in a street by the church
to the north of the theatre (Photo: A. Wilson).

in the forum, and civil basilicas or lawcourts were turned into churches at
Sabratha and, after the Justinianic conquest, at Lepcis Magna. The growing
power of the church in controlling civic space is suggested, for example, by
the double oil-press building attached to the church of Saints Gervasius and
Tryphon at Sufetula, built across and entirely blocking a street (Fig. 10.11).
The new dominance of Christianity saw a marked change in the location of
cemeteries: martyrs’ shrines, then churches, grew up around the graves of
martyrs in cemeteries, and increasingly people wanted to be buried ad
sanctos, next to the holy martyrs.79 As churches became spatially associated
with graves, the taboo on intramural burial broke down and we find burials
around churches in city centres, for example around the former civic basilica-
turned-church at Sabratha (Fig. 10.12), or in the forum at Lepcis Magna.
In the course of the fourth century there were repeated attempts to close
the pagan temples (repeated because initially unsuccessful in many regions).
In North Africa the chronology of this process remains unclear, but it must
have been largely complete by the end of the century. We find inscriptions
recording the transfer of statues ex squalidis locis, from dilapidated build-
ings (literally, ‘from filthy places’) to more public areas, and these may in

79
Yasin 2009, Chapter 2.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 429

some cases relate to the salvage of statuary from temples, although not,
presumably, the unacceptable cult images. At Thuburbo Maius, at some
undefined time in late antiquity, an olive press was installed in the basement
of the Capitolium – reflecting a more widespread phenomenon by which
agricultural processing equipment, already prominent in some cities in the
mid-Roman period (Volubilis, Madauros), was installed in formerly public
spaces in late antiquity (for example, Lepcis Magna, where a late antique oil
press is found in the gymnasium adjoining the Hadrianic Baths).
The end of the Roman aqueducts came later, although again the timing
is uncertain. Numerous fourth-century inscriptions commemorate the
repair of aqueducts,80 and even the construction of new ones.81 Many
aqueducts may have functioned until the Arab conquest: Carthage’s aque-
duct was still functioning until it was cut in the siege of 682, which saw
Carthage finally fall to the Arabs. At Leptiminus, further south along the
Tunisian coast, the eastern aqueduct was still functioning normally until
the 640s, when it was blocked immediately downstream of the reservoir
cisterns for the eastern baths, which thus received the entire flow of the
aqueduct. The baths themselves, however, were no longer serving their
original purpose: a group of amphora kilns had been built in them, and the
water was presumably therefore used for pottery production.82
Demographic decline is apparent, in North Africa as in much of the rest
of the Western Roman empire after the third century, from the shrinking
areas of cities. Lepcis Magna contracted progressively in late antiquity. The
early imperial defensive perimeter that had enclosed c.452 ha in the first
and second centuries AD was abandoned, perhaps in the fourth century,
and replaced by a defensive wall built of spolia and incorporating the arch
of Marcus Aurelius as a gateway, enclosing a total of 143 ha. In the sixth
century AD the city had further shrunk: the Byzantine fortifications
enclosed just 16.9 ha, defending the harbour, the Old Forum, and the
Severan Forum.83 A similar process of contraction happened at Sabratha,
where the Byzantine walls enclose an area around the forum and the
harbour, leaving most of the Roman city outside them. How far this
demographic contraction may have been exacerbated by the Justinianic
Plague remains an open question, as we have little direct evidence for its

80
For example, ILAlg. 1.296 Calama; CIL 8.27818 Sidi Achmed el Hachemi; ILTun 622 Henchir
Haouli, AD 339/350; CIL 8.18700 = 4766 Ksour el-Ahmar, AD 305; AE 1899, 216 Mascula, AD
367/375.
81
Cirta: ILAlg. 2.619 = ILS 5789 = CIL 8.7034, and AE 1902, 166, both dating to AD 388/392;
Henchir el-Left: AE 1949, 49 = BAC 1947, 376, AD 321/324.
82
Stirling et al. 2001. 83 Mattingly 1995, 117, figure 6.1; Wilson 2011a, 167.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 430 Andrew I. Wilson

Figure 10.13. Sullecthum (Salakta) – late defensive enclosure, belonging probably to the
Vandal period (late fifth or early sixth century AD) (Photo: A. Wilson).

effect on North Africa, although it is hard to believe that the region escaped
its impact.
Increased fortification is an especially striking feature of these late
antique towns. While some Roman towns had had wall circuits – Timgad
at its foundation, although the wall circuit had already been demolished by
the Severan period – and Tipasa in the second century, most towns were
unfortified in the second and even third centuries AD under the pax
Romana. But – at a still largely undetermined date – fortifications were
increasingly deemed necessary in late antiquity. Carthage was fortified
with a rampart under Theodosius, c.425, presumably in response to the
Vandal threat.84 But even before the Vandal conquest, and increasingly
after it, raids by Saharan tribesmen prompted the construction of fortifica-
tions at cities in the south of Roman Africa.85 At Sullecthum, a crude
defensive wall is probably to be identified with the fortification wall men-
tioned by Procopius that was built between houses on the edge of the city to
defend the town against Berber raids before the Byzantine reconquest
in AD 533 (Fig. 10.13). A similar style of fortification has been identified
at Leptiminus.86 Blockhouses, of unknown date, within the former urban
zone are found at towns in southern Tunisia and also in Cyrenaica. At

84
Wells 1980. 85 For these Moorish raids, see Fentress and Wilson 2016.
86
Fentress and Wilson 2016, 43–44. Leptiminus: Wilson et al. 2011, 536–38.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 431

Sufetula we find blockhouse construction in the outlying parts of the town,

with a fortified wall around the forum, while at Ptolemais the 3-km long
Hellenistic wall circuit was actually torn down, presumably because there
were insufficient forces to man it properly, and a number of small forts
built within the city area. These are in addition to the so-called ‘Fortress of
the Dux’, a Byzantine fort built certainly no later than the reign of
Anastasius (491–518) and possibly already by the early fifth century AD.87
During the century between the Byzantine reconquest and the first Arab
incursions of the AD 640s, forts were constructed at nearly every town in
Byzantine Africa.88 These ranged from small blockhouse-type affairs incorpor-
ating Roman honorific arches, for example at Diana Veteranorum or
Ammaedara, to large forts, with projecting corner towers, that protected
water sources, as at Timgad or Ksar Lemsa (Limisa). The distribution of
Byzantine fortifications across the former Roman territory throughout what
is now Tunisia and eastern Algeria makes it meaningless to talk of a Byzantine
limes – it was evidently expected that raids might penetrate deep into Byzantine
territory, and that the best way of holding out against them was to retreat into
fortified blockhouses, or reduced wall circuits, until the immediate threat had
passed.

Conclusions

The Roman period stands out as a phase of extraordinary urbanisation

across the Maghrib, during which levels of urban population were achieved
that were not equalled again until the late nineteenth or early twentieth
century. It is now evident that this urbanisation built on a local tradition of
Numidian villages and early towns, in addition to the much better known
Punic colonies of the coast. Africa’s famous agricultural prosperity is clearly
reflected in the monumental grandeur of the Roman public buildings and
the confident infrastructure of the aqueducts, striding across a pacified
landscape – a water supply technology suited to the pax Romana, but
vulnerable to disruption in the more turbulent centuries following the
Vandal conquest. Urban life declined under Vandal rule, and even further
in the later sixth and the seventh centuries; and after the Arab conquest of
the seventh century, the Medieval Maghrib developed an urban system
organised around new centres – either developed out of minor Roman

87 88
Reynolds 1976, 248–49. Pringle 1981.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 432 Andrew I. Wilson

towns like Tunis, or entirely new foundations like the holy city of Kairouan,
founded by the Umayyads c.AD 670.

References

Aït Kaci, A., Drine, A., Fentress, E.W.B., Morton, T.J., Rabinowitz, A. and
Wilson, A.I. 2009. The excavations. In E. Fentress, A. Drine and R. Holod
(eds), An Island through Time: Jerba Studies 1. The Punic and Roman Periods.
Portsmouth, Rhode Island: Journal of Roman Archaeology Supplement 71,
212–40.
Ben Jerbania, I., Fentress, E., Ghozzi, F., Wilson, A.I., Carpentiero, G., Dhibi, C.,
Dufton, J.A., Hay, S., Jendoubi, K., Mariotti, E., Morley, G., Oueslati, T.,
Sheldrick, N. and Zocchi, A. 2015. Excavations at Utica by the Tunisian-British
Utica Project 2014. Unpublished Field Report: University of Oxford.
Bennett, P., Wilson, A.I., Buzaian, A.M. and Kattenberg, A. 2004. The effects of
recent storms on the exposed coastline of Tocra. Libyan Studies 35: 113–22.
Beschaouch, A. 1966. La mosaïque de la chasse à l’amphithéâtre découverte à
Smirat en Tunisie. Comptes rendus de l’Académie des inscriptions et belles-
lettres 1966: 134–57.
Boardman, J. and Hayes, J. 1966. Excavations at Tocra 1963–1965. The Archaic
Deposits, vol. 1. London: British School of Archaeology at Athens.
Bomgardner, D.L. 1989. The Carthage amphitheater: A reappraisal. American
Journal of Archaeology 93.1: 85–103.
Broise, H. and Thébert, Y. 1993. Recherches archéologiques franco-tunisiennes à
Bulla Regia. II, Les architectures. Rome: Collection de l’École française de
Rome, 28.2.
Buzaian, A.M. and Lloyd, J.A. 1996. Early urbanism in Cyrenaica: New evidence
from Euesperides (Benghazi). Libyan Studies 27: 129–52.
Cherstich, L. 2008. From looted tombs to ancient society: A survey of the Southern
Necropolis of Cyrene. Libyan Studies 39: 73–93.
Cintas, J. 1956. L’alimentation en eau de Thysdrus dans l’antiquité. Karthago.
Revue d’archéologie africaine 7: 179–87.
Fantar, M. 1984–1986. Kerkouane. Cité punique de Cap Bon (Tunisie). 3 vols.
Tunis: Persée.
Fantar, M. 1998. Kerkouane: A Punic Town in the Berber Region of Tamezrat : VIth
to IIIrd Century BC. Translated by J. McGuinness. Tunis: Alif.
Fentress, E.W.B. 1990. The economy of an inland city: Sétif. In L’Afrique dans
l’occident romain Ier siècle av. J.-C. – IVe siècle ap. J.C. Actes du colloque organisé
par l’École française de Rome sous la patronage de l’Institut national d’archéologie
et d’art de Tunis (Rome 3–5 décembre 1987). Rome: Collection de l’École
française de Rome 134, 117–28.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 433

Fentress, E.W.B. 2006. Romanizing the Berbers. Past and Present 190: 3–33.
Fentress, E.W.B. 2007. Where were North African nundinae held? In C. Gosden,
H. Hamerow, P. de Jersey and G. Lock (eds), Communities and Connections:
Essays in Honour of Barry Cunliffe. Oxford: Oxford University Press, 125–41.
Fentress, E.W.B. 2009. The Punic and Libyan towns of Jerba. In S. Helas and
D. Marzoli (eds), Phönizisches und punisches Städtewesen. Mainz: Iberia
Archaeologica 13, 203–19.
Fentress, E.W.B. 2013. Diana Veteranorum and the dynamics of an inland
economy. Late Antique Archaeology 10.1: 315–42. doi:http://dx.doi.org/
10.1163/22134522-12340035 [last accessed 13 September, 2019].
Fentress, E.W.B., Ghozzi, F., Quinn, J., Wilson, A., Anastasi, M., Hobson, M.,
Leitch, V., Morley, G., Ray, N. and Rice, C. 2013. Excavations at Utica by the
Tunisian-British Utica Project 2012. Unpublished Field Report: University of
Oxford.
Fentress, E.W.B. and Wilson, A.I. 2016. The Saharan Berber diaspora and the
southern frontiers of Byzantine North Africa. In S.T. Stevens and J.P. Conant
(eds), North Africa under Byzantium and Early Islam, Washington, D.C.:
Dumbarton Oaks Byzantine Symposia and Colloquia 7, 41–63.
Fraser, P.M. 1951. An inscription from Euesperides. Bulletin de la Société royale
d’Archéologie d’Alexandrie 39: 132–43.
Göransson, K. 2004. Transport amphorae from Euesperides (Benghazi), Libya.
A presentation of preliminary results. In J. Eiring and J. Lund (eds), Transport
Amphorae and Trade in the Eastern Mediterranean. Athens: Monographs of the
Danish Institute at Athens, 137–42.
Hanson, J.W. 2016. An Urban Geography of the Roman World, 100 BC to AD 300.
Oxford: Archaeopress Roman Archaeology, 18.
Harper, K. 2015. Pandemics and passages to late antiquity: Rethinking the plague
of c.249–270 described by Cyprian. Journal of Roman Archaeology 28: 223–60.
doi:10.1017/S1047759415002470.
Hay, S., Fentress, E.W.B., Kallala, N., Quinn, J. and Wilson, A.I. 2010.
Archaeological fieldwork reports: Utica. Papers of the British School at Rome
78: 325–29.
Hitchner, R.B. 1990. The Kasserine archaeological Survey. 1986 [Institut national
d’archéologie et d’art de Tunisie-University of Virginia, USA]. Antiquités afri-
caines 26.1: 231–59.
Hitchner, R.B. 1994. Image and reality: The changing face of pastoralism in the
Tunisian High Steppe. In J. Carlsen, P. Ørsted and J.E. Skydsgaard (eds),
Landuse in the Roman Empire. Rome: Analecta romana Instituti danici
Supplement, 27–43.
Hurst, H.R. 1993a. Excavations in the southern part of the Carthage harbours,
1992–1993. CEDAC Carthage Bulletin 13: 10–19.
Hurst, H.R. 1993b. Le port militaire de Carthage. Histoire et archéologie. Les
dossiers [Paris] 183: 42–51.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 434 Andrew I. Wilson

Jouffroy, H. 1986. La construction publique en Italie et dans l’Afrique: Études et

Travaux. Strasbourg: Association pour l’étude de la civilisation romaine.
Kallala, N., Fentress, E.W.B., Quinn, J., Wilson, A.I., Ben Slimane, W., Booms, D.,
Friedman, H., Ghozzi, F., Hay, S. and Jerray, E. 2010. Survey and Excavation at
Utica 2010. Unpublished Field Report: University of Oxford.
Kane, S. and White, D. 2007. Recent developments in Cyrene’s chora south of the
Wadi bel Gadir. Libyan Studies 38: 39–52. doi:10.1017/S0263718900004234.
Lachaux, J.C. 1979. Théatres et amphithéatres d’Afrique Proconsulaire. Aix-en-
Provence: Edisud.
Lancel, S. 1995. Carthage: A History. Oxford: Blackwell.
Leone, A. 2007. Changing Townscapes in North Africa from Late Antiquity to the
Arab Conquest. Bari: Edipuglia.
Leveau, P. 1987. A quoi servaient les aqueducts romains? L’Histoire 105: 96–104.
Leveau, P. and Paillet, J.L. 1976. L’alimentation en eau de Caesarea de Maurétanie
et l’aqueduc de Cherchell. Paris: Éditions l’Harmattan.
Leveau, P. and Paillet, J.L. 1983. Alimentation en eau et développement urbain à
Caesarea de Maurétanie. In J.P. Boucher (ed.), Journées d’études sur les aqueducs
romains. Tagung über römische Wasserversorgungsanlagen, Lyons, 26–28 mai,
1977. Paris: Société d’édition ‘Les Belles Lettres’, 231–34.
Lézine, A. 1968. Carthage. Utique. Études d’architecture et d’urbanisme. Paris:
CNRS.
Lloyd, J.A., Buzaian, A.M. and Coulton, J.J. 1995. Excavations at Euesperides
(Benghazi), 1995. Libyan Studies 26: 97–100.
Lloyd, J.A., Bennett, P., Buttrey, T.V., Buzaian, A., El Amin, H., Fell, V.,
Kashbar, G., Morgan, G., Ben Nasser, Y., Roberts, P.C., Wilson, A.I. and
Zimi, E. 1998. Excavations at Euesperides (Benghazi): An interim report on
the 1998 season. Libyan Studies 29: 145–68.
Luni, M. 2001. Le temple dorique hexastyle dans le sanctuaire découvert hors de la
porte sud à Cyrène. Comptes rendus de l’Académie des inscriptions et belles-lettres
2001: 1533–52.
Luni, M. 2006. Un demi-siècle de recherches archéologiques à Cyrène. Comptes
rendus de l’Académie des Inscriptions et Belles-Lettres 2006: 2173–201.
Mackensen, M. 2012. New fieldwork at the Severan fort of Myd (. . .)/Gheriat el-
Garbia on the limes Tripolitanus. Libyan Studies 43: 41–60. doi:10.1017/
S0263718900000054.
Mackensen, M., Schimmer, F., Schmid, S., Seren, S., Stephani, M. and Weber, M.
2010. Das severische Vexillationskastell Myd (–)/Gheriat el-Garbia am limes
Tripolitanus (Libyen). Bericht über die Kampagne 2009. Mitteilungen des
Deutschen Archäologischen Instituts, Römische Mitteilungen 116: 363–458.
Marichal, R. 1992. Les ostraca de Bu Njem. Tripoli: Department of Antiquities,
Suppléments de Libya Antiqua 7.
Mattingly, D.J. 1995. Tripolitania. London: Batsford.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 435

Mattingly, D.J. 2016. Who shaped Africa? The origins of urbanism and agriculture
in Maghreb and Sahara. In N. Mugnai, N. Ray and J. Nikolaus (eds), De Africa
Romaque: Merging Cultures Across North Africa. London: Society for Libyan
Studies, 11–25.
Mattingly, D.J., Stone, D.L., Stirling, L.M., Moore, J.P., Wilson, A.I., Dore, J.N. and
Ben Lazreg, N. 2011. Economy. In Stone et al. 2011, 205–71.
Nielsen, I. 1992. Thermae et Balnea. The Architecture and Cultural History of
Roman Public Baths, 2 vols. 2nd ed. Aarhus: Aarhus University Press.
Peacock, D.P.S., Bejaoui, F. and Belazreg, N. 1989. Roman amphora production in
the Sahel region of Tunisia. In M. Lenoir, D. Manacorda and C. Panella (eds),
Amphores romaines et histoire économique. Dix ans de recherche. Actes du
colloque de Sienne (22–24 mai 1986). Rome: Collection École française de
Rome, 179–222.
Peña, J.T. 1998. The mobilization of state olive oil in Roman Africa: The evidence
of late 4th-c. ostraca from Carthage. In Carthage Papers: The Early Colony’s
Economy, Water Supply, a Private Bath, and the Mobilization of State Olive Oil.
Portsmouth, Rhode Island: Journal of Roman Archaeology Supplementary
Series, 117–238.
Pringle, D. 1981. The Defence of Byzantine Africa from Justinian to the Arab
Conquest, 2 vols. Oxford: BAR International Series.
Quinn, J. and Wilson, A.I. 2013. Capitolia. Journal of Roman Studies 103: 117–73.
Reynolds, J.M. (ed.). 1976. Libyan Studies. Select Papers by R. G. Goodchild.
London: Elek.
Saumagne, C. 1928. Notes de topographie carthaginoise: I, La «Turris Aquaria»
(d’après les fouilles de 1926); II, Les vestiges de la colonie de C. Gracchus à
Carthage. Bulletin achéologique du Comité des Travaux Historiques et
Scientifiques 1928: 629–64.
Sears, G. 2007. Late Roman African Urbanism: Continuity and Transformation in
the City. Oxford: BAR International Series 1693.
Shaw, B.D. 1981. Rural markets in North Africa and the political economy of the
Roman Empire. Antiquités africaines 17: 37–83.
Shaw, B.D. 1984. Water and society in the ancient Maghrib: Technology, property
and development. Antiquités africaines 20: 121–73.
Shaw, B.D. 1991. The noblest monuments and the smallest things: Wells, walls and
aqueducts in the making of Roman Africa. In A.T. Hodge (ed.), Future Currents
in Aqueduct Studies. Leeds: Francis Cairns, 63–91.
Smith, D.J. and Crow, J. 1998. The Hellenistic and Byzantine defences of Tocra
(Taucheira). Libyan Studies 29: 35–82.
Stirling, L., Mattingly, D.J. and Ben Lazreg, N. 2001. The East Baths and their
industrial re-use in late antiquity: 1992 excavations. In L.M. Stirling, D.J.
Mattingly and N. Ben Lazreg (eds), Leptiminus (Lamta): A Roman Port City in
Tunisia. Report no. 2. Ann Arbor: Journal of Roman Archaeology
Supplementary series 41, 29–74.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 436 Andrew I. Wilson

Stone, D.L., Mattingly, D.J. and Ben Lazreg, N. (eds). 2011. Leptiminus (Lamta),
Report no. 3: The Field Survey, vol. 3. Portsmouth, Rhode Island: JRA
Supplementary Series 87.
Swift, K. 2006. Classical and Hellenistic Coarse Pottery from Euesperides (Benghazi,
Libya): Archaeological and Petrological Approaches to Production and Inter-
Regional Distribution. Unpublished DPhil thesis, School of Archaeology,
University of Oxford.
Thomas, E. 2007. Monumentality and the Roman Empire: Architecture in the
Antonine Age. Oxford: School of Archaeology Monograph.
Wells, C.M. 1980. Carthage. The Late Roman defences. In W. Hanson and
L. Keppie (eds), Roman Frontier Studies 1979. Oxford: BAR International series,
999–1004.
Wells, C.M. 2005 A cuckoo in the nest: The Roman Odeon at Carthage in its urban
context. American Journal of Ancient History ns 3–4 [2007]: 131–42.
White, D. 1984. The Extramural Sanctuary of Demeter and Persephone at Cyrene,
Libya. Final Reports, vol. 1. Philadelphia: University Museum monographs.
White, D. 1993. The Extramural Sanctuary of Demeter and Persephone at Cyrene,
Libya. Final Reports, vol. 5. Philadelphia: University Museum monographs.
Wilson, A.I. 1997. Water Management and Usage in Roman North Africa: A Social
and Technological Study. Unpublished DPhil thesis, University of Oxford.
Wilson, A.I. 1998. Water supply in ancient Carthage. In Carthage Papers: The Early
Colony’s Economy, Water Supply, a Private Bath, and the Mobilization of State
Olive Oil. Portsmouth, Rhode Island: Journal of Roman Archaeology
Supplement, 65–102.
Wilson, A.I. 1999. Commerce and industry in Roman Sabratha. Libyan Studies 30:
29–52.
Wilson, A.I. 2000. Incurring the wrath of Mars: Sanitation and hygiene in Roman
North Africa. In G.C.M. Jansen (ed.), Cura Aquarum in Sicilia. Proceedings of
the Tenth International Congress on the History of Water Management and
Hydraulic Engineering in the Mediterranean Region. Syracuse, May 16–22,
1998. Leuven: Peeters, 307–12.
Wilson, A.I. 2001a. Timgad and textile production. In D.J. Mattingly and J. Salmon
(eds), Economies beyond Agriculture in the Classical World. London: Routledge,
271–96.
Wilson, A.I. 2001b. Urban water storage, distribution and usage in Roman North
Africa. In A.O. Koloski-Ostrow (ed.), Water Use and Hydraulics in the Roman
City. Boston, MA: Archaeological Institute of America Colloquia and
Conference Papers, New series, 83–96.
Wilson, A.I. 2002a. Marine resource exploitation in the cities of coastal
Tripolitania. Africa Romana 14: 429–36.
Wilson, A.I. 2002b. Urban production in the Roman world: the view from North
Africa. Papers of the British School at Rome 70: 231–73.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 10 Mediterranean Urbanisation in North Africa 437

Wilson, A.I. 2005. Une cité grecque de Libye: Fouilles d’Euhésperidès (Benghazi).
Comptes rendus de l’Académie des inscriptions et belles-lettres 2003: 1648–75.
Wilson, A.I. 2007a. Fish-salting workshops in Sabratha. In L. Lagóstena, D. Bernal
and A. Arévalo (eds), Cetariae 2005. Salsas y Salazones de Pescado en Occidente
durante la Antigüedad. Actas del Congreso Internacional (Cádiz, 7–9 de noviem-
bre de 2005). Oxford: BAR International Series 1686, 173–81.
Wilson, A.I. 2007b. Urban development in the Severan Empire. In S.C.R. Swain, S.
J. Harrison and J. Elsner (eds), Severan Culture. Cambridge: Cambridge
University Press, 290–326.
Wilson, A.I. 2011a. City sizes and urbanization in the Roman Empire. In
A. Bowman and A.I. Wilson (eds), Settlement, Urbanization, and Population.
Oxford: Oxford University Press, 161–95.
Wilson, A.I. 2011b. Urination and defecation Roman-style. In G.C.M. Jansen, A.
O. Koloski-Ostrow and E.M. Moormann (eds), Roman Toilets: Their
Archaeology and Cultural History. Leuven: BABesch Supplement 18, 95–111.
Wilson, A.I. 2012. Water, power and culture in the Roman and Byzantine worlds:
An introduction. Water History 4.1: 1–9. doi:10.1007/s12685-012-0050-2.
Wilson, A.I. 2013. Trading across the Syrtes: Euesperides and the Punic world. In
J. Quinn and J. Prag (eds), The Hellenistic West: Rethinking the Ancient
Mediterranean. Cambridge: Cambridge University Press, 120–56.
Wilson, A.I. 2016. The Olympian (Hadrianic) baths at Aphrodisias: Layout, opera-
tion, and financing. In R.R.R. Smith, J. Lenaghan, A. Sokolicek and K. Welch
(eds), Aphrodisias Papers 5: Excavation and Research at Aphrodisias, 2006–2012.
Portsmouth, Rhode Island: Journal of Roman Archaeology Supplementary
Series 103, 168–94.
Wilson, A.I., Mattingly, D.J., Stone, D.L., Stirling, L.M., Dodge, H. and Ben
Lazreg, N. 2011. Gazetteer of sites in the urban survey. In Stone et al. 2011,
495–628.
Xella, P., Quinn, J., Melchiorri, V. and van Dommelen, P. 2013. Cemetery or
sacrifice? Infant burials at the Carthage Tophet: Phoenician bones of contention.
Antiquity 87.338: 1199–207.
Yasin, A.M. 2009. Saints and Church Spaces in the Late Antique Mediterranean.
Architecture, Cult, and Community. Cambridge: Cambridge University Press.

https://doi.org/10.1017/9781108637978.011 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian
High Tell
joan sanmartí, nabil kallala, maria carme belarte,
joan ramon, francisco josé cantero, dani lópez,
marta portillo and sílvia valenzuela

Introduction

Il faudrait donc, en collaboration avec les archéologues et les

préhistoriens, nous interroger sur les processus qui ont fait naître sur la
surface du globe des hiérarchies nouvelles de statuts et de pouvoirs entre
des groupes sociaux qui entretenaient toujours des rapports de parenté au
sein d’une même unité sociale globale.1

For many archaeologists this quote from Godelier encapsulates one of the
main goals of our discipline. The formation of states and cities (one of the
elements normally accompanying this form of societal organisation) can-
not therefore be limited to the study of individual cases. Whatever the
specific differences, these cases may, when interpreted under the light of
well-founded hypothetical models, constitute a fundamental part in under-
standing the intercultural processes in the creation of new hierarchies.
However, there is not widespread agreement on how this may take place.
For some of our colleagues, the notion of ‘state’ (and perhaps the notion of
‘city’ as well?) is just a ‘Western’ construction, which does not have any
significance in many other parts of the world. According to this construc-
tivist point of view, the state simply did not exist in these areas.
Consequently, this notion should be eliminated from our discourse,
which should focus on the analysis of local and regional realities from
a strictly emic perspective (if we may use a term so heavily connoted by
modernist thought). For us, however, reality is not constituted (but just
designated) by words or texts, nor is it the particular vision of those who
live it, either in the past or in the present. On the contrary, we understand
that it has a completely independent existence, which can be analysed and
1
438 Godelier 1980, 662.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 439

can be objectively known (with limitations), whenever appropriate tools

and procedures are used. States, consequently, may theoretically exist
anywhere. That being said, it is true that the lack of a clear definition
about what a state actually is among historians, archaeologists and anthro-
pologists inevitably hinders any debate that is focused on this concept. For
Lull and Risch, for example, evidence of private property and elite control
over the means of production is in itself sufficient to assert the existence of
a state, regardless of the society’s territorial or demographic size.2 Johnson
and Earle, by contrast, think that a state is characterised by a multi-ethnic
and large population that runs into the hundreds of thousands of people.3
For them, administrative and institutional complexity and organisation, as
implied by large settlement sizes, are critical indicators of a state. Crumley,
however, believes that non-centralised heterarchic structures that she has
argued existed in Iron Age Europe, must be regarded as states.4 The real
issue, then, is not the term in itself, or whether it is applicable worldwide,
but rather, what is meant by its use in specific circumstances.
Others within this volume express their scepticism about the validity of
evolutionary models to explain social change, arguing that they reduce
human diversity to a single trajectory, which, moreover, has been formu-
lated from a European model.5 Therefore such a model would supposedly
be at once a Eurocentric and a dehumanising theory, one that would ignore
both human agency and history. As argued below, we believe that this view
of evolutionary theory does not take into account the development that
social evolutionism has experienced since the mid-twentieth century. For
the moment, let us just remember Harris’s lucid observations on this issue:
. . . if the causal processes of history result in predictable or retrodictable patterns of
thought and behaviour, it is not because a mysterious teleological supra-individual
force or system has imposed its will on individuals. Rather it is because individuals
who are confronted with similar constraints and opportunities tend to make
similar choices regarding their self-interest.6

His stance boils down to a profoundly humanistic idea of human psychic

unity.
If we accept that the notion of ‘the state’ is operative and that there are
objective causes and mechanisms that may explain its rise, the study of the
pre-Roman Maghrib has an obvious interest, due to the peculiar nature of
the complex autochthonous polities that developed in the region. However,

2
Lull and Risch 1995, 97–101. 3 Johnson and Earle 2000, 35, 305. 4 Crumley 1995.
5
See, for example, Scheele, Chapter 16, this volume. 6 Harris 1995, 77.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 440 Joan Sanmartí et al.

Figure 11.1. Map of the late third-century BC polities in the Central and Eastern
Maghrib.

it is hindered, as we shall see below, by the lack of solid data regarding their
formation and early development. Indeed, it is only from the third
century BC onwards that historical sources and some scarce but significant
archaeological data (mainly consisting of large mausolea and commem-
orative monuments related to monarchs and elites) shed some light on
these polities. By that time, the autochthonous states were already very
large. Leaving aside the area of Carthage’s territory, the whole of the
Maghrib was divided into only three political entities, each one apparently
corresponding to a specific ethnic group (Mauri, Masaesyli Numidians and
Massyli Numidians, see Fig. 11.1). At its peak, just before the Second Punic
War, the Masaesylian state covered an area of about 150,000 km2, compar-
able to modern England (130,000 km2), while the Massylian polity
stretched over some 80,000 km2. These are very large figures; in fact, they
are closer to the primary or pristine states attested in America and in the
Old World (Mexico valley, Oaxaca valley, pre-dynastic Egypt, Uruk
Mesopotamia) than to the early states that emerged in later stages in the
Mediterranean area: both in the east (Canaanite city-states, Minoan and
Mycenaean states, succeeded by the poleis of the Archaic and Classical
periods) and in the west (in Etruria and the eastern coast of the Iberian
peninsula, etc.). All these were comparatively small in size, often quite close
to the theoretical average extension 1,500 km2 proposed by Colin Renfrew

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 441

for the ‘early state module’.7 Quite interestingly, another exceptional

example is the Saharan Garamantian state of the Classic phase (AD
1–400), which stretched over an area of about 250,000 km2.8
Consequently, the Maghribian (and Saharan) states appear as a special
case of Iron Age state formation in the Western Mediterranean context,
and one that should be studied and explained. The closest parallels are
probably the Macedonian and Epirote states, but their formation processes
are not particularly well known either.

Some Theoretical Issues

Research on state formation and the origin of cities remains one of the
central themes of archaeological investigation, as demonstrated by the
intense contemplation that, over the last decades, has refined the models
from mid-twentieth-century neo-evolutionist North-American anthropol-
ogy. On the one hand, in the fields of both sociology and anthropology the
very notion of ‘evolution’ has been claimed as a useful, even essential tool for
the understanding of historical processes. It has also been confronted by the
often exacerbated criticism of post-modern theorists and post-
processualists.9 The latter have rightly remarked on the deficiencies of mid-
twentieth-century neo-evolutionism (to which one should add the most
dogmatic Marxist evolutionary trajectories),10 but they have ignored that it
was not a fully developed theory, with little empirical testing by
archaeologists.11 Among sociologists who have proclaimed the validity of
evolutionary thinking, E.O. Wright has clarified that the essential nature of
a social evolutionary theory does not include the criteria, or flaws, that are
criticised by contemporary anti-evolutionism.12 More specifically, he clearly
states that they do not necessarily involve functionalist approaches based on
the ‘needs’ of society or on teleologically based trends, neither do they
necessarily define a rigid and unilinear sequence of stages. Instead, they
allow for the possibility of regression, make no claim that all societies must
evolve, and, finally, they do not postulate the existence of only one mechan-
ism to explain the transition from one social type to another. Thus, each case
7
Renfrew 1986; Sanmartí 2009.
8
Mattingly et al. 2003, 351. See also Mattingly et al., Chapter 2, this volume.
9
‘In particular, we intend to urge that any notion of social evolution is theoretically flawed and
almost always embodies unwarranted ethnocentric evaluations. We suggest that evolutionary
theories, of whatever kind, need to be abandoned in favour of a theoretical framework that can
adequately cope with the indelibly social texture of change within a framework avoiding both
reductionism and essentialism.’ Shanks and Tilley 1987, 138 (emphasis added).
10
Klejn 1993, 22–24; Trigger 1989, 222–26; 1998, 93–95. 11 Marcus 2008. 12 Wright 1983.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 442 Joan Sanmartí et al.

can and should be historically explained. From the field of anthropology,

Godelier remarked that history is also evolution, ‘car tout n’est pas possible à
tout moment pour toute société . . . il y a des étapes à franchir’ and ‘si des
sociétés peuvent éviter de passer par certaines étapes c’est parce que d’autres
y sont passées’.13
From the other side of the Atlantic, Johnson and Earle have attempted to
reformulate the evolutionary sequence generated by neo-evolutionary
anthropologists.14 What is more, drawing on cultural materialism, but
also on structural Marxism and other theoretical sources, they have pro-
posed a coherent model of the mechanisms of change. In this model,
a crucial role is played by the solution to the problems created in the
subsistence economy by population increase or, as happened in Fazzan
during the Pastoral period, by a decrease of subsistence resources due to
increasing aridity.15 For these authors, indeed, the solution to these pro-
blems was the growth of political economy, that is, the creation of more
complex institutions related with the administration of economic activ-
ities. These institutions may improve the management of subsistence
resources by means of the intensification of production or by improving
storage. Alternatively, they may obtain extra resources through trade or
aggression and protect them from external attacks.
The management of these institutions, in turn, makes possible the
consolidation of hierarchical social relationships and the constitution of
elites that, at least at a first stage, must appear as beneficial or benign to the
rest of the population. Beyond the terminological differences, this
approach is essentially the same as one proposed by Godelier from struc-
turalist Marxism, when he states that consent more than violence is
necessarily at the base of social differentiation processes (though violence,
and, it goes without saying, the legitimising ideology of inequality is also
implicated).16 Without consent, it would be impossible to understand why
the dominated sectors of society accept the supremacy of the dominant. In
short, some specific historical conditions are necessary so that processes of
social differentiation occur. For Johnson and Earle, these conditions are
linked to the relationship between population size and the carrying capa-
city of a territory, in specific conditions of technological development.17
The basic premise, with which we agree, is that the acceptance of an

13
Godelier 1980, 19. This is equivalent to reasserting the fundamental role of diffusion without
necessarily opposing it to evolution.
14
Johnson and Earle 2000.
15
Mattingly et al. 2003, 339–42. See also Mattingly et al., Chapter 2, this volume.
16
Godelier 1980, 657–63; 1999, 27. 17 Johnson and Earle 2000.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 443

institutionalised regime of social inequality that accompanies the develop-

ment of the state is only possible if common people perceive there to be
objective benefits in it. Therefore, immediate real or perceived benefit is
crucial in the initial stages of the process, given that access to wealth may
have been theoretically equal until then. It does not follow that commoners
are not aware, to some extent, of the charges involved, neither that they
easily or uncritically accept them. Put another way, the crucial point is the
perception that the benefits of participating in the political economy and
delivering a part of the product of their work to the elites outweigh the
disadvantages of having to sustain them. Needless to say, the benefits lie in
the solution to the subsistence problems, which, under some particular
historical circumstances, may ultimately cause new population growth.
In the long term, this growth is bound to provoke a new situation of
stress, which can only be resolved through a further growth of political
economy. The formation of states is but the culmination of these cycles of
growth, when increased population size and territorial extent create the
need of an administrative and institutional system of this kind. From the
archaeological point of view, therefore, the analysis of these processes
should focus on long-term study of the exploitation of the environment,
on the analysis of demographic oscillations and on the technological
changes that may increase the carrying capacity of the territory, along
with indicators of social differentiation. This involves a large amount of
data that must be obtained from surveys of relatively large territories and
excavations coupling the uncovering of large areas with the analysis of long
temporal sequences.
Another important contribution of the last twenty years has to do with
a better understanding of the dynamics of archaic states, both in their
formation and in their development, and relates to the introduction of the
notion of heterarchy.18 This states that certain complex societies may have
had decentralised forms of social organisation that do not fit with the idea
that centralisation is a necessary or inevitable condition for the formation
of archaic states.19 Moreover, some authors have been able to demonstrate
that both the formation process and the development of these states often
have saw-toothed trajectories, where stages of centralisation alternate with
others characterised by the sharing of power between different actors.20
One final aspect worth commenting on is the renewed interest by some
researchers in the formation processes of secondary states, namely those
18
Crumley 1995. 19 Bondarenko et al. 2002.
20
For formation see Brun 2015; Rodríguez et al. 2010; on development, Blanton 1998; Marcus
1998.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 444 Joan Sanmartí et al.

that arose after the appearance of pristine states and in contact with them.
In this respect, the recent work of Parkinson and Galaty should be
mentioned.21 They propose, for the specific context of the Aegean
Bronze Age, different trajectories that are explained by both the structure
of pre-state societies existing at the origins of the Minoan and Mycenaean
states and by the nature of the interactions with the first generation states
of this region.
All these developments constitute an important advance on the evolu-
tionary theory of societies, and must be considered in any case study.
However, it should be noted that they in no way contradict the need for
specific historical conditions that made it possible for a given society to
accept the transition to institutionalised inequality. Put another way, the
structures of kinship and neighbourhood, for example, can decisively
determine the development of archaic states, favouring, for instance, the
formation of more or less centralised structures.22 Similarly, the integra-
tion of the regional exchange systems of earlier states can accelerate the
formation of further archaic states or condition to some extent their
structures. We believe, however, that this does not invalidate the assump-
tion according to which states are formed through the consent to an
inequality that is perceived as beneficial. This is necessary, regardless of
the specific forms and rhythms of development, or the particular structure
that a state finally adopts. Any study on the formation of the state should
start, therefore, by proving the existence, or at least the probable existence,
of the conditions that made possible the process of politogenesis. This
inevitably leads us to the study of the subsistence economy, that is, the
relationship between population size and the carrying capacity of the
territory. It should continue by searching for evidence of the existence of
institutional complexity and social segregation. We will come back to these
issues, for the specific case of Numidia, after briefly reviewing the history of
research and the state of the art in this particular area.

History of Research and State of the Art

As mentioned above, the study of state formation in Numidia (comprising
north-eastern Algeria and western Tunisia) has severe limitations due to
the absence or, at best, poor quality of available archaeological data for the
first millennium BC. Regarding settlement sites, there is almost no archae-
ological evidence for the period prior to the second century BC (leaving

21 22
Parkinson and Galaty 2007. Bondarenko et al. 2002, 65.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 445

aside our contribution at Althiburos).23 In the final stages of the Numidian

period (fourth to second/first century BC), it is limited to fragmentary
information that is often difficult to interpret. At Cirta there is a sector of
supposedly Punic type houses,24 at Thugga, a house of the second
century BC excavated a few years ago which is preserved in a quite frag-
mentary state,25 – and at Bulla Regia, hints of an orthogonal urban layout.26
We can also add the second to first century BC layers of the sanctuary of
Hathor Miskar at Mactaris, and especially the large Ionic temple recently
excavated at Zama Regia, which remains unpublished. A special case is
Simitthus, where archaeological remains that could date back to the fourth
century BC have been found, but these too remain almost completely
unpublished.27 One consequence arising from this state of affairs is the
total absence of environmental archaeological studies and the lack of
information regarding the technology applied to agricultural production.
The evidence, in short, is very poor, which is explained by two factors: on
the one hand, the difficulty for excavators to reach the lower layers in
archaeological sites that were very often occupied until at least Late
Antiquity; on the other hand, and most especially, the almost total absence
of this issue in the agenda of archaeological research. This absence has
various causes, which we have discussed in previous works.28
The difficulties involved in the study of funerary remains are of
a different kind. Leaving aside the great monuments we have already
mentioned, the funerary record is extremely rich and easily accessible. As
a matter of fact, it constitutes a characteristic element of the landscape in
many regions of the Eastern Maghrib due to the very large number of
tombs and their visibility.29 Hundreds were excavated, the vast majority
during the colonial period, from the mid-nineteenth to mid-twentieth
century. However, the scientific value of these works is largely limited by
the combination of four factors: firstly, the methodological shortcomings
of most of them (with some significant exceptions);30 secondly, the pre-
vious plundering of many structures (in particular the haouanet, but also,
23
See also Bokbot, Chapter 12 and Wilson, Chapter 10, this volume.
24
Only sketchily published in Berthier 1980. 25 Khanoussi et al. 2005. 26 Thébert 1992.
27
Excepting a general overview derived from the extremely welcome restart of the activity at this
site; see Khanousi and Von Rummel 2012.
28
Kallala 2002; Sanmartí et al. 2012, 23–24. See also Bokbot, Chapter 12, this volume, about the
widespread but poorly grounded assumption by many researchers working in the Western
Maghrib that any site where wheel-thrown pre-Roman pottery has been found must necessarily
be considered as Phoenician. This has probably biased the image about the nature of the
autochthonous populations’ economy and social organisation.
29
Camps 1961a; 1995.
30
Exceptions include the work of Camps and Camps-Fabrer (1964) at Djebel Mazela.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 446 Joan Sanmartí et al.

very often, the dolmens); thirdly, the fact that the funerary offerings are
frequently materially poor and do not, in general, include diagnostic
imported materials that can be tightly dated; finally, the small number of
research projects conducted in the last 50 years, when the improvement of
excavation techniques and the use of radiocarbon dating could have
replaced the limitations of dating based on artefacts. We still need to
add, in the case of the large megalithic necropoleis, that their huge dimen-
sions (frequently hundreds or thousands of graves spread over several
square kilometres) makes it very difficult to document them accurately,
and, therefore, to understand their possible internal structure. In short, as
in the case of settlements, the enormous potential of the funerary sites has
hardly been exploited.31
Under these conditions, studies of the formation of the Numidian states
have been almost exclusively based on indirect evidence provided by
ancient texts and epigraphic sources of the Roman imperial period.
These documents suggest that in the historical period of the Numidian
kingdoms (third to first century BC) there were also important tribal
structures that dated back, with more or less changes, to the period before
the formation of these monarchies. Drawing implicitly on this idea and
explicitly on the historically known cases of the Medieval Maghrib,
Stéphane Gsell hypothesised that the formation of the Numidian kingdoms
should be explained by the violent action of powerful tribal chiefs who had
imposed their power over other tribes.32 He also believed that the intro-
duction of the horse, and especially of iron metallurgy (which he placed
around 1000 BC) had played an important role in ensuring the military
supremacy of groups possessing them. More than 30 years later, another
great historian of the Berber world, Gabriel Camps, made Gsell’s hypoth-
esis his own, and even in the early eighties Elisabeth Smadja advocated
essentially the same.33 This model, however, has little explanatory value.
Territorial expansion in a tribal framework does not provide by itself
a clear understanding of the development of the institutional complexity
that is typical of states. In fact, Gsell himself understood, or at least
suspected, this interpretative weakness, although this did not lead him to
modify or further develop his model.34
A different model was proposed by Tadeusz Kotula, following the line of
thought initiated by Herbert Spencer in the nineteenth century (and sub-
sequently developed by other authors, in particular Carneiro). He believed
31
See now Gatto et al. 2019, for the parallel Trans-Saharan Archaeology overview volume on
funerary traditions of the Maghrib and Sahara.
32
Gsell 1927, 77–82. 33 Camps 1961b, 161; Smadja 1983, 686. 34 Gsell 1927, 80–81.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 447

that the territorial expansionism of Carthage during the fifth century BC

was the main cause of the formation of the Numidian states, since it was
likely to have promoted the centralisation of power among local popula-
tions in order to deal with this threat more effectively.35
Finally, we have to mention the Marxist inspired techno-determinist
model of Pierre Lévêque. For this scholar, the crucial triggering factor was
the development of iron metallurgy, which he believed was introduced by
the Phoenicians. This technological shift would have encouraged the
extension and intensification of agricultural production; simultaneously,
the demand of commodities by the Phoenicians would have intensified
production and trade relations. As a consequence of improved technology,
population would have increased, which in turn would have made possible
the formation of urban centres and caused the rise of intertribal violence.
The final outcome would be the emergence of more powerful tribal chiefs,
who, Lévêque believed, tended to form a rural and commercial aristocracy,
while peasants were reduced to a situation of dependency.36
All these hypotheses are plausible to some extent, or at least we can
assume that they must contain some elements of truth, but the avail-
able data are too scarce to verify them. In any case, as we said above,
we believe that, whatever the endogenous process, and whatever the
weight of contact with the Phoenicians, the acceptance of institutiona-
lised inequality and population pressure were fundamental to the
development of Numidian states. As a matter of fact, this assumption
underlies most of the abovementioned hypotheses. In the case of
Kotula’s model, it can be argued that, whatever the military pressure
of Carthage, native states could not have been constituted without
a sufficient population size. As for Lévêque’s model, we must take
into account that technological innovations are not necessarily bene-
ficial to all members and groups of a given society. They involve risks
and uncertainties that often tend to slow down their adoption until
they are imposed by elites, in order to increase their power and
wealth,37 a situation that may be favoured by population pressure on
resources. The expansion of iron metallurgy, therefore, could be the
consequence rather than the cause of the formation of elites. However,
it is true that the consequent increase of techno-environmental effi-
ciency must have resulted in a further population growth that was
presumably decisive in the long term for the development of early
states.

35 36 37
Kotula 1976, cited by Lassère 2001, 149, note 2. Lévêque 1986; 1999. Kim 2001.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 448 Joan Sanmartí et al.

The Project of the Institut National du Patrimoine and the University

of Barcelona at Althiburos
This project was specifically designed for obtaining relevant data on state
formation in an area that, according to historical sources, was occupied by
the Massyli Numidians.38 In addition, a neo-Punic inscription found in the
nineteenth century proved the existence at Althiburos (Fig. 11.2), at least in
the last centuries BC, of municipal institutions, and therefore of a pre-
Roman city.39
The research strategy adopted was based on Johnson’s and Earle’s
aforementioned model of socio-cultural change. Consequently, it was
primarily conceived to retrieve data on population growth, technological

Figure 11.2. Northern Tunisia and the location of Althiburos.

38
The project involved the Tunisian Institut National du Patrimoine and the University of
Barcelona, with the collaboration of the Catalan Institute of Classical Archaeology.
39
More specifically, it indicates the existence of a religious position (BLL ‘the sacrificer’) and
a college of three suffetes (Ennaïfer 1976, 27). The existence at Althiburos and Mactaris
(Makthar) of colleges of three suffetes (as opposed to the two that are typical of the Punic
tradition) indicates the specific nature of this local institution, Belhakia and Di Vita-Évrard
1995, 262.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 449

Figure 11.3. Schematic plan of the capitol area and location of the excavation zones.

change and hints about urbanisation. Also, given the historical context, it
considered the city’s relations with the Phoenician world. Regarding the
first of these goals, demography, it was expected to retrieve significant
information from the survey of the city itself and its closer territory, on the
assumption that ground survey would provide datable materials. Another
expected source of information was the environmental data obtained from
the excavation of the site of Althiburos, presuming that population growth
implies agricultural intensification, and that the latter should be

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 450 Joan Sanmartí et al.

recognisable in the archaeozoological and archaeobotanical record. It was

also assumed that the excavation of this site could provide more or less
strong evidence on the urbanisation process (the nature of domestic
architecture, public buildings, urban planning, defensive walls, etc.), on
the contact with Phoenician-Punic civilisation and on technological
changes. Of particular interest was the introduction of iron metallurgy
and its potential application to agricultural production. The correlation of
these aspects could provide relevant information about the specific histor-
ical conditions, in terms of the subsistence economy and the contact with
other societies, in which the formation and development of the Numidian
state took place.
Although it was not initially planned, the progress of the project also led
us to analyse the funerary record, particularly the great megalithic necro-
polis located on the El Ksour massif, which was probably related to
Numidian Althiburos. From its study it is expected to obtain data related
to demography, insofar as we are able to refine the dates of the tombs, and
to social structure. The latter can be interpreted from the spatial organisa-
tion of the necropolis (for example, by studying the differential distribution
of distinct tomb types, or their relative proximity to Althiburos) and, from
the variability of structure, size and contents of the different graves. The
study of the necropolis has provided some interesting results.40
We present below a brief overview of the results obtained in each of
these aspects, followed by a synthesis focusing on state formation and
urbanisation at Althiburos.

The Results of the Survey and the Excavation at the Capitol Area
The results of the survey have been interesting, but scarcely explicit to the
topic at hand because of difficulties of dating. These are due to the virtual
absence in collected assemblages of well characterised Greek, Punic and
Italic imported pottery and the highly fragmented nature of the Numidian
ceramics, which can only be dated when a significant part of the profile is
preserved.
Despite these limitations, it is worth noting that the survey within the
city of Althiburos led to the collection of Numidian pottery in the entire
area between the two wadis that define its core, that is, in a space of about

40
It could not be otherwise considering its enormity: over a thousand structures have been
recorded so far, stretching over an area of 30 km2, and only four of them have been excavated.
See Kallala et al. 2014; 2018; Sanmartí et al. 2015; 2019.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 451

NR1 (fourth century – 146 BC)

NM (late seventh – fifth century BC)

NA3 (eight century BC)

NA2 (ninth century BC)

NA1 (tenth century BC)

Figure 11.4. Section of sectors 3–4a in excavation zone 2 showing the stratigraphic
sequence of the Numidian period.

7 ha.41 However, this only proves that the Numidian settlement could have
stretched over this area at some time in its history. This is likely the case in
the final centuries of the first millennium, but nothing can be said, for the
time being, for the previous stages. As indicated by Mattingly, the topo-
graphy of Althiburos, an elongated ridge delimited by steep drops, is
typical of Numidian towns.42
The excavation of the Capitol area (Fig. 11.3) has provided some relevant
data regarding the subject we are dealing with.43 Firstly, it has allowed us to
document a virtually uninterrupted sequence of occupation and material
culture dating back to the tenth century calBC (Fig. 11.4).44 Secondly, for

41
Regarding the topography of Althiburos, see Kallala and Sanmartí 2011, 10 and figures
1.11–1.14; Sanmartí et al. 2012, 28–30, figures 3–5.
42
Mattingly 2016, 15. See also Wilson, Chapter 10, this volume.
43
On this dig, see Belarte 2011, and Ramon Torres and Maraoui Telmini 2011. For more detail on
the Numidian architecture, see Belarte and Ramon in Kallala et al. 2016.
44
It is worth noting that, at the opposite end of the Maghrib, Bokbot, Chapter 12, this volume,
proposes an early dating, in the first centuries of the first millennium BC, for the autochthonous
occupation of several habitation sites, such as Lixus, Mogador, Kach Kouch, Sidi Driss, Les
Andalouses and Rachgoun, most of which have been traditionally considered as Phoenician sites.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 452 Joan Sanmartí et al.

the first time it sheds some light on the structure of a Numidian settlement
and the building techniques employed. The data recovered, however, are
still very limited, as the very nature of this deep stratification, with a history
of over 2,000 years, imposes restraints on the excavated areas of the earliest
levels and hinders the understanding of structural remains. Nevertheless, it
can be said that the area excavated around the Capitol was mainly devoted
to domestic use, with walls built with solid materials (stone and earth) from
the very beginning of the sequence. Although these were plainly much
more than simple ‘huts’, their overall plan cannot be clearly defined as yet.
The buildings from the earliest phases, tenth–early seventh century BC,
had little consistency: the walls, though sometimes very thick, were often
slightly curved and built in a series of juxtaposed sections (Fig. 11.5). Often
the material employed consisted of large river stones. The nature of this
first settlement cannot be ascertained, since we do not know its total size
and the structural remains discovered so far are few. It could be either
a small hamlet or a village of larger dimensions. In any case, the type of
architecture documented does not seem the most appropriate for
a properly urban-type settlement, organised following a more or less
regular planning. This does not mean that there could not be a sizeable
population, maybe distributed in different villages. One of these could have
been located at the northern end of the Roman city, at the confluence of the
wadis, others could have been the scattered sites that are described later in
this section.
From the end of the seventh century BC, after an apparent hiatus of
several decades whose causes are unknown, there was a reorganisation of
the settlement in terms of architecture and layout, although the necessary
data to recognise the total area of occupation are still lacking. Due to the
confined space of the excavation, it remains impossible to reconstruct
the urban layout or to reconstitute the complete plan of any house, but
the available evidence hints at a more regular architecture, with rectilinear
walls clearly and solidly built (though not as wide as some dated to the
previous phase). The reasons for this change are not easy to state. Beside
the internal dynamics of the Numidian society we could take into account
an intensification of the contacts with the Phoenician world, revealed by
a slight increase in the volume of imported ceramics (which, however, are
still very rare: 1 per cent of the total ceramics) and, more clearly, by
a biapsidal cistern (CT290111) that indicates close familiarity with Punic
hydraulic technology (Fig. 11.6). To this should be added the existence of
pavements and wall coverings made with lime mortar, as well as the use of
mudbricks (but this could have already started in the eighth century BC).

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 453

Figure 11.5. Schematic Early Numidian 3 constructions in excavation zone 2.

That being said, we should also note the profoundly autochthonous char-
acter of the mid-first-millennium BC settlement in all the spheres of life,
such as cooking and the forms of food consumption, as well as most of the
architectural features, not to mention the funerary world, which will be
discussed later. The most powerful indicator of the likely urban nature of
Althiburos before the second century BC is a thick wall, probably with

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 454 Joan Sanmartí et al.

Figure 11.6. Punic-type cistern of the Middle Numidian period in excavation zone 2.

Figure 11.7. Defensive wall seen from the south-west.

defensive function, which was erected in the fourth century BC. Its remains
are located in the area immediately to the south of the Capitol (Fig. 11.7).
With regard to the peri-urban area, in the same Althiburos valley, five
more settlements, of much smaller dimensions (as compared to the total
area of Althiburos) have been located, which have also provided Numidian
potsherds of imprecise dating, and that continued to be inhabited at least
until late antiquity. It is possible that these sites were created as a result of
population growth and sedentarisation during the first millennium BC, but
the available data do not allow precision about when this occurred. It seems
likely, in any case, that all of them were occupied in the final stage of the
Numidian period.
In summary, the recovered data seem to indicate a significant increase in
the population during the first millennium BC, particularly during
its second half, and an early contact with the Phoenician settlements of

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 455

the coastal areas, which intensified greatly from the fourth century BC.
This is proved not only by the acquisition of table-ware and other kinds of
pottery (including amphoras), but also by the adoption of Punic hydraulic
techniques already in the sixth century BC. The erection of the defensive
wall in the fourth century BC may also indicate instability, which could
have been caused by internal troubles or related to Carthage’s expansionist
policy. In any case, the huge effort required for its construction confirms
the existence of a sizeable community at Althiburos.

The Environmental Data

Althiburos has provided a large amount of palaeoenvironmental data,
thanks to the development of high-resolution integrated sampling strate-
gies within the site. However, palynological research conducted in different
parts of the region has not given usable results. We present in this section
the conclusions reached from the archaeozoological and archaeobotanical
studies in the area of the Capitol, in order to recognise signs of economic
intensification that could be the result of a significant population increase;
that is to say, to verify one of the predictions of the model. We are aware
that the absence of palynological data represents a serious limitation for
delineating vegetation change at the regional level. Nevertheless we believe
that the available data are sufficiently significant to propose well-grounded
hypotheses.
Anthracological data provide some (albeit incomplete) information on
the changing characteristics of surrounding woodlands.45 From the earliest
excavated phase (NA1), at the beginning of the first millennium BC, the
charcoal mainly came from pines rather than holm-oak woodlands, which
are native to this region, with a considerable presence of wild (maybe also
domestic?) olive-trees. The replacement of oak woods by Aleppo pine
suggests the existence of a dry Mediterranean forest, as a consequence of
an important anthropic action. It is possible, therefore, that during
the second millennium BC the area had already been intensively exploited
and the native woodland modified. This is consistent with the results of
palynological studies carried out in the region of Aïn Draham, in north-
western Tunisia, that indicate a decline in deciduous oak forest due to
a combination of anthropic activities (clearance) and climatic changes (a
peak in aridity) that can be recognised as early as c.2000 BC.46 Given this
situation, it is particularly striking that, in the present state of research,

45 46
Cantero and Piqué 2016. Stambouli-Essassi et al. 2007.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 456 Joan Sanmartí et al.

Fôret Méditerranéenne sèche Fôret hydrophile Cultivars

Ta ace p. icin en
is
lic s s of gre
al

ta
ns s

la
pe nu

Sa inu us ver
is

gl p. icu
le m

f
ea
ro p.

ax rin . e
le su a

ha ha

Ju us s art
nu a p.
O res lae
eu s s
pa

ra
Fr a sp
Pi re s

.
.

.
m ae
sp
s /R

sp

sp

vi p.
fe
lm is
illy us
e

tis s
os us
C hym

U clin

ni
ix
ia

s
Ph iper

Vi nus
an
R rc

ar
ac

tra
a

ue
.T
up

st
n

u
Te
Ju

Pr
Q
Pi
cf

MED

VAN

HE

NR

NM

NA3

NA2

NS1

5 5 20 5 20 40 60 80 100 20 5 5 20 20 20 20 40 10 10

Figure 11.8. Percentage of charcoal taxa per phase, from a diachronic perspective (Early
Numidian to Medieval times).

there are virtually no archaeological remains datable to the second

millennium BC across the Eastern Maghrib.47
The results of the wood charcoal analyses are consistent with paleocar-
pological data (Figs 11.8–11.9).48 They indicate the preponderance of
cultivated species during the tenth century calBC (82.1 per cent of the
remains) and, among these, of cereals (92.3 per cent), mainly barley
(Hordeum vulgare) and free-threshing wheat (Triticum aestivum/durum).
There is also a considerable presence of cultivated grape seeds (Vitis
vinifera). This is not surprising by any means, since the cultivated vine is
attested in Fazzan in the early first millennium BC.49 Vines are a delayed-
yield crop, whose cultivation demands a considerable initial effort.

47
Only two recently discovered tombs at Thugga; see Khanoussi et al. 2005. It is also worth noting
that the Bronze Age is relatively well attested in western Algeria and Morocco (Camps 1960; see
also Bokbot, Chapter 12, this volume).
48
López Reyes and Cantero 2016.
49
Mattingly et al. 2003, 342; van der Veen and Westley 2010, 507, 509.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 457

fourrage textiles/oléag.
0,4% 0,1%
fruitiers
AU
4,7%
PS 11,2% légumineuses
6,6% 2,5%

PC céréales
82% 92,3%

1 2

60 60
50 50
40 40
30 30
20 20
10 10
0 0
.
/d e

ic m

(fo um

ilia )
um

is

ba

ra

um

iva
sp
ta
um gar

ric
ar
u

Fi inife
fa
ue

at
im
c

ce
ur

s
lin

ca
oc
l

ra ullu

.s
a
vu

ar .
ol sp.

.
rq

e
ic iss

sp

sp
.

Av sp.

rm sp.

Bi sp.

m .
um ae

po um

Po e

a
cu

v
ci

sp

sp

ea
s

cf

Fu cea

ce
Lo ia s
p.

cu
Vi
um

ni Citr

t
d

ty

m
M sita

no alb
ns

ac
a

s
os ena

um

ra
o

R era
ss

po llac

tu

i
um

Pa lta

lu
ug
ag

tru
a
m
iv

ar
de

fo
Le

ilo
di
ga

rtu hal
.u
st

liu
cu
e

no hy
Aj

is
ic
or

el
sp
ae

fe

tra
ed
cf

ap
it

ni

pe

ca
p

P
H

di

M
Tr

yo
um

As
um

la
vi

he
ar
nu

th
ic

tis
ic

C
Li
it

Li

Po
it

Vi

he
Tr
Tr

3 4

C
ra

iva
um

.
e

um

um

is

um
sp
ar

ric
ife
ar

at
fa
ur

m
cc

ce
lg

p.

.
s

ea
in
lin

ca

sp

sp
.

.
.s

pe

e
sp

sp

sp

sp

sp

sp
lu
/d

si
vu

Ph s s
ea
co

v
ilia

cu

Fu lbu

ac
ci

cf
ul

tis

ty
s

ris

m
p.
um

um

ra

ia
di

cu
Vi

ac
um

tu
itr
m

er
o

lu
ug

en
ss

tru
ta
ns

m
ar

a
fo
ag

ilo
C
um

Fi

Po
iv

ol
ga

al
rm
si
um

um

liu
de

Av
Aj
Le

Bi

is
ra
st

el
ic

.u

ca
tra

ap
Lo
ic

pe
or

fe
ae

ic

di

M
ed

cf
it

la
ni

As
n

R
H

po
os
Tr

M
Pa
um

rtu
m
vi

th

no
nu
tis

Po
ic

Li

he

6
Li
it

Vi

5
Tr

Figure 11.9. 1) Percentage of cultivated plants (PC), wild plants (PS) and other types
(AU) over the total of the remains (1,294 individuals and fragments); 2) Percentage of
different types of cultures over the total number of individuals (836); 3) and 4)
Percentage of the total number of cultivated and wild plants; and 5) and 6) estimation of
the distribution of frequencies (25 and 16 mentions respectively).

Although it does not produce fruits for five or six years, once mature, it can
subsequently be exploited for a long period of time. The logical conse-
quence is that the population who cultivated it had to be fully sedentary.
From the perspective of socio-cultural evolution, these data suggest that in
the early first millennium BC, if not before (maybe long before?), the
population had grown beyond the limits within which a semi-sedentary
lifestyle is feasible. Extensive swidden agriculture and widespread exploita-
tion of wild resources, both animal and vegetable would have become
impossible as the territory became completely occupied and exploited,
leading to a fully sedentary way of life. Additional evidence comes from

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 458 Joan Sanmartí et al.

the study of microfossil livestock dung assemblages.50 From the tenth

century calBC onwards, these remains contain significant concentrations
of calcitic dung spherulites and grass phytoliths, mostly of leaves and
stems, probably from Pooid domestic varieties. This may suggest that
fodder was obtained from agricultural by-products (such as wheat and
barley, chaff and straw). This grass-rich diet for livestock may be linked to
the expansion of cereal crop production. Yet, the study of faunal remains of
the tenth–ninth centuries BC indicates a large supply of bovines, whose
slaughter age pattern suggests a focus on meat production and rapid herd
renewal.51 This, in turn, implies that despite the human pressure on the
environment, these herds were viable. Grazing was probably available
either on fallow crop fields or grasslands (their use is still ethnographically
documented in the area), although the possibility that cattle were brought
in from somewhere else cannot be excluded.
The eighth century BC marks a major turning point. At this time, the
relative frequency of bovines decreases significantly, while the number of
cattle over four years old slaughtered increases (Fig. 11.10). This may be

Althiburos %NISP change - main domesticates

100
Cattle
Suids
Caprines
80

60
%NISP

40

20

0
p.

p.
A1

A2

A3

al
N

Em

Em

nd
N
N

Va
rly

te
La
Ea

Periods

Figure 11.10. Relative frequency of the three main faunal taxa all over the occupation.
The vertical lines indicate twice the standard deviation (confidence level of 95 per cent).

50 51
Portillo and Albert 2011; 2016; Portillo et al. 2012. Valenzuela-Lamas 2016.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 459

due to the need to employ them for a longer period working in the fields,
maybe as a result of agricultural intensification. This trend increased even
further during the sixth–fifth centuries, when animals slaughtered at over
six years’ of age predominated, while the presence of calves severely
decreased. From the eighth century BC, and most particularly from the
sixth century BC, meat supply also became largely dependent on sheep,
goats and, especially pigs. These are species that are better adapted to low
quality pastures, typical of dry environments, or in the case of pigs, can be
fed with scraps of human food and, therefore, away from agricultural land.
This is an important indicator of intensification that is confirmed to some
extent by the archaeobotanical data. Indeed, the anthracological remains
indicate, for the sixth century BC and onwards, an increase in the number
of species and the exploitation of new ecological niches. In particular, the
presence for the first time of Ulmus sp., Salix/Populus and Fraxinus sp.,
three hydrophilic trees, may indicate that pine, olive and other trees that
grow in dry environments were no longer enough. Perhaps this was
because there were more people exploiting them, perhaps because increas-
ing agricultural activity had reduced the extent of the forest (unless some
climatic shift may account for this). The same could be indicated by the
presence of Juniperus sp., a shrub/tree that grows within the forests of
Aleppo pine. This may suggest that pine woods were more open, due to
more intensive exploitation. Likewise, grapevines are documented for the
first time as a fuel source. In short, there is evidence to suggest an extension
of the catchment areas for wood, and that those sources that were pre-
viously used were more intensively exploited. Finally, the carpological data
indicate the presence of new plants that may be used as fodder, particularly
clover during the eighth century BC, and during the sixth century BC
alfalfa, Coronilla, horseshoe vetch, Astragalus, Melilotus, etc. This supports
the idea of a landscape with less grasslands and large areas of crops.
In summary, the environmental data seem to support the idea of
a considerable economic intensification from the eighth century BC
onwards, and that this might have been the consequence of sustained
population increase. Given that there is reason to assume that the area
was already well populated by the end of the second millennium BC, it is
plausible to suppose that this further growth entailed high pressure on the
environment. This in turn urged the development of political economy and
thereby opened the way to the formation of state-like polities. Obviously,
one cannot exclude that this process was influenced by a demand for
farming products by the Phoenicians, but even in this case we would
conclude the existence of a sizeable population and a pressure on the

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 460 Joan Sanmartí et al.

environment that would have led to an increase in social and institutional

complexity. Interestingly enough, botanical remains prove that a similar
pattern of agricultural intensification occurred in Fazzan in the early first
millennium BC, in the form of irrigated cultivation, with comparable social
impact.52

Technological Change
The oldest iron fragment excavated from the site is dated to the late ninth
century calBC, but is unfortunately shapeless. Since it could be an extre-
mely early Phoenician import, it cannot definitely prove the existence of
local iron production at that time, though the latter possibility is altogether
likely. Two further small undiagnostic fragments of iron slag can be dated
to the eighth century BC; for the moment we cannot state whether they
were produced by smelting or smithing. Therefore, this material does not
prove that there was local iron production, though in our opinion this is
quite probable.
From these data we may infer the possibility that iron instruments were
used for agricultural work from the eighth century BC onwards. This
technology would have facilitated an increase in the territory’s carrying
capacity. Indeed, iron tools would have made possible the expansion of
cultivation into areas that were difficult to exploit with wood or stone
instruments. It could also have led to the progressive implementation of
a Eurasian grain-farming model, based on permanent fields and rotating
fallow.53 Given that in a dry area such as the Althiburos region, the fallow
lands must be ploughed and the natural vegetation removed to facilitate the
accumulation of water, the use of iron tools is necessary for an efficient
permanent use of the fields.
It is logical to assume that, as in Europe, agricultural iron tools were
widely used during the first millennium BC, and that they constituted one
of the key elements of the economic and demographic potential of the
Numidian kingdoms. While we do not have direct evidence of iron agri-
cultural implements, nor is it possible, a fortiori, to know when they began
to be used on a large scale, in any case, the data recovered at Althiburos
indicate that iron production could have started in the eighth century BC
(if not before). However, we have to keep in mind that knowledge of
a particular technology does not necessarily mean its immediate

52
See Mattingly et al., Chapter 2, this volume; Mattingly and Sterry, Chapter 19, this volume.
53
Wolf 1966, 30–32.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 461

application to production as has been observed in several case studies.54

Application is usually linked to the interests of elites, among who the needs
of the domestic economy may be included. A better understanding of the
spread of this technology as applied to agriculture is therefore highly
desirable to facilitate our understanding of the formation processes of
cities and states.

Figure 11.11. View of monument 53 during the excavation.

The Funerary Data

The development of the Althiburos project has led to the study of a large
megalithic necropolis located to the south of the city and stretching over
much of the El Ksour massif.55 This complex, multi-site, funerary land-
scape, which had not previously been recorded as such, has been inten-
sively surveyed in the western and central zones of the massif, and more
54 55
Kim 2001. Kallala et al. 2014; 2018; Sanmartí et al. 2019.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 462 Joan Sanmartí et al.

Figure 11.12. Chamber of monument 53.

rapidly explored in the remaining areas. To date 1,035 structures have been
recorded, most of which relate to the type of monuments that Camps desig-
nated as ‘dolmens sur socle et dolmens à manchon’, which are typical of the
interior areas of the Eastern Maghrib.56 We prefer to call them ‘zenithal access
dolmens’. As mentioned above, in spite of the fact that many examples of these
structures were excavated between 1850 and 1960, the available information
about them is very scarce, particularly with regard to their dating. For this
reason, and also to try to understand the apparent variability existing among
them, three zenithal access dolmens of the El Ksour necropolis (monuments
42, 53 and 647) have been excavated. The 14C dating for the latter (Beta-
333228) has provided a chronology in the Hallstatt plateau, with a strong
likelihood for the fifth century BC, whilst that of monument 53 (Beta-283142),
also in the Hallstatt plateau, corresponds to the eighth to mid-fifth
century BC.57 This dating is consistent with the associated Numidian pottery,
whose chronology begins to be known following our excavations at
Althiburos. As for monument 42, the few pottery sherds recovered from the
layer on which it was built indicate a construction date in the second half of the
first millennium BC. The parallels with well-dated ceramics from Althiburos
allow a proposed dating between the eighth and the fifth century BC for at

56
Camps 1961a, 130–36.
57
Conventional radiocarbon age: 2480±40 BP. 2 Sigma calibrated result: (95 per cent probability)
775–430 calBC (Beta-283142).

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 463

least five more tombs excavated in the 1950s by G. Camps and H. Camps-
Fabrer in the necropolis of Djebel Mazela (Bou Nouara, Algeria).58 The rest of
the tombs excavated at Djebel Mazela cannot be precisely dated from parallels
at Althiburos, but it is worth indicating that the ceramic shapes typical of the
fourth–first century BC on our site are completely absent there. This may
suggest that they were erected no later than the fifth century BC. To this we
should add the dating of two dolmens of the necropolis of Henchir Mided, at
least one of which belongs to a different type from those at El Ksour and
Djebel Mazela, as it is a free-standing dolmen (that is, with no tumulus
surrounding it). The 14C dating for the latter also corresponds to the
Hallstatt plateau,59 while the other one contained an Attic cup of Vicup
type, whose production is dated to the second quarter of the fifth
century BC.60 To sum up, in spite of the very low number of (to some extent)
well-dated monuments, it seems to us remarkable that all of them belong to
a period spanning from the eighth to the fifth century BC. This could also be
a clue, though certainly still very tenuous, of population increase in the central
centuries of the first millennium BC.
A second aspect regarding the three excavated monuments concerns
their variability in size, construction quality, number of depositions and
other matters related to the funerary ritual. Indeed, one of the three
excavated tombs (number 53) stands out for its remarkable size (diameter
c.13 m, with a burial chamber measuring 1.40 × 1.20 m) (Figs 11.11–11.12),
whilst monuments 42 and 647 are much smaller: diameters respectively
5.50 m and 5 m; dimensions of the chambers are 0.97 × 1.12 m and 0.85
× 0.60–0.70 m respectively (Figs 11.13–11.15). The latter figures are among
the most common in the whole necropolis. Indeed, 77 per cent of these
monuments measure between 4 and 7.9 m diameter, and the modal peak is
5 to 5.9 m (30 per cent). Only 2 per cent measure more than 12 m in
diameter. Monument 53 is also remarkable due to the careful construction
of the chamber which was built from large, relatively well-cut blocks that
were superimposed in several courses. Conversely, monument 647 was
composed of vertical slabs and monument 42 a combination of both
systems. Added to this is the fact that in monument 53 the circular wall
delimiting the tumulus was double-faced and rather carefully built, while in
the other two cases it was made up of only one row of slabs or blocks.
Another noticeable feature is the existence in monument 53 of two ‘anten-
nas’ or ‘arms’, that is, projecting rectilinear walls (Fig. 11.16), whose sig-
nificance remains obscure, but that lead us to suspect a high degree of ritual

58 59 60
Kallala et al. 2014, 55. Marras et al. 2009, 188. Ferjaoui 2010, 344.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 464 Joan Sanmartí et al.

Figure 11.13. Plan of monument 42.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 465

Figure 11.14. Plan of monument 647.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 466 Joan Sanmartí et al.

Figure 11.15. Chamber of monument 647.

complexity; stronger, at any rate, than in the other two monuments.

A further important difference lies in the fact that, while the burial chamber
of monument 647 contained the remains of a single individual (an adult),
monument 53 has produced the remains of at least six bodies of different
ages,61 three of them infants (the chamber of monument 42 was completely
plundered). Finally, it is very important to note that a relatively large
quantity of pottery vessels, mainly tableware, was recovered in the space
between the chamber and the wall that delimitated monument 53. They
suggest a remarkable ritual activity, of which there is no sign in the other
tombs we have excavated.
It goes without saying that the small number of data points makes
interpretation very delicate. However, considering that the three monu-
ments are roughly contemporary, it is a logical assumption that the
observed variability reflects the appearance of a hierarchical structure of
Numidian society by the mid-first millennium BC. More specifically,

61
In every case, human remains are few, due to previous excarnation of the bodies (Kallala et al.
2014, 29, 30, 56). This practice had already been clearly attested by Camps and Camps-Fabrer at
Djebel Mazela (Camps and Camps-Fabrer 1964, 75–79; see also Camps 1961a, 481–501 for
a general discussion on this issue).

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 467

Figure 11.16. Plan of the first phase of monument 53 with the two ‘antennas’ or ‘arms’.

monument 53 seemingly contained the remains of the members of a high

status family or even of a larger kinship group.

Conclusions
To summarise, we can say, firstly, that the environmental data are appar-
ently consistent with the predictions of our model that urbanisation and
state formation were contingent on the emergence of hierarchy, population
growth and an intensification of land-use leading to the development of the

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 468 Joan Sanmartí et al.

political economy. The evidence presented indicates, from the eighth

century BC, a pressure on the environment that can be reasonably attrib-
uted to population increase. Unfortunately, the lack of palynological data
makes it impossible to analyse the situation at a regional or even micro-
regional scale, but it is reasonable to assume that this trend was not strictly
limited to Althiburos and its immediate surroundings. This is also con-
sistent with available information on the funerary world, as the few more
or less well-dated tombs all belong to the central centuries of the first
millennium BC. However, this may be purely accidental, since we are
dealing with only seven tombs among the hundreds of thousands that
comprise the megalithic necropoleis of Numidia. This possible population
increase, if it really resulted in food shortages and difficulties for the
domestic economy, could have caused a growth of political economy. In
the sense of Johnson and Earle’s model, a higher institutional complexity
resulted and, in the end, the consolidation of social hierarchies. The most
obvious evidence to support this is the observed variability between the
three tombs excavated in the El Ksour necropolis, since monument 53
evidently stands out with respect to the other two (and, at least for its size,
in relation to the vast majority of tombs identified through the survey).
Finally, evidence on the restructuring of the settlement of Althiburos in the
sixth century BC, as well as the introduction of some elements of technical
sophistication, specifically the Punic-type cistern, suggest that the urbani-
sation process was already underway by the sixth century BC. This is
chronologically consistent with what has been said about economic inten-
sification and the signs of social differentiation in the megalithic necropo-
lis. The existence of a powerful defensive wall from the fourth century BC
can also be an indication of the importance and possible urban category of
Althiburos.
The final conclusion to be drawn bringing together, in the most prudent
way, all these pieces of information, is that the formation of the Numidian
city of Althiburos, and likely of the archaic state to which it belonged, began
by the mid-first millennium BC. This was probably the final result of a long
process of population growth and consequent pressure on the environment,
decisively accentuated in the eighth century BC. This process, however,
must certainly have started in the course of the second millennium BC,
prior to the arrival of the first Phoenician settlers along the North African
coast. Research on this earlier period, completely unknown at present in the
Eastern Maghrib, is long overdue. But it is a crucial topic if we are to paint
the great picture of North African Protohistory.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 469

It is also clear that Althiburos did not stand alone in the first
millennium BC, but must in some sense be illustrative of ‘Numidian’
developments more generally. Nor was this the only autochthonous tran-
sitioning to urbanisation and statehood. State formation in Fazzan see-
mingly followed a similar pathway, with increasing demographic pressure
on the environment as the key element that may explain the formation of
the vast Garamantian kingdom.62
It goes without saying that, in the present state of research, little can be
stated about the specific ways in which state formation developed. To do
so, it would be necessary to gather more extensive information on the
nature and internal organisation of the settlements, as well as on the spatial
structure of the necropoleis, which would hopefully provide information
about the specific forms of social organisation. As a matter of fact, all we
can say in this regard is that the classical texts and epigraphic documents of
the Roman period attest the persistence of powerful tribal structures at the
end of the Numidian period, and even in the Roman Imperial period.63 It is
possible, or even likely, that these tribal chiefdoms constituted, at some
moment in the mid-first millennium BC, polities of an intermediate stage
between the local group (which we think was still dominant by the early
first millennium BC) and the monarchic states attested by the Greco-
Roman written sources. The latter were strongly constituted in the third
century BC, but their origins must certainly date back at least to the fourth
century.64 This view is consistent with Justin’s indication that the kingdom
of the Mauri already existed by the mid-fourth century BC.65 The same
probably holds true for the Massyle kingdom, since Massinissa’s father,
Gaia, was the heir of a dynasty. It is likely that one of his ancestors was
Aylimas, the ‘Libyan king’ mentioned by Diodorus as an ally of
Agathocles.66 The precise moment when these kingdoms were constituted
is obviously unknown, but it is not unreasonable to think that, as proposed
by Kotula many years ago, it was related to the territorial Carthaginian
expansion from the fifth century BC. Maybe this is the differentia specifica
that may explain the particular pathway followed by the North African
states as opposed to the much smaller polities that developed on the north
shores of the western Mediterranean.

62
Mattingly et al. 2003. See now, Mattingly and Sterry, Chapter 19, this volume.
63
Fentress 2006; Gsell 1927; Lassère 2001.
64
A similar dating is proposed by Bokbot, Chapter 12, this volume for the rise in the Western
Maghrib of mighty tribal confederations, or even kingdoms.
65
Justin, Epitome 18.6. 66 Diodorus Siculus, Bibliotheca Historica 20.17.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 470 Joan Sanmartí et al.

May we assume that the contact with the Phoenicians had other effects on
urbanisation and state formation processes? Maybe the aspect where it could
have been the most decisive is the diffusion of iron metallurgy. However, as
noted above, the data recovered at Althiburos suggest the possibility of
a local development of this technology. Another aspect where it could
have been important, is in the introduction of prestige goods that could
have been given a high status and could have been used as dowries in
matrimonial exchanges, as well as in the creation of social debts, particularly
through feasting.67 This could have caused a rise of the price of dowries
among certain lineages that had easier access to imported products, and thus
promoted the creation of closed circuits of marriage exchange of superior
status. However, even in this situation we should be able to explain the
reasons why the society would have accepted the consolidation of these
hierarchies and institutionalised inequality, which leads us back to the role
of elites in solving the problems of the subsistence economy.
On the other hand, and returning to the specific material data, the
hypothesis of Phoenician imports used in the context of a prestige goods
economy can be perfectly consistent with their extremely small number
(no more than 1 per cent of ceramic materials at Althiburos) until the
fourth century BC, by which period hereditary inequality must have
been well established. However, it seems difficult to understand their
absence (although there may exist ideological reasons to explain it)
among the tableware found in the tumulus of a seemingly elite grave
such as monument 53, which, as already mentioned, has delivered
exclusively locally hand-made vessels. It is also worth recalling that
one dolmen at Henchir Mided has produced a fifth century BC Attic
black-glazed cup.
Despite the contribution of the Althiburos project, the state formation
process in pre-Roman North Africa is still, at the beginning of the twenty-
first century, the most poorly understood in the whole western
Mediterranean area. Our work has provided some information that may
be useful to its understanding, but it is only a small glimmer on an area of
hundreds of thousands of square kilometres that remain in absolute dark-
ness. To turn this glimmer into a powerful spotlight will require the
continuation of this project for many years. The next steps are to better
understand the structure and chronology of the city and the megalithic
necropolis, and to contextualise this information by digging some of the

67
Regarding the model of prestige-goods economy and its application to an archaeological case
study, see Frankenstein and Rowlands 1978.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 471

periurban settlements. But even if this desideratum were accomplished

(which is, at present, very uncertain), the information recovered would still
have a purely local character, and could hardly be considered as represen-
tative of the process of state formation on the whole of the territory
occupied by the Numidian kingdoms. Other projects, and probably other
insights, are, therefore, necessary (preferably within the framework of an
international well-coordinated programme) in order to bring to light the
evidence that may clarify the emergence of pre-Roman states in both the
Maghrib and Sahara.68

References

Belarte, M.C. 2011. Les sondages dans la zone 1. In Kallala and Sanmartí 2011,
45–110.
Belkahia, S. and Di Vita-Évrard, C. 1995. Magistratures autochtones dans les cités
pérégrines de I’Afrique romaine. In P. Trousset, P. (ed.), Monuments funéraires,
institutions autochtones en Afrique du Nord antique et médiévale, VIe colloque
international sur I’histoire et l’archéologie de l’Afrique du Nord. Paris: Editions
du CTHS, 255–74.
Berthier, A. 1980. Un habitat punique à Constantine. Antiquités africaines 16: 13–26.
Blanton, R.E. 1998. Beyond centralization. Steps toward a theory of egalitarian
behaviour in archaic states. In Feinman and Marcus 1998, 135–72.
Bondarenko, D.M., Grinin, L.E. and Korotayev, A.V. 2002. Alternative pathways of
social evolution. Social Evolution & History 1.1: 54–79.
Brun, P. 2015. L’évolution en dents de scie des formes d’expression du pouvoir
durant l’âge du Fer en Europe tempérée. In C. Belarte, D. Garcia and J. Sanmartí
(eds), Les estructures socials protohistòriques a la Gàllia i Ibèria (Homenatge
a Aurora Martín i Enriqueta Pons), VII Reunió Internacional d’Arqueologia de
Calafell (Calafell, del 7 al 9 de març de 2013). Barcelona: Institut Català
d’Arqueologia Clàssica, 49–59.
Camps, G. 1960. Les traces d’un Âge du Bronze en Afrique du Nord. Revue
africaine 104: 31–55.
Camps, G. 1961a. Monuments et rites funéraires protohistoriques: Aux origines de la
Berbérie. Paris: Arts et Métiers Graphiques.
Camps, G. 1961b. Aux origines de Ia Berbérie. Massinissa ou les débuts de I’histoire.
(Libyca 8). Algiers: Service des Antiquités.
Camps, G. 1995. Les nécropoles mégalithiques de l’Afrique du Nord. In P. Trousset
(ed.), Monuments funéraires, institutions autochtones en Afrique du Nord

68
This research has been funded by the Spanish Ministry of Economy and Competitiveness
(projects HAR2012-39189-C02-01 and HAR2012-39189-C02-02).

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 472 Joan Sanmartí et al.

antique et médiévale, VIe colloque international sur I’histoire et l’archéologie de

l’Afrique du Nord. Paris: Editions du CTHS, 17–31.
Camps, G. and Camps-Fabrer, H. 1964. La nécropole mégalithique du Djebel
Mazela à Bou Nouara. Mémoires du Centre de recherches anthropologiques,
préhistoriques et ethnographiques, III. Paris: Arts et métiers graphiques.
Cantero, F.J. and Piqué, R. 2016. Ressources forestières à partir de l’étude des
charbons de bois. In Kallala et al. 2016, 491–515.
Crumley, C. 1995. Building an historical ecology of Gaulish polities. In B. Arnold and
D.B. Gibson (eds), Celtic Chiefdom, Celtic State. The Evolution of Complex Social
Systems in Prehistoric Europe. Cambridge: Cambridge University Press, 26–33.
Ennaïfer, M. 1976. La cité d’Althiburos et l’ëdifice des Asclëpieia. Bibliothèque
Archéologique, vol. I. Tunis: lnstitut National d’Archéologie et d’Art.
Feinman, G.M. and Marcus, J. (eds). 1998. Archaic States. Santa Fe: School of
American Research Press.
Fentress, E. 2006. Romanizing the Berbers. Past and Present 190: 3–32.
Ferjaoui, A. 2010. Les relations entre Carthage et l’intérieur de l’Afrique. Le cas de
Zama Regia et sa région. In A. Ferjaoui (ed.), Carthage et les autochtones de son
empire du temps de Zama. Hommage à Mhamed Hassine Fantar. Tunis: Institut
National du Patrimoine, 341–52.
Frankenstein, S. and Rowlands, M.J. 1978. The internal structure and regional
context of Early Iron Age society in south-western, Germany. Bulletin of the
Institute of Archaeology 16: 73–112.
Gatto, M.C., Mattingly, D.J., Ray, N. and Sterry, M. (eds). 2019. Burials, Migration
and Identity in the Ancient Sahara and Beyond. Trans-Saharan Archaeology, vol.
2. Series editor D.J. Mattingly. Cambridge: Cambridge University Press.
Godelier, M. 1980. L’État: Les processus de sa formation, la diversité de ses formes
et de ses bases. Revue internationale des sciences sociales 32.44: 657–71.
Godelier, M. 1999. Chefferies et États, une approche anthropologique. In Ruby
1999, 19–30.
Gsell, S. 1927. Histoire ancienne de l’Afrique du Nord, t. V. Les royaumes indigènes.
Organisation sociale, politique et économique. Paris: Librairie Hachette.
Harris, M. 1995. Anthropology and postmodernism. In M.F. Murphy and M.
L. Margolis (eds), Science, Materialism and the Study of Culture. Gainesville:
University Press of Florida, 62–77.
Johnson, A.W. and Earle, T. 2000. The Evolution of Human Societies from Foraging
Group to Agrarian State. Second Edition. Stanford, California: Stanford
University Press.
Kallala, N. 2002. Archéologie romaine et colonisation en Afrique: Enjeux et
pratiques. (L’exemple du Kef (Sicca Veneria) et de sa région, dans le nord-
ouest de la Tunisie). Africa 19: 57–81.
Kallala, N. and Sanmartí, J. (eds). 2011. Althiburos I. La fouille dans l’aire du
capitole et dans la nécopole méridionale (Documenta 18). Tarragona: Institut
Català d’Arqueologia Clàssica.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 473

Kallala, N., Sanmartí, J., Jornet, R., Belarte, M.C., Canela, J., Chérif, S., Campillo, J.,
Montanero, D., Miniaoui, S., Bermúdez, X., Fadrique, T., Revilla, V., Ramon, J.
and Ben Moussa, M. 2014. La nécropole mégalithique de la région d’Athiburos,
dans le massif du Ksour (Gouvernorat du Kef, Tunisie). Fouille de trois monu-
ments. Antiquités africaines 50: 15–52.
Kallala, N., Sanmartí, J. and Belarte, M.C. (eds). 2016. Althiburos II. L’aire du
capitole et la nécropole méridionale: Etudes (Documenta 18). Tarragona: Institut
Català d’Arqueologia Clàssica.
Kallala, J., Sanmartí, J. and Belarte, M.C. (eds). 2018. Althiburos III. La nécropole
protohistorique de Althiburos (massif du Ksour). Tarragona: Institut Català
d’Arqueologia Clàssica.
Khanoussi, M. and Von Rummel, P. 2012. Simitthus (Chimtou, Tunesien).
Vorbericht über die Aktivitäten 2009–2012. Mitteilungen des deutschen
archäologischen Instituts, römische Abteilung 118: 179–222.
Khanoussi, M., Ritter, S. and Von Rummel, P. 2005. The German-Tunisian Project
at Dougga: First results of the excavations south of the Maison du Trifolium.
Antiquités africaines 40–41: 43–66.
Kim, J. 2001. Elite strategies and the spread of technological innovation: The
spread of iron in the Bronze Age societies of Denmark and Southern Korea.
Journal of Anthropological Archaeology 20.4: 442–78.
Klejn, L. 1993. La Arqueología soviética: Historia y teoría de una escuela descono-
cida. Barcelona: Crítica.
Kotula, T. 1976. Masynissa. Warsaw: PIW.
Lassère, J.-M. 2001. La tribu et le monarche. Antiquités africaines 37: 140–55.
Lévêque, P. 1986, L’émergence des pouvoirs structurés dans I’Afrique mineure
de l’Age du Fer. In Gli interscambi culturali e socio-economici fra I’Africa
settentrionale e I’Europa mediterranea, Atti del Congresso Internazionale di
Amalfi (5–8 dicembre 1983). Naples: Istituto Universitario Orientale,
637–51.
Lévêque, P. 1999. Avant et après les Princes. L’Afrique mineure de l’Age du fer. In
Ruby 1999, 153–64.
López Reyes, D. and Cantero, F.J. 2016. Agriculture et alimentation à partir de
l’étude des restes de graines et des fruits. In Kallala et al. 2016, 449–90.
Lull, V. and Risch, R. 1995. El estado argárico. Verdolay 7: 97–109.
Marcus, J. 1998. The peaks and valleys of ancient states: an extension of the
dynamic model. In Feinman, and Marcus 1998, 59–94.
Marcus, J. 2008. The archaeological evidence for social evolution. Annual Review of
Anthropology 37: 251–66.
Marras G., Doro L., Floris R. and Zedda M. 2009. Il dolmen 102. Nota preliminare.
In G. Tanda, M. Ghaki and R. Marras (eds), Storia dei paesaggi preistorici
e protostorici nell’Alto Tell tunisino. Missioni 2002–2003. Cagliari: Università
degli Studi di Cagliari and Ministère de la Culture et de la Sauvegarde du
Patrimoine (Tunisia), 179–200.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 474 Joan Sanmartí et al.

Mattingly, D.J. 2016. Who shaped Africa? The origins of urbanism and agriculture
in Maghreb and Sahara. In N. Mugnai, N. Ray, and J. Nikolaus (eds), De Africa
Romaque: Merging Cultures across North Africa. London: Society for Libyan
Studies, 11–25.
Mattingly, D.J., Reynolds, T. and Dore, J. 2003. Synthesis of human activities in
Fazzan. In D.J. Mattingly (ed.), The Archaeology of Fazzan, vol. 1, Synthesis.
London: Society for Libyan Studies, 327–73.
Parkinson, W.A. and Galaty, M.L. 2007. Secondary states in perspective: an
integrated approach to state formation in the prehistoric Aegean. American
Anthropologist 109.1: 113–29.
Portillo, M. and Albert, R.M. 2011. Husbandry practices and livestock dung at the
Numidian site of Althiburos (el Médéina, Kef Governorate, northern Tunisia):
the phytolith and spherulite evidence. Journal of Archaeological Science 38.12:
3224–33.
Portillo, M. and Albert, R.M. 2016. Les activités domestiques de la période numide
à travers l’étude des micro-restes végétaux et fécaux: phytolithes et sphérolithes.
In Kallala et al. 2016, 517–27.
Portillo, M., Valenzuela, S., and Albert, R.M. 2012. Domestic patterns in the
Numidian site of Althiburos (northern Tunisia): The results from a combined
study of animal bones, dung and plant remains. Quaternary International 275:
84–96.
Ramon Torres, J. and Maraoui Telmini, B. 2011. Les sondages dans la zone 2. In
Kallala and Sanmartí 2011, 153–262.
Renfrew, C. 1986. Peer polity interaction and social change. In C. Renfrew and
J.F. Cherry (eds), Peer Polity Interaction and Social Change. Cambridge:
Cambridge University Press, 1–18.
Rodríguez, A., Pavón, I. and Duque, D. 2010. Población, poblamiento y modelos
sociales de la Primera Edad del Hierro en las cuencas extremeñas del Guadiana
y Tajo. Arqueología Espacial 28: 41–64.
Ruby, P. (ed.). 1999. Les princes de la protohistoire et l’émergence de l’État. Actes de
la table ronde internationale de Naples. Naples/Rome: Collection du Centre Jean
Bérad 17/Collection de l’École Française de Rome, 252.
Sanmartí, J. 2009. From the archaic states to romanization: A historical and
evolutionary perspective on the Iberians. Catalan Historical Review 2: 9–32.
Sanmartí, J., Kallala, N., Belarte, M.C., Ramón, J., Maraoui, B., Jornet, R. and
Miniaoui, S. 2012. Filling gaps in the Eastern Maghreb’s Protohistory: The
Althiburos archaeological Project (el Kef, Tunisia). Journal of African
Archaeology 10.1: 21–44.
Sanmartí, J., Kallala, N., Jornet, R., Belarte, M.C., Canela, J., Chérif, S., Campillo, J.,
Montanero, D., Bermúdez, X., Fadrique, T., Revilla, V., Ramon, J., Ben
Moussa, M. 2015. Roman Dolmens? The Megalithic necropolises of Eastern
Maghreb revisited. In M. Díaz-Guardamino, L. García Sanjuán and D. Wheatley

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 11 Numidian State Formation in the Tunisian High Tell 475

(eds), The Lives of Prehistoric Monuments in Iron Age, Roman, and Medieval
Europe. Oxford: Oxford University Press, 287–304.
Sanmartí, J., Cruz Folch, I., Campillo, J. and Montanero, D. 2019. Numidian burial
practices. In Gatto et al. 2019, 249–80.
Shanks, M. and Tilley, C. 1987. Social Theory and Archaeology. Cambridge: Polity
Press.
Smadja, E. 1983. Modes de contact, sociétés indigènes et formation de l’Etat
numide au second siècle avant notre ère. In Modes de contacts el processus de
transformations dans les sociétés anciennes, Actes du Colloque de Crotone (1981).
Pisa/Rome: Scuola normale superiore di Pisa/École française de Rome, 685–702.
Stambouli-Essassi, S., Roche, E. and Bouzid, S. 2007. Evolution of vegetation and
climatic changes in North-Western Tunisia during the last 40 millennia. Geo-
Eco-Trop 31: 171–214.
Thébert, Y. 1992. Bulla Regia, Encyclopédie Berbère 11, 1647–53.
Trigger, B. 1989. A History of Archaeological Thought. Cambridge: Cambridge
University Press.
Trigger, B. 1998. Sociocultural Evolution: Calculation and Contingency. Oxford:
Blackwell Press.
Valenzuela-Lamas, S. 2016. Alimentation et élevage à Althiburos à partir des restes
fauniques. In Kallala et al. 2016, 421–48.
Van der Veen, M. and Westley, B. 2010. Palaeoeconomic studies. In D.J. Mattingly
(ed.), The Archaeology of Fazzan. Volume 3, Excavations Carried out by C.M.
Daniels. London: Society for Libyan Studies, 488–522.
Wolf, E.R. 1966. Peasants. Englewood Cliffs, New Jersey: Prentice-Hall, Inc.
Wright, E.O. 1983. Giddens’s critique of Marxism. New Left Review 138: 11–35.

https://doi.org/10.1017/9781108637978.012 Published online by Cambridge University Press

 12 The Origins of Urbanisation and Structured
Political Power in Morocco
Indigenous Phenomenon or Foreign Colonisation?

youssef bokbot

Introduction

This chapter reviews the evidence for the emergence of urbanisation in

Morocco and evaluates the contributions of Phoenician settlers and of
local communities to this process. In line with the new evidence from
Tunisia, the previously emphatic emphasis on the role of outsiders
in introducing urbanism and sparking the initial steps towards state
formation now seems less convincing, with a much more significant
role for local actors.1
After first reviewing the ancient source evidence and the history of
archaeological research, I shall present the latest archaeological findings
based on my personal experience of a number of key settlements, as well as
some impressive funerary monuments that illustrate the emergence of
hierarchy within early Berber (Imazighen) society (Fig. 12.1).
The rarity of identified Iron Age settlements in North Africa has led
historians and archaeologists to suggest that a nomadic pastoral economy
predominated in the Protohistoric period, despite hints that the sedentary
agrarian society began as early as the Neolithic period.2 Both ancient and
modern authors alike have not hesitated to paint a dark picture of the
populations of the Maghrib, who they characterise as barbarians, still
limited to producing tools from flint, condemned to a state of stagnation
and isolation that only the arrival of the Phoenicians and Romans could
interrupt.
Some classical authors presented the populations of the Maghrib as
eternal wanderers who ignored agricultural activity and sedentary life.
Strabo, Valerius Maximus, Appian and Polybius all claimed that Numidia

1 2
476 See Sanmartí et al., Chapter 11, this volume. Martínez-Sánchez et al. 2018.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 477

Figure 12.1. Map of sites mentioned in this chapter.

was valueless before Massinissa and incapable of agricultural production.3

According to Strabo, Massinissa rendered the Numidians sociable and made
them into farmers.4 Similarly, Appian wrote that divine favour allowed him
to develop a vast country where previously the Numidians ate only grass
because they did not cultivate the land themselves.5
Such attitudes were readily embraced by colonial-era scholarship, which
was naturally inclined to the view that innovations were brought to Africa
by outsiders.6 According to this discourse, the ancestors of the Imazighen
were condemned to a completely passive role. They are presented as having
received a ready-formed civilisation from the East that was transported by
a handful of Phoenician seafarers from the Levant.
Herodotus, on the other hand, specified that ‘at the river Triton, [the Gulf
of Gabes in the south-east of Tunisia], there are Libyan farmers . . . They

3
Gsell 1927, 169–212, especially 186–90. 4 Strabo, Geography 17.3.15.
5
Appian, Historia Romana 106; Gsell 1927, 187.
6
See Mattingly 2011, 43–72; 2016 for a full discussion of the impact of the colonialist discourse.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 478 Youssef Bokbot

have houses and are called Maxyes’.7 Herodotus contrasted eastern Libya,
where nomads lived, to western Libya, which was mountainous, wooded, full
of wild animals and inhabited by sedentary agriculturalists.
What then was the actual state of the Imazighen civilisation before
the arrival of the Phoenicians? Evidence for agricultural activity and
pre-Phoenician settlements is well established. With the development of
archaeological research, it has now been demonstrated that at the arrival of
the Phoenicians, the Imazighen were not uncivilised or primitive barbarians.
Rock art discovered in the Central High Atlas dating back to the Bronze Age
depicts scenes of agricultural activities. A ploughing scene is engraved on the
cliffs of the Yaggour at Azib n’Ikkis, and Jean Malhomme interprets the bent
lines present at Oukaïmedden as representing sickles.8Archaeobotanical
evidence for the introduction and cultivation of cereals in Morocco dates
as early as the Neolithic as recent finds demonstrate.9 Cereals were an
established resource already in the eighth century BC, suggested by the
frequent representation of iron sickles among the funerary offerings in
indigenous burials in the hinterland of Tangier at that time.10
Data from archaeological excavations thus often contradict the ancient
sources and raise several questions. Can one speak of Phoenician trading
posts everywhere? Were there only Phoenician trading posts on the coast?
What was the nature of these settlements? What differentiated them from
other indigenous settlements?
The existence of Phoenician trading posts is well attested in texts. In
particular, the Greek geographer Pseudo-Scylax enumerates a series of
relevant points, towns and trading posts on the coasts of the Maghrib.11
When Pseudo-Scylax wrote about Phoenician trade with the Atlantic
regions of Morocco, he drew a detailed, and very different, picture from
that described by Herodotus. The Phoenicians no longer had dealings with
primitive and fearful natives who fled all contact with civilised peoples. The
Ethiopians (a term used to describe the Imazighen) lived in a city and
imported a variety of goods, indicating a relatively developed and complex
society.12 This passage from Pseudo-Scylax may relate to trade with the
ancient city of Lixus, especially since, in another passage, the same author
indicates that the Ethiopians also had a great city where Phoenician
merchant ships went to trade.13

7
Herodotus, Histories 4.191; Gsell 1916, 29.
8
Malhomme 1953, figure 1, 384; de Torres and Ruiz-Gálvez 2014.
9
Martínez-Sánchez et al. 2018. 10 Ponsich 1967; 1969; 1970.
11
Gsell 1927, 160; Shipley 2012. 12 Psuedo-Skylax, Periplous 112; Villard 1960, 21.
13
Gsell 1927, 113.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 479

Therefore, we have strong reasons to believe that these trading posts

were not founded by the Phoenicians alone, but rather were established
alongside existing indigenous settlements. However, much of the
research in the last century, instead of relying on physical evidence
revealed by archaeological excavations, continued to deny any urban
character to the paleo-Amazigh world. In perpetuating this narrative,
some local and foreign researchers continue to describe any ancient
settlements where imported material is discovered as Phoenician, Punic
or Roman.14
Unfortunately, the excavations carried out at Moroccan sites during
the first half of the twentieth century did not produce any valid data able
to refute these arguments. During this period, the emphasis was placed
on monumental archaeology, rather than stratigraphic excavations. The
archaeological syntheses produced during this period were too greatly
influenced by the writings of the Greek and Latin authors. Pierre Cintas,
one of the great scholars of Phoenician/Punic culture has drawn a
contentious conclusion. According to him, the excavations confirm
the existence of Phoenician trading posts discussed in the texts: ‘an
excellent mooring in the centre of an island, cape or estuary could not
have escaped the eye of the Phoenician sailors’.15 Everything is
described as if these coastal areas, overrun by the Phoenicians, were
completely unoccupied and these foreigners from the Lebanese coast
settled where they wanted without worrying about the resistance of local
populations.
In recent years the development of archaeological research in Morocco
and western Algeria has sparked new questions concerning the origins of
some cities and trading posts, usually considered as Phoenician or Punic
foundations. In the second half of the twentieth century, archaeological
excavations had already begun to reveal occupation levels beneath the
Punic and Roman layers belonging to an indigenous culture possessing
agriculture knowledge, a sedentary lifestyle and even early urbanisation.16
In order to reach a more objective and complete reconstruction of the
history of these coastal sites we must put the ancient texts to the test using
archaeological data, studying them in their entirety, from an emic perspec-
tive, limiting bias and influence from outside sources such as ancient texts.
This chapter will focus on a selection of sites (Lixus, Mogador, Kach
Kouch, Sidi Driss, Mzora and Sidi Slimane), but it needs stating at the
14
See Mattingly et al. 2017 for discussion of imported material in the Maghrib and Sahara more
generally.
15
Cintas 1954. 16 Camps 1961a, 49.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 480 Youssef Bokbot

outset that knowledge of the Protohistoric development of a number of

other sites is also advancing in important ways (Rirha, Volubilis, Sala,
Thamusida, Tingi, Tamada, Ceuta, Melila).17

The Ancient City of Lixus

The earliest traditions refer to Lixus as being among the oldest cities
in the western Mediterranean basin. The ancient texts trace its founda-
tion to the same period as that of Gades in the eleventh century
BC. These sources are contradicted, however, by the evidence from
archaeological excavations. Surveys carried out by Miguel Tarradell in
the ancient city of Lixus have provided a basic stratigraphy: a level of
occupation characterised by an abundance of hand-made ceramic
material, which he described as belonging to a ‘Neolithic tradition’.18
Excavations undertaken at the same site by Michel Ponsich confirmed
the existence of this distinct archaeological layer.19 The pottery recov-
ered in this layer was hand-made, generally smooth or polished and
its surface was rarely decorated, although sometimes a decorative
horizontal band appears surrounding the top of the vessel. When
present, this band was either embossed using fingers, or decorated
with an alignment of impressed patterns.
The most determinative piece of evidence for a pre-Phoenician
phase, however, is the discovery of a new type of pottery, labelled
‘graffito’ (Fig. 12.2, 1–2), previously unknown in North Africa, and
for which analogies are found in pre-Phoenician ceramic traditions
of the southern Iberian Peninsula.20 As a whole, the forms and
patterns of the pottery from the lower levels of Lixus are almost
identical to those of the Bronze Age levels of the Ghar Cahal and
Khaf Taht el Ghar caves (Figs 12.2, 3–11).21 Similar examples of
‘grafitto’ style pottery were also discovered during the
recent excavations undertaken by the joint Moroccan and Spanish
team, INSAP-Universidad de Valencia (Figs 12.2, 12–13). This dis-
covery further demonstrates the relative abundance of this pottery
type in the pre-Phoenician levels of Lixus.
This archaeological material did not exist in isolation, but was rather
found in association with relatively archaic domestic structures. Michel

17
Callegarin et al. 2016; Papi 2019; Papi and Akerraz 2008. 18 Tarradell 1954, 790.
19
Ponsich 1981, 131. 20 Bokbot 1991, 198, 321. 21 Bokbot and Onrubia-Pintado 1992, 20.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 481

Figure 12.2. Ceramics from Lixus: 1)–2)‘Graffito’ pottery from Lixus (Ponsich
Survey); 3)–5) Pre-Phoenician pottery from Lixus; 6)–11) Bronze Age pottery,
Ghar Cahal and Kahf Taht el Ghar caves; 12) Vase with incised decoration
applied after firing, Lixus (recent survey); 13) Lixus (recent survey) sherd with
graffito.

Ponsich pointed out that certain wall construction techniques and certain
pre-Roman walls at Lixus do not correspond with those found in the
Phoenician east. Furthermore, Ponsich pointed out that the city opened
onto the countryside through a megalithic door preceding the Punic gate,
which led to a road lined with ancient tombs.22
The presence of these settlement structures with megalithic character-
istics, as well as the hand-made ceramics, make it possible to envisage
the prospect of a pre-Phoenician settlement at Lixus. The assertion of
early local occupation is sustainable, especially since the site of Lixus and
its immediate surroundings produced metallic objects attributable to the
Bronze Age. At the Second Pan-African Congress on Prehistory, Bernardo
Sáez Martín announced the discovery of a bronze sword at the mouth
of the Loukkos (Fig. 12.3).23 Although then lost, this was subsequently
re-located, some 30 years later, in the Museum for Pre- and Early History,
Berlin. This sword is generally described as a ‘Ballintober’ type, which dates

22 23
Ponsich 1988, 86. Sáez Martín 1955, 659.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 482 Youssef Bokbot

Figure 12.3. Sword from the Loukkos, used with permission (Brandherm 2007).

to the Atlantic Late Bronze Age c.1300–1150.24 Georges Souville argues

instead that it can only be compared to examples coming from the deposi-
tion of the Ria de Huelva and is more akin to the ‘Ballintober’ type. In
addition, Enrique Gozalbes-Gravioto reported the presence of a flat bronze
axe in the collections of the Tetuan Museum, the source of which may be
the site of Lixus.25
It is likely that these metal objects are directly related to the funerary
megaliths from the site of Lixus and its surroundings. Paul Pallary recorded
the presence of so many dolmens between the town of Larache and the ruins
of Lixus that it led him to conclude that dolmen burials were indigenous to
Lixus.26 Tarradell reported two burials composed of large blocks of stone at
Lixus and according to his description, one of them could potentially coin-
cide with the extramural tomb discovered in the nineteenth century by the
French diplomat Charles Tissot and cleared by Henri De la Martinière.27 It
consists of a megalithic tomb composed of a corridor made from large
vertically placed slabs and covered by five transverse juxtaposed tiles.28
Locally named ‘al-Quantara’ (Fig. 12.4), this monument, which corresponds
typologically to an ‘allée couverte’, remains a unique excavated specimen in
Morocco. However, on the hillock where it stands, I noticed several large
slabs of similar type which raises the possibility that the al-Quantara ‘allée
couverte’ monument is one of a group of megalithic tombs.

The Pre-Roman Settlement on the Island of Mogador

Pierre Cintas reported the recovery of hand-made ceramic material in the

oldest layers of the site of Mogador and concluded that it was locally
24
It was originally described as a ‘Rosnoën’ blade by Ruiz-Gálvez 1983, 64. Souville 1995, 248
argued in favour of the ‘Ballintober’ classification. The most up-to-date publication of the
sword is Brandherm 2007, 34, 134 – A 1.
25
Gozalbes-Gravioto 1975, 14. 26 Pallary 1907, 308; 1915, 195. 27 Tarradell 1960, 167.
28
Bokbot 1991, 181.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 483

Figure 12.4. The ‘allée couverte’ tomb of al Quantara.

produced and preceded the Punic trading posts (Fig. 12.5a).29 The pottery
usually has a smooth or polished surface and little variation in form –
mainly large vases with a flat bottom with applied cords decorated either
digitally or incised with a nail. These cords were placed horizontally
around the neck and serpentiform on the body. The vessels have similar
characteristics to those found in the Bronze Age levels of the El-Khill, Ghar
Cahal, Khaf Taht el Ghar and Dar es Soltane caves,30 as well as those found
in the lower layers of Lixus.31 André Jodin has drawn attention to the
resemblance of this pottery type to those of the European Bronze Age.32
Excavations recently carried out at the site of Mogador by the Moroccan-
German mission (INSAP-DAI Madrid) confirmed the presence of this type
of pottery (Fig. 12.5b). Although the stratigraphic position of their prove-
nance is not entirely clear, it seems probable that they came from the
deepest levels.

29 30 31 32
Cintas 1954, 41. Jodin 1966, 166. Bokbot 1991. Jodin 1957, 37.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 484 Youssef Bokbot

Figure 12.5. a) Mogador: hand-made pottery with decoration applied with fingers; b)
Mogador (recent survey): hand-made pottery.

Excavations have also uncovered the remains of furnaces, nozzles and

iron slag, representing the oldest testimonies of metallurgical activities
related to iron in Morocco. According to recent surveys, the iron ore
originated from sites approximately 10 km north of Essaouira.33
Moreover, geomorphological studies carried out by a Moroccan–German
team have yielded important results concerning changes in the configura-
tion of coastal shores from the early Neolithic to the present. During the
period of Phoenician expansion, Mogador was not an island, but rather a
peninsula contiguous with the continent. This new piece of information
has opened the debate on the geographical nature of the location of
Phoenician trading posts on the Moroccan coast and on the nature of
relations between colonial and indigenous populations.

The Hilltop Settlement at Kach Kouch

The summit of the hill of Kach Kouch, also known as Dhar el-Moudden,
consists of a series of calcareous outcrops, partly blocked by sediment,
which terminate in steep slopes. These outcrops delimit a roughly circular

33
El Khayari 2007, 57.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 485

Figure 12.6. Kach Kouch plateau, overlooking the lower Oued Laou valley.

area, with a diameter of approximately 40 m and provide the site with

natural fortifications. The hill of Kach Kouch is located on the right bank of
the Oued Laou, at the beginning of the meander indicating the approach of
the mouth. Located about 70 m above sea level, the hill dominates the great
alluvial plain of the lower valley of Oued Laou towards the north-east
(Fig. 12.6).
The first investigations carried out on the site in 1988 produced
surface finds, notably fragments of indigenous hand-made ceramic
traditions, mixed with rare fragments of amphorae produced in the
Phoenician tradition, which led to the conclusion that there was a
broad phase of Protohistoric occupation. The subsequent archaeological
surveys in 1992 were carried out with the intention of determining the
presence of archaeological structures dating to that period, as well as to
ascertain if the sedimentary fill showed signs of stratified anthropogenic
deposits. Four trenches were dug perpendicular to the axis of a recent
threshing floor that occupied the centre of the hill. Only two of the
trenches produced evidence of structures and allowed for observations
on the stratigraphy.34 The stratigraphic profiles of test trenches B and D
are as follows:

34
Bokbot and Onrubia-Pintado 1995.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 486 Youssef Bokbot

• Level 1: A disturbed upper deposit of agricultural soil comprising a fine

brown sediment and redeposited archaeological material from deeper
deposits.
• Level 2: A middle layer created distinguishable in part by the accumula-
tion of thick fragments of orange-coloured daub, composed of reddish
clay mixed with stems of herbaceous plants and retaining the imprints of
the wooden rods on which it was applied.
• Level 3. An occupation level, slightly offset from the line of the steep
slopes, composed of earthy brown to greyish compacted local sediments
resting directly on the limestone substrate. This bedrock, which has
many cracks, has been artificially modified by the installation of struc-
tures, and the digging of pits and funerary trenches.
The structures revealed in the trenches can be grouped into three
distinct categories: combustion features (hearths), semi-cylindrical pits,
possibly for storage, and burial trenches. The hearths were characterised
by two small paving stones, roughly circular in form and reddened by
thermal action. The semi-cylindrical storage pits, which were on average
0.4 m in diameter and 0.3 m in depth, appeared in both trenches (Fig.
12.7a). Eight of these pits were located in trench D and three in trench B.
The six burial trenches, on the other hand, were exclusively found in trench
B, across its entire surface (Fig. 12.7b). These burial trenches were

Figure 12.7. a) Kach Kouch: storage structures; b) Kach Kouch: archaeological

structures revealed during excavation.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 487

Figure 12.8. Kach Kouch, vase decorated with ‘graffito’.

elongated and shallow and were carved into the bedrock. They contained
the remains of three adults and three children buried in extended right
lateral position. The burials were all oriented north-south, heads turned
towards the east and hands resting on the pelvis.
Among the ceramic repertory, indigenous hand-made types coexist with
Phoenician wheel-thrown pottery. The hand-made pottery is represented
by a series of fairly open flat-bottomed storage jars, often decorated at the
shoulder with incisions or impressions made either directly on the wall of
the vessel or on an applied cord. Some vessels were equipped with gripping
elements in the form of an inverted crescent. A small bowl decorated in
‘graffito’, similar to those from Lixus, was also found. It has a concave font
and its thin walls were carefully polished (Fig. 12.8).35 In the absence of
absolute dates, the ceramics allow this Protohistoric settlement to be dated
within a chronological range of the ninth to the sixth century BC.

The Protohistoric Settlement of Sidi Driss

The site of Sidi Driss is located on the Mediterranean coast, halfway

between Al-Hoceima and Nador. It occupies the last hill on the left bank
of the Oued Amekrane, overlooking the mouth of the river.36 Sidi Driss
was a coastal settlement contemporary with the Phoenician expansion,
overlaid by the current village of Sidi Driss. The archaeological surface
material collected makes it possible to attribute the oldest occupation to the
eighth and seventh centuries BC. Amphorae types from Sidi Driss are
possibly related to the Rachgoun 1 and 4 types, their closest typological
parallel being those from Mersa Madakh on the neighbouring Orense
coast.37 This imported material coexists with indigenous hand-made

35 36
Bokbot and Onrubia-Pintado 1995, 223. Kbiri-Alaoui et al. 2004.
37
Vuillemot 1965, 27–28.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 488 Youssef Bokbot

pottery decorated with finger impressions, analogous to those found in the

Bronze Age levels of the caves of north-west Morocco.38

Other Sites in Northern Morocco and Western

Algeria

If we extend our geographical scope to the nearby regions along the north-
eastern coast of Morocco, we find the site of The Andalusians, west of
Oran, which occupies the edge of a steep cliff at the bottom of a wide bay
and consists of a settlement and a necropolis partially excavated by G.
Vuillemot. One of the investigations carried out on the settlement revealed
that the oldest level, representing a settlement characterised by hearths and
hand-made pottery, rests on the sandstone substrate of the cliff. Above this
level, ceramic material dated to the fifth century BC was found. Vuillemot
concludes that the first level seems to reveal a self-enclosed world.39 Several
circular tumuli, which seem to always contain a cremation burial, have
been excavated on the slopes of the plateau that closes the bay of the
Andalusians towards the west.
Further to the west of Oran, 2 km from the coast opposite the mouth of
the Oued Tafna, is the small island of Rachgoun (c.15 ha). Theoretically,
the position is ideal for a Phoenician/Punic site, however the archaeologi-
cal data reflects the reality of occupation. The oldest unproblematic datable
ceramic fragments come from an Attic amphora from the second half of
the seventh century BC. However, the excavation reports mention the
presence of local ceramics contemporary to, and even prior to, the Attic
amphora.
Vuillemot called these sites Punic, under heavy influence of the theories
of his mentor Pierre Cintas, for whom no coastal point favourable for
defense must have slipped the attentive eye of the Phoenicians. He did
make a very significant remark, however, when he stated that certain
essential aspects of the inhabitants of the island of Rachgoun were not
consistent with those expected of a Phoenician population.40 The funeral
rites of the groups that occupied these trading posts do not correspond to
those of the Phoenicians, but are rather evidence of Protohistoric Amazigh
sepulchral practices.

38 39
Bokbot 1991; Kbiri-Alaoui et al. 2004, 596–97. Vuillemot 1965, 42.
40
Vuillemot 1965, 93.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 489

Discussion

Indigenous and Phoenician Urbanisation in the Western

Mediterranean
If we broaden our scope to the northern shore of the western Mediterranean
where similar problems arise, we will find the answers to our questions.
Archaeological work in Spain and France has demonstrated the complexity
of the phenomenon of Hellenisation of the littoral zone and its hinterland.
There were certainly colonies and trading posts, but, equally certainly, there
were also numerous indigenous settlements where trade was carried out and
where an original civilisation evolved through varied contact with the Greek
and Punic world.
At the coastal sites of the Maghrib, whenever Punic or Phoenician
imported ceramics are found, the majority of excavators automatically
think of these sites as being occupied by eastern seafarers. We must pay
tribute to Gabriel Camps, who aggressively opposed this interpretation
of the archaeological data when he stated ‘That coastal villages received
Mediterranean productions from their origins is so normal that it
cannot be presented as a scientifically valid argument about their own
origins’.41
Paul Albert Février questioned whether all the trading posts or small
settlements had actually been founded by Phoenicians or whether,
instead, they should be regarded as indigenous. For him there was
no doubt that indigenous villages existed long before the creation of
trading posts, which could very well have been newly established,
but alongside older settlements previously founded by indigenous
populations.42
Pseudo-Scylax distinguished between two cities on the Loukkos, one
Phoenician and the other populated by Libyans.43 This statement more
accurately reflects, in my view, the ancient reality. In order to trade
their products, the Phoenicians would logically have chosen to settle
where there was already a fairly large human occupation. Moreover,
Pseudo-Scylax, discussing Phoenician trade with the Atlantic regions
of Morocco, may have been referring to trade with the city of Lixus,
since the author mentions in another passage that the Ethiopians
made a great deal of wine from their vines, which the Phoenicians
exported.44

41 42 43 44
Camps 1979, 48. Février 1967, 108. Carcopino 1949, 89. Villard 1960, 22.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 490 Youssef Bokbot

François Villard concluded that the cities of the North African coast
labelled as Phoenician may have had an earlier origin. He also noted that
the birth of urbanism in Morocco appears not to have been, properly
speaking, the result of colonisation.45 Maurice Euzennat, for his part,
drew attention to the fact that it is inappropriate to make the Punic
influence play too important a role.46
Similarly, Gabriel Camps observed that many pre-Roman cities in North
Africa had megalithic necropolises at the gates of the ancient centre,
corresponding to funerary rites that were foreign to those of the Romans
and Phoenicians.47 These burials were the work of the Amazigh popula-
tions who inhabited these cities before any contact or foreign occupation.
This observation probably applies to the ancient city of Lixus as well.
For François Villard, the process of urbanisation in Morocco, which he
placed during the sixth century BC, was the result of the intensification of
commercial contacts with the Phoenicians of Gades, the Carthaginians and
the Greeks, but was not the result of colonisation.48 According to Villard,
when ancient tradition describes a city as Phoenician, it only means that its
inhabitants had adopted the language and manners of the Phoenicians, but
does not necessarily imply that it was founded by them.49
This Protohistoric occupation at sites that have previously been char-
acterised as primarily Phoenician or Punic leaves open the question of the
date of origin of Palaeo-Amazigh villages of proto-urban character, for
which we currently lack a precise chronology. It is tempting to link this
emergence to the development of metallurgy, which enabled increased
cultivation and productivity and which in turn offered the potential for
the population growth that was necessary for any increase in the socio-
economic complexity of these communities. Openness to Mediterranean
trade must also be taken into consideration, even if such trade is still ill-
defined.50
The necropoleis of the Tangier hinterland contribute to a clearer vision
of the populations of the Late Bronze Age. They indicate a presence, as
early as the sixth century BC, of indigenous rural settlements, still char-
acterised by Bronze Age traditions.51 In these same necropoleis, appearing
alongside the megalithic cists of the Middle and Final Bronze Age, burials

45
Villard 1960. 46 Euzennat 1965, 261.
47
Camps 1993; cf. Sanmartí et al., Chapter 11, this volume. 48 Villard 1960, 22.
49
Villard 1960, 23.
50
There are striking similarities of evidence and interpretation with the situation of the Eastern
Maghrib laid out by Sanmartí et al., Chapter 11, this volume.
51
Ponsich 1988, 87.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 491

with somewhat different architecture and funerary rites are also found.
This evidence for the coexistence, without a transition, of the Libyan
civilisation of the Late Bronze Age and the ‘phoenicianised’ Libyans of
the eighth and seventh centuries BC provides an image that best corre-
sponds to the protohistoric settlements of Kach Kouch and Sidi Driss, and
to the oldest phases of occupation at Lixus and Mogador.

Figure 12.9. Aerial view of the Mzora tumulus (Camps 1961, pl. I.2).

Monumental Tombs and Social Hierarchy

Mzora is one of the best-known Moroccan megalithic monuments attri-
butable to the Protohistoric period. This large burial mound, surrounded
by a ring of stone monoliths, is currently the only known example of its
kind in North Africa (Fig. 12.9). Because of the lack of an excavation
report, however, its dating poses a real problem. The tumulus of Mzora
is distinguished from the other burial mounds in north-west Morocco by
the larger dimensions of the monoliths of its enclosure. Gabriel Camps
noted that although this characteristic was not typically African, it repre-
sents the development, in a vertical dimension, of annular arrangements of
stones around the base of other North African tumuli.52
The megalithic nature, which was used as a criterion for linking Mzora
to Iberian megalithic tradition, is not unknown in the funerary monu-
ments of north-west Morocco. Other large tumuli have been reported in

52
Camps 1961b, 76–78.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 492 Youssef Bokbot

northern Morocco: the tumulus of Sidi Khelili, which is 90 m long, 30 m

wide and 8 m high; the tumulus of Nouillat Kebira measuring 41 m in
diameter and 8 m in height; and finally, the large tumulus identified by
Danilo Grébénart in the region of Taza measuring 100 m in diameter at
the base and 15 m in height.53 The Mzora monument also displays
evidence of Atlantic influences. The location of the two large monoliths
seems to suggest an orientation towards the west. Such an orientation is
rare, if not totally absent, in the Protohistoric funerary monuments of
Morocco since burials were oriented either east or south-east. Miguel
Tarradell and Gabriel Camps considered the tumulus of Mzora as an
indigenous monument presumably built in pre-Roman times for a
regional chief or Moorish king. This raises the question of social hier-
archy in north-west Africa.54
At Sidi Slimane of Gharb, the removal of a tumulus, located in the centre
of a weekly market (souk), revealed a funerary monument in the form of a
dwelling (Fig. 12.10). This atypical house had a corridor, a courtyard and a
funerary chamber covered with cedar logs. Two bodies rested in the
chamber, and two others in the corridor and in the court, these were
doubtless servants buried after immolation to guard the burial of the
chiefs.55 Finds included pieces of ivory that must have originally been
part of a box. Nearby, a Libyan inscription refers to the burial of two
individuals, a father and his son. Using the amphorae present in the tomb,
a date for the monument can be assumed to be between the fourth and
third centuries BC.
If it is true that two servants were sacrificed to accompany their masters
to the afterlife, these human sacrifices may represent an essential rite of
power in emerging class societies. The question of the emergence of the
Mauritanian kingdoms is therefore probably related to these great tombs,
which have tended to be associated with princely societies.56 Their impor-
tance should to be taken into account because they imply an authority
powerful enough to control considerable surplus labour, but also because
they denote a legitimisation of the tribal chiefs in the process of becoming
kings.
Do the several large burials of this type in north-west Morocco demon-
strate an intermediate stage where powerful leaders of tribal confederations
were engaged in displays of power? In any case, it is clear that these chiefs
benefited from a marked increase in productive forces, probably connected

53
Grébénart 1967. 54 Camps 1961b, 78; Tarradell 1952, 233. 55 Camps 1961b, 196–99.
56
For developed argument along these lines, see further, Papi 2019.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 493

Figure 12.10. Monument in the form of a dwelling under the mound of Sidi Slimane:
a) general view; b) plan and cross-section (Camps 1961, pl. X.3 and Fig. 81).

with the use of a plough, however archaic, which led to considerable

development in production and significant demographic growth. The
organisation of agricultural labour and the distribution of social products
required greater authority and the ability to exercise their power and
influence over wider territories.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 494 Youssef Bokbot

Everything suggests that powerful confederations, if not kingdoms, were

formed in Morocco in the fourth century BC. The origins of the
Mauritanian dynasties remain mysterious, but this is hardly surprising.
Their power, which eventually spread to cover immense territories, was, in
fact, derived from that of tribal leaders who succeeded in widening their
field of action and influence.
We can therefore imagine the urgency for Moroccan and Maghribian
archaeology in general to consider more excavations extending into the
pre-Roman and pre-Phoenician levels of ancient cities and trade posts.
These excavations are indispensable to the development of knowledge
about the Protohistoric civilisations of the Maghrib.

References

Aranegui-Gasco, C. 2001. Lixus: Colonia fenicia y ciudad punico-mauritana.

Valencia: SAGVNTVM, Papeles del Laboratorio de Arqueologia de Valencia.
Extra 4.
Brandherm, D. 2007. Las espadas del Bronce Final en la Península Ibérica y
Baleares. Prähistorische Bronzefunde, vol. IV. Stuttgart: Franz Steiner, 16.
Bokbot, Y. 1991. Habitats et monuments funéraires du Maroc protohistorique.
Thèse de Doctorat, Université de Provence.
Bokbot, Y. 1998. Une céramique à graffito à Lixus. Bulletin d’Archéologie
Marocaine 18: 321–23.
Bokbot, Y. and Onrubia-Pintado, J. 1992. La basse vallée de l’Oued Loukkos à la fin
des temps préhistoriques. In Actes du Colloque International: la ville antique de
Lixus. Larache 8–11 Novembre 1989. Rome: Collection de l’Ecole Française de
Rome 166, 17–26.
Bokbot, Y. and Onrubia-Pintado, J. 1995. Substrat autochtone et colonisation
phénicienne au Maroc. Nouvelles recherches protohistoriques dans la
péninsule tingitane. In Actes du 118 Congrès National des Sociétés Historiques
et Scientifiques. Paris: CNRS, 219–31.
Callegarin, L., Kbiri Alaoui, M., Ichkhakh, A. and Roux, J.C. (eds). 2016. Rirha: Site
antique et medieval du Maroc. II. Période Maurétanienne (Ve siècle av. J.-C. – 40
ap. J.-C.). Madrid: Casa de Velázquez.
Camps, G. 1961a. Aux origines de la Berbérie. Massinissa ou les débuts de
I’histoire. Libyca 8. Algiers: Service des Antiquités.
Camps, G. 1961b. Monuments et rites funéraires protohistoriques: Aux origines de la
Berbérie. Paris: Arts et Métiers Graphiques.
Camps, G. 1979. Les Numides et la civilisation punique. Antiquités africaines 14:
43–53.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 495

Camps, G. 1993. Réflexions sur l’origine protohistorique des cités en Afrique du

Nord. In L. Serra (ed.), La città mediterranea: Eredita antica e apporto arabo-
islamico sulle rive del Mediterraneo occidentale e in particolare nel Maghreb: atti
del Congresso internazionale di Bari, 4–7 maggio 1988. Napoli: Istituto universi-
tario orientale, 73–81.
Carcopino, J. 1949. Le Maroc antique. Paris: Gallimard.
Cintas, P. 1954. Contribution à l’étude de l’expansion carthaginoise au Maroc. Paris:
Publication de l’Institut. des Hautes Etudes Marocaine 56.
Chifman, C. 1963. Naissance de la puissance carthaginoise. Moscow.
de Torres, J. and Ruiz-Gálvez, M., 2014. Unravelling patterns in Oukaïmeden rock
art. Complutum 25.2: 167–87.
El Khayari, A. 2007. La présence phénicienne au Maroc. Les Dossiers d’Archéologie
(Hors Série 13, La Méditerranée des Phéniciens): 56–59.
Euzennat, M. 1965. Héritage punique et influences gréco-romaines au Maroc à la
veille de la conquête romaine. In Le rayonnement des civilisations grecque et
latine sur les cultures périphériques. VIIIème Congrès International d’Archéologie
Classique. Paris: de Boccard, 265–66.
Février, P-A. 1967. Origine de l’habitat urbain en Maurétanie césarienne. Journal
des Savants 2: 107–23.
Gozalbes-Gravioto, E. 1975. Las edades del cobre y bronce en el N-O de Marruecos.
Cuadernos de la biblioteca espanola de Tetuan 12: 7–32.
Grébénart, D. 1967. Prospection archélogique dans la region de Taza (Maroc),
préhistoire et protohistoire. Libyca 15: 147–55.
Grillo W.F. 2003. Recherches archéologiques maroco-italiennes 2002–2003. Rome:
Edizioni Quasar, Istituto Italiano di Cultura.
Gsell, S. 1916. Hérodote. Textes relatifs à l’Histoire ancienne de l’Afrique du Nord.
Algiers and Paris: Ernest Leroux.
Gsell, S. 1927. Histoire ancienne de l’Afrique du Nord, vol V. Paris: Librairie
Hachette.
Jodin, A. 1957. Note préliminaire sur l’établissement préromain de Mogador.
Bulletin d’Archéologie Marocain 2: 9–40.
Jodin, A. 1966. Mogador: Comptoir Phénicien du Maroc atlantique. Rabat: Etudes
et Travaux d’Archéologie Marocaine 2.
Kbiri-Alaoui, M., Siraj, A. and Vismara, C. 2004. Recherches archéologiques
maroco-italiennes dans le Rif. L’Africa romana 15: 567–604.
Malhomme, J. 1953. Les représentations anthropomorphes du Grand Atlas. Libyca
anthropologie-préhistoire-ethnographie 1: 373–85.
Martínez-Sánchez, R.M., Vera-Rodríguez, J.C., Pérez-Jordà, G., Peña-Chocarro, L.
and Bokbot, Y. 2018. The beginning of the Neolithic in northwestern Morocco.
Quaternary International 470 B: 485–96.
Marzoli, D. and El Khayari, A. 2010. Vorbericht Mogador (Marokko) 2008.
Deutsches Archäologisches Institut, Abteilung Madrid 51: 61–108.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 496 Youssef Bokbot

Mattingly, D.J. 2011. Imperialism, Power and Identity. Experiencing the Roman
Empire. Princeton: Princeton University Press.
Mattingly, D.J. 2016. Who shaped Africa? The origins of agriculture and urbanism
in Maghreb and Sahara. In N. Mugnai, J. Nikolaus and N. Ray (eds). De Africa
Romaque. Merging Cultures across North Africa. London: Society for Libyan
Studies, 11–25.
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017. Trade in the Ancient Sahara and Beyond. Trans-Saharan
Archaeology Volume 1. Series editor D.J. Mattingly. Cambridge: Cambridge
University Press and The Society for Libyan Studies.
Pallary, P. 1907. Recherches palethnologiques sur le littoral du Maroc en 1906.
L’Anthropologie 18: 301–14.
Pallary, P. 1915. Recherches préhistoriques effectuées au Maroc (1912–1913).
L’Anthropologie 26: 193–217.
Papi, E. and Akerraz, A. (eds). 2008. Sidi Ali Ben Ahmed – Thamusida 1. I contesti.
Rome: Quasar.
Papi, E. 2019. Revisiting first millennium BC graves in north-west Morocco. In M.
C Gatto, D.J. Mattingly, N. Ray and M. Sterry (eds), Burials, Migration and
Identity in the Ancient Sahara and Beyond. Cambridge: Cambridge University
Press, 281–312.
Ponsich, M. 1967. Nécropoles phéniciennes de la région de Tanger. Rabat: Etudes et
Travaux d’Archéologie Marocaine 3.
Ponsich, M. 1969. Influences phéniciennes sur les populations rurales de la région de
Tanger. In Tartessos y sus problemas. V Symposium Internacional de Prehistoria
Peninsular. Barcelona: Universidad de Barcelona, 173–84.
Ponsich, M. 1970. Recherches archéologique à Tanger et dans sa région. Paris:
CNRS.
Ponsich, M. 1981. Lixus: Le quartier des temples. Rabat: Etudes et Travaux
d’Archéologie Marocaine 9.
Ponsich, M. 1988. Implantation rurale du Maroc phénicien. Dossiers Histoire et
Archéologie 132: 84–87.
Ruiz-Gálvez Priego, M. 1983. Espada procédante de la ria de Larache en el Museo
de Berlin Oeste. In Homenaje al Prof Martin Almagro-Basch. Madrid: Ministerio
de Cultura, vol. 2, 63–68.
Sáez Martín, B. 1955. Sobre la supuesta existencia de una edad del bronce en Africa
menor. In IIe Congrès Panafricain de Préhistoire, Alger 1952. Paris: Arts et
Métiers Graphiques, 659–62.
Shipley, G. 2012. Pseudo-Skylax’s Periplous: The Circumnavigation of the Inhabited
World. Text, Translation and Commentary. Exeter: Bristol University Press.
Souville, G. 1995. Pénétrations atlantiques des influences ibériques au Maroc
protohistorique. In IIe Congreso Internacional: El Estrecho de Gibraltar, Ceuta
Novembre 1990. Madrid: Universidad Nacional de Educación a Distancia, vol. 2,
245–92.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 12 The Origins of Urbanisation in Morocco 497

Tarradell, M. 1952. El tumulo de Mezora (Marruecos). Archivo de Prehistoria

Levantina 3: 229–39.
Tarradell, M. 1954. Las excavaciones de Lixus. In IV congreso internacional de
ciencias prehistóricas y protohistóricas. Zaragoza: La Académica, 789–96.
Tarradell, M. 1960. Historia de Marruecos: Marruecos Punico. Tetuan: Cremades.
Villard, F. 1960. Céramique grecque du Maroc. Bulletin d’Archéologie Marocaine 4:
1–26.
Vuillemot, G. 1965. Reconnaissances aux échelles puniques d’Oranie. Autun: Musée
Rolin.

https://doi.org/10.1017/9781108637978.013 Published online by Cambridge University Press

 13 Architecture and Settlement Growth
on the Southern Edge of the Sahara
Timing and Possible Implications for Interactions
with the North

kevin c. macdonald

Introduction

Since the 1970s, there has been a significant increase in regional data sets
relating to tell sites in the West African Sahel. Previous colonial notions
that Trans-Saharan commerce in the Islamic era created the first Sub-
Saharan towns and polities have now been largely abandoned; it is all too
apparent that complex settlement systems were widespread in the south
long before the eighth century.1 Yet questions of Trans-Saharan interac-
tions with the Berber world in preceding centuries have not been so
effectively addressed. In this chapter, I examine archaeological evidence
for architectural development and settlement growth within the Hodh and
Middle Niger basins: the Tichitt, Walata, Tagant and Néma escarpments,
the Méma, the Inland Niger Delta, the Lakes Region, the area of Timbuktu
and the Gourma (Fig. 13.1). The advent of different types of dry stone,
coursed earth or mudbrick structures and the regional timing of settlement
growth will be the primary factors under consideration. This analysis will
focus on data between 1600 BC and AD 800 and propose multiple points of
change during this timeframe. It is intended that this micro-synthesis will
provide a means for the systematic comparison of Sahelian developments
with early Berber settlements in the Sahara.
In this text, concepts of urbanism and statehood are intentionally not
invoked as defined categories. This is for two reasons:
1) Definitions of urbanism go beyond mere size of settlement (although
typically one does not even think of urbanism in sites of less than 20 ha).
Yet there is a paucity of adequate temporally defined data for most
regions concerning factors such as occupational specialism, public

1
498 See McIntosh and McIntosh 1988.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 499

Figure 13.1. Map of regions and key sites discussed in this chapter.

works and trade networks, which makes qualitative assessment of urba-

nisation difficult. Jenné-jeno is an important exception to this caveat.2
More recently, tentative cases have been advanced for still earlier urban-
ism at Dia and Tombouze.3 With more data, cases for urban status
should eventually be made for many localities in the first-millennium
AD Middle Niger, in some places potentially even earlier. However to
argue about these cases in advance of suitable data is unwise.
2) The definition of statehood is a contentious issue and largely depends
on how, and if, one buys into existing social evolutionary frame-
works. Tichitt – for example – has been described in the literature
as a ‘Chiefdom’ or as a ‘Complex Chiefdom’,4 conceptual frameworks
borrowed from historic Polynesia and of doubtful direct applicability
to Africa.5 It is therefore better to discuss polities in more qualitative
and individual terms rather than in exteriorly derived and arbitrary
categories. There is insufficient space to do so here. The case of
Ghana/Wagadu alone, given its connections to both Arabic and

2
McIntosh 1995. 3 Dia: Bedaux et al. 2001; 2005; Tombouze: Park 2010.
4
Holl 1985; Munson 1971. 5 See Pauketat 2007.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 500 Kevin C. MacDonald

oral literature and other data sets, could easily consume a chapter in
its own right. Thus, I have limited my concerns to changes in built
environment and periods of settlement growth.

Architectural Development

The potential influence of pre-Islamic North African architectural forms

on the Sahel has had a long currency. Prussin, for example, asserts
a Roman origin for the Berber ‘courtyard house’ or fortified farms, seen
as leading to subsequent allied Islamic forms in the Sahel.6 However she
also claims an ill-defined syncretic aspect for the ‘internal courtyard
house’, with an ‘indigenous prototype being known from southern
Mauritania by the mid-second millennium BC’ (that is Tichitt and
Tagant). Indeed, were it not for the ample and early architecture of the
Tichitt Tradition an argument could have been advanced for a primarily
North African role in the creation of the West African compound.
Instead, when Tichitt is taken into account, we are left with a much
more nuanced and complex sequence of architectural change and accom-
modation in the Trans-Saharan region.

The Tichitt Architectural Sequence

Tichitt and its neighbouring settled escarpment ranges in southern
Mauritania remains one of African archaeologies greatest enigmas. It is one
of the continent’s earliest agricultural landscapes and first polities – potentially
a pristine one. It also possesses a four-tier hierarchy of stone-built settlements
interspersed over 200,000 km2 and has strong material connections to the
foundation of key first-millennium BC Middle Niger settlements.7 As such,
Tichitt could be viewed as the western counterpart of Kerma, but it remains
much less celebrated and discussed in world archaeology. The reasons for this
obscurity might be explained by its very remoteness, the small number of
(largely Francophone) researchers who have worked at these sites, and the
reluctance of some scholars to place confidence in the ceramics-based
sequences constructed for its deflated and sometimes unstratified settlements.
Critically, its mortuary archaeology, while manifestly abundant and at times
monumental, has gone almost entirely unexcavated.8

6
Prussin 1986, 105–8. 7 Holl 1993; MacDonald 2011a.
8
Amblard-Pison 2006; Vernet 1993.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 501

Although the massive architectural sites of the Tichitt Tradition have

undergone a range of different field enquiries, Munson’s original work
in the Tichitt region remains the only one that attempts to trace
the temporal evolution of its architecture.9 When considered via recent
re-evaluations of the Tichitt ceramic sequence, 14C dates and associated
phasing of sites, it is possible to consider two broad architectural
periods: Classic (1600–1000 BC) and Late (1000–400 BC).10 To these
can be added a Terminal phase from the Tagant, where the building of
smaller, if more internally complex sites, continued over the period of
400 BC to AD 350.11
From the outset of the Classic Tichitt phase (1600–1000 BC), defensive
considerations played an important part in settlement design: high (c.2 m)
and robust (c.1–1.5 m thick) stone walls surrounded major settlements,
with means of access including gateways capped with lintel stones.12 The
internal space of sites was divided into tens, sometimes hundreds, of
irregularly shaped curvilinear compounds defined by relatively low
(c.1 m high) stone walls. The interior dimensions of these compounds
ranged widely, falling anywhere between 70 and 4,050 m2.13 The reason for
this wide variation is likely because some of these enclosures were domestic
while others served as livestock pens or cultivation areas. A remarkable
aspect of Classic Tichitt phase sites is the relative scarcity of signs of
internal dwellings within these compounds. Most common are so-called
‘pillar’ or nine-stelae structures which are variously interpreted. These
groupings of stone pillars (of c.1–2 m in height) may have served as granary
bases or the frames for shelters covered in mats or hides.14 While both uses
appear possible, the dimensions of most would be in line with traditional
granary styles still extant amongst the Soninke and Bamana of Mali.15
Much rarer during this initial phase are circular stone structures.
Munson perceived these as being stone-capped ‘igloo’ – or ‘beehive’ –
shaped dwellings or granaries chinked with mud.16 Amblard-Pison prefers
to compare them to circular hybrid stone-thatch dwellings, chinked with
mud and dung on their interiors only, a tradition still known from eastern
Senegal.17 The comparatively small numbers of such enclosed dwelling
places in Classic Tichitt sites may argue for either relatively low population
density on sites or the extensive use of more mobile and less durable

9
Amblard Pison 2006; Holl 1986; Munson 1971; Ould-Khattar 1995; Vernet 1993.
10
MacDonald 2011a; 2015. 11 Ould-Khattar 1995. 12 Munson 1971, II, 334–35.
13
Holl 1993, Table 16. 14 Amblard-Pison 2006, 79–90; Munson 1971, II, 338.
15
MacDonald et al. 2009. 16 Munson 1971, II, 340.
17
Amblard-Pison 2006, 78–79; compare Dujarric 1981.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 502 Kevin C. MacDonald

Figure 13.2. Plan of Tichitt ‘Village 72’ (Akreijit) (after Vernet 1993, 274).

supplementary structures of thatch, mat and pole, such as those built by

recent historic Tuareg or Fulani.18
The best studied of Classic Tichitt sites is Akreijit, also known as Village 72
and Seyyid Ouinquil (Fig. 13.2). It stands as a good example of a medium-
sized (15 ha) Tichitt settlement and is well-dated to c.1500–900 BC.19 The site,
oblong in shape, is situated atop an escarpment and composed of 177
compounds (including notional livestock pens), themselves grouped into
12 contiguous clusters.20 Domestic stone structures (not including the nine-
stelae granary type, which is almost ubiquitous) occur in only 60 of these
compounds. These 146 single-room stone houses are of three sub-types:
circular isolated structures of 9 to 10 m2 (51), semi-circular structures built
into the corner of a concession of 9 to 12.5 m2 (48), and oblong structures
backing onto the long wall of a compound of up to 16 m2 (47).21 Rarely, those
backing onto compound walls could form up to three contiguous structures,

18
David 1971; Nicolaisen and Nicolaisen 1997, I, 413–35.
19
Amblard-Pison 2006, Figure 11. The largest known Tichitt settlement is that of Dakhlet el
Atrouss I, with a remarkable 540 compounds and walled livestock pens, as well as extensive
tumuli, covering a surface of 80.5 ha, see Holl 1993.
20
Holl 1993, 116–17. 21 Ambard Pison 2006, 75–77.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 503

but most were isolated. Thus, the clustering of modular room blocks so
common in traditional West African courtyard compounds was only margin-
ally extant, while the overall tendency of wall-lines was curvilinear rather than
rectilinear.
The Late Tichitt phase (1000–400 BC) saw the increase of individual
stone ‘dwellings’ within compounds. Munson notes that average
compounds were at this time 30 to 40 m in diameter with surrounding
dry-stone walls c.1.5 m in height.22 Internal stone structures now
numbered between one and ten per compound and were equally
divided in number between ‘stelae’ structures and so-called ‘rubble-
circle houses’. Settlements in this phase tend to have been placed higher
on escarpments in positions of less visibility and greater natural
defensibility. In the Terminal Tichitt phase (400 BC–AD 350), known
only from the Tagant, there was increasing occurrence of internal stone
dwellings within compounds. A well-dated and thoroughly studied
example is site T150 (c.150 BC–AD 70).23 The site comprises approxi-
mately 200 compound enclosures formed of dressed stone usually
attaining 1 m in height, though the evident subsequent robbing-out of
many of these walls makes estimating their true height rather difficult.
Two forms of smaller (one room) stone domestic structures are appar-
ent, circular and rectangular, with the former being more common
and at least equal in number to the compound enclosures. The rarer
rectangular structures were quite substantial, measuring between 6 and
12 m in length. Both types of internal structure occurred in approxi-
mately half the compounds, the rest were seemingly empty. The circular
dwellings clustered in twos and threes and frequently abutted rectan-
gular structures (Fig. 13.3). Stelae structures were not present. It is
notable that sites such as T150 in the Tagant, like those of the Late
Tichitt phase in the Néma, exhibit evidence of both iron-working and
iron objects.24
In summary, post-1000 BC Tichitt sites have clearer architectural evi-
dence of compounds serving extended family units, with multiple gran-
aries and multiple single room dwellings within each. The classic Sahelian
curvilinear compound form now seems fully developed. The appearance of
rectangular dwellings by 150 BC–AD 70 in the Tagant is a particularly
interesting development and the earliest yet documented in Sub-Saharan
Africa. It may represent evidence for the advent of a Trans-Saharan sphere

22 23
Munson 1971, II, 340–41. Ould-Khattar 1995, 241–51, figure 12.
24
MacDonald et al. 2009.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 504 Kevin C. MacDonald

Figure 13.3. Plan of the Tagant site T150 (after Ould-Khattar 1995, figure 12).

of interaction, given the appearance of comparable rectilinear structures at

Fewet and elsewhere in southern Libya in this same era.25 Further temporal
indicators for such connections will be considered below.

Pre-Islamic Architecture in the Middle Niger

The lack of raw materials suitable for stone construction over most of the
Middle Niger’s floodplain means that despite the material culture association
of many ‘founder’ sites with the Tichitt Tradition one is obligated to search
for traces of earthen architecture.26 Generally speaking such structures were
made of either layers of coursed earth (termed variously ‘banco’, ‘bogoton’,
‘sana’, etc. according to local usage) or out of mudbrick (usually loaf-shaped
in early periods, but also cylindrical).
Only two examples of earthen architecture are known from the first
millennium BC Middle Niger, both incompletely published. The earlier of
the two is the site of Kolima Sud-Est, a low 10-ha tell in the multi-period

25 26
Mattingly et al., Chapter 2, this volume; Mori 2013. MacDonald 1996; 2011a; 2015.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 505

Kolima cluster at the centre of Mali’s Méma region. The site was first
surveyed and surface collected by Téréba Togola and the author in 1990.
We recorded the traces of several eroding curvilinear ‘tauf’ wall lines at the
site, notable by their colour and heavy clay composition, being of the
dimensions of both compound and individual round house walls.
Although these walls were mentioned in my doctoral dissertation,27 and
remain better described in my notes, they were not recorded when the site
was test excavated by Takezawa and Cissé in 2000. The deposits, featuring
late Tichitt Tradition pottery and evidence of cattle and fonio agriculture,
were dated by three 14C dates across a range from 910 to 540 BC.28
Unfortunately, the Méma has been a region of high insecurity since the
1990s, access is risky and no further field research has been completed
there since 2000.
Just to the south, in the Macina region, further traces of early earthen
architecture were found and dated at the site of Dia-Shoma by a multi-
national 1998–2003 excavation team.29 The earliest layers of this vast
50-ha tell were heavily eroded – perhaps by intermittent inundation –
and such traces of building as there were often took the form of hardened
or burned floors of indeterminate shape with occasional postholes.
However burned loaf-shaped brick fragments were recovered from the
lowest layer of exposure A/B. A single charcoal date from the conflagra-
tion which burned them gave the result of 9 calBC–calAD 67.30 If these
were indeed burned mudbricks, instead of broken fragments of a burned
coursed earth wall, they would be amongst the oldest in Sub-Saharan
Africa.
After these early manifestations, patchy evidence shows considerable
variation in the material and form of earthen architecture across the
Middle Niger. Back in the Méma, the next dated area of earthen construc-
tion comes from Mound B of the Akumbu complex (unit AK3) where
a wall constructed with ‘irregular mudbrick’ was dated by an associated
hearth to calAD 342–442.31 Subsequent documented constructions at
Akumbu were all of roughly rectangular, loaf-shaped, mudbrick with
structures themselves being both round and rectangular in outline.
Curvilinear earthen walls (of both coursed earth and mudbrick) were
frequently recorded during the 1989 Méma survey at sites attributed to the
‘Early Period’ (AD 200–600) via their ceramics.32 Thus, very provisionally,
it appears that curvilinear built forms were dominant in the Méma from
27
MacDonald 1994, 92. 28 MacDonald 2011a; Takezawa and Cissé 2004.
29
Bedaux et al. 2005. 30 Bedaux et al. 2005, 130–31. 31 Togola 2008, 34.
32
MacDonald 1994, 92; Togola 2008.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 506 Kevin C. MacDonald

the first millennium BC through to the mid-first millennium AD when

rectilinear structures appeared.
Relatively early mudbricks are also documented from the Malian Lakes
Region at the tell site of Mouyssam II (aka KNT 2).33 A curvilinear wall of
irregular loaf-shaped mudbricks was excavated at a depth of 7.59–7.79 m
comprising a structure exceeding 6 m in diameter – more likely
a segment of a compound wall than a domestic structure. This feature
lies stratigraphically between two 14C samples of calAD 340–540 and
calAD 570–690.
On the opposite side of the Inland Delta and east of the Lakes Region
there is the urban or proto-urban settlement mound complex of
Tombouze, near historic Timbuktu. The 2008 excavations at the mound
of TBZ1 revealed curvilinear ‘banco’ (coursed earth) structures at the
margin of ‘Phase 1’ and ‘Phase 2’ layers, the nearest associable 14C date
being calAD 120–330.34 More extensive excavations at TBZ2 in the follow-
ing years have revealed compound clusters of such round ‘banco’ dwellings
dating from the mid-first millennium AD.35 No mudbrick or rectilinear
structures were encountered in any of the Tombouze excavations, which
overall cover a period of AD 100–1000.36
Directly south of Timbuktu there is the well-preserved architectural
sequence of Tongo Maaré Diabel, situated between the Bandiagara and
Dyoundé escarpments at the western limit of the Inland Delta floodplain.
As yet incompletely published,37 intermittent excavations at the site
between 1993 and 2010 by Togola, MacDonald and Gestrich have
exposed a detailed multi-phase architectural sequence.38 The earliest
occupational horizon at the site (AD 500–650) featured circular coursed
earth structures very similar in size and layout to those of Tombouze
(Fig. 13.4). However there was a sudden shift in the next horizon
(AD 650–750) to rectilinear structures built in both coursed earth and
loaf-shaped mudbrick. Since there is no corresponding material culture
change between these two occupations otherwise, and because coursed
earth forms then co-existed with mudbrick structures at the site until its
abandonment (c.AD 1150), this appears to represent continuity in the
face of new architectural ideas.39
It should be noted that the loaf-shaped mudbricks of Tongo Maaré
Diabel, despite being hand-formed, were remarkably regular in their size
33
Raimbault and Sanogo 1991, 301–23. 34 Park 2010. 35 Park et al. 2010.
36
Park, personal communication.
37
MacDonald 1998; Walicka Zeh and MacDonald 2004. See now Gestrich and MacDonald 2018.
38
Gestrich 2013. 39 Gestrich and MacDonald 2018.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 507

Figure 13.4. Linked round structures with coursed earth walls from Tongo Maaré
Diabel, Horizon I, Unit B, AD 500–650 (photo: K. MacDonald).

Figure 13.5. Loaf-shaped mudbrick from Tongo Maaré Diabel, Horizon II, AD
650–750 (scale marked at 10 cm intervals) (photo: K. MacDonald).

and shape, being 300–400 mm in length, 170–200 mm in width and

60–80 mm in thickness (with a tendency towards the smaller end of this
range in earlier periods). Their upper surfaces were invariably convex and
their lower surfaces mildly concave to facilitate stacking (Figs 13.5 and
13.6). Interestingly, they are comparable in length and width, although
thicker, to the mould-formed bricks at Gao-Saney that date to c.AD 1000.40
From the foregoing paragraphs it should be apparent that earthen
architecture was widespread in the Middle Niger from the early first

40
See Cissé et al. 2013; McIntosh, Chapter 14, this volume.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 508 Kevin C. MacDonald

Figure 13.6. Plan of excavations at Unit(s) A-B-C, Tongo Maaré Diabel, Horizon
II, AD 650–750. Note combination of coursed-earth and mudbrick structures.

millennium AD. The key urban tell complex of Jenné-jeno is therefore

curious for its relatively tardy and idiosyncratic architectural sequence,
perhaps due to local ecological and cultural factors. Simple mat and pole
structures daubed with clay seem to have prevailed in the site’s initial

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 509

occupation phases (250 BC–AD 400). Curvilinear ‘tauf’ (coursed earth)

structures only appear in Phase III (AD 400–850), with mudbrick (of
a locally distinctive cylindrical variety) beginning in round structures
after AD 850.41 Indeed, rectilinear structures, also built using cylindrical
brick, began only after AD 1200.
At first glance, from the results of 45 years of Middle Niger archaeology,
it is difficult to perceive a neat pattern in the spread of coursed earth,
mudbrick, round and rectilinear structures. There is good reason to believe
that coursed-earth walled compounds with multiple circular interior struc-
tures were ubiquitous along the margins of the Inland Delta in the first few
centuries AD. The fact that the only early first millennium BC settlement
with such structures yet known is a derivative of the Tichitt Tradition
(Kolima Sud-Est), suggests that the development of such earthen structures
stemmed from the need to find a local substitute for sandstone. The spread
of such modular, round, coursed earth compounds may also have been part
of the wider Tichitt diaspora and its knock-on influence. Even well to the
south, in central Burkina Faso, such structures appear by c.AD 100 (the
Kirikongo complex in the Mohoun Bend).42
If curvilinear coursed earth structures may be taken as the Sahelian
status quo for the period of AD 100–900, we are left with a puzzling series
of local architectural changes over the second half of that period, with the
advent of both rectilinear forms and mudbrick. Mudbrick is an important
innovation in building technology facilitating both greater planning and
more monumental construction. Formerly, Sahelian architectural specia-
lists like Labelle Prussin were satisfied, via supposition and historical
linguistic linkages, that ‘the most persuasive argument for the introduction
of masonry techniques into the African savannah by Islamic agency lies
in the building technologies associated with them’.43 In other words,
mudbrick was evidence of Islamic intervention in the South. Yet even if
we put aside the questionable early mudbricks of Dia, our evidence for this
innovation in building technology appears to go in a clockwise direction
around the pre-Islamic periphery of the Middle Niger: the Méma c.AD 400,
the Lakes Region c.AD 550, and the Gourma c.AD 650. Still farther east,
loaf-shaped mudbricks are present in the earliest layers of Gao-Saney c.AD
700.44 However sparse this pattern might be, it disassociates Islamic trade
from the first appearances of early Sahelian (loaf-shaped) mudbrick and is
suggestive of the gradual spread of a fashion coming from the north and

41
McIntosh 1995, 64–66. 42 Dueppen 2012, 277. 43 Prussin 1986, 38.
44
Cissé et al. 2013; McIntosh, Chapter 14, this volume.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 510 Kevin C. MacDonald

west towards the south and east between AD 400 and 700 – whether driven
by the innovations of a nascent empire (Ghana) and/or via the Trans-
Saharan sphere of interaction.45

Early Settlement in the Sub-Sahara: Perceiving Its Historic

Amplitude

It is clear from the archaeological record of the West African Sahel that
there has not been a simple, gradual growth in regional settlement, but
rather peaks and troughs in settlement size and density. For the pre-Islamic
period we witness a peak in Tichitt-Walata settlement size and density
between c.1200–1000 BC together with the Tichitt tradition’s spread into
the Tagant, Néma and Méma regions.46 After an apparent scattering of
populations in the Western Sahel during the first millennium BC, pub-
lished sequences from the Middle Niger show substantial settlement
growth in the mid-first millennium AD: Jenné-jeno from AD 450, Dia
from AD 500, the Méma from AD 500, or for the area of Timbuktu
beginning sometime between AD 200 and 600.47 The consequence has
therefore been a notion of post-Tichitt decline and obscurity in the first
millennium BC, followed by gradual re-emergence of small settlements in
the early first millennium AD succeeded by a rapid reflorescence from AD
500. But is this really so?
One of the primary difficulties in estimating settlement dynamics in the
Middle Niger is the nature of earthen-settlement-mound archaeology.
Naturally what we see and record on the surface of sites are abandonment
assemblages – so, in fact, moments of crisis. Additionally, our temporal
estimates for surface survey are based upon association with pottery
phases, with a best resolution of 400 to 600 year time periods. Site sizes
and distributions so recorded are not only not punctual, they do not speak

45
It should be noted that the curious cylindrical mudbricks of Jenné-jeno and its successor city,
Djenné (known as Djenné-ferey or Djenné-wéré) are as yet archaeologically unrecorded in Mali
outside of the central Inland Niger Delta, however they are also known in second-millennium
AD Kano and contemporary Hausa urban architecture in Northern Nigeria (where they are
termed tubali; Dmochowski 1990). While outside the time range of this volume, this suggests
some connection between castes of masons across the Northern Sahel during the Islamic
period. Mould-formed rectangular mudbricks (today called toubobou-fery – the foreigner’s
brick) are rarely documented in the Sahel until recent historic times (specifically, the nineteenth
century), but see McIntosh, Chapter 14, this volume, who notes the presence of mould-made
bricks in the more recent (c.AD 1000) layers of Gao-Saney.
46
Holl 1985; 1993; MacDonald 1996; 2011a; 2015.
47
Bedaux et al. 2005; McIntosh 1995; Park 2010; Togola 2008, 23.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 511

to periods of prosperity, when there may have been few abandonments.

These highpoints in the life of settlements lie buried beneath abandonment
layers. In other words, our methodology is likely to be biased towards
recording abundant settlement only at times of collapse, periods after those
of initial settlement growth.
A secondary difficulty is how one quantifies the surface area of tell
clusters. Do we quantify the area of sites on the basis of individual mounds
(often separated by no more than 10 m of ‘cordon sanitaire’) or the total
area of a mound cluster? This has a role in how we evaluate the growth of
site hierarchy but could also be largely illusory. Historically we know that
large singular tells can come about through the gradual joining of many
distinct quarters.48 The appearance of tell clusters may merely be the
artefact of the stage of site formation at the moment of abandonment,
and not necessarily a representation of the cognition of connectedness in
the minds of the ancient inhabitants. In other words, would there really be
a qualitative difference, in terms of urbanisation, between a single 25-ha
settlement mound and ten 2.5-ha settlement mounds no more than 100 m
apart from one another?
Let us take as a case in point the Méma region as surveyed by Téréba
Togola and the author in 1989–1990. If we were to merely innumerate
individual mounds from the Early (AD 200–600) Middle (AD 600–1400)
and Late (AD 1600–1800) Periods (Fig. 13.7) we can see that ‘Early’ sites
are most numerous, but usually quite small. If, however, site areas are
grouped and summed by their clusters we see much more concentrated
Early settlement, with traces of size hierarchy (Fig. 13.8). If ‘dead space’ in
between mounds is included, the size of early clusters increases by at least
a third. The real question, however, is how much Early settlement lies
beneath Middle Period abandonment deposits and thus un-detected and
un-counted? One notable proof of this is the principal mound of Toladié –
at 76 ha and with 15-m depth of stratigraphy the largest single occupation
mound yet recorded in Mali. A 14C sample taken from midway down an
erosion gully at the site produced a date of 1465 ± 60 BP (calAD 429–663)
broadly within the Early Period.49 If such major sites abandoned during the
Middle Period were indeed widely occupied in the Early Period, settlement
dynamics extrapolated from mound abandonment surfaces would clearly
be highly misleading.
Likewise, one of the Akumbu settlement mounds (Mound B) excavated
by Togola and myself had Early Period ceramics with a calAD 342–442 14C

48 49
Walicka Zeh 2000. Togola and Raimbault 1991; T. Togola, personal communication.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 Mema: Mound by Mound

80

70

60

50
Ha
40 Early
Middle
Late
30

20

10

0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47
Individual Mounds

Figure 13.7. Chart showing the size distribution (ha) of individual settlement mounds
surveyed in the Méma by Togola and MacDonald in 1989–1990.

Mema Clusters

120

100

80

I 60 Early
Middle
Late

40

20

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Clusters

Figure 13.8. Chart showing the size distribution (ha) by aggregated settlement mound
clusters – each mound’s dimension summed by the cluster to which they belong –
surveyed in the Méma by Togola and MacDonald in 1989–1990.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 513

date and first millennium BC Faita facies ceramics in the next strata
below. Additionally, eroding from flanks of other mounds at the site
were these same first millennium BC Faita facies ceramics.50 As Togola
noted, ‘We believe that many of these deeply stratified Early Assemblage
sites rest upon Terminal LSA [first millennium BC] deposits.’51 Thus,
sites like Kolima Sud-Est (10 ha between c. 900–500 BC) rather than
being the top examples of early settlement in the Méma, may instead be
the unsuccessful sites of that period – those which did not get deeply
buried beneath subsequent centuries of continuing settlement. So, in
summary, important settlement growth may in fact have begun in
the Méma during the first millennium BC, and it is very likely to have
begun by AD 400.
To this same end we can re-examine the growth of settlement at Dia one of
Mali’s great ‘Medieval’ cities. The site monograph places the initial Horizon
I occupation (800–1 BC) of the site as comprising two habitation zones – one
of 19 ha and the other of 3 ha.52 However sampling coverage is far from
complete and main occupation area could be easily redrawn as 22 ha (an
addition of 3 ha). Likewise, if one assesses occupation on the basis of the
spread of Horizon I diagnostic material collected by systematic field walking,
they occur in 11 of 16 transects, and are indicative of an early occupation
covering 50 per cent or more of the mound’s 50-ha surface.53 Additionally,
surrounding the site there are at least nine satellite mounds of 0.4 to 3 ha sur-
face, abandoned in Horizon I.54 Once again, substantial first millennium BC
occupation seems a greater prospect at Dia than has thus far been asserted.
The Dia sequence is also of interest because of the area’s almost complete
abandonment (notionally due to a period of drought) between AD 1–500.
Likewise, the western Gourma seems to have experienced a period of aban-
donment at this time until settlement growth recommenced in the fifth
century, perhaps for similar ecological reasons. Proliferation of settlement
at the eastern edge of the IND and Lakes Region, around Timbuktu, seems
only to have taken place in the pre-Islamic period between AD 200 and 600.55
In summary, we can tentatively advance three periods of settlement
growth for our study regions: 1200–1000 BC in the area of Tichitt-
Walata and its margins, 800–1 BC in the Méma and Macina regions of
the old IND and the Mauritanian Tagant, and a settlement transformation
with abandonment, growth and nucleation around AD 400 across most of
our zone.
50
MacDonald 2011a; MacDonald and Schmidt 2004; Togola 2008. 51 Togola 2008, 83.
52
Bedaux et al. 2005, 446. 53 See Wilson and Schmidt 2005, 35–41, Figure 3.I.3.
54
Schmidt 2005, Table 10.1.2. 55 Park 2010.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 514 Kevin C. MacDonald

Assessing Parallels with the Trans-Saharan World

What might all of this mean for the Sahel’s relationship with its northern
neighbours? At a broad, synthetic level, I prefer to see the Trans-Saharan
region as a sphere of interaction in trade and ideas – much like the
Mediterranean. These ideas go both ways. It is to our advantage to inves-
tigate both potential syncretisms in architecture and relationships in set-
tlement dynamics. It is possible to make a number of preliminary
observations concerning both local innovations and connections within
our sphere of interaction.
1) The flourishing of Tichitt appears to long pre-date any comparable
proto-urban, proto-state development on the northern margins of the
Sahara. True, there remain relatively un-prospected and poorly dated
areas in southern Morocco and Algeria, but at 1200–1000 BC Tichitt
and Walata still remain without peer. As such Tichitt represents that
great rarity – a pristine complex society – making its relative lack of
global status within archaeology all the more remarkable.
2) The peak of Classic Garamantian civilization in southern Libya was
between the first century BC and the fourth century AD, with related
social forms dating back to the early first millennium BC.56 The outset of
this period corresponds with the decline and dissolution of settlement
around Tichitt-Walata and the end of this period falls shortly before the
greatest flowering of early urbanism in the Middle Niger.57 Conflict with
Berber populations, made manifest in fortified sites and rock art, has been
claimed as one of the reasons for the decline of Tichitt.58 Others have also
put forward a case for some element of syncretism between incoming
Berber populations and local populations during the Late Tichitt Phase
(1000–400 BC).59 During the earliest Garamantian heyday (100 BC),
Sahelian settlement is only known to have been vibrant in the
Tagant, Méma, Macina and the Inland Niger Delta. Between 100 BC
to AD 300 one can only point to the Jenné-jeno and the Tombouze
cluster near Timbuktu as areas experiencing documented settlement
growth. That Tombouze, in a more proximate zone, may have to some
extent been stimulated by Trans-Saharan interactions is not outside the
realm of possibility, but it appears to lack any raw material indices of such

56
Liverani 2003; Mattingly 2011; Mori 2013.
57
Though the late antique expansion of Garamantian fortified settlements noted by Mattingly
et al., Chapter 2, this volume overlaps.
58
Munson 1980. 59 MacDonald et al. 2009.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 515

interaction.60 Likewise, the initial floruit of urban settlement in Gao

appears to be much later – in the seventh century AD.61 Thus, one is
left with a surprising lack of linkages between Classic Garamantian age
Fazzan and sites at the Sahelian margins – areas where correspondence
should have been at its strongest.
3) Architecturally there is no reason to see an external hand behind the
round compounds and structural forms of Classic Tichitt. Likewise
there is no reason to view the transformation of this stone block and
shingle built technology to similar forms in coursed earth as anything
but a localised development. However it is interesting to note the
advent of rectangular built forms in the Tagant (c.150 BC–AD 350)
and much later in the Gourma (c.AD 600). It is tempting to view these
developments as syncretic phenomena tied to long-distance interac-
tions with the wider ‘Berber’ world. Parallels between the rectilinear
coursed earth structures of Aghram Nadharif (Liverani 2006) and those
of Tongo Maaré Diabel are very interesting in this regard. Similarly,
there is the question of the advent of Sahelian mudbrick (AD 400–600),
which may either be part of the exchange of ideas at the Saharan margin
or a local development linked to the local (Sahelian) growth of settle-
ment during this period.
4) It is apparent that the first millennium BC is not the obscure period of
low settlement it was once thought to be. There appears to have been
concentrated settlement growth and individual sites exceeding 10 ha in
three portions of the wider Sahel: the Méma (Kolima, Akumbu, etc.),
the Macina (the Dia-Shoma cluster) and, much further to the east, the
southern margins of Lake Chad (the Zilum cluster).62 With such dates
as we have, growth seems to fall closer to the middle of the millennium
(c.600–400 BC) than to either end. Such developments pre-date com-
parable nucleated settlements in south-west Fazzan around Ghat, but
overlaps with occupation at Zinkekra in the Garamantian heartlands.63
At the current state of evidence I would agree with Magnavita et al. that
in these areas we are witnessing the ‘first steps towards urbanism’ with
growing site hierarchies and densities, evidence for specialists and (at
least in the case of Zilum) defences.64 I would also agree that ‘increasing

60
Park, personal communication. 61 Cissé et al. 2013.
62
Kolima, Akumbu, etc.: MacDonald 2011b; Togola 2008; the Dia-Shoma cluster: Bedaux et al.
2005; MacDonald 2011b; and the Zilum cluster: Magnavita et al. 2006.
63
For the Ghat area, see Mori 2013 (Fewet) and Liverani 2006 (Aghram Nadharif). For the
Garamantian heartlands, see Mattingly 2003, 136–42; Mattingly 2010, 19–119.
64
Magnavita et al. 2006, 168.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 516 Kevin C. MacDonald

social complexity evolved in Sub-Saharan Africa at an early date from

a fully indigenous cultural background’.65
5) Liverani speculates that Garamantian Trans-Saharan trade routes
might extend back as far as the sixth century BC, even if supporting
evidence for this assertion is fairly thin on the ground.66 As I have
written elsewhere, there is persistent, scattered evidence for down-
the-line trade in Saharan stone beads (especially carnelian and ama-
zonite) since the second millennium BC, but there is no physical
evidence at southern sites for an acceleration of such interaction until
the fourth century AD.67 From AD 300 to AD 700 indices of long
distance exchange increase in variety and ubiquity, if not in quantity:
copper and chickens at Jenné-jeno, glass beads at Dia and chickens at
Tongo Maare Diabel; glass and carnelian beads and wool cloth at
Kissi, and so forth.68 The timing of this greater flow of materials may
align with the advent of the Trans-Saharan gold trade, as posited by
Garrard,69 but it is most likely that such trade grew in partnership
with emergent Sub-Saharan urban or proto-urban networks. In sum,
it is difficult to engage with long-distance trading partners who are
not already adept at assembling goods from their own regional
networks.
From the foregoing it should be apparent that there is no dramatic
or straight-forward parallel between the Classical Garamantean era
and the growth of settlement in the West African Sahel. However,
there are many disparate elements which indicate a Trans-Saharan net-
work of communication (exchanges of ideas) since at least the first
millennium BC, intensifying post AD 300. The largest missing pieces
in our puzzle – which covers the coalescing Trans-Saharan zone in the
first millennia BC/AD – lie in other areas of the Saharan rim, particularly
southern Algeria. We must envy the future generations to whom that
terrain will be open.

References

Amblard-Pison, S. 2006. Communautés villageoises néolithiques des Dhars Tichitt

et Oulata (Mauritanie). Oxford: Archaeopress.

65
Magnavita et al. 2006, 168. 66 Liverani 2003. 67 MacDonald 2011b.
68
MacDonald 2011b; Magnavita 2017; McIntosh 1995.
69
Garrard 1982; Phillipson 2017; cf. Nixon 2017.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 517

Bedaux, R., MacDonald, K.C., Person, A., Polet, J., Sanogo, K., Schmidt, A. and
Sidibé, S. 2001. The Dia Archaeological Project: Rescuing cultural heritage in the
Inland Niger Delta (Mali). Antiquity 75: 837–48.
Bedaux, R., Polet, J., Sanogo, K. and Schmidt, A. (eds). 2005. Recherches
archéologiques à Dia dans le Delta Intérieur du Niger (Mali): Bilan des saisons
de fouilles 1998–2003. Leiden: CNWS Publications.
Cissé, M., McIntosh, S.K., Dussubieux, L., Fenn, T., Gallagher, D. and Chipps
Smith, A. 2013. Excavations at Gao Saney: New evidence for settlement growth,
trade and interaction on the Niger Bend in the first millennium CE. Journal of
African Archaeology 11: 9–37.
David, N. 1971. The Fulani compound and the archaeologist. World Archaeology 3:
111–31.
Dmochowski, Z.R. 1990. An Introduction to Nigerian Traditional Architecture,
Volume I: Northern Nigeria. London: Ethnographica.
Dowler, A. and Galvin, E.R. (eds). 2011. Money, Trade and Trade Routes in Pre-
Islamic North Africa. London: British Museum Press.
Dueppen, S.A. 2012. Egalitarian Revolution in the Savanna: The Origins of a West
African Political System. Sheffield: Equinox.
Dujarric, P. 1981. L’architecture traditionnelle au Sénégal oriental. Objets et
Mondes 21: 141–48.
Garrard, T. 1982. Myth and metrology: The early Trans-Saharan gold trade.
Journal of African History 23: 443–61.
Gestrich, N. 2013. The Archaeology of Social Organisation at Tongo Maaré Diabel.
Unpublished PhD Dissertation, University College London.
Gestrich, N. and MacDonald, K.C. 2018. On the margins of Ghana and KawKaw:
Four seasons of excavation at Tongo Maaré Diabal (AD 500–1150), Mali.
Journal of African Archaeology 16.1: 1–30.
Holl, A. 1985. Background to the Ghana Empire: Archaeological investigations on
the transition to statehood in the Dhar Tichitt region (Mauritania). Journal of
Anthropological Archaeology 4: 73–115.
Holl, A. 1986. Economie et société néolithique du Dhar Tichitt (Mauritanie). Paris:
Editions Recherche sur les Civilisations.
Holl, A. 1993. Late Neolithic cultural landscape in southeastern Mauritania: an
essay in spatiometrics. In A.F.C. Holl and T.E. Levy (eds), Spatial Boundaries
and Social Dynamics: Case Studies from Food-Producing Societies. Ann Arbor:
International Monographs in Prehistory, 95–133.
Liverani, M. 2003. Aghram Nadharif and the southern border of the Garamantian
kingdom. In M. Liverani (ed.), Arid Lands in Roman Times: Papers from the
International Conference (Rome, July 9th–10th 2001), Arid Zone Archaeology
Monograph no.4. Firenze: All’insegna del Giglio, 23–36.
Liverani, M. 2006. Aghram Nadharif. The Barkat Oasis (Sha’abiya of Ghat, Libyan
Sahara) in Garamantian Times. Arid Zone Archaeology Monographs no.5.
Firenze: All’insegna del Giglio.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 518 Kevin C. MacDonald

MacDonald, K.C. 1994. Socio-Economic Diversity and the Origins of Cultural

Complexity along the Middle Niger (2000 BC to AD 300). Unpublished PhD
dissertation, University of Cambridge.
MacDonald, K.C. 1996. Tichitt-Walata and the Middle Niger: Evidence for cultural
contact in the second millennium BC. In G. Pwiti and R. Soper (eds), Aspects of
African Archaeology: Papers from the 10th Congress of the Pan-African Association
for Prehistory and Related Studies. Harare: University of Zimbabwe Publications,
429–40.
MacDonald, K.C. 1998. More forgotten tells of Mali: An archaeologist’s journey
from here to Timbuktu. Archaeology International 1: 40–42.
MacDonald, K.C. 2011a. Betwixt Tichitt and the IND: The pottery of the Faïta
facies, Tichitt tradition. In Identity Fashion and Exchange: Pottery in West Africa
(special issue edited by A. Haour and K. Manning) = Azania: Archaeological
Research in Africa 46.1: 49–69.
MacDonald, K.C. 2011b. A view from the south: Sub-Saharan evidence for contacts
between North Africa, Mauritania and the Niger, 1000 BC–AD 700. In Dowler
and Galvin 2011, 72–82.
MacDonald, K.C. 2015. The Tichitt tradition in the West African Sahel. In
G. Barker and C. Goucher (eds), The Cambridge World History, Volume II:
A World with Agriculture 12000 BCE – 500 CE. Cambridge: Cambridge
University Press, 499–513.
MacDonald, K.C. and Schmidt, A. 2004. The Faïta facies and the ‘obscure millen-
nium’ of the Middle Niger. In Sanogo and Togola 2004, 222–29.
MacDonald, K.C., Vernet, R., Martinon-Torres, M. and Fuller, D.Q. 2009. Dhar
Néma: From early agriculture to metallurgy in southeastern Mauritania. Azania:
Archaeological Research in Africa 44: 3–48.
Magnavita, C., Breunig, P., Ameje, J. and Posselt, M. 2006. Zilum: A mid-first
millennium BC fortified settlement near Lake Chad. Journal of African
Archaeology 4: 153–69.
Magnavita, S, 2017. Track and trace. Archaeometric approaches to the study of
early Trans-Saharan trade. In Mattingly et al. 2017, 393–413.
Mattingly, D.J. 2003. The Archaeology of Fazzan, Volume 1 Synthesis. London:
Society for Libyan Studies.
Mattingly, D.J. 2010. The Archaeology of Fazzan, Volume 3, Excavations Carried
out by C.M. Daniels. London: Society for Libyan Studies.
Mattingly, D. 2011. The Garamantes of Fazzan: An early Libyan state with trans-
Saharan connections. In Dowler and Galvin 2011, 49–60.
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017. Trade in the Ancient Sahara and Beyond. Trans-Saharan
Archaeology Volume 1. Series editor D.J. Mattingly. Cambridge: Cambridge
University Press and The Society for Libyan Studies.
McIntosh, R.J. and McIntosh, S.K. 1988. From ‘siècles obscurs’ to revolutionary
centuries on the Middle Niger. World Archaeology 20: 141–65.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 13 Architecture and Settlement Growth on the Southern Edge of the Sahara 519

McIntosh, S.K. and McIntosh, R.J. 1980. Prehistoric Investigations in the Region of
Jenné, Mali: Part I. Oxford: British Archaeological Reports.
McIntosh, S.K. (ed.). 1995. Excavations at Jenné-Jeno, Hambarketolo, and Kaniana
(Inland Niger Delta, Mali), the 1981 Season. Berkeley: University of California
Press.
Mori, L. (ed.). 2013. Life and Death of a Rural Village in Garamantian Times:
Archaeological Investigations in the Oasis of Fewet (Libyan Sahara). Arid Zone
Archaeology Monographs no. 6. Firenze: All’insegna del Giglio.
Munson, P.J. 1971. The Tichitt Tradition: A Late Prehistoric Occupation of the
Southwestern Sahara. Unpublished PhD dissertation, University of Illinois at
Urbana-Champaign.
Munson, P.J. 1980. Archaeology and the prehistoric origins of the Ghana Empire.
Journal of African History 21: 457–66.
Nicolaisen, J. and Nicolaisen, I. 1997. The Pastoral Tuareg: Ecology, Culture and
Society. 2 vols. London: Thames and Hudson.
Nixon, S. 2017. Trans-Saharan gold trade in pre-modern times: Available evidence
and research agendas. In Mattingly et al. 2017, 156–88.
Ould-Khattar, M. 1995. La fin des temps préhistoriques dans le sud-est Mauritanien.
Unpublished PhD dissertation, Université de Paris-I, Pantheon-Sorbonne.
Park, D.P. 2010. Prehistoric Timbuktu and its hinterland. Antiquity 84: 1076–88.
Park, D.P., Coutros, P., Mahmoud Abdallahi, M. and Ould Sidi A. 2010. La campagne
de recherche archéologique dans la région de Tombouctou et la région des lacs:
Rapport sur la troisième campagne de recherche à Tombouctou préhistorique.
Report to the DNPC (Mali), La Mission Culturelle de Tombouctou, and Yale
University. Available online at: www.academia.edu/1497424/_3 [last accessed
13 September, 2019].
Pauketat, T.R. 2007. Chiefdoms and Other Archaeological Delusions. Lanham: Alta
Mira.
Phillipson, D.W. 2017. Trans-Saharan gold trade and Byzantine coinage. The
Antiquaries Journal 97: 145–69.
Prussin, L. 1986. Hatumere: Islamic Design in West Africa. Berkeley: University of
California Press.
Raimbault, M., and Sanogo, K. (eds). 1991. Recherches archéologiques au Mali: Les
sites protohistoriques de la zone lacustre. Paris: ACCT-Karthala.
Sanogo, K. and Togola, T. (eds). 2004. Acts of the XIth Congress of the Pan-African
Association Prehistory and Related Fields, Bamako, February 7–12, 2001.
Bamako: Soro Print Color.
Schmidt, A. 2005. Prospection régionale autour de Dia. In Bedaux et al. 2005,
401–22.
Takezawa, S. and Cissé, M. 2004. Domestication des cereales au Méma, Mali. In
Sanogo and Togola 2004, 105–21.
Togola, T. 2008. Archaeological investigations of Iron Age sites in the Méma Region,
Mali (West Africa). Oxford: British Archaeological Reports.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 520 Kevin C. MacDonald

Togola, T. and Raimbault, M. 1991. Les missions d’inventaire dans le Méma, Karéri
et Farimaké (1984–1985). In Raimbault and Sanogo 1991, 81–98.
Vernet, R. 1993. Préhistoire de la Mauritanie. Nouakchott: Centre Culturel
Français A. de Saint Exupéry, Sépia.
Walicka Zeh, R. 2000. Building Practice and Cultural Space amongst the Bamana,
Senufo and Bozo of Mali: An Ethnoarchaeological Study. Unpublished PhD
Dissertation, University College London.
Walicka Zeh, R. and MacDonald, K.C. 2004. An ethnoarchaeological study of
architectural remains and spatial organization, an example from the site of
Tongo Maaré Diabal, Mali. In Sanogo and Togola 2004, 353–64.
Wilson, J. and Schmidt, A. 2005. La prospection de Dia-Shoma. In Bedaux et al.
2005, 35–41.

https://doi.org/10.1017/9781108637978.014 Published online by Cambridge University Press

 14 Long-Distance Exchange and Urban Trajectories
in the First Millennium AD
Case Studies from the Middle Niger and Middle Senegal River
Valleys

susan keech mcintosh

Introduction

It has been 40 years since the first publication of results of archaeological

investigations at the massive settlement mound of Jenné-jeno challenged
traditional historiography concerning towns and trade in West Africa.
Prior assumptions that Arab-initiated Trans-Saharan trade provided the
impetus for town growth along the Middle Niger have yielded to an
explanatory model grounded in the local development of regional, inter-
regional and long-distance exchange and interaction by the mid-first
millennium AD.1 These findings coincided with an active political and
scholarly agenda to decolonise Africa’s past, which provided an incentive
for archaeologists to seek additional instances of early trade and town
growth along the Middle Niger (Méma, Dia, Timbuktu, Gao, Bentia) and
the Middle Senegal (Fig. 14.1).2
The deployment of comparable research methodologies by many of
these projects has made possible the construction of preliminary, evidence-
based frameworks for discussing trade and exchange, interaction, and
town growth in different areas, although data density remains notably
low and patchy. Central to these discussions are definitions of towns,
urbanism, and trade. Back in the 1980s, Rod McIntosh and I grappled
with the question of urbanism in the context of Jenné-jeno’s unexpected
pattern of early settlement growth and close clustering of other, contem-
poraneous settlement mounds, at which we found no evidence for public

1
McIntosh S. and McIntosh R. 1980; McIntosh, R. and McIntosh, S. 1981.
2
Méma: McIntosh S. 2017; Togola 1996; 2008; Dia: Bedaux et al. 2005; Haskell et al. 1988;
McIntosh, R. and McIntosh, S. 1987; Timbuktu: Insoll 2000a; McIntosh, S. and McIntosh,
R. 1986; Park 2010; 2011; Gao: Cissé et al. 2013; Insoll 1996; 2000b; Bentia: Arazi 1999; Middle
Senegal: McIntosh, S. et al. 1992; McIntosh, R. et al. 2016; McIntosh and Bocoum 2000. 521

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 522 Susan Keech McIntosh

Figure 14.1. Map of sites, regions and major trade routes from the tenth century.

architecture or detectable ranking or stratification during the first millen-

nium of occupation. We proposed expanding the definition of urban
beyond the then-dominant classical, evolutionary models in order to
include ‘cities without citadels’.3 Key elements in our definition of urban-
ism were site size, population density and heterogeneity, and functional
specialisation, in line with Trigger’s definition of a city as providing
specialised services to a broader hinterland.4 As our thinking developed,
we theorised that both site clustering and absence of monumental archi-
tecture or clear social ranking at Jenné-jeno could be related to an effective
resistance to the monopolisation of power. The result would be power
relations that were counterpoised and horizontally distributed (that is,

3 4
McIntosh, S. and McIntosh, R. 1993. Trigger 1972.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 523

heterarchy), rather than effectively consolidated vertically in a hierarchy of

elites.5 The debate surrounding urban definitions continues, with archae-
ologists forging new frameworks for understanding population aggrega-
tion on a substantial scale in other areas of the world where it was likewise
not visibly accompanied by one or more of the classical markers of
urbanism.6 Africanist archaeology continues to provide important inputs
to this debate.7
Definitions of trade and exchange figure importantly in the discussions
in this chapter and in this volume. Analytic programmes focused on
identifying source areas of non-local raw and manufactured materials at
African sites have increased significantly over the past two decades.
Consequently, we have some insight, albeit preliminary, into the networks
along which certain categories of goods – especially glass, copper-based
metals, and some kinds of stone – moved in West Africa and their
chronology. Reports on foreign goods have been an important component
of the recent Africanist literature,8 reflecting archaeology’s recent global
turn. Ironically, this has occurred as theorising about trade in the broader
archaeological literature has waned. Bauer and Agbe-Davies provide
a useful overview of the shifting position of trade studies in archaeology.9
As they point out, most of the influential archaeological literature on trade,
exchange and interaction appeared in the 1970s and 1980s. The subsequent
decline in trade-oriented theoretical studies probably reflects the influence
of post-processual concerns and the association of trade with the positivist
programme of processualism. In twenty-first-century Africanist archaeol-
ogy, however, trade has remained a strong theme, emphasising the role of
African agency in trade relationships ranging from the Roman period to
the Atlantic trade and the mapping of global intersections both within and
beyond these historically-documented systems.
In the Africanist literature, ‘trade’ is frequently invoked to account for
exotics present in an archaeological site, without specifying what the term
encompasses. As Renfrew pointed out 40 years ago, goods can move over
distances by many different modalities with very different levels of orga-
nisation, from hand-to-hand, down-the-line reciprocal exchange, to cen-
tral place redistribution, to market-based trade.10 His proposition that the
5
Crumley 1995; McIntosh, S. 1999b.
6
For example, Birch 2013; Chapman et al. 2014; Creekmore and Fisher 2014; Fletcher 2009;
Lekson 2018.
7
For example, Haour 2005; Kusimba 2008; Manyanga et al. 2010; McIntosh, R. 2015; Wynne-
Jones and Fleisher 2014. See also, Magnavita, Chapter 15, this volume.
8
For recent summaries see Dowler and Galvin 2011; Magnavita 2013; Mattingly et al. 2017.
9
Bauer and Agbe-Davies 2010, 28–47. 10 Renfrew 1975.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 524 Susan Keech McIntosh

patterned distribution and scale of exotics would permit identification of

specific modalities ultimately proved unrealistic. It not only required a very
high density of reliably representative data from a large number of sites,
but it also could not resolve the problem of equifinality – that different
processes could result in similar patterns. In West Africa, the exceptionally
sparse, patchy nature of the data on exotics is a challenge for anyone who
wishes to investigate the changing landscape of interaction and exchange
in the centuries preceding historical documentation of market-dominated
Trans-Saharan trade. Central to such studies is an understanding that
material exchange is embedded in larger webs of communication that
facilitate the exchange of ideas, values, and information along any or all
directions of the network.11 Here, I am using exchange in the sense of
interaction, as outlined by Oka and Kusimba.12 It is a key point because it
implies that even if material evidence of foreign goods or raw materials is
lacking, other clues to exchange relationships and interactions may be
present. We must be attentive to evidence for changes in style and taste,
technological practice, symbolic systems, and social organisation, in addi-
tion to the presence of exotics. In the absence of all of these, it is difficult to
make a case for interaction, let alone trade.
In this chapter, I follow Bauer and Agbe-Davies’ definitions for
exchange and trade:

Exchange refers to the transfer of goods through a wide range of mechanisms, from
ritualised gift exchange to the negotiated transactions of barter and markets and
the one-way exchange of coercion and piracy. Trade is a more specific category of
activity in which the exchange is more formalised and market based, both in the
individual interaction and on a systemic scale. Trade is thus one type of exchange
relationship in which each interaction is usually ‘closed’, or completed in a single
moment of exchange of x for y, and which often occurs across otherwise powerful
social and geographic boundaries.13

This definition of trade corresponds to our understanding of the historic

Trans-Saharan trade, in which commodities (gold, slaves, salt, cloth, cop-
per) were exchanged at entrepôts and market towns. The trade was sup-
ported by broader networks linking traders and towns into webs of
interaction that involved many different kinds of exchange.
The case studies in this chapter outline the evidence for exchange at
different scales and its temporal relation to settlement growth. At Gao on
the eastern Niger Bend, a surge in the scale and intensity of long-distance

11 12
Bauer and Agbe-Davies 2010, 19. Oka and Kusimba 2008, 340.
13
Bauer and Agbe-Davies 2010.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 525

exchange after the seventh century AD was accompanied by rapid settle-

ment expansion. In the western middle Niger at Jenné-jeno, settlement
growth is associated with regional exchange systems that are present from
initial colonisation. Networks expanded to include the Southern Sahara
by AD 500. Only after AD 900 is there material evidence of interactions
that extend across the Sahara, by which time, the settlement mound had
reached its maximum extent of over 30 ha. In the Middle Senegal, exchange
interactions and agropastoral mobility linked the river with mobile popu-
lations as far north as the Mauritanian Adrar during the later first
millennium BC, as attested by the distribution of Mauritanian copper
and related pottery styles. These interactions appear to have largely ceased
c.100 BC and did not reappear for almost a millennium. Evidence for long-
distance exchange and interaction reappears after c.AD 800, including
copper and glass working, new building technologies, and pottery styles,
along with newly founded Saharan entrepôts. Arabic sources identify the
existence of the Takrur polity along the Middle Senegal by the eleventh to
twelfth centuries; unexpectedly, we found no evidence of urban centres
within our study region.
Before continuing on to the case studies, I would emphasise the critical
role of paleoclimate. Shifting patterns of rainfall, wind and evaporation,
groundwater recharge, and river regimes fundamentally affect food secur-
ity and exchange in the marginal lands of the Sahel and neighbouring
Sahara.14 Movement in the desert is dependent on aquifer and water table
recharge to supply wells and oases. The limits of agricultural and pastoral
production fluctuate within different regions depending on local rainfall,
which varies both interannually and at longer scales of decades and cen-
turies. During the last 3,000 years, the Sahel-desert boundary has shifted
200 km or more north of its present latitude on several occasions, most
recently, in the seventeenth century, as documented by Webb (Fig. 14.2).15
Conversely, intense arid episodes analogous to the Sahelian drought of the
1970s and 1980s have occurred multiple times in the past 3,000 years. At
these times, the Senegal and Niger river valleys attracted diverse popula-
tions from surrounding areas, creating significant interaction zones.
The colonisation of the Middle Niger and Middle Senegal floodplains by
cultivators and agropastoralists in the early first millennium BC was linked
to an arid period in an already dry Sahara. Early farming/herding/fishing
communities were established on the northern margins of the Niger and

14 15
Maley and Vernet 2015. Webb 1995.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 526 Susan Keech McIntosh

WESTERN SAHARA

WESTERN SAHARA
Ijil Ijil

ADRAR ADRAR
Cap Blanc Cap Blanc

WESTERN SAHEL

Senegal River
WESTERN SAHEL
Senegal River
Timbuktu Timbuktu
WESTERN SAVANNA
Cubalel Cubalel

Dia Niger River WESTERN SAVANNA Dia Niger River

Cap Vert Cap Vert
Jenne-jeno Jenne-jeno

Plateau regions Sahelian cattle zone Rock salt deposit

0 200 400km
Settled agricultural zone Great camel zone Town

Figure 14.2. Change in land use zones from 1600 (left) to 1850 (right) as reconstructed
from historical documents by Webb (1995, 6, 10).

Senegal middle valleys at this time.16 Subsequently, a wetter period

between 800–400 BC in Mauritania appears to correspond to the extension
of cattle herding activities northward towards the Adrar. Copper ores at
Akjoujt were processed and widely exchanged among these pastoral/agro-
pastoral groups. This may have been a period in which movement over
long distances both north-south and east-west was facilitated. Evidence for
exchange between the Middle Senegal valley and the Northern Sahel/
Southern Sahara disappears sometime between 100 BC and AD 100. This
possibly coincided with a dramatic arid episode in the course of an erratic,
erosive phase that lasted until AD 300–400. During this period, river flood
regimes and rainfall would have been very unpredictable. Until c.AD 600,
rainfall in the Sahel was very low, even as the rainfall further south was high
and ‘out of phase’ with the Sahel,17 supplying generous floods to the Middle
Senegal and Niger floodplains. Higher rainfall and lower evaporation
returned to the Sahel and Southern Sahara in the seventh century AD,
coincident with the appearance of glass and copper on a significant scale
along the eastern Niger Bend and further east in Niger at Marandet.18
The current state of our knowledge about the three areas discussed
here – the upper Inland Niger Delta around Jenné-jeno, the eastern Bend
of the Niger at Gao Saney and Gao Ancien, and the Middle Senegal Valley
at Cubalel and Sincu Bara – is lamentably partial. However, the sequences

16
Bedaux et al. 2005; Deme and McIntosh 2006; MacDonald, Chapter 13, this volume.
17
Maley and Vernet 2015, 191.
18
Grébénart 1985, 354–79; Magnavita 2013; Magnavita et al. 2007.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 527

MIDDLE SENEGAL MIDDLE NIGER SOUTHERN SAHARAN CENTRAL SAHARA

TRADE TOWNS FAZZAN

1600

1400

KUMBI SALEH
PHASE IV

GAO ANCIEN

ESSUK
1200

TEGDAOUST
GAO SANEY
1000

MARANDET

POST-
800
JENNE-JENO

PHASE III
SINCU BARA

LATE
600
CUBALEL

400

GARAMANTIAN
?
PHASE I/II

CLASSIC
JARMA
200
BC AD

200

EARLY
WALALDE

400

600

800

1000
YEAR

Figure 14.3. Timeline showing excavation sequences discussed in the text.

excavated thus far provide a preliminary framework for evaluating the case
for multi-scalar interaction, exchange, and settlement growth at each
(Fig. 14.3).

Jenné-jeno: An Early Town Centre for Regional Exchange in the

Inland Niger Delta

Jenné-jeno shares with its descendant settlement, Djenné, a location on the

Inland Niger Delta (IND) floodplain that is optimal for numerous sub-
sistence pursuits, including flood cultivation of African rice, dry season

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 528 Susan Keech McIntosh

pasturage for livestock, rain-fed cultivation of millet on non-inundated

lands, and fishing, plus nearby access to the Bani River for transport and
trade. Beginning in 1977 and continuing in 1980–1981, 1994, 1997–1998,
2008, and 2010, the occupation mounds in the vicinity of the town of
Djenné have been investigated through multiple phases of excavation,
coring, and surface survey, with a primary focus on the 30 ha mound of
Jenné-jeno.19 Eighteen units of sizes up to 10 × 6 m have been excavated
down to sterile earth on Jenné-jeno (maximum depth, 6 m) with ten
additional units excavated on neighbouring mounds. A robust material
culture sequence has been established and confirmed by analysing the
material culture in each excavated unit separately to test the proposed
time-sensitive attributes and diagnostic pottery types for the four occupa-
tion phases we defined.20 Excavation proceeded according to identifiable
deposition contexts, seeking to maintain maximal stratigraphic integrity.
Excavated contexts included houses, workshops, burials, refuse pits, mud
wall melt accumulations, and abandonment surfaces. A series of 29 radio-
carbon dates anchors the sequence in time, beginning with initial occupa-
tion at 112 calBC–calAD 65,21 and continuing until final abandonment by
1400. From initial settlement, the subsistence economy included domestic
cattle and ovicaprids, fish, wild bovids, domestic rice, millet, and wild
cereals, with little evidence for major change in emphasis over time.
Exploration of urbanism at Jenné-jeno has focused on the evidence for
rapid settlement mound expansion in the first millennium AD accompa-
nied by the development of numerous other mounds in a distinctive,
tightly clustered pattern. The absence of evidence for large public buildings
or monuments, or wealthy elites, and the settlement’s participation in
widening networks of exchange and interaction have been of particular
interest. The major architectural feature at the site is a 3.7 m wide

19
Initial instrument survey in 1977 indicated an area of 33 ha. The reduction to 30 ha according to
instrument survey in 2008 by D. Park reflects significant erosion, especially on the western side
of the site, in addition to differences in defining site boundaries on the descending slope of the
mound. The excavations at Jenné-jeno and regional surveys have been published in two
volumes, McIntosh, S. and McIntosh, R. 1980; McIntosh, S. 1995.
20
A typological approach to identifying and recording pottery was, however, avoided, see
McIntosh, S. and McIntosh, R. 1980; McIntosh, S. 1995, 130–32. Rather than organising pottery
recording on the basis of pre-established types or wares, feature sherds (rims, bases) were
processed individually, with 15 formal variables recorded for each sherd, in the interest of
preserving potentially significant data on variability, both intra-site and inter-site.
21
McIntosh, S. 1995, 60; McIntosh, S. and McIntosh, R. 1980, 93, 195–97. A new AMS 14C date of
2018±36 (AA94432) for the initial occupation deposit in Unit LX-N narrows the date range
previously available from low precision standard radiocarbon dates.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 529

mudbrick wall that encircled the northern three quarters of the site.22
Unlike Takrur and Gao, Djenné was not mentioned in any written trade
itinerary or account until the fifteenth century. Located on an internal
waterway, it apparently existed outside the reach of Arab and Berber
traders who plied Saharan trade.
Several aspects of the Jenné-jeno data have been the focus of our inter-
pretive efforts.

Rapid Settlement Growth

The presence of the earliest, Phase I pottery types at the bottom of six
excavation units (ALS, LXN, LXS, M1, M2, CTR, DT) in the central and
south-eastern sectors of the site suggests that the site was quite large by AD
400 although we cannot be sure that occupation was continuous across the
c.10 ha expanse these units encompass (Fig. 14.4). Models of site growth
have included the possibility of two or more small mounds accreting over
time.23 Furthermore, all excavated units, including neighbouring
Hambarketolo, have painted polychrome pottery that is diagnostic for
Phase III, dated to AD 400–900. We interpret this as indicating that Jenné-
jeno had reached its maximum expanse by the end of this phase. The
massive city wall was built at this time, enclosing 3,000 or more
inhabitants.24 By the end of Phase IV, c.AD 1400, Jenné-jeno was aban-
doned for reasons about which it is possible only to speculate.

Exchange and Interaction

The presence of iron and stone beads and grindstones from the beginning of
occupation indicates probable exchange relationships, as these materials are
not present in the floodplain. Copper appeared c.AD 500 and gold was
present at the end of the first millennium AD. Lead isotope analysis by T.
Fenn (personal communication) indicates that the Phase IV (AD 900–1400)

22
Although the existence of the Jenné-jeno city wall has been called into question by Schmidt
et al. 2005, 124, the wall has been mapped several times since 1977 and, as a result of
accelerating erosion, is now elevated 10 cm or more above the surface at multiple points around
the circumference of the site.
23
McIntosh, S. 1995, 23.
24
Estimates of population size are fraught with pitfalls, see McIntosh, S. and McIntosh, R. 1993,
633 and McIntosh, S. 1999b, 71–73 for discussion of methods. Our early estimates, based on
average population densities at occupied mounds in the Inland Delta historically, were
undoubtedly too high, since Jenné-jeno lacked the two-storey buildings that characterise
settlements today and have been revised.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 530 Susan Keech McIntosh

5300

NWS
5200

ALS

5100 M2 CP1
LXN
WFL 196
KIS LXS

M1
5000 196 199
195 197 198
194 197

193
192 196
CP2 195
CTR
4900

TK

194

4800
193

JF1
192

6
5
4700 4
3
2 SB DT
1
0

PHASE IV
PHASE III
4600
PHASE I/II
WALL REMAINS
CORE LOCATION 1994
EXCAVATION UNITS 1977,1981
HK
EXCAVATION UNITS 1997,1999

4500

1600 1700 1800 1900 2000 2100 2200 2300

Figure 14.4. Excavation and augur coring locations on Jenné-jeno, showing the depth
of deposits and phase chronology for the excavation units.

copper alloy artifacts are all from a Moroccan source. The earliest, unalloyed
copper artifact (sample 1460), on the other hand, appears to derive from an
unknown West African source. While luxury exotics (glass, copper, stone
beads) were rare throughout the sequence (Table 14.1), imports of stone
grinders number in the hundreds, and iron slag is nearly ubiquitous in the
deposits, suggesting regional movement of iron blooms on a significant scale
or, less likely, the import of ore for smelting. The beginnings of large-scale

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 531

Table 14.1 Imported materials excavated from Jenné-jeno in 1977 and 1981

Copper-Based Glass Stone Sandstone Unit

Period Iron Objects Objects Gold Beads Beads LX-N (43 Kg)

I/II
100 BC –AD 400 1125 g 3 7 83 pieces
III
AD 400–900 1654 g 10 (12 g) 1 (3.9 g) 2 2 83 pieces
IV
AD 900–1400 1739 g 23 (73 g) 8 5 67 pieces

regional iron smelting have been dated to the seventh century at Fiko,
100 km downriver near Mopti.25
Based on the historical role of Djenné in provisioning drier areas downriver
to the north with surplus staple goods such as rice and dried fish,26 we have
speculated that Jenné-jeno in the first millennium similarly engaged in
exchange at multiple scales that included regions downriver and outside the
floodplain, as well as long-distance exchange for salt and copper.27 We propose
that exchange is implicated in the rapid growth of the settlement, and note that
evidence of North African cultural influences (rectilinear house plans, technol-
ogy transfer (spinning, weaving, cotton seeds), and imported brass) all post-
date AD 900. The date at which Jenné-jeno’s networks became effectively
linked to Trans-Saharan trade is not definitively known. Interaction with the
Lakes Region and the town of Kumbi Saleh is suggested by the presence of
distinctive white-on-red geometric pottery, which dates to AD 800–1000 at
Jenné-jeno (Fig. 14.5).
The earliest painted pottery at Jenné-jeno (100 BC–AD 200) used cross-
hatched red paint on a burnished, unslipped surface.28 Around AD 200
polychrome pottery (white and black paint on red slip) appeared contem-
poraneously with polychrome in both the Lakes Region and at Tombouze
near Timbuktu.29 The inspiration for this new decorative style is not

25
Robion-Brunner 2010.
26
Transport of these goods by large canoes to Timbuktu was reported by Leo Africanus in the
sixteenth century and René Caillié in the nineteenth century.
27
McIntosh, S. 1995; 2018; McIntosh, S. and McIntosh, R. 1993.
28
Rare finds of similarly decorated sherds in the Garamantian sphere have been proposed as evidence
of interaction between the Garamantes and Jenné-jeno in the early first millennium AD by Gatto
2006 and Liverani 2006, 446. Chemical analysis confirmed that tested sherds were not trade items –
Artioli et al. 2006. With regard to both vessel forms and fabrication technique, I find the
Garamantian pottery assemblage to be fundamentally unlike Phase I Jenné-jeno pottery.
29
Raimbault and Sanogo 1991; Park 2011.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 532 Susan Keech McIntosh

Figure 14.5. White-on-red geometric pottery links the Lakes Region (left), Jenné-jeno
(centre), and Kumbi Saleh (right; in IFAN, Dakar collection) (photographs by S. McIntosh).

known.30 The chronologies are not firm enough to determine whether it

appeared earlier along the Niger Bend or in the Inland Niger Delta. Yet its
distribution along 500 km of the Middle Niger indicates a robust interac-
tion zone.

Absence of Monumental Architecture or Evidence of Elites

Of the 34 burials excavated, none has provided more than an iron finger ring
by way of grave goods, regardless of burial context (funerary urn or simple
inhumation). This is not to conclude that ranking or inequality were not
present and potentially materialised in non-durables such as salt or livestock,
textiles, or pirogues for fishing or river transport. Rather, it simply points out
that the frequently used archaeological scale of differential grave goods cannot
be applied. Nor has differential access to prestige goods been detected in
residential contexts. In addition, no possible public or religious buildings
have been identified. We have repeatedly emphasised that this may be
a sampling artifact, given the minute area of Jenné-jeno that we have exca-
vated, yet the more we dig and uncover the same pattern, the more deserving
of serious consideration it appears to be. Grave goods have been reported from
urn burials of Jenné-jeno type at other mound sites, including one recently
excavated by A.C. Stone.31 Is the absence of grave goods at Jenné-jeno
a sampling artifact or perhaps an indication that material distinctions in
death ritual were consciously minimised at Jenné-jeno in particular?

30
There is no obvious Garamantian connection. Contemporaneous painted pottery from Jarma
uses red paint over a white painted base or slip, see Mattingly 2013. Black geometric and cross-
hatched paint is documented on terminal Late Stone Age pottery of ‘wasa’ type from Niger,
dated c.1000 BC, Grébénart 1985, 88–98.
31
Stone 2015.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 533

Djenné

Jenne-jeno
0.5 1.0 km

Figure 14.6. Clustered mounds around Jenné-jeno and Djenné.

Unusual Settlement Pattern of Clustered Mounds

In the immediate vicinity (within 4 km) of Djenné, there are 65 occupation
mounds, creating a distinctive and predominantly anthropogenic land-
scape (Fig. 14.6). Surface survey at 32 of these mounds showed that three-
quarters had Phase IV (AD 900–1400) surface pottery, but nothing more
recent, suggesting abandonment during that period. Since many mounds
have more than 2 m of deposits, we proposed that their occupation
spanned several centuries.32 This was confirmed by the presence at many
sites, including the largest mounds near Jenné-jeno, of Phase III surface
pottery as well, from which we infer that population in the area reached
peak density during Phase III and IV, before a major settlement pattern
shift occurred, concentrating population in single, non-clustered sites,
such as the modern town of Djenné.33 Basal levels at Djenné have recently
been radiocarbon dated to calAD 1297–1409,34 indicating very little over-
lap between the two occupation sequences. The area sampled by excavation

32
McIntosh, S. 1995; McIntosh, S. and McIntosh, R. vol. 2, 1980.
33
However, over the past two decades, new housing has increasingly spread beyond Djenné onto
long-abandoned mounds.
34
Beta 416884, 600±30 BP.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 534 Susan Keech McIntosh

is extremely small, however. The presence of Phase III and IV deposits at

satellite mounds has been confirmed by 17 radiocarbon dates on surface
and sub-surface features at six of these mounds.35 The inhabitants of
Djenné consider that all these mounds were, in the past, an integral part
of the town. Given these various elements, we have proposed that the
contemporaneously occupied mounds in Phase III and IV functioned as
part of a single, urban complex.

An Urban Landscape Reconsidered

Both R. McIntosh and I have argued that Jenné-jeno grew as a centre for
river-based exchange in foodstuffs for iron, stone, and very plausibly salt,
with copper and gold as rare markers of expanding interaction networks.
R. McIntosh has proposed that individual sites in the mound clusters
around Jenné-jeno were occupied by a proliferating set of subsistence
and craft specialists, forming an urban network of specialised parts inte-
grated into a generalised economy.36 In its latest iteration, this process of
proliferation has become the basis of a hypothesised series of transforma-
tions that trace the emergence of a self-organising urban landscape, begin-
ning with expansion of subsistence specialists and continuing with craft
specialists.37 The clustered form of the town was instrumental in resisting
monopolisation of power by any one group. The total population settled on
the mounds within 1 km for Jenné-jeno is estimated at between 6,000 and
13,000.38 Evidence for population heterogeneity has recently emerged
from Stone’s isotopic studies on human teeth from burials on Jenné-
jeno, which indicate a variety of non-local origins, in contrast to indivi-
duals buried on small, rural settlements who show a local isotopic
signature.39
The original hypothesis of functional interdependence among mounds
relied heavily on surface features and artifacts that were argued to be
markers for particular specialisations. These included brick ring features
interpreted as granary bases (= farmers), furnace bases (= blacksmiths),
netweights (= fishers), loomweights (= weavers).40 Mary Clark’s outstand-
ing doctoral dissertation reported on her surface survey of all 67 mounds in
the Urban Cluster. It included a closer examination of intersite feature and
artifact variability attributed to differences in site function. She also

35
Clark 2003, 136–39. 36 McIntosh, R. 1991; 1993; 1998. 37 McIntosh, R. 2005.
38
McIntosh, S. 1999b, 73. 39 Stone 2015. 40 McIntosh, R. 1991; 1993; 1998; 2005; 2015.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 535

excavated 35 surface features in order to ascertain chronology and inves-

tigate function. Several of her findings are particularly relevant:
• Certain features and artifacts used in the model are ambiguous with
regard to function. The brick ring features identified as granaries may
instead be fish-drying platforms and netweights and loomweights
proved to be difficult to differentiate unambiguously.41
• Surface features produced a wide variety of dates between AD 400 and
1400, forcing reconsideration of any assumption that all surface ele-
ments are relatively contemporaneous. Differential erosion is likely
implicated.
• Some features originally thought to be forges turned out to have walls of
dried clay or clay fired at low temperatures, and are thus not
metallurgical.42 Others were confirmed to be hearths used for melting
copper or forging iron. At three sites, these forges occur in linear
alignments suggestive of smithing precincts.
Additionally, Clark noted that five mounds inventoried in 1977 had
disappeared, likely due to fluvial erosion, and five new mounds of <1 ha
had been formed by floodwaters carving up mounds into smaller pieces.43
Given the amount of erosion that has occurred in recent decades at the
41 ha site of Kaniana to the north of Jenné-jeno, she predicts that it will
soon be divided into a cluster of mounds. The highest mounds, including
Jenné-jeno and Hambarketolo, are being remodelled by erosion, but are
less threatened by dissection. This serves as an important reminder that the
surface and margins of sites are constantly being reshaped by dynamic
processes. Because we began our research at Jenné-jeno at the height of the
Sahelian drought, we did not witness the power of rainwater and floods on
the mounds until the 1990s, and it was a considerable shock to see how
much change could take place in a very short time.44
While the functional interdependence hypothesis for clustering
around Jenné-jeno remains open for investigation and ‘good to think
with’, Clark’s findings underscore the need for excavation at multiple
sites in the cluster in order to evaluate the model appropriately.
Regardless of potential post-depositional partitioning of lower eleva-
tion mounds, the existence of closely spaced settlement mounds in
greater or lesser numbers is a phenomenon that demands explanation.
Clark’s 14C dates confirm active deposition at the same time that
41
Clark 2003, 82, 108. 42 Killick 2003. 43 Clark 2003, 190–91.
44
We inaugurated an erosion monitoring programme on Jenné-jeno in 2008, and re-surveyed the
mound again in 2011, but have not been able to return to continue the project.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 536 Susan Keech McIntosh

Jenné-jeno was occupied. Why were people creating occupation

mounds in places that limited proximity to their rice fields? The
functional interdependence hypothesis is an excellent starting point
for investigating this question, but we have barely scratched the sur-
face in terms of the work required to evaluate it.
The organic nature of the urban cluster is thus less clear than
previously thought. In the absence of obvious power differentials,
centralisation, or specialisation by site, what bound the mounds of
the cluster together? Functional interdependence remains a strong
candidate for the principle integrating the mounds, but spatial separa-
tion of occupational specialists may not have been fundamental to its
operation. How did the inhabitants resolve disputes over access to
agricultural land and grazing rights or any of the myriad disagree-
ments that inevitably arise in large populations? Where did effective
power reside? The question has been raised as to the relevance of
historical Mande (Bamana) concepts of occult power and the land-
scape, as discussed by R. McIntosh, for our understanding of Jenné-
jeno’s urban landscape in the first millennium AD.45 Yet how can
their degree of relevance ever be determined? A more appropriate
question is whether they serve as a useful heuristic for transcending
entrenched Western assumptions about the nature of power.46 I think
they do. The concept of heterarchy gives us some insight,47 but
a deeper understanding of the details will require considerable addi-
tional research.

Gao/Gawgaw: Elite and Commercial Centre on the Niger River

Located at the intersection of the Niger River and Tilemsi valley, which
extends north to the Adrar des Ilforas, Gao commands a strategic point for
trade (Fig. 14.7). In contrast to Djenné/Jenné-jeno, Gawgaw appears early
in historical sources. In the ninth century, al-Yaqubi describes it as ‘the
greatest of the realms of the Sudan’. Al-Muhallabi a few decades later
provides details of dual towns: the market town, Sarnah, and the royal
town, residence of the ‘king’, whose treasure was salt.48 Archaeological
research at the badly looted, 32 ha mound of Gao Saney has established
a preliminary framework for the history of occupation, from AD

45
McIntosh, R. 2005; Smith 2006. 46 Fleisher and Wynne-Jones 2010.
47
McIntosh, S. 1999b. 48 Levtzion and Hopkins 1981.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 537

LEY
VAL
KOIMA

M SI
TILE
YY
WEST BANK NEY
Y
GAO SA
GAO ANCIEN GS 3 GS 1, ACGS

ER
A

W
DI 0 7 Km

AB
GAN G
GAO
Floodplain of ancient channels
Niger floodplain
Niger River
Y Y Y Saney cemetery

Figure 14.7. The location of Gao Ancien and Gao Saney.

700–1100,49 revealing craft activity on a scale that supports its identifica-

tion as Sarnah.50 Again in contrast to Jenné-jeno, imports of glass and
copper were reaching Gao-Saney in considerable quantities from the ear-
liest period of documented occupation. While the data in hand are abso-
lutely inadequate for the task of reconstructing the history of this massive
mound, they do suggest that the initial occupation was extensive. Hunwick
proposed that Sarnah originated as a settlement of North African Muslim
merchants,51 but the archaeological evidence suggests a considerable
Berber influence with perhaps a certain hybridisation of ceramic influences
with local Songhay groups. Hunwick’s argument, that the Songhay Za
dynasty (in the Ta’rikh al-Sudan) relocated from Kukiya to Gao in the
ninth century to provision (and control) North African merchants trading
through Tadmakka, aligns fairly closely with the evidence for North
African, Berber, and Songhay interaction at Gao-Saney.52
The initial settlement of Gao Saney coincides with a period of reduced
aridity in the Sahel. This may have produced increased rainfall in the Adrar
des Ilforas, raising the water table in the Tilemsi Valley and facilitating
long-distance movement along it. The Wadi Gangaber, on which Gao
Saney is located, may have flowed seasonally during this period, but we
have no evidence regarding this.

49
The surface is likely to be significantly deflated; Insoll 1996 found imported and local ceramics
on the surface that likely date somewhat later.
50
Cissé et al. 2013. 51 Hunwick 1999, xxxiv. 52 Hunwick 1994, note 21; 1999.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 538 Susan Keech McIntosh

The excavations at Gao-Saney have recently been published, so only

a brief summary is presented here.53 Three excavation units in stable,
unlooted areas of the mound revealed 7 m of deposit representing occupa-
tion between the eighth and twelfth centuries, with unfired mudbrick
architecture throughout the sequence. The lower deposits have large, loaf-
shaped bricks of a type known earlier in the Niger Bend, along with walls of
irregularly shaped banco lumps that recall construction technology else-
where on the Niger Bend and at Jarma.54 Architectural features in later
deposits have standardised mould-formed rectangular bricks (400 × 200
× 45 mm). The inhabitants were involved in the secondary processing of
both glass beads and copper. Excavations recovered hundreds of glass bead
wasters from the earliest deposits and over 400 copper crescent fragments
from the upper 3 m of deposits, along with 150 copper crucibles. Lead
isotope analysis of several of the copper-alloy artefacts indicates, tenta-
tively, a Tunisian source. Similar material also reached other Sub-Saharan
sites, including Kissi, Marandet, Essouk-Tadmakka, and Igbo Ukwu along
what increasingly appears to be an early Trans-Saharan route from Ibadite
centres, such as Tahert in North Africa.55 Other copper objects appear to
derive from different sources, possibly located in Morocco.56 The glass
beads at Gao Saney include both plant ash glass manufactured east of the
Euphrates and natron glass from Syro-Palestinian sources.57
The entire Saney sequence is dominated by distinctive, polychrome,
funnel-neck jars and a smaller number of red slipped, burnished and
comb-impressed bowls. A virtually identical assemblage appears in the
Tombouze sequence near Timbuktu in the seventh/eighth century AD.58
The origin of this distinctive assemblage is of considerable interest, as it
appears to combine vessel forms and channelled decoration documented at
earlier sites in north-eastern Burkina Faso with polychrome decoration
that appeared in the third century in other excavated sequences along the
Middle Niger.59
The same funnel-neck jar assemblage was excavated at Gao Ancien
in association with a pillared structure that was constructed in dressed

53
Cissé 2017; Cissé et al. 2013.
54
At several sites in the Méma and Lakes Region, discussed by MacDonald, Chapter 13, this
volume. For Jarma, see Mattingly 2013, 141, though note that the mud lump construction is
a later Medieval development there, postdating mudbrick.
55
Fenn 2011; Fenn et al. 2009; Nixon 2009. 56 Fenn 2011. 57 Dussubieux 2011.
58
Park 2011.
59
Czerniewicz 2004; Magnavita, personal communication. See also, Jenné-jeno: McIntosh,
S. 1995; occupation mounds in the Lakes Region: Raimbault and Sanogo 1991; and Tombouze:
Park 2011.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 539

stone in the tenth century.60 From inside the structure, fragments of

gold and over 6,000 glass beads were recovered. The nature of this
building is a puzzle, but the pillars are a distinctly North African
architectural feature and are evocative of religious structures, such as
mosques and tombs.61 More recent excavations in adjacent areas have
revealed other stone walls in the immediate vicinity, one with a pillar
visible in the section, plus a basin lined with fired brick, and a toilet
and washing area. Radiocarbon dates indicate that they were all built
between calAD 880 and 1020.62 Because of the extent of wall robbing
and later construction, it will take considerable ingenuity to recon-
struct how all these stone walls related to one another. Were they part
of a complex of structures, or part of a single structure? At the
moment, we do not know. But it is clear that a singularly important
building or building complex was constructed in a North African style
in the tenth century. These excavations add considerable detail on this
period, which was initially documented by Insoll’s work in the same
sector of Gao Ancien.63 Several aspects of the Gao Saney evidence
merit particular emphasis here.

Rapid Settlement Growth

The earliest occupation at Gao Saney appears to have been extensive, with
functional differentiation between residential areas (unit GS3) and work-
shops (units GS1, ACGS).

Trade and Interaction

The data from Gao-Saney powerfully situate the site within much more
extensive networks of trade and interaction. Funnel-necked, polychrome
pots accompanied by red burnished comb-impressed bowls are wide-
spread along the Niger during the period AD 700–1100; they are present
near Timbuktu, Gourma Rharous, and Bentia in addition to Gao, clearly
indicating river-based cultural interaction.64 However, the Saney pottery
is very unlike Late Garamantian pottery or the contemporaneous, pre-
sumably Berber pottery in the early deposits at Essouk-
Tadmakka (Essuk on Fig. 14.1).65 Overland trade networks extending
60
Loaf-shaped banco bricks underlie this structure but earlier levels were not excavated.
61
Cissé 2017; Cissé et al. 2013, 14. 62 M. Cissé, personal communication.
63
Insoll 1996; 2000b. 64 Arazi 1999; McIntosh, S. and McIntosh, R. 1986; Park 2011.
65
Mattingly 2013; Nixon 2008; 2017.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 540 Susan Keech McIntosh

north-south and east-west are certain and historically documented by the

ninth century. The introduction of mould-formed bricks in the ninth–
tenth century in unit GS3 may indicate the presence of North African
merchants or styles, a hypothesis supported by the presence of oil lamps
(albeit of unusual form), which have almost never been found in Middle
Niger archaeological sites. Their domed form is unusual and matched by
an example recovered at Sijilmasa.66 A number of sherds of glazed, wheel-
made pottery were also recovered.67
For earlier periods, the evidence is more tenuous. Polychrome wares
earlier than the seventh century have not yet been documented at Gao; the
levels under the massive stone architecture complex at Gao Ancien remain
unexcavated, so we may find some relevant information there. However,
across the river at Koima, an assemblage of short everted rims and ogival
rims, often embellished with paint (predominantly black) resembles the
Middle Assemblage (AD 100–650) at Tombouze.68 Neither glass nor
copper has been documented in excavations at Niger Bend sites prior to
the seventh century, unlike at Kissi (Burkina Faso) where both were
present between the second and sixth centuries AD.69 In the western
Niger Bend, at Tongo Maaré Diabel, copper and glass appear only in the
late first millennium AD, but the presence of chicken from the fifth
century AD and mudbrick building technology provide evidence of
wider interaction earlier in time.70
Gao-Saney joins the growing ranks of archaeological sites that
document an expansive increase during the first millennium AD in
the movement of exotics, both interregionally, between the Sahara
and the Sahel, and across the Sahara. Locally, the relationship
between Gao Saney and Gao Ancien corresponds closely to the
Arab accounts of the dual towns of Gawgaw, despite conflicting
details on locations with regard to the river.71 The construction of

66
Fauvelle-Aymar 2016. 67 Cissé 2011.
68
Based on Koima pottery collected by S. McIntosh, see Park 2011 for the assemblage from
Tombouze.
69
The copper and beads from burials at Kissi have been abundantly published: Magnavita 2003;
2013; 2015; Robertshaw et al. 2009. For the Niger Bend, Park 2010 reported large amounts of
carnelian and glass beads in an early first millennium AD context from Tombouze 1 Unit 6.
However, the full excavation report indicates that these materials came from a different
excavation unit (Unit 5) in which the sequence begins no earlier than the seventh century –
Park 2011. In that unit, ten carnelian beads, eight glass beads and 103 bone beads were
recovered near a burial with two copper bracelets and a copper ring – Park 2011, 397–98. No
glass beads or carnelian were recovered from any of the levels in Unit 6 – Park 2011, 403–15.
70
Gestrich 2013; MacDonald, Chapter 13, this volume. 71 Discussed in Hunwick 1994.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 541

the dressed stone pillar structure complex falls strikingly close in

time to the conversion of the ruler of Gawgaw to Islam by
the second half of the tenth century.72

Gao: Summary

The archaeological evidence thus far excavated from Gao Saney and
Gao Ancien corresponds remarkably well to historical descriptions
and reconstructions based on the Ta’rikh al-Sudan and Songhay oral
traditions. These two sites have identical ceramic assemblages in the
late first millennium AD, but evidence of very different functions, as
Insoll first proposed: elite consumption and architecture occurs at Gao
Ancien and craft and manufacture at Gao Saney.73 Although both sites
might appear superficially to conform to an Arab-led-trade model
stimulating town growth, the earliest material culture suggests
a Songhay-Berber, rather than an Arab presence, reflecting interaction
along the Niger Bend from Bentia to Tombouze. The well-developed
glass processing industry in the earliest levels raises questions about
possible antecedents to the glass and copper trade at Gao Saney,
perhaps extending back two or more centuries and linked to the
transit of the copper and glass in dated mid-first millennium burials
at Kissi, as well as Sirba gold.74 Luckily, the levels under the massive
stone architecture complex at Gao Ancien remain unexcavated, so we
may find some relevant information there.

Takrur: A Riverside Trade Town/Polity in the Middle Senegal

River Valley

The Senegal River has for centuries been a vital artery for the transport of
gold from the Bambuk source area along the Faleme River, which flows
into the Senegal south of Bakel. The French established forts at the Faleme
confluence in the late seventeenth century to control the gold trade and
transport downriver to their fort at St. Louis on the Atlantic Ocean. But
prior to the establishment of the Atlantic trade, gold was traded to towns in
the Middle Senegal Valley for transport north by caravan across the desert.

72
Hunwick 1980. 73 Insoll 1996.
74
This eastern Niger route has been proposed by Insoll and Shaw 1997 and Nixon 2017, 268–69.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 542 Susan Keech McIntosh

In the twelfth century, Takrur was the preeminent trade town and polity on
the Senegal River for the export of gold and slaves.75 A century earlier, Silla
held that distinction. Despite al-Bakri’s description of Silla as the capital of
a ‘vast kingdom’ led by a ruler almost as grand as Ghana’s,76 archaeological
interest in identifying and illuminating the history of either Silla or Takrur
has been dwarfed by the attention focused on Ghana. This disparity may be
attributed to several factors: the earlier historical prominence of Ghana as
a source of gold, the detail provided by al-Bakri on Ghana and its trade
networks, and the visible and monumental nature of archaeological sites
such as Tagdaoust (Tegdaoust) and Kumbi Saleh that are identified with
these trade networks.77 Whatever the reasons, little systematic research had
been conducted to locate Takrur and Silla and to document their develop-
ment and interaction networks over time.
Throughout the 1990s, the Middle Senegal Valley Archaeological
Project undertook a series of investigations in the region thought to be
the traditional heartland of Takrur. I will review the evidence these
projects provide on exchange, trade, and town growth. Most of the
relevant data comes from excavation and survey in a 460 km2 study
area between the towns of Walaldé and Saldé and upriver at Sincu
Bara.78 The occupation sequences documented by the project at more
than a dozen different sites date to the period 800 BC–AD 1200. To this
database, we can add other excavated and dated occupation first millen-
nium sequences at Tulel Fobo (AD 700–1000),79 Ogo (AD 900–1200)80
and, further upriver near the confluence of the Senegal and Faleme rivers,
Arondo (AD 500–1000)81 (Fig. 14.8). These sequences document two
periods of interaction with regions to the north of the Middle Senegal
Valley: 600–300 BC, and post-AD 800. In marked contrast to Middle
Niger sites, there is no evidence of first millennium AD exchange and
interaction with regions to the north or west earlier than AD 800, and the
quantity of imports prior to AD 1000 is extremely limited. This is
unexpected in an area of the river that has historically been important
to Saharan trade, not only as a conduit for the gold trade, but also as
a grain-producing area that provisioned Saharan trade towns. This sug-
gests a different trajectory than either of our earlier town case studies. In

75
Al-Idrisi, in Levtzion and Hopkins 1981, 107.
76
Al-Bakri in Levtzion and Hopkins 1981, 77–78.
77
Nixon, Chapter 17, this volume, discusses the archaeology of Tagdaoust, presumed site of
historical Awdaghust, and Kumbi Saleh.
78
Deme 2003; Deme and McIntosh 2006; McIntosh, R. et al. 2016; McIntosh and Bocoum 2000.
79
Bocoum 2000. 80 Chavane 1985. 81 Thiaw 1999.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 16° 14° 12°

ITS
Lac Rkiz

lil

OS
Walaldé

P
Aw
Podor
Senegal River

DE
Walaldé l

LT
Dagana
ILE A MO
u go
or
G

SA
(TAK RPH Cubalel
RUR? IL Kaedi
Lac de
16° St. Louis
) SSaldé
ldé
Siwré Guiers Tulel Fobo
SE Fe SENEGAL Sincu Bara
NE rlo Matam
G
AL Ogo
RI
VE
R
Bakel
Dakar Arondo

0 100km
Juudé Jaabé
D 14°
5x
Falémé River

RI OU
VE É
R

Madina Ndiatebé
7x MAURITANIA

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

7x Cubalel
6x 3x

SENEGAL

Saldé
Plages
N Mounds 5x
Towns
Ngoui
Meander Trains
9-10 m levee
0 2 5KM Floodplain Margin
Boundary of Study Area

Figure 14.8. Middle Senegal study area with inventoried sites and geomorphological features. Inset shows other sites mentioned in the text.
 544 Susan Keech McIntosh

addition, evidence is sparse for town growth in the first millennium AD

until copper and brass, spinning and weaving, glass, and other exotics
appeared in the region, along with sorghum.

Overview of the Middle Senegal Valley Archaeological Project

and Results
The middle sector of the Senegal Valley is 20–30 km wide in the area called
the Ile à Morphil, where the river bifurcates into a southern branch, the
Doué. Historically, this sector has been the most populous because of the
concentration of clay basins suitable for recession cultivation of
sorghum.82 This region today is dominated by Hal-pulaar’en (Pulaar
speakers), a broad grouping that includes fishers (subalbe), sedentary
Tukolor,83 and pastoral Fulbe. All three groups cultivate rain-fed millet,
but the Tukolor also practice recession cultivation of sorghum. Due to high
interannual variability in flood and rainfall, the extent of the area flooded
may vary from year to year by a factor of five or more.84
In our study area within the Ile à Morphil, there are two broad categories
of archaeological site: mound sites 2–6 m high located on the high levees
(9–10+ m above mean sea level), and surface scatters known as ‘plages’
with negligible (<1 m) above-ground accumulation (Fig. 14.8). Most of
these plages have scatters of sherds and fired earth or ash. Our excavations
focused on 11 mound sites in the vicinity of the villages of Cubalel, Siwré
and Walaldé. The bulk of the excavated deposits date to the period 500 BC–
AD 1000. All of the excavated mounds revealed a quite consistent picture
of episodic agropastoralist occupation of variable but generally low inten-
sity, punctuated by periods of abandonment lengthy enough for the devel-
opment of recognisable erosion surfaces in a number of cases. The most
common features encountered in all units were pits and fired earth of
various kinds associated with domestic structures or wall melt from such
structures, and domestic debris. Most of the structures appeared to be
circular pole-and-daub houses 3–4 m in diameter. Mound area is generally
less than 3 ha, but some sites exceed that, reaching up to 12 ha. Subsistence
economy was relatively generalised, with a primary emphasis on ovica-
prids, cattle, and dog (apparently a regular food item), pearl millet and
fonio, plus small amounts of fish and wild bovids. Sorghum, so important

82
Boutillier 1962.
83
Tukolor, a French transcription of the Arabic Takrur or Takruri (Bâ 2002, 62) originally
referred to ‘people of Takrur’.
84
Michel 1973, 135.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 545

Figure 14.9. Distribution of first-millennium BC copper in Mauritania extending to

Walaldé (after Vernet 2012, 112).

to the Middle Senegal Valley economy historically, is absent for most of the
first millennium AD in all units within the study area, at Sincu Bara and at
Arondo. This absence is significant, because a systematic flotation protocol
that sampled every context was followed at all these sites.
The earliest evidence of floodplain settlement dates to 800–500 calBC at
the 5 ha mound of Walaldé, which was occupied by agropastoralists with
herds dominated by cattle.85 This coincides with a very dry episode sig-
nalled by a sharp spike in dust accumulation off the Mauritanian coast and
an inferred southward retraction of the cattle zone.86 During the wetter

85 86
Deme and McIntosh 2006. Hanebuth and Henrich 2009.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 546 Susan Keech McIntosh

period c.500–300 BC, copper from the ore source at Akjoujt was present at
Walaldé and throughout much of central and western Mauritania (Fig.
14.9).87 We have suggested that copper moved with and among pastoral-
ists/agropastoralists in their seasonal migrations.88
Interactions, likely indirect, between Mauritanian sites, Walaldé, and
North Africa are indicated by Phoenician-style copper earrings at Walaldé
and around Akjoujt similar to those found in Moroccan graves of the
eighth to sixth centuries BC.89 In addition, perforated A. senilis shells at
Walaldé indicate possible interaction with coastal areas.
We thus have concrete evidence of linkages between the Mediterranean
and the Senegal River in the first millennium BC, but little compelling
reason to think in terms of regular Trans-Saharan exchange or trade
relations. The only identifiable northern import reported from
Mauritania is a Punic-style bronze fibula with an iron pin discovered
near Akjoujt.90 Other finds of bronze are extremely rare, and brass arti-
facts, also rare, are of more recent date.91
The end of occupation at Walaldé c.200 BC appears to coincide with the
onset of an unstable arid episode with sporadic flash flooding.92 Settlement
at Cubalel at the turn of the first millennium AD marks the next docu-
mented occupation. The settlement pattern during Cubalel Phase I (AD
1–200) of small (<3 ha), regularly spaced settlements on the high (10+ m)
levees indicates that population density was low. However, an unknown
number of unexcavated mounds may have Phase I deposits at their base,
and we have no data regarding seasonal influxes of pastoralists and their
herds.
At some point during this arid period, an exceptionally arid episode of
unknown duration occurred, and seawater flowed into the Ferlo valley
(normally a tributary of the Senegal River) for 350 km.93 Its duration was
sufficiently prolonged (from decades to a century or more) to support
saltwater shellfish all along the valley. A 14C date on the shell deposits
indicates a date between 100 calBC and calAD 300. Possible contributing
factors include prolonged reduction in rainfall, increased evaporation,

87
Vernet 2012.
88
Deme and McIntosh 2006; Vernet 1993. Horses may have been among the livestock herded in
these pasturelands; rock art depicting horses occurs in the Adrar, conventionally attributed to
invading ‘Libyco-Berbers’. The high number of copper spears and arrows is certainly suggestive
of violent confrontations; alternatively, these may have been used in hunting.
89
Compare Deme and McIntosh 2006, 335; Jodin 1966, 71; Ponsich 1967, 34, 91, 196; Vernet
1993, 347.
90
Lambert 1972. 91 Lambert 1983. 92 Bouimetarhan et al. 2009; Nizou et al. 2010.
93
Monteillet et al. 1981.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 547

and/or a drop in the water table. This raises the question of a possible
connection between the apparent cessation of Akjoujt copper exploitation
by the end of the first millennium and this exceptional arid episode.94 If
this episode produced an emphatic southward movement of the cattle
pastoralism zone, the Akjoujt copper mines could have become remote
and inaccessible. A climate episode of such severity could have been
devastating for cattle herds in the Sahel, resulting in huge losses and
wholesale abandonment of formerly occupied areas, entirely disrupting
prior interaction networks.
It may be at just this time that Berber groups began developing and/or
expanding camel herds into the southern desert, ultimately reaching the
Senegal River at Siwré. A single camel bone was recovered there from
deposits dating to AD 200–400.95 Without speculating on the linguistic
affinity of the Siwré occupants, it may be noted that their material culture
was quite different from that of Phase I at Cubalel. They had highly
decorated pottery that shares certain distinctive features (pastilles, linear
appliqués, high frequency of slip) with the pottery several centuries later at
Tagdaoust.96
Occupation throughout the first millennium is well represented in our
excavations. Survey shows that occupation on small mounds (0.5–3 ha in
area) located on the highest levees (10+ m) was common. If goods were
moving between the river and the desert in the early first millennium AD,
no evidence for it has yet turned up. Indeed, it is noteworthy that no
evidence of copper, glass or any other northern import has been found in
first millennium deposits prior to c.AD 800 at any excavated site in the
Middle or Upper Senegal Valley.97 Also absent are any signs of contact with
the coast, such as A. senilis beads. In contrast with the Middle Niger, where
domestic chicken appears between the fifth and seventh centuries AD,

94
Dating is regrettably imprecise; Lambert 1983 records a single date of calAD 28–537 (1760 ±
110 BP) on a smelting furnace at Lemdena that may indicate exploitation in the first
millennium AD. The argument that the surface finds of copper in Mauritania all date to the first
millennium BC is based on typological criteria linked to the date range for the operation of the
Grotte aux Chauves Souris mine at Akjoujt and several copper smelting furnaces – Lambert
1971; 1975; 1983; Vernet 2012. Recovery of copper artifacts from closed and dated contexts is
needed to confirm this chronology.
95
MacDonald and MacDonald 2016. 96 For example, Robert-Chaleix 1989, 227–32.
97
The claim by Hatté et al. 2010 that Sincu Bara pottery was slipped or otherwise surface-treated
with bitumen imported from the Dead Sea, Egypt, or Nigeria from AD 400 onward cannot be
evaluated until the surface slip is analysed by gas chromatograph–mass spectroscopy to confirm
the presence of bitumen. 14C dates of over 10,000 BP for the slip on two potsherds prompted the
search for possible sources of old carbon. These results certainly demand deeper investigation;
the case for bitumen is not yet established.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 548 Susan Keech McIntosh

there is no chicken in the first millennium AD assemblages.98 Because so

little is known of first millennium AD material culture north of the Senegal
River and downstream to the coast, it is unfortunately impossible to assess
any other potential evidence for interaction during this period.99
Interaction along the river expanded by AD 600 to establish a common
material culture with distinctive collared pottery over much of the middle
valley, from Cubalel to Matam, with variants found in the Falémé at
Arondo.100 In the Cubalel study area, copper appears after AD 800 in
association with this collared ware assemblage. Only three pieces of copper
were recovered from excavations at Cubalel sites; all were pure copper, in
levels dating between AD 800 and 950.101 Upriver, at Sincu Bara and
Arondo, copper and brass appear at approximately the same date.102
Also at this time, there is the first evidence of new architectural technol-
ogies, including banco and unfired mudbrick, supplementing the earlier
wattle-and-daub technology. Shortly thereafter, cordoned pottery in
a distinctive style documented at Tagdaoust and Kumbi Saleh appears at
sites in the Cubalel study region, at Sincu Bara, and at Arondo (Fig. 14.10).
These developments inaugurated a period of intense interaction with
Saharan trade towns and North Africa. Cordoned pottery at Sincu Bara
and Ogo developed a uniquely exuberant style of raised geometric motifs
on the body, including oculi, ovals, lozenges, chevrons, and hearts that may
mimic patterns on ninth- and tenth-century Fatimid glassware, several
examples of which were recovered at Tagdaoust.103 The forms include new
vessel types, such as footed goblets.104 This very distinctive pottery dates to
between AD 1000–1200 at Ogo and is associated with disturbed deposits at
Sincu Bara that appear to date to this same time period.105 Also recovered
from the disturbed deposits was a fragment of green glazed ware from
North Africa, 150 pendants and beads made on marine shell (A. senilis,
Conus), plus eleven cowries (Cypraea moneta).106 No marine shell was
found in the earlier, undisturbed deposits. Glass beads or vessel glass have
not yet been recovered in any significant numbers from excavations at any
site of any date on the Senegal River.
After analysing copper-based metals from a variety of Senegal Valley sites,
Garenne-Marot concluded that there was a trade circuit involving Tagdaoust
and the Middle Senegal sites of Sincu Bara and Saré-Coffi (Podor) beginning in

98 99
MacDonald and MacDonald 2016. Vernet 2004. 100 Thiaw 1999.
101
Fenn and Killick 2016. 102 Bocoum and McIntosh 2002, 82; Thiaw 1999.
103
Chavane 1985, 127; Vanacker 1979, 163–66. 104 Thilmans and Ravisé 1980, 49–51.
105
Chavane 1985; Bocoum and McIntosh 2002 refer to this pottery as assemblage IVb.
106
Thilmans and Ravisé 1980, 66, 70–75.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 549

Tegdaoust-style cordoned pottery AD 800?-1100?

Cubalel Sincu Bara Arondo Tegdaoust

Cubalel-style collared pottery AD 600-900 MAURITANIA Sincu Bara cordoned

Tichitt
ware AD 750-1100
Middle
Senegal
Valley Tegdaoust
Podor
Kaédi
Cubalel

Sincu
Bara

SENEGAL Arondo MALI

Ba
Cubalel

m
buk
GUINEA Sincu Bara
BISSAU
Fuuta
Jalon
GUINEA

Sincu Bara Arondo

Figure 14.10. Ceramic styles on the Middle Senegal from AD 600 to 1100. The known
distribution areas are: Cubalel collared ware – Cubalel to Sincu Bara; Sincu Bara
cordoned ware – Kaédi to Arondo; Tagdaoust-style cordoned ware – Podor to Arondo.

the ninth century and gaining in intensity towards the eleventh and twelfth
centuries.107 Similarities are particularly marked in the brass clochettes (small
bells), trefoils, and bar links that may have been decorative and harness
elements for horses.108 These suggest the presence of an elite culture associated
with horses and horse riding. Copper and brass bracelets found at Saré-Coffi
and Sincu Bara have engraved geometric decoration that resembles Fulbe
motifs historically.109 A number of these were created using lost-wax casting.
Cotton, absent from the hundreds of first millennium AD flotation
samples from Cubalel sites, has been identified in the early second millen-
nium deposits at Ogo; spindle whorls co-occur.110 In the Middle Senegal
Valley study zone, 42 spindle whorls were recovered from survey, primarily
at second-millennium sites. The three spindle whorls from excavated
contexts all post-date the ninth century.

107
Garenne-Marot 1993; Garenne-Marot et al. 1994
108
Chavane 1976, 63, 67; Thilmans and Ravisé 1980, 30–38.
109
Chavane 1976; Thilmans 1977; Thilmans and Ravisé 1980, 22.
110
Chavane 1985, 111–12, 150; Murray n.d..

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 550 Susan Keech McIntosh

The available evidence for exotics, spinning and weaving technology,

lost-wax casting, and interaction with Saharan trade towns indicates that
they first appeared all along the river around or shortly after AD 800. This
correlates with climate amelioration between AD 700/800 and 1100/1200
due to a ‘longer rainy and cloudy period during the year and decreased
evaporation’.111 The result would have been northward displacement of
ecological zones, consistent with the report by al-Bakri that Awdaghust
(arguably the site of Tagdaoust) in the eleventh century abounded with
date palms, henna trees, many cattle and sheep, and good well-water.112
It goes without saying that our limited excavation sample from the
Senegal Valley does not permit any firm conclusions regarding the scale
and distribution of imports over time. The patterning of durable goods
recovered thus far is suggestive, but hardly definitive.

Towns and Polities: Where and When Did Takrur

and Silla Emerge?
The absence of evidence for trade or exchange with the north along the
Middle Senegal for most of the first millennium is broadly consistent with
historical accounts. Ibn Hawqal in the eleventh century mentions the
coastal salt source, Awlil,113 and links it directly with Awdaghust without
any intermediary stops along the Senegal River.114 He does not mention
either Takrur or Silla. Both first appear historically in the eleventh century.
Al-Bakri discusses Takrur mainly in reference to its leader, Warjabi
b. Rabis, who converted to Islam in the early eleventh century, then
effected the conversion of the inhabitants of both Takrur and Silla, and
allied Takrur with the Almoravids in a battle against the Gudala Berbers in
the Mauritanian Adrar.115 No specific role in trade is mentioned, nor is
Takrur linked to any trade network or itinerary. The main trade town in al-
Bakri’s time appears to have been Silla, which he links to Ghana. He notes
that its inhabitants use sorghum, salt, copper rings, and lengths of cotton
manufactured in nearby TRNQH for money.116 The viroles recovered in
vast quantities from Sincu Bara may correspond to the copper rings he
mentions.

111
Maley 2010, 250; Maley and Vernet 2015. 112 Levtzion and Hopkins 1981, 68.
113
Awlil is thought to be located in the sebkhas of the Aftout es Saheli in southern Mauritania –
MacDougall 1990; Robert-Chaleix 1991.
114
Levtzion and Hopkins 1981, 46. 115 Levtzion and Hopkins 1981, 73.
116
Levtzion and Hopkins 1981, 77–78.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 551

A century later, al-Idrisi provided significant new information.117 He

places the town of Takrur upriver from Silla on the main river route for the
salt trade from Awlil, and describes it as the capital of a powerful trading
polity to which Silla was now subject. Its rise was likely related to its alliance
with the Almoravids and control of the coastal salt deposit at Awlil. The salt
was transported down the coast to the Senegal River and then by boat to
Silla, Takrur, Barisa and other towns. While Silla was still ‘a meeting place
for the Sudan and a good market’, Takrur was the larger town and had
more trade. ‘The people of the Maghreb al-Aqsa go there with wool,
copper, and beads, and they export from there gold and slaves.’118 This is
the first mention of gold trade in connection with the Middle Senegal. The
itineraries link Takrur, Silla, and Awlil with Sijilmasa via the Almoravid
stronghold in the Adrar.119
Since our undisturbed excavation sequences largely end at the beginning
of the second millennium AD, archaeology provides only limited insights
into the processes of town growth and polity formation referenced by the
historical sources. The largest site excavated, Sincu Bara, has surface
material covering an area of 67 ha, most of it dating to AD 900–1200,
according to the pottery sequence developed for the site.120 From the
quantity and quality of copper-alloy artifacts recovered by Thilmans and
Ravisé it is clear that Sincu Bara was an important regional centre with
extensive trade connections.121 Our excavations there encountered mainly
first millennium deposits that provided no evidence for earlier exchange,
or for agricultural intensification,122 but the sample is small. An elephant
tusk was recovered from the excavated deposits, suggesting at least one
possible item traded north. How Sincu Bara was positioned strategically for
trade is unclear. A great deal more research at the site and its region is
required in order to understand Sincu Bara’s development and its involve-
ment in trade.
Within the Cubalel study area, population aggregation appears to have
increased in the second millennium, based on surface surveys. Several sites
exceed 10 ha in area, in contrast to the period between AD 1 and 900,123 in
which occupation mounds ranged in size from 0.5 to 3 ha.124 There is at
present no evidence for an urban network extending back into the past as
far as the fourth century, specifically in regard to Takrur, as proposed by

117
Levtzion and Hopkins 1981, 106–7. 118 Levtzion and Hopkins 1981, sv al-Idrisi, 107.
119
Devisse 1988. 120 Bocoum and McIntosh 2002. 121 Thilmans and Ravisé 1980.
122
Murray 2008. 123 Ceramic phases I, II, and III – McIntosh, S. 2016.
124
The 12 ha mound at Siwré is a possible exception, but the upper deposits are second
millennium AD in date and we do not know the extent of the early first-millennium deposits.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 552 Susan Keech McIntosh

Becker and Bocoum.125 It must be recalled, however, that the largest sites in
the Middle Senegal Valley have historically been on the jeejegol, the lands
bordering the floodplain that are transitional to the dryland jeeri. While we
surveyed the jeejegol in the study region, the Mauritanian side of the river
was not open to us for survey. We are undoubtedly missing a significant
component of the settlement pattern of the populations that lived and
interacted along this stretch of the river. Understanding the full dimen-
sions of human settlement in the Middle Senegal Valley at any period
requires additional in-depth investigation of the jeeri and jeejegol on both
sides of the Ile à Morphil, as well as the floodplain sites studied intensively
during the Middle Senegal Valley Archaeological Project. Subsurface
geoarchaeology could well reveal settlements under the sands or alluvium
that are unsuspected from surface traces. Our current conclusions about
trade are based on the expectation that floodplain sites should reflect larger
regional integration in the exchange networks that sustained the towns on
either side of the river, even if we have not yet located those towns.
A final piece of evidence for the emergence of new social, economic, and
political configurations in the early second millennium is the production of
iron on an industrial scale within the study region and immediately across
the river. At the site of Juude Jaabe just north-west of Cubalel, we inven-
toried 150 melting furnaces arranged in sinusoidal alignments. Dates on
excavated furnaces suggest two periods of use: AD 1000–1200 and
1200–1400.126 Immediately across the river, Robert-Chaleix and Sognane
inventoried more than 40,000 furnaces in alignments that occur close to
laterite ore outcroppings. Dates on two furnaces also fall in the early second
millennium.127 The emergence of specialised smelting precincts at this
time appears to be a new pattern that contrasts with the preceding two
millennia, in which iron smelting furnaces occur on habitation sites and
iron slag is ubiquitous in the occupation deposits at all sites except Siwré.
Without comparable second millennium occupation sequences, we cannot
say whether domestic production of iron ceased when smelting precincts
appeared. As polities such as Takrur emerged, it would have been advanta-
geous to control the production of iron for weapons. However, outcrops of
laterite occur frequently enough along the Mauritanian side of the river
that it is unlikely that a monopoly on production could be maintained.
Oral histories link the origins of political power in the Senegal Valley
with the Jaa-Ogo, who legendarily introduced metallurgy to the Senegal

125
Becker and Bocoum 1998. 126 Killick 2016.
127
Robert-Chaleix 1994; Robert-Chaleix and Sognane 1983.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 553

Valley. According to one account of the Jaa-Ogo (and there are many),128
the ability of the Jaa-Ogo to smelt iron established their spiritual authority,
which permitted them ultimately to reserve the right to mine and smelt ore.
Bâ links Takrur and the Jaa-Ogo in an ‘Etat-forgéron’ in which the emer-
gence of craft specialisation provided the functional interdependence
necessary for the formation of the state.129
In this unfolding scenario of towns, trade, and craft specialisation, the
appearance of sorghum in the late first millennium AD plays a critical role.
Without sorghum as a recession crop, agricultural production was depen-
dent on rain-fed cultivation of millet and fonio. The potential for produ-
cing surpluses was greatly increased when a second annual cultivation
cycle was adopted that exploited the clay-rich depressions in the floodplain
where millet cannot be grown but sorghum thrives. Agricultural schedul-
ing then occupied the entire year, resulting in an increase in sedentarism by
a previously agropastoral sector. This group took on the historical identity
of Tukolor. Tochoror and Felle both appear on Angelino Dulcert’s 1339
map of western North Africa. I suggest that the adoption of sorghum
agriculture underwrote the expansion of specialisations of all types – smi-
thing, weaving, trading – that were in existence by the time of al-Bakri’s
account. If the evidence we have for the Ile à Morphil in the first
millennium AD is correct, then an agropastoral system with low agricul-
tural productivity was substantially transformed between the ninth and
eleventh centuries, concomitant with the historical emergence of Takrur
and Silla. This involved incorporation into the Trans-Saharan trading
sphere after a period of almost a millennium with negligible evidence for
desert-side exchange or interaction.

Conclusions

The three archaeological case studies outlined here focus on areas of the
Senegal and Niger rivers that have historically been nodes for trade and
interaction networks linking regions along the rivers with the desert, the
savanna, and beyond. Attempts to establish the time depth and shifting
configurations of these networks are hampered by the patchiness of our
current database. Where exotics are present, we can often identify source
areas, but can only speculate on the routes or modes of their transport.
Where exotics are absent, changes in style, technology, or spatial and social

128 129
Summarised in Thilmans and Ravisé 1980, 176–78. Bâ 2002, 91–93.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 554 Susan Keech McIntosh

organisation might be the material manifestations of interaction, exchange,

and trade that are otherwise invisible. The difficulty is that we often lack
archaeological data from neighboring regions in the desert and savanna
with which to compare our river valley sequences. In that case, the infer-
ences we can draw about networks that predate historical documentation
of trade beginning in the ninth century AD are quite tentative.
For the first millennium BC, the best-documented interaction zone extends
from the Mauritanian Adrar to the Middle Senegal Valley at Walaldé. The
distribution of copper artifacts with the distinctive chemical signature of the
Akjoujt ores as well as related pottery styles throughout central and southern
Mauritania describes an area within which ideas and goods circulated between
c.500–200 BC. Evidence that networks extended to Punic settlements in North
Africa is slim, however, and suggests mainly stylistic borrowings of jewellery
fashions by mobile Berber groups. Evidence for movement of copper from
Akjoujt or shells from the coast ends by c.100 BC, and, despite the presence of
camels by AD 400, no further exotics are encountered in Middle Senegal
Valley excavated sites before c.AD 800–900. At that time, new building
technologies appear (mudbrick and banco). River-based interaction between
Sincu Bara and Arondo is attested by the presence of a distinctive style of
cordoned pottery; subsequently, another style of cordoned ware associated
with Tagdaoust becomes widespread along the river from Cubalel to Arondo.
This coincides with the appearance of brass, glass, cotton, and spindle whorls,
at approximately the same time that Arab sources first mention Takrur in
relation to trade. Reliable evidence for village or town settlements exceeding
10 ha also dates to this period.
Along the Middle Niger valley, regional and interregional exchange in
stone and iron were essential to settlements in the Inland Niger Delta, such
as Jenné-jeno. A trickle of exotics from much greater distances included
several glass beads, one in Hellenistic style and one with a chemical signature
best known from south-east Asia.130 Both came from deposits dated between
100 BC and AD 200. Taken in conjunction with the appearance after AD 100
of polychrome pottery at Jenné-jeno and the Lakes Region, the glass beads
seem to be associated with increased interaction along the river, but their
mode of transport and exchange cannot be determined. For this reason,
I have not identified the glass beads as a sign of Trans-Saharan trade. There
are few, if any, signs of interaction with desert regions at this time, and none
suggesting Fazzan–Jenné-jeno contacts.

130
Brill 1995; a Phoenician/Punic glass eye bead has recently been identified in a first-millennium
BC context east of the Bandiagara plateau, Giachet et al. 2019.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 555

More extensive Niger River networks from the fourth to seventh centuries
are indicated by the presence of copper at Jenné-jeno in addition to the
continuing presence of glass beads, matched by evidence for copper, brass,
glass beads and wool textiles at Kissi in Burkina Faso. These elements are
absent in deposits of this period at Tombouze, near Timbuktu, and there is
likewise no evidence for exotics in the pre-AD 750 deposits at Tadmakka.
A novel faunal element, domestic chicken, appears at Jenné-jeno, Tongo
Maaré Diabel and Kirikongo, Burkina Faso.131 Along the Middle Niger, new
building technologies appear, including mudbrick and banco. At Jenné-jeno,
basal deposits with pottery dating from 100 BC to AD 400 occur over an area
500 m in extent, suggesting robust settlement growth.
From the eighth century onward, Middle Niger sequences provide much
richer evidence of interaction and trade. The region from Timbuktu to
Bentia appears to be an expansive interaction zone, with common pottery
types and a proliferation of imported glass and copper, especially at Gao.
Other North African elements at Gao include building technologies in
mould-formed mudbrick and dressed stone, imported glazed pottery, and
oil lamps. Gold is present at Gao, Jenné-jeno, and Tadmakka after AD 850.
The large mound site of Gao Saney grew very rapidly after initial settlement
in the eighth century, reaching over 30 ha before abandonment around AD
1100–1200. Jenné-jeno reached its maximum extent (33 ha) by AD 900,
when its interaction networks included the Lakes Region and Kumbi Saleh,
as indicated by the distribution of white-on-red painted pottery. The
notable increase in the scale and intensity of interaction, especially
after AD 900, is consistent with historical accounts of the establishment
of regular Trans-Saharan trade after AD 700.

References

Arazi, N. 1999. An archaeological survey in the Songhay heartland of Mali. Nyame

Akuma 52: 25–43.
Artioli, G., Laddaga, L. and Dapiaggi, M. 2006. Archaeometry of pottery:
Mineralogical and chemical analyses. In Liverani 2006, 249–57.
Bâ, A. 2002. Le Takrur: Des origines à la conquête par le Mali (VIe–XIIIe siècles).
Dakar: IFAN/Université Cheikh Anta Diop; Nouakchatt: CRIAA, Université de
Nouakchott.

131
Dueppen 2011.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 556 Susan Keech McIntosh

Bauer, A.A. and Agbe-Davies, A.S. 2010. Trade and Interaction in Archaeology.
In A.A. Bauer and A.S. Agbe-Davies (eds), Social Archaeologies of Trade and
Exchange. Walnut Creek: Left Coast Press, 29–47.
Becker, C. and Bocoum, H. 1998. Manifestations of urbanism in Senegambia. In B.
W. Andah, M.A. Sowunmi, A.I. Okpoko and C.A. Folorunso (eds), Africa: The
Challenge of Archaeology. Ibadan: Heinemann Educational Books, 149–70.
Bedaux, R., Polet, J., Sanogo, K. and Schmidt, A. (eds). 2005. Recherches
archéologiques à Dia dans le Delta intérieure du Niger (Mali): Bilan des saisons
de fouilles 1998–2003. Leiden: CNWS Publications.
Birch, J. (ed.). 2013. From Prehistoric Villages to Cities: Settlement Aggregation and
Community Transformation. London: Routledge.
Bocoum, H. 2000. L’age du fer au Sénégal: Histoire et archéologie. Dakar: IFAN/
Université Cheikh Anta Diop; Nouakchatt: CRIAA.
Bocoum, H. and McIntosh, S. 2002. Fouilles à Sincu Bara, un site de l’age de fer dans
la moyenne vallée du Sénégal. Dakar: IFAN/Université Cheikh Anta Diop;
Nouakchott: CRIAA.
Bouimetarhan, I., Dupont, L., Schefuss, E., Mollenhauer, G., Mulitza, S. and
Zonneveld, K. 2009. Palynological evidence for climatic and oceanic variability
of NW Africa during the late Holocene. Quaternary Research 72.2: 188–97.
Boutillier, J.-L. 1962. La moyenne vallée du Sénégal: Étude socio-économique. Paris:
INSEE Servicé de Coopération; Presses Universitaires de France.
Brill, R. 1995. Chemical analysis of some glasses from Jenné-Jeno. In McIntosh
1995, 252–56.
Chapman, J., Videiko, M., Gayadarska, B., Burdo, N., Hale, D., Villis, R.,
Swann, N., Thomas, N., Edwards, P., Blair, A., Hayes, A., Nebbia, M. and
Rud, V. 2014. The planning of the earliest European proto-towns: A new
geophysical plan of the Trypillia mega-site of Nebelivka, Kirovograd Domain,
Ukraine. Antiquity 88, Project Gallery.
Chavane, B.A. 1976. Le site protohistorique de Saré-Thioffi (Podor, Sénégal).
Unpublished mémoire de maîtrise. Université Cheikh Anta Diop, Dakar.
Chavane, B.A. 1985. Villages de l’ancien Tekrour. Paris: Editions Karthala.
Cissé, M. 2011. Archaeological Investigations of Early Trade and Urbanism at Gao
Saney (Mali). Unpublished PhD thesis, Rice University.
Cissé, M. 2017. The Trans-Saharan trade connection with Gao (Mali) during the
first millennium. In Mattingly et al. 2017, 101–30.
Cissé, M., McIntosh, S., Dussubieux, L., Fenn, T., Gallagher, D. and Smith, A. 2013.
Excavations at Gao-Saney. Journal of African Archaeology 11: 9–37.
Clark, M.E. 2003. Archaeological Investigations at the Jenné-jeno Settlement
Complex, Inland Niger Delta, Mali, West Africa. Unpublished PhD thesis,
Southern Methodist University.
Creekmore, A. and Fisher, K.D. (eds). 2014. Making Ancient Cities: Space and Place
in Early Urban Societies. Cambridge: Cambridge University Press.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 557

Crumley, C.L. 1995. Heterarchy and the analysis of complex societies. In

R.M. Ehrenreich, C.L. Crumley and J.E. Levy (eds), Heterarchy and the
Analysis of Complex Societies. Archaeological Papers of the American
Anthropological Association 6.1: 1–6.
Czerniewicz, M. von. 2004. L’évolution de la décoration et de la forme des
céramiques pendant l’Age du Fer au Nord du Burkina Faso. Préhistoire
Anthropologie Méditerranéennes 13: 127–32.
Deme, A. 2003. Archaeological Investigation of Settlement Evolution and Emerging
Complexity in the Middle Senegal Valley. Unpublished PhD thesis, Rice
University.
Deme, A. and McIntosh, S. 2006. Excavations at Walaldé: New light on the
settlement of the Middle Senegal Valley by iron-using peoples. Journal of
African Archaeology 4.2: 317–47.
Devisse, J. 1988. Trade and trade routes in West Africa. In UNESCO General
History of Africa III: Africa from the Seventh to the Eleventh Century. Oxford:
Heinemann, 367–435.
Dowler, A. and Galvin, E.R. (eds). 2011. Money, Trade and Trade Routes in Pre-
Islamic North Africa. London: British Museum Press.
Dueppen, S. 2011. Early evidence for chickens at Iron Age Kirikongo, Burkina
Faso. Antiquity 85: 142–57.
Dussubieux, L. 2011. LA-ICP-MS Analysis of glass beads from Gao-Saney. In Cissé
2011, Appendix III.1.
Fauvelle-Aymar, F.-X. 2016. Why and how did most African Islamic sites disap-
pear in the 15th century? Paper presented at the workshop ‘Did the Plague
impact sub-Saharan Africa before 1899?’ 22 April 2016, College of William and
Mary.
Fenn, T. 2011. Preliminary report: lead istopic ratio analysis of copper objects
excavated from Gao Saney. In Cissé 2011, 334–49
Fenn, T. and Killick, D. 2003. Metallurgical report on Walaldé copper. In Deme
2003, Appendix C3.
Fenn, T. and Killick, D. 2016. Copper alloys. In McIntosh, R. et al. 2016, 281–98.
Fenn, T., Killick, D., Chesley, J. and Ruiz, J. 2009. Contact between West Africa and
Roman North Africa: Archaeometallurgical results from Kissi, northeastern
Burkina Faso. In Magnavita et al. 2009, 119–46.
Fleisher, J. and Wynne-Jones, S. 2010. Authorisation and the process of power: The
view from African archaeology. Journal of World Prehistory 23.4: 177–93.
Fletcher, R. 2009. Low-density, agrarian-based urbanism: A comparative view.
Insights (University of Durham) 2.4: 2–19.
Garenne-Marot, L. 1993. Archéologie d’un métal: Le cuivre en Sénégambie (Afrique
de l’Ouest) entre le Xe et le XIVe siècle [thèse nouveau régime, Paris 1], 2 vols.
Paris: Université de Paris I Pantheon-Sorbonne, UER d’Histoire.
Garenne-Marot, L., Wayman, M. and Piggott, V. 1994. Early copper and brass in
Senegal. In S. Terry Childs (ed.), Society, Culture, and Technology in Africa,

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 558 Susan Keech McIntosh

MASCA Research Papers in Science and Archaeology. Philadelphia: University

of Pennsylvania Museum of Archaeology and Anthropology, 45–62.
Gatto, M.C. 2006. The local pottery. In Liverani 2006, 201–40.
Gestrich, N. 2013. The Archaeology of Social Organisation at Tongo Maaré Diabal.
Unpublished PhD thesis, University College London, Institute of Archaeology.
Giachet, M.T., Gratuze, B., Ozainne, S., Mayor, A. and Huysecom, E. 2019.
A Phoenician glass eye bead from 7th–5th c. cal BCE Nin-Béré 3, Mali:
Compositional characterisation by LA–ICP–MS. Journal of Archaeological
Science: Reports 24: 748–58.
Grébénart, D. 1985. Le néolithique final et les débuts de la métallurgie. La région d’In
Gall – Tegidda N Tesemt (Niger). Etudes Nigeriennes no. 49. Niamey: Institut de
Recherches en Sciences Humaines.
Hanebuth, T. and Henrich, R. 2009. Recurrent decadal-scale dust events over
Holocene western Africa and their control on canyon turbidite activity
(Mauritania). Quaternary Science Reviews 28: 261–70.
Haour, A. 2005. Power and permanence in precolonial Africa: A case study from
the central Sahel. World Archaeology 37.4: 552–65.
Haskell, H., McIntosh, R.J. and McIntosh, S.K. 1988. Archaeological
Reconnaissance in the Region of Dia, Mali. Final report to the National
Geographic Society, Washington DC.
Hatté, C., Saliège, J.-F., Senasson, D. and Bocoum, H. 2010. Cultural and trade
practices in Sincu Bara (Senegal): A multi-proxy investigation. Journal of
Archaeological Science 37: 561–68.
Hunwick, J.O. 1980. Gao and the Almoravids: A hypothesis. In B. Swartz and
R. Dumett (eds), West African Cultural Dynamics: Archaeological and Historical
Perspectives. The Hague: Mouton, 413–30.
Hunwick, J.O. 1994. Gao and the Almoravids revisited: Ethnicity, political change
and the limits of interpretation. Journal of African History 35.2: 251–73.
Hunwick, J.O. 1999. Timbuktu and the Songhay Empire: Al-Sa’di’s Tarikh al-Sudan
down to 1613 and other Contemporary Documents. Leiden: Brill.
Insoll, T. 2000a. The origins of Timbuktu. Antiquity 74: 483–84.
Insoll, T. 2000b. Urbanism, Archaeology and Trade: Further Observations on the
Gao Region (Mali), the 1996 Field Season Results. Oxford: British Archaeological
Reports Ltd.
Insoll, T. 1996. Islam, Archaeology and History: Gao Region (Mali) ca. AD
900–1250. Oxford: Tempus Reparatum.
Insoll, T. and Shaw, C.T. 1997. Gao and Igbo-Ukwu: Beads, interregional trade,
and beyond. African Archaeological Review 14.1: 9–24.
Jodin, A. 1966. Bijoux et amulettes du Maroc Punique. Bulletin d’Archéologie
Marocaine 6: 55–90.
Killick, D. 2016. Iron working in the Middle Senegal Valley. In McIntosh, R. et al.
2016, 191–280.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 559

Killick, D. 2003. Examination of archaeological samples from Djenné, Mali. In

Clark 2003, 415–25.
Kusimba, C. 2008. Early African cities: Their role in the shaping of urban and rural
interaction spheres. In J. Marcus and J. Sabloff (eds), The Ancient City: New
Perspectives on Urbanism in the Old and the New World. Santa Fe: School for
Advanced Research Press, 229–46.
Lambert, N. 1971. Les industries sur cuivre dans l’ouest Saharien. West African
Journal of Archaeology 1: 9–21.
Lambert, N. 1972. Objets en cuivre et néolithique de Mauritanie occidentale. In
H.J. Hugot (ed.), Actes de 6e session congrès panafricain de préhistoire, Dakar
1967. Pan-African Congress on Prehistory and Quaternary Studies, 1967, Dakar,
Senegal. Chambéry: Les Imprimeries Réunies de Chambéry, 159–74.
Lambert, N. 1975. Mines et métallurgie antiques dans la région d’Akjoujt. Annales
de l’Institut Mauritanien de Recherche Scientifique 1: 6–24.
Lambert, N. 1983. Nouvelle contribution à l’étude du Chalcolithique de
Mauritanie. In N. Echard (ed.), Métallurgies africaines: Nouvelles contributions.
Paris: Mémoires de la Société des Africanistes 9, 63–87.
Lekson, S. 2018. A Study of Southwestern Archaeology. Salt Lake City: The
University of Utah Press.
Levtzion, N. and Hopkins, J.F.P. 1981. Corpus of Early Arabic Sources for West
African History. Cambridge: Cambridge University Press.
Liverani, M. 2000. Looking for the southern frontier of the Garamantes. Sahara 12:
31–44.
Liverani, M. (ed.). 2006. Aghram Nadharif: The Barkat Oasis (Sha’abiya of Ghat,
Libyan Sahara) in Garamantian Times, Arid Zone Archaeology Monographs 5.
Florence: All’Insegna del Giglio.
MacDonald, K.C. 1998. More forgotten tells of Mali. Archaeology International 1:
40–42.
MacDonald, K.C. and MacDonald, R.H. 2016. Mammalian, avian and reptilian
remains. In McIntosh, R. et al. 2016, 311–34.
Magnavita, S. 2003. The beads of Kissi, Burkina Faso. Journal of African
Archaeology 1.1: 127–38.
Magnavita, S. 2013. Initial encounters: Seeking traces of ancient trade connections
between West Africa and the wider world. Afriques: débats, methods, et terrains
d’histoire 4. Available at: http://afriques.revues.org/1145 [last accessed
15 September, 2015].
Magnavita, S. 2015. 1500 Jahre am Mare de Kissi: Eine Fallstudie zur
Besiedlungsgeschichte des Sahel von Burkina Faso. Frankfurt: Africa Magna
Verlag.
Magnavita, S., Maga, A., Magnavita, C. and Idé, O.A. 2007. New studies on
Marandet (central Niger) and its trade connections: An interim report.
Zeitschrift für Archäologie Außereuropäischer Kulturen 2: 147–65.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 560 Susan Keech McIntosh

Magnavita, S., Kolé, L., Breunig, P. and Idé, O.A. (eds.). 2009. Crossroads
Carrrefour/ Sahel. Cultural and Technological Developments in the First
Millennium BC/AD West Africa. Frankfurt: Africa Magna Verlag.
Maley, J. 2010. Climate and palaeoenvironment evolution in north tropical Africa
from the end of the Tertiary to the Upper Quaternary. Palaeoecology of Africa 30:
227–78.
Maley, J. and Vernet, R. 2015. Populations and climatic evolution in north tropical
Africa from the end of the Neolithic to the dawn of the modern era. African
Archaeological Review 32.2: 179–232.
Manyanga, M., Pikirayi, I. and Chirikure, S. 2010. Conceptualising the urban mind
in pre-European Southern Africa: Rethinking Mapungubwe and Great
Zimbabwe. In P.J. Sinclair, G. Nordquist, F. Herschend and C. Isendahl (eds),
The Urban Mind: Cultural and Environmental Dynamics. Uppsala Studies in
Global Archaeology 15. Uppsala: Uppsala University, 573–90.
Mattingly, D.J. (ed.). 2003. Archaeology of Fazzan, Volume 1, Synthesis. London:
Society for Libyan Studies.
Mattingly, D.J. (ed.). 2013. Archaeology of Fazzan, Volume 4, Survey and
Excavations at Old Jarma. London: Society for Libyan Studies.
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017. Trade in the Ancient Sahara and Beyond. Trans-Saharan
Archaeology Volume I. Series editor D.J. Mattingly. Cambridge: Cambridge
University Press and The Society for Libyan Studies.
McDougall, E.A. 1990. Salts of the Western Sahara: Myths, mysteries, and histor-
ical significance. International Journal of African Historical Studies 23.2: 231–57.
McIntosh, R.J. 1991. Early urban clusters in China and Africa: The arbitration of
social ambiguity. Journal of Field Archaeology 18: 199–212.
McIntosh, R.J. 1993. The pulse model: Genesis and accommodation of specializa-
tion in the Middle Niger. The Journal of African History 34.2: 181–220.
McIntosh, R.J. 1998. The Peoples of the Middle Niger: Island of Gold. Oxford:
Blackwell.
McIntosh, R.J. 2005. Ancient Middle Niger: Urbanism and the Self-Organizing
Landscape. Cambridge: Cambridge University Press.
McIntosh, R.J. 2015. Different cities: Jenne-jeno and African urbanism. In The
Cambridge World History: Volume 3, Early Cities in Comparative Perspective,
4000 BCE–1200 CE. Cambridge: Cambridge University Press, 364–80.
McIntosh, R.J. and McIntosh, S.K. 1981. The Inland Niger Delta before the Empire
of Mali: Evidence from Jenne-jeno. Journal of African History 22: 1–22.
McIntosh, R.J. and McIntosh, S.K. 1987. Prospection archéologique aux alentours
de Dia, Mali 1986–1987. Nyame Akuma 29: 42–45.
McIntosh, R.J., McIntosh, S.K. and Bocoum, H. (eds). 2016. The Search for Takrur:
Archaeological Excavations and Reconnaissance along the Middle Senegal Valley.
New Haven: Yale University Publications in Anthropology.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 561

McIntosh, S.K. (ed.). 1995. Excavations at Jenne-jeno, Hambarketolo and Kaniana

(Inland Niger Delta, Mali): The 1981 Season. University of California
Monographs in Anthropology. Berkeley: University of California Press.
McIntosh, S.K. 1999a. Beyond Chiefdoms: Pathways to Complexity in Africa.
Cambridge: Cambridge University Press.
McIntosh, S.K 1999b. Modeling political organization in large-scale settlement
clusters: A case study from the Inland Niger Delta. In McIntosh 1999a, 66–79.
McIntosh, S.K. 2016. The Cubalel and Siwré pottery assemblages. In McIntosh,
R. et al. 2016, 117–73.
McIntosh, S.K. 2017. The Akumbu mound complex (Méma, Republic of Mali):
Ceramics, change, and continuity in the first millennium AD. In N. Rupp,
C. Beck, G. Franke, and K.P. Wendt (eds), Winds of Change: Festschrift for
Peter Breunig. Frankfurt am Main: Afrika Magna Verlag, 193–204.
McIntosh, S.K. 2018. Surplus and complexity in the inland Niger Delta, Mali: The
invisibility conundrum. In H. Meller, and R. Risch (eds), Surplus without the
State: Political Forms in Prehistory. Halle: Tagungen des Landesmuseums für
Vorgeschichte Halle Band 18, 373–86.
McIntosh, S.K. and Bocoum, H. 2000. New perspectives on Sincu Bara, a first
millennium site in the Senegal Valley. African Archaeological Review 77.1:
124–78.
McIntosh, S.K. and McIntosh, R.J. 1980. Prehistoric Investigations in the Region of
Jenne, Mali: A Study in the Development of Urbanism in the Sahel. Cambridge
Monographs in African Archaeology 2, 2 vols. Oxford: British Archaeological
Reports.
McIntosh, S.K. and McIntosh, R.J. 1986. Archaeological reconnaissance in the
region of Timbuktu, Mali. National Geographic Research 2: 302–19.
McIntosh, S.K. and McIntosh, R.J. 1993. Cities without citadels: Understanding
urban origins along the middle Niger. In T. Shaw, P. Sinclair, B. Andah, and
A. Okpoku (eds), The Archaeology of Africa: Food, Metals and Towns. London:
Routledge, 622–41.
McIntosh, S.K., McIntosh R.J. and Bocoum, H. 1992. The Middle Senegal Valley
Project: Preliminary results from the 1990–91 field season. Nyame Akuma 38:
47–61.
Michel, P. 1973. Les basins des fleuves Sénégal et Gambie: Étude géomorphologique.
Paris: Mémoire de l’ORSTOM, 63.
Monteillet, J., Faure, H., Pirazzoli, P. and Ravisé, A. 1981. L’invasion saline du Ferlo
(Sénégal) à l’Holocène supérieure (1900 B.P.). In J.A. Coetzee and E.M. van
Zinderen Bakker (eds), Palaeoecology of Africa and the Surrounding Islands,
Volume 13. Rotterdam: A.A. Balkema, 205–15.
Murray, S.S. 2008. A report on charred botanical remains from Sincu Bara, a
mid-first millennium AD Middle Senegal Valley site. Nyame Akuma 69: 56–63.
Murray, S.S. n.d. Report on Charred Botanical Remains from Cubalel Sites and
Siwré. Unpublished report, Rice University Department of Anthropology.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 562 Susan Keech McIntosh

Nixon, S. 2008. The Archaeology of Early Islamic Cross-Saharan Trading Towns in

West Africa: A Comparative View and Progressive Methodology from the
Entrepot of Essouk-Tadmekka. Unpublished PhD Thesis, University College
London.
Nixon, S. 2009. Excavating Es-Souk–Tadmakka (Mali): New archaeological inves-
tigations of early Islamic trans-Saharan trade. Azania: Archaeological Research
in Africa 44.2: 217–55.
Nixon, S. (ed.). 2017. Essouk-Tadmekka: An Early Islamic Trans-Saharan Market
Town. Leiden: Brill.
Nizou, J., Hanebuth, T.J., Heslop, D., Schwenk, T., Palamenghi, L., Stuut, J.-B. and
Henrich, R. 2010. Senegal river mud belt: A high-resolution archive of paleocli-
matic change and coastal evolution. Marine Geology 278.1–4: 150–64.
Oka, R. and Kusimba, C. 2008. The archaeology of trading systems, part I: Towards
a new trading synthesis. Journal of Archaeological Research 16.4: 339–95.
Park, D.P. 2010. Prehistoric Timbuktu and its hinterland. Antiquity 84: 1076–88.
Park, D.P. 2011. Climate Change, Human Response and the Origins of Urbanism at
Timbuktu. Unpublished PhD thesis, Yale University.
Ponsich, M. 1967. Nécropoles phéniciennes de la région de Tanger. Tanger: Editions
Marocaines.
Raimbault, M. and Sanogo, K., 1991. Recherches archéologiques au Mali. Paris:
Editions Karthala.
Renfrew, C. 1975. Trade as action at a distance: Questions of integration and
communication. In J.A. Sabloff and C.C. Lamberg-Karlovsky (eds), Ancient
Civilization and Trade. Albuquerque: University of New Mexico Press, 3–59.
Robert-Chaleix, D. 1989. Une concession médiévale à Tegdaoust: Implantation,
evolution d’une unité d’habitation. Paris: Editions Recherche sur les
Civilisations.
Robert-Chaleix, D. 1991. Sel, coquillages et jarosite: À propos de la
saline médiévale d’Awlil. Journal des Africanistes 61.2: 169–85.
Robert-Chaleix, D. 1994. Métallurgie du fer dans la Moyenne Vallée du Sénégal:
Les bas fourneaux de Silla. Journal des Africanistes 64.2: 113–27.
Robert-Chaleix, D. and Sognane, M. 1983. Une industrie métallurgique ancienne
sur la rive mauritanienne du Fleuve Sénégal. In Métallurgies Africaines:
Nouvelles contributions. Paris: Mémoires de la Société des Africainistes 9: 45–62.
Robertshaw, P., Magnavita, S., Wood, M., Melchiorre, E., Popelka-Filcoff, R.S. and
Glascock, M.D. 2009. Glass beads from Kissi (Burkina Faso): Chemical analysis
and archaeological interpretation. In Magnavita et al. 2009, 105–18.
Robion-Brunner, C. 2010. Forgerons et sidérurgie en pays dogon. Vers une histoire
de la production du fer sur le plateau de Bandiagara (Mali) durant les empires
précoloniaux. Frankfurt: Africa Magna Verlag.
Schmidt, A., Sanogo, K. and Keita, D. 2005. Les fortifications. In Bedaux et al. 2005,
117–24.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 14 Exchange and Urban Trajectories. Middle Niger and Middle Senegal 563

Smith, M. 2006. Review of R.J. McIntosh, ‘Ancient Middle Niger: the Self-
Organizing Landscape’. Journal of African Archaeology 4.2: 357–59.
Stone, A.C. 2015. Urban Herders: An Archaeological and Isotopic Investigation into the
Roles of Mobility and Subsistence Specialization in an Iron Age Urban Center in Mali.
Unpublished PhD thesis, Washington University, Department of Anthropology.
Thiaw, I. 1999. Archaeological Investigation of Long-Term Culture Change in the
Lower Falemme (Upper Senegal Region): A.D. 500–1900. Unpublished PhD
thesis, Rice University.
Thilmans, G. 1977. Sur les objets de parure trouvés à Podor (Sénégal) en 1958.
Bulletin de l’Institut Fondamental d’Afrique Noire (BIFAN) Series B 39.4:
669–94.
Thilmans, G. and Ravisé, A. 1980. Protohistoire du Sénégal: Recherches
archéologiques. Vol. 2, Sinthiou-Bara et les sites du fleuve. Mémoires de
l’Institut Fondamental d’Afrique Noire 91. Dakar: IFAN.
Togola, T. 1996. Iron Age occupation in the Méma region, Mali. African
Archaeological Review 13.2: 91–110.
Togola, T. 2008. Archaeological Investigations of Iron Age Sites in the Meme Region,
Mali (West Africa). Cambridge Monographs in African Archaeology no. 73.
Oxford: B.A.R.
Trigger, B. 1972. Determinants of urban growth in pre-industrial societies. In
P.J. Ucko, R. Tringham and G.W. Dimbleby (eds), Man, Settlement and
Urbanism. London: Duckworth, 575–99.
Vanacker, C. 1979. Tegdaoust II: Recherches sur Aoudaghost. Fouille d’un quartier
artisinale. Paris: ADPF.
Vernet, R. 1993. Préhistoire de la Mauritanie. Nouakchott: Centre culturel français
A. de Saint-Exupéry, Sepia.
Vernet, R. 2000. Un habitat de l’âge du cuivre (2500 BP) de la région de Nouakchott
(Mauritanie occidentale: Imbich-Est). Sahara 12: 83–90.
Vernet, R. 2004. L’archéologie sur la rive mauritanienne du fleuve Sénégal: Un
chantier à ouvrir. MASADIR: Cahiers des Sources de l’Histoire de la Mauritanie
4: 21–31.
Vernet, R. 2012. Le Chalcolithique de Mauritanie (3000–2500 cal BP): État de la
question. Sahara 23: 7–28.
Webb, J.L. 1995. Desert Frontier: Ecological and Economic Change along the
Western Sahel 1600–1850. Madison: University of Wisconsin Press.
Wynne-Jones, S. and Fleisher, J.B. 2014. Swahili urban spaces of the eastern
African coast. In Creekmore and Fisher, 2014: 111–44.

https://doi.org/10.1017/9781108637978.015 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements
at Lake Chad
Implications for the Origins of Urbanisation and State
Formation in Sub-Saharan Africa

carlos magnavita

Introduction

Between 2004 and 2009, a programme of ground-based geophysical

surveying revealed a number of large ancient fortified settlements at the
south-western margins of Lake Chad.1 Dating to the first millennia BC/
AD, these archaeological sites are amongst the oldest evidence for com-
munal-level protection in Sub-Saharan Africa. On the one hand, settlement
defences and dense populations, estimated in the low thousands, clearly
indicate that security must have been a major concern for people living in
the area at that time. On the other, these sites raise a series of questions that
are fundamental for understanding the emergence of urban centres and
state-level societies in the Lake Chad area and in Sub-Saharan Africa. This
chapter introduces the archaeological evidence pertaining to those settle-
ments and discusses their implications for the current debate on the origins
of urbanisation and state formation in Sub-Saharan Africa.
The scholarly interest in the ancient city as an object of analysis mainly
goes back to the early twentieth century.2 In archaeology, the growing
interest in the city as a unit of analysis came after V. Gordon Childe (in
1934) coined the term ‘urban revolution’ for highlighting what he regarded
as one of the major socioeconomic turning-points in human history.3
Beyond Childe’s ideas, theory-backed systematic archaeological investiga-
tions of early urban settlements and the landscapes in which they are
embedded are a relatively recent phenomenon rooted in the Processual

1
The central goal of this chapter is to present and discuss archaeological data collected over
a period of six years (2004–2009) in the scope of an archaeological research programme based at
the Goethe-University Frankfurt and generously funded by the German Research
Council (DFG).
2
564 Most famously through Weber 1958. 3 Trigger 1989, 253–54.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 565

Archaeology (New Archaeology) of the 1960s to 1970s.4 Though much

criticised from the 1980s onwards due to the often inflexible and one-
dimensional neo-evolutionary approach used by many processualist
researchers, the practical results of the archaeological work conducted on
urban sites in the scope of that school of thought was of great scientific
value. For the first time, there was indeed a major effort in revealing and
understanding the origins and the function that cities played within a given
archaeological landscape.5
Embracing a processual and landscape-wide archaeological research
agenda, the origins and function of urban settlements likewise became
central themes of archaeological research on Sub-Saharan African urbanisa-
tion in the late 1970s and early 1980s.6 In fact, that research approach
developed into a major paradigm change in the historiography of Sub-
Saharan Africa. Departing from the traditional, document-based view that
evoked a strong role for white North Africans and Near Eastern immigrants
in the foundation of cities and states south of the Sahara, archaeological
research in the 1980s began to show that at least the roots of Sub-Saharan
urban life were much older and much more closely linked to local and
regional socio-economic and political developments than earlier thought.7
Likewise, rigid and universal definitions of what the ancient city is or what it
must have looked like gave way to a more flexible definition that emphasised
what a city does, that is, what do they provide in terms of specialised
production and services for its hinterland.8 In West Africa, the best studied
archaeological instance demonstrating a gradual and indigenous process of
settlement growth culminating in an ancient urban landscape comes from
the hierarchical cluster of sites around Jenné-jeno in Mali.9 However, there
are at least three other, older, archaeological regional occurrences in West
Africa that lend support to the indigenous roots and long-term devel-
opment of urbanisation within Sub-Saharan Africa. The first is the
hierarchical cluster of settlements at Dhar Tichitt (1600–1000 BC) in
southern Mauritania.10 The second is the hierarchical cluster of settle-
ments in the Macina region of Mali, dated to c.900–400 BC.11 The third
is a number of, in part hierarchically organised, fortified settlement sites

4
See, for example, Sabloff 1994 for the Mayan case.
5 6
Compare Creekmore and Fisher 2014, 3. McIntosh and McIntosh 1984.
7
See McIntosh, S., Chapter 14, this volume; McIntosh, S. and McIntosh R. 1984; 1993.
8
Cowgill 2004, 527; McIntosh, S. and McIntosh, R. 1993, 625; Trigger 1972, 577.
9
See McIntosh, S., Chapter 14, this volume; McIntosh, S. and McIntosh, R. 1993.
10
Holl 1993; MacDonald 2003; Munson 1980; see also MacDonald, Chapter 13, this volume.
11
Bedaux et al. 2001; 2005.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 566 Carlos Magnavita

near Lake Chad in Nigeria, dated to the first millennia BC/AD.12 This
paper presents and discusses the archaeological evidence recovered at
the latter sites and considers their implications for understanding the
rise of urbanisation and state-level societies in the Lake Chad Basin and
in Sub-Saharan Africa.

The Origins of Food Production and Settled Life in the Chad

Basin: The Gajiganna Complex

The deepest foundations of urbanisation are clearly linked to food produc-

tion. Urbanisation occurred as part of long-term processes set in train by the
initial domestication of plants, but dependent also on the human capacity for
developing techniques to intensify and diversify the production of food for
a growing number of people. In the Lake Chad Basin, approximately 2,500
years elapsed between the beginnings of food production and the appearance
of historically known urban communities in the southern margins of Lake
Chad. Since the 1960s, the growing commitment to archaeological research
around the Chad Basin has greatly expanded our knowledge of the cultural
developments in the region. It is true that the archaeological data so far
recovered are only mere snap-shots of those developments, and details about
the processes leading from one point to the other are broadly unknown.
Nevertheless, it can be acknowledged that we are now better informed about
the cultural trajectories of the human groups that inhabited the Chad Basin
in pre-historical times than we were 20 years ago.
In our present state of knowledge, the first food-producing communities
that appeared in the Lake Chad region occupied the lake’s south-western
margins at the onset of the second millennium BC.13 Archaeologically,
those groups of people came to be known as the Gajiganna Complex
(c.1800–400 BC) in view of their shared material culture, settlement pat-
terns and other aspects.14 At the beginning of this period, those groups
seem to have been relatively mobile, since their archaeological remains are
characterised by mere scatters of potsherds over some few square metres.
At that stage (Gajiganna Phase I) their economy was mainly based on
livestock-keeping but a minor contribution of (domesticated) plants can-
not yet be ruled out. From c.1500 BC there are the first clues for a growing
reliance on domesticated vegetal food compared to livestock farming. This
change was gradually accompanied by a more settled way of life in hamlets

12 13 14
Magnavita, C. et al. 2006; 2009. Breunig and Neumann 2002. Breunig et al. 1996.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 567

and village-like communities with clay houses, the long-standing locations

of which resulted in the formation of the first Gajiganna settlement
mounds (Gajiganna Phase IIa/b). This settled way of life can be traced
back until about 1000 BC. After that, and for about 400 years, the archae-
ological evidence merely suggests the presence of small Gajiganna human
groups temporarily living in the area (Gajiganna Phase IIc). The reason(s)
for the abandonment of their permanently settled living spots in the
previous millennium is unclear, but there is the suspicion that environ-
mental factors may have played the major role in this development. In fact,
it is known that a vast, formerly year-round flooded area south of Lake
Chad must have become seasonally dry from this time on, thus allowing
the beginning of a possibly cyclic and, later on, permanent occupation by
settled farmers.15 This event may indicate a major change of rainfall
patterns that led to an extensive period of drought in the Chad Basin and
elsewhere. In view of the absence of climatic proxy data for the region, the
explanation of a drastically changed environment remains tentative.
Alternatives for understanding the abandonment of the Gajiganna settle-
ment mounds have been offered recently,16 but hard data that might
support them has yet to be found.
However, the archaeological evidence reveals a revival of the settled
way of life of Gajiganna groups in the area some centuries later, in the
middle of the first millennium BC (Gajiganna Phase III, c.600–400 BC).
Local faunal data indicates that this phase of occupation might coincide
with a short humid period in the south-western Lake Chad region, which
possibly explains the return to a fully settled way of life in the area.17
Chronologically, this evidence correlates with an occupation hiatus
recorded in the stratigraphies of a number of settlement mounds in the
southern fringes of Lake Chad.18 Characterised by a succession of sterile
clay bands that formed within a period of 110–350 14C-years, this site
abandonment event was originally thought to have resulted from a major
transgression of Lake Chad with water levels rising to 290 m asl, that is,
3 m above c.1000 BC levels.19 Though such a large lake oscillation is now
regarded as improbable,20 there are at least some indications that
Gajiganna Phase III settlements might have emerged in an environmen-
tally comparatively favourable phase in the middle of the first
millennium BC.

15
Brunk and Gronenborn 2004, 108–9. 16 Breunig 2013, 562. 17 Breunig et al. 2006.
18
Gronenborn 1998, 250. 19 Brunk and Gronenborn 2004, 109; Gronenborn 1998, 250.
20
Brunk and Gronenborn 2004, 108–9.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 568 Carlos Magnavita

Current regional proxy data indicate major millennial and centennial

environmental trends for the Lake Chad Basin during the period dis-
cussed in this paper. For the time between c.1600 and 100 calBC, the data
available point to a relatively drier environment compared to previous
centuries and millennia.21 Conversely, a trend to wetter conditions is
attested for the period 100 calBC–calAD 500/700.22 Though I recognise
that those major environmental trends may have been a major factor
in shaping the long-term dynamics of settlement foundation and aban-
donment at the Lake Chad margins, as referred to above, shorter-term
environmental changes not detected by those studies also seem to
have played a non-negligible role in those developments. Since regional
environmental studies generally lack specific, reliable and finer-grained
evidence of actual local environmental conditions, I will abstain from
making any serious attempts at correlating particular cultural develop-
ments attested archaeologically with the environmental data available on
a regional scale.
The following narrative is necessarily focused on three settlements, as
they are presently the only substantial source of information related to the
archaeological themes, time period and region discussed. Where it appears
meaningful, other archaeological sites in the Chad Basin and elsewhere are
mentioned for the sake of comparison.

First Millennium BC Fortified Settlements at Lake Chad

Altogether 14 Gajiganna Phase III sites are so far known at the south-
western margins of Lake Chad.23 They all consist of flat sites with no, or
just a minimal, succession of cultural deposits. This implies a relatively
short period of occupation of each site that probably did not exceed two or
three generations. The estimated and confirmed site sizes vary widely from
1 ha to about 12–14 ha (Fig. 15.1). Particularly interesting is the fact that
the small to mid-sized sites are located as satellites of the two largest
settlements known, Galgalkura A (12–14 ha) and Zilum (12–13 ha).
In fact, the clustering indicates some kind of interrelation between
settlements of different sizes. Though the Gajiganna Phase III sites are
smaller than their counterparts in second-millennium BC Mauritania and
first-millennium BC Mali, like those they display the very first regional
evidence for a sort of trait that seems deeply connected with the rise of

21 22 23
Holmes et al. 1997, 317. Holmes et al. 1997, 317. Magnavita, C. et al. 2006.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 569

Figure 15.1. Map showing the location of sites discussed in the text.

West African urbanisation: site-size hierarchy or hierarchical settlement

clustering (see discussion below).
In a sub-continental scale, the specific reasons and processes leading to
the formation of early hierarchical settlement clusters may have greatly
varied. In most cases, however, the developments were related to the
deliberate foundation of smaller settlements near previously established
larger ones, either by people from within the larger communities or by
outsiders. At least in the particular case of the first millennium BC
Gajiganna III sites, there is strong evidence suggesting that site-size hier-
archy was triggered, or at least catalysed, by an increased need of security.

Zilum
Of the 14 Gajiganna Phase III sites known, the best yet investigated and
published is Zilum. Research at that location comprised a geophysical
survey, mapping and excavation programme carried out over three field
seasons. Geophysical work was conducted using fluxgate gradiometry
(magnetometry) and resulted in the mapping of the site’s entire subsurface
extent;24 topographic surveys and plotting of specific find types and

24
Magnavita, C. et al. 2006.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 570 Carlos Magnavita

particular features brought about an idea of the distribution of craft-related

tools and facilities. Excavations produced detailed insights into archaeolo-
gical features that were related to the occupation at the site. Beyond
confirming the total site area that had been previously identified by test
pitting,25 geophysical survey provided fundamental information that con-
tributed to understanding the nature of Zilum. A large part of the flat site
had its upper archaeological deposits reduced by erosion, thus exposing
numerous finds’ concentrations and archaeological features at the surface;
geophysical survey was also able to detect and locate subsurface features
that were not immediately discernible.26
Figure 15.2, the magnetometry plot of Zilum, shows the shape and
extent of the site as well as the location of the numerous subsurface features
(magnetic anomalies) detected and related to past human activity at the
settlement. The most frequent kind of magnetic anomaly is dot-like in
shape. They mainly represent the location of pits sunk into the ground
during the time the settlement was occupied and are scattered all over the
site extent. Their relatively high density and distribution across the site
indicates both the area of the site which was settled in the past and,
likewise, how nucleated Zilum was, since virtually no vacant space existed
within the settlement. Other features visible from the magnetogram are
areas within the settlement that were once enclosed by fences or earthen
walls, thus implying the former existence of residential compounds. This
interpretation is supported by the find of a series of anomaly-free circular
spaces with diameters of up to 5 m and surrounded by pits; this strongly
suggests that these were the location of former round houses (Fig. 15.2).
Though excavations carried out at the site did not locate a single post hole,
finds of burnt daub with impressions of wattle indicate the main construc-
tion technique of residential structures.27 Perhaps the most surprising and,
likewise, most revealing subsurface feature detected by geophysical work at
Zilum was a number of peripheral ditches that once enclosed large portions
of the site. Though magnetometry was able to detect long segments of those
features, some sectors remain barely discernible (Fig. 15.2). However, there
is little doubt that the segments are parts of continuous ditch features that
once surrounded the site at different times. At least three phases of con-
struction and expansion of the surrounding ditch are discernible at Zilum,
which hints at an increase of the settlement size over time. While direct
evidence of a superstructure in the form of an earthen rampart or wall is no
longer visible, there is strong geophysical support for the former existence

25 26 27
Magnavita, C. 2008, 49. Magnavita, C. et al. 2006. Magnavita, C. 2008, 32.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 571

Figure 15.2. Magnetogram of Zilum, with indication of features discussed.

of such a construction. Evidence for this is the presence of elongated

anomaly-free areas adjacent to some ditch segments (Fig. 15.2) that hints
at the location of such a structure. Excavations of the enclosing ditch
conducted at three different sectors of the site, reveal that its shape and
dimensions were not uniform throughout.28 With a U-shape and a volume
of only about 3 m3/1 m section, the ditch was smaller in the south sector
than at the east and north sectors, where it had a V-profile and volumes of
c.7 m3 and 8–10 m3/1 m section. Altogether, the evidence collected at
28
Magnavita, C. et al. 2006.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 572 Carlos Magnavita

Zilum suggests the construction of a sizeable earthen rampart or wall

behind the ditch. In tandem, the ditch and rampart/wall must have
worked as a real obstacle that protected the settlement and its inhabitants
from external threat, whether robbery or armed attacks. Meanwhile, the
existence of former ditch-wall systems are attested at each of the five
Gajiganna Phase III settlements investigated by geophysical prospection
in the past years. In at least one further case, anomaly-free areas along an
enclosing ditch support the existence of now decayed earthen walls. The
results of that work are about to be published, but they fully agree with
the view that at least some of the food-producing communities living at
Lake Chad in the first millennium BC did seek protection by erecting
earthen walls.
Beyond walls, another strategy for coping with insecurity (felt or real)
caused by external threat was nucleation and aggregation. The results of
geophysical surveys at Zilum demonstrate how frequent the settlement
ground was transformed by repeated digging activities, ultimately demon-
strating the nucleated nature of the settlement. They also provide evidence
that can be used to calculate the number of people living in Zilum at the
time the settlement achieved its maximum size. Based on the area and
number of houses within a particular residential compound in the south-
eastern part of Zilum, it was estimated that the whole settlement once had
between about 1,750 and 2,500 inhabitants or c.140–210 people/hectare.29
Since large human groups are, due to their sheer size, relatively stronger
than smaller ones and in a better position to defend themselves, it is not
surprising to see a trend towards large and relatively dense populated
settlements as in the case of Zilum. In the same line of reasoning, the
need of security also explains why smaller settlements tended to aggregate
near larger ones, thus producing the familiar picture of a hierarchical
settlement system.
In addition to the benefits of security, a secondary effect of nucleation
and aggregation of human groups at Zilum pertains to an apparent change
in socio-economic relations within and, potentially, between communities.
To judge from the lack of evidence, neither metal tools nor any form of
metallurgy was known by the inhabitants of Zilum. Whilst they were
mainly farmers growing pearl millet and cowpeas, raising cattle, sheep
and goats, they also did some hunting and fishing.30 There is archaeologi-
cal evidence for craft production on the site that hints at some occupational
differentiation. The mapping of craft-related artefacts and features at

29 30
Magnavita, C. 2004, 87. Linseele 2007; Magnavita, C. et al. 2004.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 573

Zilum resulted in the discovery that pottery (at least certain forms such as
large storage vessels), stone beads or bone points and, possibly, leather
were produced in distinct parts of the site by different people.31 Given the
relatively large residential population of the settlement, it seems indeed
probable that at least some items (whose production required enhanced
skills or special knowledge) became the domains of (part-time) specialists.
In view of Zilum’s relatively large population and, therefore, the relatively
high demand for certain products, it is reasonable to think at least of
a seasonal (dry-season) occupational differentiation of people. Inevitably,
demand for certain products from nearby settlements may have formed the
basis for economic relations beyond an exchange of natural produce.
Although inter-community exchange remains archaeologically elusive,
the archaeological evidence of nucleation and aggregation around Zilum
may be seen as an instance of how the initial stages of socio-economic and
occupational differentiation took place in Sub-Saharan Africa.
Apart from the regional procurement of stones for domestic use in the
Mandara Mountains and Biu Plateau,32 very little evidence from Zilum and
contemporary sites hints at long-distance contacts. In fact, the only finds
that might have any significance in this regard are two polished lip-plugs
made of amazonite, one of them fragmented.33 The origin of the rocks used
in their manufacture is unknown, but regional (potentially, the Mandara
Mountains c.140 km away or the Biu Plateau c.190 km away) or Saharan
sources (Air Mountains c.900 km away, Tibesti c.1600 km away) are
conceivable options.
Zilum and adjacent sites are the earliest walled settlements currently
known in the Chad Basin. Whilst similarly large or even larger settle-
ments (10–30 ha) are known to have existed c.120 km further south
(Maibe and Malankari), they consist of rather dispersed non-walled
locations.34 Along with Zilum, only two other West African fortified
sites are dated to the last two millennia BC: Akrejit in Mauretania and
In Begouen in Mali, are known to have been enclosed by, respectively,
stone walling and wooden palisades.35 Instead, most of the archaeologi-
cally known enclosed settlements in Sub-Saharan Africa date from the
last thousand years,36 but a number of them such as Jenné-jeno in Mali
and Namudu in Sudan as well as Zubo and Dorota, both in Nigeria, date
to the first millennium AD.37 As is shown by the fieldwork at Zilum and
31
Magnavita, C. 2004; 2008, 121–25; Magnavita, C. et al. 2006. 32 Rupp 2005.
33
Magnavita, C. and Breunig 2008. 34 Breunig 2009, 18.
35
Gaussen and Gaussen 1988, 202; Holl 1993, 111. 36 Connah 2000, 21.
37
McIntosh, S. and McIntosh, R. 1984, 91; Mohammed 1986, 185–86.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 574 Carlos Magnavita

the other archaeological sites described below, it can be supposed that the
comparatively low number of as yet known first millennia BC/AD
Sub-Saharan enclosed settlements is probably the result of the poor
archaeological visibility of enclosures rather than their general
absence.

First Millennium AD Fortified Settlements at Lake Chad

Following the abandonment of the fortified Gajiganna III sites (c.600–

400 BC), no archaeological evidence hinting at a settled way of life is known
at the south-western margins of Lake Chad for the next four to five
centuries. Whilst permanent community life is attested in the southern
margins of Lake Chad (in Nigeria, Cameroon and Chad) by the rise
of settlement mounds in the second half of the first millennium BC,38
the former Gajiganna area appears to have been virtually deserted.
Environmental constraints may explain the abandonment of the area for
several centuries, but there is as yet no (climatic) proxy data that might
support such a claim. This situation changed dramatically around the
beginning of the first millennium AD when people with a pottery tradition
distinct from that of the former Gajiganna population enter the stage. It is
unknown from whence they originated, but some decorative elements of
their pots are not dissimilar to ceramics from the southern margins of Lake
Chad.39
Several dozens of sites connected to the novel pottery tradition are known
from the former Gajiganna area (Fig. 15.3). Their sizes range from 1 ha to
almost 50 ha and, unlike Gajiganna III sites, they all consist of low settlement
mounds. Preliminary test excavations and analysis of the archaeological
remains recovered from four of those sites provided first insights into their
chronology and the economy practised by the former residents. Nine radio-
carbon dates place their rise and development into the time between the first
and the early seventh centuries AD.40 As in the case of Gajiganna Phase III
settlements, the inhabitants of the sites were mainly farmers who grew pearl
millet and cowpeas, but a new sort of cultivar is also attested in the period:
sorghum.41 Domestic animals were by far the most important source of
protein, with hunting and fishing playing secondary roles in the economy.42
Compared to Gajiganna III settlements, however, ovicaprines grew in
38
Connah 1981; Gronenborn 1998; Holl 2002. 39 Magnavita, C. et al. 2009.
40
Magnavita, C. et al. 2009, table 1. 41 Magnavita, C. 2002; Magnavita, C. et al. 2009.
42
Linseele 2007, 97 and figures 35–36.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 575

Figure 15.3. Map showing the location of Iron Age sites in the Gajiganna area.

economic importance relative to cattle, a fact thought to be a strategic adapta-

tion to a drier environment during the first millennium AD.43 Unlike at
Zilum, the inhabitants of the first-millennium AD sites used and produced
iron artefacts in or near their settlements, although the required iron in the
form of blooms, bars or objects must have been brought from somewhere else.
The nearest source of iron ore is the plains around the Mandara Mountains
(c.100 km away), where evidence from iron working dates to as early as the
mid-first millennium BC.44 The Mandara Mountains are also the main rock
source for the grinding stones used in the settlements of the period.45

Zubo and Dorota

The two largest sites known in the area (Zubo and Dorota) were the focus
of intensive geophysical survey and subsequent archaeological work by the
author between 2007 and 2009. The younger of the two sites, Dorota, dated
to the second to sixth centuries AD, is the best investigated. It is located
c.40 km south-west of Zilum. The outcome of topographical and geophy-
sical mapping (employing fluxgate gradiometry) indicates that the settle-
ment had a maximum size of c.33.5 ha.46 The geophysical plan of the site
(Fig. 15.4) displays a myriad of magnetic anomalies within the settlement

43 44 45
Linseele 2007, 147, table 14, figure 49. MacEachern 1996. Rupp 2005.
46
Magnavita, C. et al. 2009.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 576 Carlos Magnavita

850

800

D8
750

Access D3
700 points

Modern
650 masakwa
dams

600

Earth road
550

D2 5
500
4.5
4
3.5
450 3
2.5
2
D5 1.5
400
1
0.5
0
350 –0.5
–1
–1.5
–2
Access
300 points
–2.5
–3
–3.5
–4
250 –4.5
–5

200

150

N
D4 D1 Pottery firing place?
100
Access
points
50

Scatter of iron slag

0
D6
Probable fumaces

–50
–50 0 50 100 150 200 250 300 350 400 450 500 550 600

Figure 15.4. Geophysical survey of Dorota.

area, revealing both intensive digging activities and the presence of major
subsurface archaeological features. The most obvious of these is an
elongated magnetic anomaly situated along the edges of the mound. As
in the case of Zilum and other Gajiganna Phase III sites, the elongated
anomaly is interpreted as being a peripheral ditch, potentially associated

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 577

with a now decayed earthen wall or rampart. With a total length of about
2,200 m, it marks the limits of the main settlement area. While the anomaly
is discernible for several hundred metres along the site’s fringes, the
magnetic contrast between the feature and the surrounding sediment is
in places so low that the ditch seems to be non-existent. This, however, is
not the case. A closer look at the image shows that it is an almost
continuous feature, only interrupted by a series of anomaly-free passages
that were probably access points to the settlement (Fig. 15.4). Two archae-
ological sondages that sectioned the ditch at different locations provided
information on its dimensions and shape. The first sondage (D6), set at the
northernmost margin of the site, revealed that the ditch was a V-shaped
feature with a maximum width of about 4 m and a depth of at least 1.9 m.
The second sondage (D8) at the south-eastern limits of the site showed that
the ditch was likewise V-shaped, but a bit smaller than at sondage D6,
being about 3 m wide and 1.4–1.6 m deep. Altogether, a volume of earth
c.2.1–3.8 m3/1 m section was once quarried from Dorota’s ditch. These
figures are somewhat smaller than those calculated for Zilum and Zubo
(see below). Though it is suggested that a substantial earthen rampart or
wall was once associated with the ditch in the past, there is no clear
geophysical evidence for its presence at Dorota. As Figure 15.4 shows,
there are no anomaly-free areas along the interior border of the ditch,
which should be the area where such a structure once stood. A probable
explanation for this relates to the chronology of the ditch digging.
Extensive zones at the edges of the site, especially at the northern, western
and southern portions, have the same pattern of magnetic anomalies as
those visible within the enclosure. The location of those anomalies suggests
that the settlement once comprised areas beyond the ditch, which appears
to be a secondary construction. In this respect, a closer look at the magnetic
plan reveals that the ditch was quarried after the pits in the vicinity of the
ditch were dug. This strongly suggests that the ditch was only built some
years after the settlement attained its maximum extent.
Apart from their chronological implications, the enormous number and
density of the dot-like anomalies that are interpreted as pits within the
settlement area demonstrate the intensive use of Dorota’s inner space for
activities possibly related to food storage, building and other human
actions such as burying the dead and disposing of domestic refuse. Their
distribution likewise points to the nucleated character of the settlement. In
fact, virtually no open space is left within the area enclosed by the ditch,
suggesting that people must have lived closely together within the settle-
ment. Unlike at Zilum, the longer occupation and the repeated digging of

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 578 Carlos Magnavita

Dorota’s cultural deposits hinder the identification of the location of

former houses and living areas in the magnetic plan. This in turn pre-
cludes calculating the size of Dorota’s population directly from on-site
evidence. If, however, the population density of first-millennium BC
Zilum (140–210 persons per ha) is taken as a measure for estimating
the number of people living at Dorota,47 a figure of 4,700–7,000 people
seems a realistic approximation, especially if compared with those for
historically known settlements with similar physical characteristics (see
discussion below).
The analysis of the magnetic plan and an inspection of the terrain
suggest that some craft activities like iron working and the firing of
pottery may have taken place outside the area confined by the ditch
(see Fig. 15.4). Unlike at Zilum, no evidence of occupational differentia-
tion could be discerned in the few excavations carried out at Dorota, but
it is hardly imaginable that a settlement with a population numbering in
the low thousands would lack people working as specialists, at least in
a part-time or seasonal manner.
The second settlement, Zubo, is dated to the first to third century AD.
Located c.10 km north of Dorota, the site consists of a sizeable, low
mound widely surrounded by clay depressions. Research at Zubo
involved extensive geophysical and topographical surveys and a test-
excavation. The geophysical survey encompassed a total area of 41.5 ha
and provided information on the location of the major archaeological
features as well as the size and shape of the settlement, though c.12.5 ha
at its centre remains unexplored. The geophysical plan (Fig. 15.5) shows
an overview of the site’s subsurface extent. As at Dorota, the most
conspicuous magnetic anomaly is a long subsurface archaeological fea-
ture that roughly marked the site’s limits. Identified as a ditch, this
feature has a total length of around 2,600 m and enclosed an area of
about 46 ha. It is interrupted by a small number of breaches, probably
access points, particularly at the settlement’s south-eastern limits.
Another similar anomaly, with an estimated length of 1,300 m, is visible
in the north-western central part of Zubo. Encompassing a smaller area
of the site (approximately 18 ha), it partly runs parallel to the peripheral
feature, joining the latter at two distinct points along the settlement’s
western and northern sides. The smaller ditch was obviously constructed
sometime after the larger one, and as in the case of Dorota, both ditches
seem to have been dug after a large part of the settlement had reached its

47
Magnavita, C. et al. 2009, 51.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 579

700

650

600 Earth road

550

500
Scatter of iron slag
450

400 3
Furnaces? 2.5
350 2
1.5
300 Test 1
trench 0.5
250 Access point 0
–0.5
200 –1
–1.5
Scatter of iron
150 –2
slag-
–2.5
100 –3

50

–50

–100

–150

–200
N
–250

–300

–350

–400

–450
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750

Figure 15.5. Magnetogram of Zubo showing the location of features discussed.

maximum extent. This is indicated by continuous scatters of dot-like

anomalies, which in several places are apparently cut by the ditches.
However, a few locations such as the southern and northern extremities
of the settlement seem to display some anomaly-free areas along the
interior border of the ditch. Contrary to former assertions,48 the
48
Magnavita, C. et al. 2009, 55.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 580 Carlos Magnavita

anomaly-free sectors at Zubo can be taken as evidence that its ditches

were once accompanied by now decayed earthen walls/ramparts. A test-
cutting made at the south-eastern sector of Zubo’s peripheral ditch
revealed a massive V-shaped feature with a width of 4.5 m and a depth
of about 2.3 m. Those figures indicate that a total soil volume of about
5 m3/1 m section was once quarried, suggesting the construction of
a sizeable structure associated with the ditch.
As at Dorota, both features enclose numerous dot-like magnetic anoma-
lies that most likely indicate the location of former pits, the functions of
which certainly encompassed activities such as grain storage, clay extrac-
tion for building purposes, refuse disposal and interments. Similarly to
Zilum and Dorota, their large number and density express the intensive use
of space and the nucleated character of the settlement. As with Dorota, no
houses or living spaces can be unmistakably identified from the geophysi-
cal plot available, hence, there is no direct archaeological evidence that can
be used for estimating the resident population of Zubo. Based on the
population density (c.140–210 persons/ha) calculated for the c.500 years
older fortified settlement of Zilum, however, we can suggest that the
number of people living in the settlement may have ranged from about
6,500–9,700 individuals.
The only evidence for craft activities performed at Zubo relates to iron
working. The eastern periphery of the site is characterised by a wide scatter
of iron slag fragments, indicating that iron was being worked in that area
(Fig. 15.5). Another extensive scatter of iron slag, visible both on the
magnetic plan and on the surface, is likewise located inside the outer
ditch in the southern sector of the site. This area is furthermore related
to large, strong magnetic anomalies that possibly indicate the presence of
buried iron-working hearths or forges. Considering the large resident
population, it is again reasonable to suppose the existence of at least
seasonal occupational differentiation among the inhabitants of the settle-
ment. In particular, crafts that required expertise, such as the production of
large storage vessels, might conceivably have been performed by skilled (or
at least part-time) specialists. However, as at Dorota, direct archaeological
evidence remains elusive.
An interesting point to be made relating to Zubo and Dorota is that,
unlike at Zilum, field surveys were unable to locate small- or medium-sized
satellite sites in their immediate surroundings. Provisionally, the lack of
satellite settlements around Zubo and Dorota could be considered a matter
of visibility and as an artefact of the surveys carried out around the sites.
Future fieldwork in the area may certainly change this initial impression.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 581

The significance of this and other facets of the archaeology of the first
millennia AD/BC fortified settlement at Lake Chad are discussed below
within the scope of explanations of the origin of urbanisation and the state
in the Lake Chad region and in Sub-Saharan Africa.

Origins of Urbanisation in the Lake Chad Region

and Sub-Saharan Africa

Other than livestock-keeping, the cultivation of plants was, in many parts

of the world, the critical step that enabled a sustained process towards the
development of urbanisation. The archaeological sequences now avail-
able for the Dhar Tichitt region of Mauritania, the Macina region of Mali
and the Chad Basin demonstrate that only about 500 to 1,000 years
elapsed between the initial emergence of plant cultivation locally
and the first appearance of a fundamental social feature associated with
later West African urbanisation: hierarchically organised settlement
clusters.49 The archaeological record also shows that other social and
economic traits later linked to mature urbanisation made their initial
appearance at a relatively early date in West Africa (c.1600–500 BC).
These include the relatively large size of central settlements, compara-
tively large resident populations and (at least seasonal) occupational
differentiation of some persons. Most of those traits, while occurring
early in Mauritanian, Malian and Chad Basin communities, did of course
not materialise on the scale and complexity that later was commonplace
in and around late first-millennium AD Jenné-jeno: the earliest and best
researched archaeological instance of a fully urban settlement and its
hinterland in West Africa.50
The hierarchical settlement clusters of Dhar Tichitt (Mauritania),
Macina (Mali) and the Chad Basin may not be labelled by most researchers
as urban, as there is indeed no convincing evidence yet that their central
settlements were, by definition, providing specialised services and goods
for its hinterland.51 However, in keeping with the questions debated in this
volume, the issue over where the threshold lies between non-urban and
urban settlement clusters is open to debate. Or, to put it more precisely,
what is the scale and how diverse must be the services and goods provided

49
MacDonald 2013; Magnavita, C. et al. 2006; McIntosh, S. and McIntosh, R. 1993.
50
See McIntosh, Chapter 14, this volume. 51 McIntosh, S. and McIntosh, R. 1993, 625.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 582 Carlos Magnavita

by a central settlement to its hinterland for prehistoric settlement clusters

to be recognised as urban?
Despite inherent problems with the practical use of definitions in
archaeology in general, what can be indisputably stated is that the early
archaeological instances mentioned above add to the increasing aware-
ness that some of the fundamental characteristics of late urbanisation in
Sub-Saharan Africa appeared very early in the West African archaeolo-
gical record. In fact, those instances remind us that the rise of the fully
urban settlements of late first- and second-millennia AD West Africa
were preceded by centuries of social, economic and political processes
that eventually led to the emergence of what might be seen as the
forerunners or even the earliest stages of urbanisation in Sub-Saharan
Africa.52 Whether the early hierarchical settlement clusters from
the three regions mentioned are regarded as such or not, two major
conclusions can be drawn. The first relates to the cause or causes leading
to the rise of those hierarchical settlement clusters. The other is linked
to their autochthonous versus allochthonous nature and, by extension,
the autochthonous versus allochthonous nature of Sub-Saharan Africa
urbanisation in general.
Whilst the reason(s) for hierarchical settlement clustering in the
Mauritanian–Malian borderland remains contentious,53 the cause of simi-
lar aggregations in the Chad Basin seems related to a greater concern with
security. Considering the fact that each of the settlements investigated was
once surrounded by ditches and, probably earthen walls or ramparts,
protection seems to have been a major factor in the emergence of
Zilum’s site clustering. In this sense, both nucleation and site aggregation
must be seen as human responses or strategies for coping with imminent or
latent threats. It is much more probable that local, rather than long-
distance factors,54 were the major contributor in instigating the mixture
of nucleation, fortification and site clustering detected in the Gajiganna
area. Possible triggers for that are disputes for land between settled farmers,
conflicts for resources between nomadic cattle keepers and settled farmers
or banditry and raids.55
For decades, the emergence of urban life in Sub-Saharan Africa was
explained as resulting from the direct or indirect influence exercised by
Arab-driven long-distance Trans-Saharan trade on Sub-Saharan peoples.
Since the 1980s and 1990s at least, when archaeological research was
52
McIntosh, S. and McIntosh, R. 1984, 89–90.
53 54
See discussion in MacDonald, Chapter 13, this volume. Magnavita, C. and Breunig 2008.
55
Magnavita, C. and Breunig 2008, 77.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 583

conducted by Susan and Roderick McIntosh at Jenné-jeno and its

surroundings in the Inland Niger Delta of Mali, it became clear that that
was not the case. Instead, the picture that emerged was one of a gradual
development of urbanisation that pre-dated Arabic Trans-Saharan trade
and possibly derived from, amongst other factors, increasing local and
regional interaction and commerce along the Niger River. The evidence of
developments leading to the emergence of hierarchical settlement clusters
in three different West African regions (Mauritania, Mali, Chad Basin)
preceding the events around Jenné-jeno by several centuries clearly under-
lines the view that Sub-Saharan Africa urbanisation resulted from local and
regional processes rather than from long-distance (North African) influ-
ences. This notion is further supported by the virtual paucity of archae-
ological finds from extraneous (North African or Nile Valley) origin in
those and other Sub-Saharan Africa sites until the mid-to late first
millennium AD and, locally, perhaps even much later.56

Origins of the State: A Hypothesis for the Lake Chad Region

Just as with urbanisation, the foundation of the earliest West African states
and empires were for a long time accredited to foreigners or to their direct
or indirect influence.57 In attempting to detect ‘the early state’ or ‘the
process of state formation’ from archaeological remains, archaeologists
face the problem that the identification of political institutions and deci-
sion-making individuals from material evidence is rarely straightforward.
For instance, in the absence of highly visible signs of status differences
assigned to state-level societies, trying to infer the level of socio-political
integration of non-literate human groups from other kinds of archaeolo-
gical data – such as the presence/absence or scale of site hierarchies – is
regarded as both conceptually and methodologically problematic.58
Similarly, the almost obsessive attempts of researchers to fit prehistoric
societies into simplistic neo-evolutionary ‘categories of human progress’
(band, tribe, chiefdom and state) have been highly criticised for their
correctness and questionable value for archaeology as a whole.59 Instead
of over-emphasising classification, it has been advocated that archaeolo-
gists would be better off doing what they do best: evidencing diachronic
56
Gronenborn 2001a; MacEachern et al. 2001; Magnavita, C. and Breunig 2008; see also Connah
1981.
57
Compare the discussion in Munson 1980, 458–59. 58 Cowgill 2004, 543.
59
McIntosh, S. 1999; Yoffee 2005, 31–32.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 584 Carlos Magnavita

changes in the archaeological record over lengthy time spans and trying to
explain those changes in terms of local and regional economic, social and
political trajectories.60
Notwithstanding whether (and which) labels are used to describe pre-
colonial human groups, the fact is that highly hierarchical and stratified
(state-level) societies were a historical reality in the Chad Basin and West
Africa in the late first to second millennium AD.61 It is also certain that
the latter undoubtedly developed out of societies that were relatively
undifferentiated socio-economically and were politically structured pri-
marily by kin-relations. Following the above mentioned methodological
and theoretical principles, the ultimate goal of archaeologists and histor-
ians interested in understanding the emergence of stratified (state-level)
societies is, therefore, to reveal the specific developmental trajectories
of the latter. In the Chad Basin, this can be done by considering the
socio-political developments of the late first and second millennia AD,
when stratified societies demonstrably emerged. That glimpse can serve
as a basis for achieving an understanding of the forces that may explain
both the origins of early state societies in the Chad Basin and the
emergence of first millennium BC/AD sites such as Zilum, Zubo and
Dorota, described above.
Written and oral historical sources suggest that at least two types of states
emerged in the Lake Chad region at the end of the first and the beginning of
the second millennium AD. The first and older of these appears to be
Kanim-Borno (from around the eighth century AD), a regional expansio-
nist state whose core area was first situated at the north-eastern and, later, at
the north-western margins of Lake Chad.62 There is strong evidence that
Kanim-Borno’s early economic and political success is tightly linked to the
early Arabic Trans-Saharan trade, but its socio-political roots may predate
those large-scale commercial linkages.63 The archaeology of early Kanim-
Borno remains elusive as its early administrative centres do not seem to
have been densely populated cities.64 Only later, from the fifteenth century
onwards, did its putative capitals (the contemporaneous sites of Birni
Gazargamo and Garumele) develop into large fortified urban centres dis-
playing monumental architecture.65 The second and later state-level society
is represented by the series of small independent Kotoko (or Makeri)
principalities or city-states on the southern margins of Lake Chad.66
Dated by oral traditions, king lists and archaeology as emerging in the
60
Yoffee 2005, 5. 61 Forkl 1983; Hunwick 1985; Lange 1977; Levtzion 1973.
62
Barkindo 1985. 63 Lange 1977, 141–43. 64 Gronenborn 2001b: 103–4.
65
Haour and Gado 2009. 66 Forkl 1983.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 585

Figure 15.6. Aerial view of Goulfei.

time between the twelfth and the fourteenth century AD, each of those
polities was characterised by a single fortified town (Fig. 15.6 for the
example of Goulfei) and a territory of one to several hundred km2.67
Alongside their perimeter walls, architectural evidence for the existence
of a central government comes in the form of a palace. Just as the extent
of the territory (and villages) controlled by each town provides evidence
of the ruler’s power, so do the size of the town and its population. In
this context, the walled town of the Houlouf principality (twelfth–
sixteenth century AD) and those of Goulfei and Logone-Birni attained
sizes of about 16, 20 and 57 ha, respectively.68 The population size of
Houlouf at its political climax in the early sixteenth century AD is
unknown, but Goulfei’s and Logone-Birni’s populations were estimated
to have respectively amounted to about 8,000 and 15,000 individuals in
the second half of the nineteenth century.69
Unlike the case of the regional state of Kanim-Borno which appears to
have been non-urbanised in its early days, the Kotoko principalities seem
to have been essentially urban from their beginning, thus justifying the

67
Lebeuf 1981, 213. 68 Griaule and Lebeuf 1951, 1; Holl 2002, 139; Lebeuf 1969, figure 8.
69
H. Barth and G. Nachtigal in Lebeuf 1969, 30.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 586 Carlos Magnavita

label city-state for the latter polities. The question that has to be posed in
this context is why did early Kanim-Borno not develop urban centres,
while the Kotoko did? Ninth-century and later Arabic written sources
describing the involvement of Kanim-Borno in the Trans-Saharan
trade suggest that the main commodity given by that state in exchange
for products imported from North Africa was slaves.70 While in the
ninth–tenth century AD the supply of slaves may have been guaranteed
by raids carried out in or around the territories under the state’s
dominions,71 it appears that such targeted attacks extended later into
independent regions. Thus, the areas of the southern margins of Lake
Chad appear to have been attacked for such purpose at the latest from
around the eleventh–twelfth centuries AD onward.72 In the thirteenth
century several expeditions are also said to have been carried out against
areas around Lake Chad and between the thirteenth and fourteenth
century the western margins of Lake Chad (Borno) appear to have
been under Kanim-Borno authority.73 The available evidence thus lar-
gely suggests that political pressure exerted by Kanim-Borno on south-
ern Lake Chad populations alongside intercommunity conflicts were
probably main factors in the emergence of the urbanised Kotoko princi-
palities historically known from the twelfth century on.74
It is today largely acknowledged that the large majority of the earliest
state-level societies worldwide were not large and expansionist territor-
ial states. Instead, they consisted of comparatively small polities centred
on single urban settlements, which, greatly differing in size, complexity
and wealth, were variously called city-states, town-states, petty states,
micro-states or statelets.75 A major factor leading to the emergence of
the earliest city-states is regarded as something called ‘defensive coales-
cence’, that is, the convergence of people to a central settlement to seek
protection against raids from neighbouring polities.76 In fact, city-states
did not rise in isolation, but in clusters from a few to several dozens
and hundreds. The analysis of several archaeological cases suggests
that their emergence was intimately linked both to competition and
hostility among nascent peer-polities,77 but also by alliances among
some of them. It is within such a scenario of external threat that the
southern Lake Chad Kotoko city-states emerged around the twelfth
century AD.
70
Al-Yaqubi in Levtzion and Hopkins 1981, 22.
71
Al-Yaqubi and Yaqut in Levtzion and Hopkins 1981, 22, 171. 72 Barkindo 1985, 235.
73
Barkindo 1985, 238–39; Lange 1993, 270–72. 74 Gronenborn 1998, 246–47; Holl 2002, 213.
75
Cowgill 2004, 542; Gat 2002, 125; Yoffee 2005, 44. 76 Gat 2002, 125. 77 Gat 2002, 129.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 587

As in the case of Zilum, the first millennium AD fortified settlements

of Dorota and Zubo may be, by definition, hardly labelled as urban, the
more so as we do not know if they even had satellite settlements.
Similarly, the question of the kind of political organisation of their
inhabitants cannot be fully answered, as reliable archaeological evidence
for that is as yet lacking. However, Zilum as well as Dorota and
Zubo display a number of physical traits (fortifications, dimensions
and population sizes) that are strikingly similar to those of the
Kotoko central settlements. The rise and development of the second
millennium AD Kotoko settlements are, of course, chronologically and
culturally unrelated to those of first-millennia BC/AD Zilum, Dorota
and Zubo. Nevertheless, the archaeological evidence recorded at these
fortified settlements attests that the human groups that inhabited the
region must have experienced comparable levels of socio-political pres-
sure generated by external threats to those suggested as triggering the
rise of the Kotoko city-states. Whether people sought protection due to
warfare among peer-polities, due to attacks of an expansionist state or
merely due to the insecurity caused by marauding bands, the now
available archaeological evidence points out that real or latent violence
was for centuries and millennia a central feature of social life and politics
in the Chad Basin. The first millennia BC/AD archaeological settlements
discussed do not yet show evidence of the kind of political organisation
(consensus-based decision within groups or formalised central authority)
that had to face and manage the consequences of external threat.
Nevertheless, their existence demonstrates that the first historically
known state-level societies in the Lake Chad region – both eighth
century AD Kanim-Borno and the twelfth century AD Kotoko principa-
lities – postdate centuries of social, economic and politic ‘experimenta-
tion’ that produced densely populated physical and social spaces that were
intriguingly similar to the central settlements of second millennium AD
Kotoko city-states.
These parallels allow us to draw three major conclusions. First,
far-reaching economic and socio-political developments (the foundation
of the densely populated fortified settlements of Zilum, Zubo and Dorota)
that, in favourable circumstances, might have paved the way towards the
rise of state-level societies appear to have been recurrent events from the
middle of the first millennium BC to at least the middle of the first
millennium AD. Second, those early economic and socio-political devel-
opments occurred several centuries before Trans-Saharan contacts and
trade might have had any significant influence on Lake Chad Basin

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 588 Carlos Magnavita

human groups, thus strongly indicating that they were entirely autochtho-
nous developments. Third, and perhaps most important, the origins of
early urban centres in the Chad Basin appears to be strongly linked to the
rise of state-level societies, namely those labelled by some as city-states.
Regarding the latter, it can be hypothesised that the political structure of
the Kotoko states and their possibly less complex ancestral counterparts,
may have developed years or decades after the formation of their walled
towns. That would mean that state-like political structures and its symbols
arose in the wake of the process of nucleation or ‘defensive coalescence’
devised in the archaeological and historical record of the Chad Basin in the
course of the last 2,500 years.

Summary and Conclusions

Looking back at the relevant archaeological and historical evidence now

available for the Lake Chad region in the course of the last 2,500 years, it
seems that a first hypothesis or local trajectory that might account for the
origins of urbanisation and the state in the Lake Chad Basin can be put
forward.
The most recent archaeological data demonstrate that the emergence
of relatively large, densely populated (nucleated) and fortified settle-
ments were a conspicuous feature in the economic, social and political
trajectory of Lake Chad societies. Though smaller sites did exist, some
even figuring as satellites of the larger ones, the largest sites in particular
and their fortifications tell a story that may explain their own rise. That
story highlights that, in the first millennia BC/AD and for reasons that we
do not fully understand, human groups inhabiting the south-western
Lake Chad margins were under recurring threat. This threat, be it real or
merely perceived, led them to gather together in large settlements around
which they built protective ditches and earthen ramparts/walls.
Altogether, their relatively large areas, fortifications and population
estimates in the low thousands bear a strong resemblance to the walled
central settlements of the Kotoko principalities or city-states of the
early second millennium AD. Those polities are the earliest urbanised
state-level societies presently known in the Lake Chad region. Though it
is clear that, chronologically and culturally, the Kotoko city-states are
unrelated to the fortified settlements of the first millennia BC/AD, the
physical similarities (fortification, dimensions and population size) of
those central social spaces are so remarkable that at least a common

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 589

factor might be seen as having triggered their emergence. Local historical

and oral sources, the analysis of the circumstances leading to the appear-
ance of city-states in other regions of the world as well as the results of
more recent archaeological and geophysical research presented in this
paper reveals that that common factor was various forms of local and
regional threats and conflicts, either real or latent.
Just as historically attested regional or local conflicts were able to
trigger the appearance of the Kotoko city-states, so it is justifiable to
suggest that, by analogy, the economic and socio-political developments
around the foundation of the densely populated fortified settlements of
Zilum, Zubo and Dorota might also have led, under propitious circum-
stances, to the rise of state-like polities. Though there is no archaeolo-
gical evidence that hints at the kind of political organisation of their past
inhabitants, the sheer size of those settlements, their early dates and the
paucity of archaeological finds that demonstrate external influences
largely support the suggestion that they resulted from local economic,
social and political processes. Finally and again based on the archae-
ological evidence, it can be suggested that the emergence of the first
urban centres in the Lake Chad region occurred almost in tandem with
the rise of state-level societies, with state structures possibly appearing
just years or decades after settlement growth.
While the archaeological evidence appears to suggest that this trajec-
tory might have led to the formation of the first urban centres and, in
fact, the first urbanised states in the Lake Chad Basin, it is also plausible
that similar conflict driven trajectories may have led to the emergence of
comparatively compact central and defended settlements elsewhere in
Africa. This hypothesis adds to the growing awareness among
prehistorians78 that the processes and trajectories linked to the origins
and development of Sub-Saharan Africa urbanisation may have been
more diverse than hitherto thought.

References

Ajayi, J.F.A. and Crowder, M. (eds). 1985. History of West Africa. Vol.1. Harlow:
Longman.
Barkindo, B. 1985. The early states of the Central Sudan: Kanem, Borno and some
of their neighbours to c.1500 AD. In Ajayi and Crowder 1985, 225–54.

78
See MacDonald and Camara 2012, 186–88.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 590 Carlos Magnavita

Bedaux, R., MacDonald, K.C., Person, A., Polet, J., Sanogo, K., Schmidt, A. and
Sidibe, S. 2001. The Dia archaeological project: Rescuing cultural heritage in the
Inland Niger Delta (Mali). Antiquity 75: 837–48.
Bedaux, R., Polet, J., Sanogo, K. and Schmidt, A. (eds). 2005. Recherches
archéologiques à Dia dans le Delta intérieur du Niger (Mali): bilan des saisons
de fouilles 1998–2003. Leiden: CNWA Publications.
Breunig, P. 2009. Cultural change in the first millennium BC: Evidence from
Nigeria, West Africa. In Magnavita, S. et al. 2009, 15–26.
Breunig, P. 2013. Pathways to food production in the Sahel. In Mitchell and Lane
2013, 555–70.
Breunig, P. and Neumann, K. 2002. From hunters and gatherers to food producers:
New archaeological and archaeobotanical evidence from the West African Sahel.
In F.A. Hassan (ed.), Droughts, Food and Culture: Ecological Change and Food
Security in Africa’s Later Prehistory. New York: Kluwer, 123–55.
Breunig, P., Neumann, K. and Van Neer, W. 1996. New research on the Holocene
settlement and environment of the Chad Basin in Nigeria. African Archaeological
Review 13.2: 111–45.
Breunig, P., Eichhorn, B., Kahlheber, S., Linseele, V., Magnavita, C., Neumann, K.,
Posselt, M. and Rupp, N. 2006. G(l)anz ohne Eisen: Große Siedlungen aus der
Mitte des ersten Jahrtausends BC im Tschadbecken von Nordost-Nigeria. In H.-
P. Wotzka (ed.), Grundlegungen. Beiträge zur europäischen und afrikanischen
Archäologie für Manfred K.H. Eggert. Tübingen: Narr Francke Attempto Verlag
GmbH & Co. KG, 255–70.
Brunk, K. and Gronenborn, D. 2004. Floods, drought and migrations: The effects
of Late Holocene lake level oscillations and climate fluctuations on the settle-
ment and political history in the Chad Basin. In Krings and Platte 2004, 101–32.
Connah, G. 1981. Three Thousand Years in Africa: Man and His Environment in
the Lake Chad Region of Nigeria. Cambridge: Cambridge University Press.
Connah, G., 2000. African city walls. A neglected source? In D.M. Anderson and
R. Rathbone (eds), Africa’s Urban Past. Oxford: James Currey, 36–51.
Cowgill, G.L. 2004. Origins and development of urbanism: Archaeological
perspectives. Annual Review of Anthropology 33: 525–49.
Creekmore, A.T. and Fisher, K.D. (eds). 2014. Making Ancient Cities: Space and
Place in Early Urban Societies. Cambridge: Cambridge University Press.
Forkl, H. 1983. Die Beziehungen der zentralsudanischen Reiche Bornu, Mandara
und Bagirmi sowie der Kotoko-Staaten zu ihren südlichen Nachbarn unter
besonderer Berücksichtigung des Sao-Problems. Münchner ethnologische
Abhandlungen 3. München: Minerva Publikation.
Gat, A. 2002. Why city-states existed? Riddles and clues of urbanization and
fortifications. In M.H. Hansen (ed.), A Comparative Study of Six City-State
Cultures. Copenhagen: Reitzels Forlag, 125–39.
Gaussen, J. and Gaussen, M. 1988. Le Tilemsi prehistorique et ses abords. Sahara et
Sahel Malien. Paris: Éditions du CNRS.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 591

Griaule, M. and Lebeuf, J.P. 1951. Fouilles dans la region du Tchad (III). Journal de
la Société des Africanistes 21.1: 1–95.
Gronenborn, D. 1998. Archaeological and ethnohistorical investigations along the
southern fringes of Lake Chad, 1993–1996. African Archaeological Review 15.4:
225–59.
Gronenborn, D. 2001a. Beads and the emergence of the Islamic slave trade in the
southern Chad Basin (Nigeria). The Bead Forum 38: 4–11.
Gronenborn, D. 2001b. Kanem-Borno: A brief summary of the history and archae-
ology of an empire of the central bilad-al-sudan. In C. DeCorse (ed.), West
Africa during the Atlantic Slave Trade: Archaeological Perspectives. Leicester:
Leicester University Press, 101–30.
Haour, A. and Gado, B. 2009. Garumele, ville médiévale du Kanem-Bornu? Une
contribution archéologique. Journal of African History 50: 1–21.
Holl. A.F.C. 1993. Late Neolithic cultural landscape in southeastern Mauritania:
An essay in spatiometrics. In A.F.C. Holl and T.E. Levy (eds), Spatial Boundaries
and Social Dynamics: Case Studies from Food-Producing Societies. Ann Arbor:
International Monographs in Prehistory, 95–133.
Holl, A.F.C. 2002. The Land of Houlouf. Genesis of a Chadic Polity, 1900 BC–AD
1800. Ann Arbor: The Museum of Anthropology.
Holmes, J.A., Street-Perrott, F.A., Allen, M.J., Fothergill, P.A., Harkness, D.D.,
Kroon, D. and Perrott, R.A. 1997. Holocene palaeolimnology of Kajemarum
Oasis, Northern Nigeria: An isotopic study of ostracodes, bulk carbonate and
organic carbon. Journal of the Geological Society 154: 311–19.
Hunwick, J.O. 1985. Songhay, Borno and the Hausa states 1450–1600. In Ajayi and
Crowder 1985, 323–65.
Krings, M. and Platte, E. (eds). 2004. Living with the Lake: Perspectives on Culture,
Economy and History of Lake Chad. Köln: Köppe.
Lange, D. 1977. Le diwan des sultans du (Kanem-)Bornu: Chronologie et histoire d’un
royaume africain (de la fin du Xe siècle jusqu’à 1808). Wiesbaden: Franz Steiner
Verlag.
Lange, D. 1993. Ethnogenesis from within the Chadic state: Some thoughts on the
history of Kanem-Borno. Paideuma 39: 261–77.
Lebeuf, A.M.D. 1969. Les principautés Kotoko. Essai sur le caractère sacre de
l’autorité. Paris: CNRS.
Lebeuf, A.M.D. 1981. L’origine et la constitution des principautés Kotoko
(Cameroun septentrional). In C. Tardits (ed.), Contribution de la recherche
ethnologique à l’histoire des civilisations du Cameroun. Paris: CNRS, 209–18.
Levtzion, N. 1973. Ancient Ghana and Mali. London: Methuen.
Levtzion, N. and Hopkins, J.F.P. 1981. Corpus of Early Arabic Sources for West
African History. Cambridge: Cambridge University Press.
Linseele, V. 2007. Domestic Livestock, Subsistence Strategies and Environmental
Changes in Sahelian West Africa during the Past 4000 years: Evidence from
Archaeofaunal Remains. Oxford: BAR, Archaeopress.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 592 Carlos Magnavita

MacDonald, K. 2003. New light on the Tichitt tradition: A preliminary report on

survey and excavation at Dhar Néma. In P.J. Mitchell, A. Haour and J. Hobart
(eds), Researching Africa’s Past: New Contributions from British Archaeologists.
Oxford: Oxford University School of Archaeology, 73–80.
MacDonald, K. 2013. Complex societies, urbanism and trade in the western Sahel.
In Mitchell and Lane 2013, 829–44.
MacDonald, K. and Camara, S. 2012. Segou, slavery and Sifinso. In J.C. Monroe
and A. Ogundiran (eds), Power and Landscape in Atlantic West Africa:
Archaeological Perspectives. Cambridge: Cambridge University Press, 169–90.
MacEachern, S. 1996. Iron Age beginnings north of the Mandara Mountains,
Cameroon and Nigeria. In G. Pwiti, and R. Soper (eds), Aspects of African
Archaeology. Harare: University of Zimbabwe Press, 489–96.
MacEachern, S., Bourges, C. and Reeves, M. 2001. Early horse remains from
northern Cameroon. Antiquity 75: 62–67.
McIntosh, S.K. 1999. Beyond Chiefdoms: Pathways to Complexity in Africa.
Cambridge: Cambridge University Press.
McIntosh, S.K. and McIntosh, R.J. 1984. The early city in West Africa: Towards an
understanding. The African Archaeological Review 2: 73–98.
McIntosh, S.K. and McIntosh, R.J. 1993. Cities without citadels: Understanding
urban origins along the middle Niger. In T. Shaw, P. Sinclair, B. Andah and
A. Okpoko (eds), The Archaeology of Africa. Food, Metals and Towns. London:
Routledge, 623–41.
Magnavita, C. 2002. Recent archaeological finds of domesticated Sorghum bicolor
in the Lake Chad region. Nyame Akuma 57: 14–20.
Magnavita, C. 2004. Zilum – towards the emergence of socio-political complexity
in the Lake Chad region. In Krings and Platte 2004, 73–100.
Magnavita, C. 2008. Studien zur endsteinzeitlichen und früheisenzeitlichen
Besiedlung im südwestlichen Tschadbecken (1300 BC – 700 AD). Unpublished
PhD thesis, Goethe-University Frankfurt.
Magnavita, C. and Breunig, P. 2008. Facts and speculations: A reply to D. Lange’s
The emergence of social complexity in the southern Chad Basin towards 500 BC:
archaeological and other evidence. Borno Museum Society Newsletter 74–75:
71–83.
Magnavita, C., Kahlheber, S. and Eichhorn, B. 2004. The rise of organisational
complexity in mid-first millennium BC Chad Basin. Antiquity 78: 1–17.
Magnavita, C., Breunig, P., Ameje, J. and Posselt, M. 2006. Zilum: A mid-first
millennium BC fortified settlement near Lake Chad. Journal of African
Archaeology 4.1: 153–69.
Magnavita, C., Breunig, P., Ishaya, D. and Adebayo, O. 2009. Iron Age begin-
nings at the south-western margins of Lake Chad. In Magnavita, S. et al. 2009,
27–58.
Magnavita, S., Koté, L., Breunig, P. and Idé, O.A. (eds). 2009. Crossroads/Carrefour
Sahel: Cultural and Technological Developments in First Millennium BC/AD

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 15 First Millennia BC/AD Fortified Settlements at Lake Chad 593

West Africa. Journal of African Archaeology Monograph Series 2. Frankfurt:

Africa Magna Verlag.
Mitchell, P. and Lane, P. (eds). 2013. The Oxford Handbook of African Archaeology.
Oxford: Oxford University Press.
Mohammed, I.M. 1986. The Archaeology of Central Darfur (Sudan) in the 1st
Millennium AD. Oxford: BAR S285.
Munson, P.J. 1980. Archaeology and the prehistoric origins of the Ghana Empire.
The Journal of African History 21.4: 457–66.
Rupp, N. 2005. Land ohne Steine: die Rohmaterialversorgung in Nordost-Nigeria
von der Steinzeit bis zur Eisenzeit. Unpublished PhD thesis, Goethe-University
Frankfurt.
Sabloff, J.A. 1994. The New Archaeology and the Ancient Maya. New York:
Scientific American Library.
Trigger, B. 1972. Determinants of urban growth in pre-industrial societies. In
P.J. Ucko, R. Tringham and G.W. Dimbleby (eds), Man, Settlement and
Urbanism. London: Duckworth, 575–99.
Trigger, B.G. 1989. A History of Archaeological Thought. Cambridge: Cambridge
University Press.
Weber, M. 1958. The City. Translated by D. Martindale and G. Neuwirth.
New York: The Free Press.
Yoffee, N. 2005. Myths of the Archaic State: Evolution of the Earliest Cities, States
and Civilizations. Cambridge: Cambridge University Press.

https://doi.org/10.1017/9781108637978.016 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa
New Historical Focus on the Process of Emergence of a Saharan
State and a Caravan City

chloé capel

Preamble: On the Origins of the State?

The French sociologist Pierre Bourdieu taught us in the early 1990s that to
question the ‘genesis of the state’ is one of the most arduous tasks for the
historian.1 Indeed, in essence, the state defines itself when it is already
accepted and firmly installed. Then it produces instruments of recognition
and domination, including symbolic ones, to confirm the social norm, the
social order that it supervises or even imposes, and to validate this ‘funda-
mental consensus on the social world’ of which it is the guarantor.2 For
instance, it is customary for historians to define a state by the establishment
of taxation, which at first sight would be preliminary in any policy con-
struction. However, Bourdieu, and Weber before him, pointed out that for
this step to occur, it is necessary that the taxing authority has already been
recognised by the social body and that the legitimacy of such an initiative
has already been accepted by society.3 The collection of taxes is a result of
the formation of the state, not its founding and like other means of
conceptualising and structuring political power, it always occurs after the
strengthening and the recognition of that power. This is why the produc-
tion of these bureaucratic instruments can take place after, or even much
later than, the birth of the state, whether they are texts (constitution, legal
powers and rules, received religious texts, historical accounts) or have
a material scope (currency issue, monumental architecture, territorial
and environmental markers). Therefore, there are no contemporary
records of the emergence of the state produced by itself. This is enough
to explain the difficulties of studying such a historical moment. But,
archaeologists and historians who wish to study the birth of the state

1 2 3
594 Bourdieu 2012. Genêt 2014, 101; Lenoir 2013, 134. Genêt 2014, 100.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 595

precisely use these kinds of data and documents – the ones that the state
produces to legitimate its position once firmly established – to build their
arguments. As a consequence, they are continuously prisoners of what
Bourdieu calls ‘the thought of the state’.4 These mental and cognitive
structures, promoted by the state because they strengthen its authority,
are necessarily anachronistic compared to the society in which this power
emerged.5 In order to study the birth of political authority or the birth of
the state, one should reconstitute the social space in which this conceptual
changeover was made possible. Comparing sources from different origins,
including external sources that escape the shackles of the thought of the
state, at least partially, may be a way. But because of the current state of
these sources, this approach is still a challenge and this is why it is so
complex to theorise the birth of states, especially for ancient periods, for
which documents are so scarce.

Introduction: Reconsidering the Foundation of Sijilmasa

Studying the emergence of urban societies in the Medieval Sahara does not
escape the epistemological obstacles raised by Bourdieu. When the histor-
ian is for instance interested in the development of Sijilmasa,6 the most
famous of these northern cities, the only historical sources concerning the
birth of the city are Arabic texts. All of them were written far from the city,
mostly by authors who had never been there and who had no idea of
environmental, social and political conditions of this remote Saharan
place. Above all, they were, consciously or not, part of a political system
of thought centred on Mediterranean Islamised societies and involved in
a state-controlled (re)construction of the history of the city. The only
external data, produced outside of the Islamic scholarly circles, which
could illustrate the state of society on the eve of the birth of Sijilmasa are
archaeological in nature. But to this day, they remain poorly studied and
underappreciated. Studies on the development of the state and the city of
Sijilmasa have therefore not made much progress over the last 30 years,

4
Bourdieu 2012. 5 Lenoir 2013, 132–33.
6
To conform with the volume style, a simplified transliteration of Arabic is used for proper names
(e.g., al-Bakri). Only words commonly used in English comply with fully westernised spelling
(e.g., Sijilmasa, Cordoba). Anglicised nouns and adjectives stemmed from Arabic or Berber
terms do not respect the transliteration rules either (filalian). On the etymology of Sijilmasa, see
Mezzine 1984.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 596 Chloé Capel

since the work of Charles Pellat in 1960, Isma’il Mahmud ‘Abd al-Raziq in
1985 and Mohamed Mellouki in the same year.7
These three studies were predictably and almost exclusively based on the
review of al-Bakri’s text, which is the main reference for understanding the
early periods of the city. Writing in Cordoba, al-Bakri completed his Kitab
al-Masalik wa‘l-Mamalik around 460 AH/AD 1068. But in the note he
devoted to the city of Sijilmasa, his main source of information was the
work of the much earlier Andalusian al-Warraq. This writer was born in
North Africa in 292 AH/AD 904 and probably visited Sijilmasa in his
youth.8 His text, which is now lost but largely revived by al-Bakri, offers
a unique historical depth. It is from al-Warraq that al-Bakri took the
majority of his information about the dynastic power of the first centuries
of Sijilmasa. Indeed, his sovereign list stops in the third quarter of the tenth
century, precisely when al-Warraq was writing his work. The text of al-
Bakri presents several versions to the founding story of the city, which, as
a scrupulous historian, he listed successively (though he declared one
version more likely than others). Al Bakri’s favoured version was the one
of a foundation precisely occurring in 140 AH/AD 757 and being the act of
a nomadic shepherd of the Miknasa tribe. This man, as he was traversing
the area now called Tafilalt (or Tafilalat) managed to unite a number of
Miknasa companions in converting to Islam and to the belief in khariji
faith. From this human and spiritual gathering arose a political authority
and a city. Another version was that Sijilmasa was founded in the aftermath
of the revolt of the suburbs of Cordoba in 202 AH/AD 818, by a Miknasi
man, settled in al-Andalus, but driven out from his city at the end of the
Cordoba rebellion. He brought with him into the Maghrib the art of
blacksmithing and, after settling in Tafilalt, he won the trust and respect
of the Berber shepherds grazing in the area, selling them weapons and tools
of his own production and was recognised as their leader. The city and the
state of Sijilmasa arose shortly afterwards. In both versions, the founders
established a dynastic political system wherein Sijilmasa acquired the status
of an autonomous state, being governed by an hereditary prince. In both
cases, al-Bakri also reported that nomadic people were grazing in Tafilalt
(though he does not name the region in this way) before the foundation of
Sijilmasa. He also described an annual gathering of people, based on
a market fair, this being the same one where the blacksmith used to sell
his production. This gathering would have represented the population base

7
‘Abd al-Raziq 1985, 109–43, 210–29; Mellouki 1985, 51–98, 109–58; Pellat 1960.
8
Gilliot 2002.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 597

from which Sijilmâsa was built. To this double myth, other indications
from al-Bakri must be added, complicating this hitherto fairly linear frame.
First, the Miknasi shepherd, founder of the city in the 140s AH, was
evidently the grandfather of the Andalusian blacksmith, also known as
the founder of the city 60 years later. In another variant, the first of the two
founders for a period transferred authority over his community to a third
character. The reign of 15 years of this latter, very likely not a member of
the Miknasa tribe, did not prevent the return to power of the founder of the
dynasty. This first dynasty is often known as the Midrarids (or Banu
Midrar), named after the Andalusian blacksmith Midrar. Arabic literature
also sometimes refers to this family as the Banu Wasul, named after the
first character, Abu al-Qasim Samgu ibn Wasul, the shepherd and grand-
father of Midrar.9
In this complex history, with multiple sides, most historians adopt al-
Bakri’s position where the story of Samgu, the shepherd and khariji
preacher, seems the most plausible version of the foundation of Sijilmasa.
Midrar’s version is generally excluded mainly for chronological reasons:
this myth would place the development of the city too late as it is men-
tioned in Arabic cosmogonies written in the Abbasid court of Baghdad at
precisely the same time.10 With these references it is customary to consider
that the reputation of the Saharan city, located at the other end of Dar al-
Islam, took quite a long time to reach the Caliphate Iraqi court and that it
was not simultaneous with its founding. Furthermore, the version of
Midrar is also dismissed because al-Bakri provides many indications
about Sijilmasa in connection to the period before the time of the revolt
of the suburbs of Cordoba. It does not make sense to invalidate this
information and consider it as a complete fabrication on the grounds
that the city was supposedly founded later. Among this information,
topographic clues, about the wall and the great mosque for instance, are
valuable for the study of the city and thus incline historians to definitively
rule out the later dating of the foundation of Sijilmasa.
The warning of Pierre Bourdieu about the terms of the study of the birth
of states suggests that we need to look into these foundation myths once
more. Even if they may not report actual historical events, they carry clues
about the societies and the powers that produced these texts. For this
reason, all versions of the Sijilmasa foundation must be taken into account
because they certainly contain many elements clarifying the social space in
9
Al-Bakri 1911, 148–52, 167; 1913, 282–90, 315–16.
10
Al-Mas‘udî 1865, iv.39, working in the mid-tenth century but stemming his information from
Kitab al-zij of al-Fazari, who wrote in the first quarter of the ninth century.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 598 Chloé Capel

which the city emerged. This is why a new reflection on the birth of
Sijilmasa is required. Firstly, al-Bakri must be reconsidered beyond the
simple examination of the historical facts he relates in order to detect social
or cultural elements in connection with primeval times. Secondly, the fact
that al-Bakri got most of his historical information from a writer who was
active in al-Andalus a century before him must be kept in mind. Finally,
original archaeological data have to be included into this new reflection.
These archaeological data represent new information about Sijilmasa: they
are the results of surveys carried out between 2011 and 2013.11 This
considerable renewal of sources of information, in association with the
written sources previously exploited, allows a leap forward in understand-
ing the emergence of the city.

Sijilmasa: Paragon of Triumphant Kharijism in the Maghrib and

Construction of a Proselyte State
Modern historians of Sijilmasa adopt al-Bakri’s point of view explaining
the emergence of the city in terms of sufri proselytism having found in
pastoral lands of the southern Maghrib a fertile ground for its preaching.
The Miknasi Abu al-Qasim Samgu, bearing the khariji fire, managed to
unite around him some nomadic shepherds in Tafilalt. These companions,
freshly converted to the egalitarianism of khariji faith, rapidly formed
a model state, where only the best Muslims are worthy of community
responsibility. Forty men formed the first circle which, in the sufri system,
elected as a leader the wisest of them: ‘Isa ibn Mazid, who was black,
a characteristic sufficiently exceptional for al-Bakri to mention it. This
pigmentary detail emphasises that human qualities and piety were the only
requirements to weigh in the balance of power, without any interference of
social, economic, tribal or ethnic issues. The Sijilmasian state was thus
based on khariji doctrinal exemplarity. But after 15 years of governance,
‘Isa ibn Mazid, severely criticised for irresponsible management of com-
mon affairs and whose skin colour is again highlighted, but now as an
aggravating element of his faults, was violently manoeuvred out of power.
Then, Samgu was called to lead in turn but, breaking with the khariji ideal,

11
My doctoral thesis (2016) fell within a framework of Saharan archaeology initiated in 2009 at
the University Paris 1 Panthéon-Sorbonne under the supervision of Professor Jean Polet. This
doctoral work is also based on the exploitation of the archives of the Moroccan American
Project at Sijilmasa (MAPS), conducted between 1988 and 1998 under the supervision of
Ronald Messier (MTSU). For some initial results, see Capel 2016; 2018; Capel and Fili 2018.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 599

he started a hereditary dynastic cycle of Miknasi princes, which ruled until

it was finally deposed in 366 AH/AD 976 by Maghrawa, allies of the
Umayyads of Cordoba.
Twentieth-century studies emphasise that the growth of this sufri hub
south of the Atlas can only be understood in the light of the religious
situation affecting the whole Maghrib at the same time. The mid-eighth
century was actually a period of considerable development of kharijism,
whether it is Sufri or Ibadi, against the caliphal Kairouanese power. Several
major events followed within the space of a generation: the uprising of
Maysara in 122 AH/AD 739–740 in Tangier, followed by the victory of
Khalid over the Kairouanese forces during the ‘Battle of the Nobles’; then,
after a time of setbacks, such as the storming of Kairouan in 140 AH/AD
757 by Warfajjuma people, invoking sufrism, and the same year the
proclamation of the Ibadi imamate in Tripolitania, which began the con-
quest of the region under the authority of Abu al-Khattab and then,
strengthened by a new wave of proselytism, took the Ifriqiyan capital.12
Thus, kharijism not only emerged as a dogmatic branch of Islam in North
Africa but also as a powerful armed movement and a real political force
confronting the Umayyad and then Abbasid entity of Kairouan. The
foundation of Sijilmasa precisely occurred in this context. But by balancing
this point of view, our work has already pointed out that if this intellectual
and conquering environment was partly responsible for the birth of the
Sijilmasian state, this is probably as much the result of successes of the
movement as defeats or setbacks. These latter could have caused migration
waves of khariji people to distant and protected lands south of the Atlas, as
in Tafilalt, where people, as fighters enduring reprisals following the
assassination of Maysara by his own allies or even Warfajjuma expelled
from Kairouan by the confrontation with the Ibadi, could have found
refuge.13 Thus, Sijilmasa was equally built on a local development of
religion and an influx of population, which contextualises the presence of
Miknasa in this remote region, considered as peripheral to their main areas
of development, which were primarily in the Atlas moutains.

The Myth of Samgu: Hypothesis of a Strong Political Structure in

Tafilalt before Sijilmasa
Without questioning al-Bakri’s text, Tafilalt, the preaching area of Samgu,
is usually considered as a land abandoned to pastoralism in pre-Islamic

12
‘Abd al-Raziq 1985, 48–49, 62, 70, 77; Mellouki 1985, 55–59, 110. 13
Capel and Fili 2018.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 600 Chloé Capel

times. It is portrayed as a region that is relatively hostile and devoid of

natural resources, outside of any state-controlled or authority framework
and with a low and limited population density.14 The foundation of the
Sijilmasian state, though certainly idealised as most historians recognise it,
marks the opening of the region to Islamic social structuring and to an
economic and political expansion era. Without compromising the histor-
ical frame which was recently built about the khariji process in Tafilalt,
a careful reading of al-Bakri’s text and its comparison to archaeological
sources invite a complementary approach to the question of the birth of
power in Sijilmasa.
Historians have questioned the name of the first ruler of Sijilmasa ‘Isa
ibn Mazid, called ‘the black’, stressing that it suggested that political
egalitarianism had led to the emergence of the city.15 But no one has
dwelt upon the name of the preacher Abu al-Qasim Samgu ibn Wasul,
whose ancestry appears predominantly Arab, but also contains a Berber
term: Samgu which also can be spelt Samghû. The SMX root of this name
means in modern Berber ‘slave’, or even ‘black slave’ in a pejorative sense.16
This name perhaps throws light on the history of Tafilalt on the eve of the
emergence of Sijilmasa. The success of kharijism in the Maghrib is usually
attributed to the defense of religious egalitarianism that allowed Berber
people as well as Arab conquerors to claim supreme religious and com-
munity responsibilities. Perhaps one should combine with this reason the
possibility for men to access the highest responsibility whatever their social
status, prince of noble blood or simple slave. It is through the prism of this
possibility of social emancipation, and not just through ethnolinguistic
matters, that we should read the zeal of Samgu, slave or freedman in
Tafilalt. Appropriating Islam and its new intellectual and dogmatic uni-
verse allowed low social classes, Samgu among them, to reverse the logic of
the ‘natural’ authority structuring the social body in the region. Thus, the
name of Samgu is certainly a window on filalian society (of Tafilalt) and its
forms of power at the time of emergence of Sijilmasa.
Therefore, the origins of ‘Isa, the first governor of the city, coupled with
the assumption of the servility of Samgu can be reconsidered on a new grid
of analysis. If ‘Isa ‘the Black’ was elected at the head of the young sufri
community, if the authority was ultimately entrusted to him to the detri-
ment of Samgu the spiritual figure, maybe it is not so much for purely
ethical or religious reasons (to the credit of khariji egalitarianism). Instead,

14
‘Abd al-Raziq 1985, 113; Mellouki 1985, 55–59.
15
‘Abd al-Raziq 1985, 116; Mellouki 1985, 111. 16
Amanis 2009, 493.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 601

it was much more for strategic and pragmatic aims, in order to respect
a local social order, which was firmly rooted and where Samgu, the slave of
subordinate condition, could not aspire to any upper position or social
power in this society. By installing ‘Isa at the head of the community, the
new circle of believers led by Samgu would only have respected and have
submitted to a power structure effective in Tafilalt before the establishment
of kharijism. This authority, embodied by ‘Isa, was possibly exercised not
by Berber people, but by another sociolinguistic group and was based, at
least partly, on a pastoral economy that included servile schemes. The
prevalence of this authority in the region was certainly so important that
a new social system, based on khariji faith, could not overcome it easily.
Although the filalian society, or at least a part of it, may have been won over
kharijism and Islam, it preserved its old political system, at least for a time.
Consequently, the social and economic context in which Samgu asserted
himself was probably not as poor as historians suggested it and the
presence of a hierarchical society with a strong political authority estab-
lished in pre-Islamic Tafilalt should be reconsidered. It is often argued that
Samgu was from a more northerly region, which would explain his
hypothetical involvement in the uprising of Maysara in 122 AH/AD 740,
as proposed by historians, and his intellectual formation in Kairouan with
the famous preacher ‘Ikrima, a figure of oriental kharijism, as al-Bakri
relates. We already emphasised the historical weakness of such a biography
and underlined the logic of such an intellectual parentage towards the
requirements of power and recognition of the Midrarid authority.17 These
biographical elements illustrate the structuring instruments that the state
creates a posteriori as Bourdieu analysed them. To push further our argu-
ments, we propose to see Samgu as a filali, that is a person from Tafilalt,
fully integrated into the social structure of his region, but taking part in
a considerable change in the established order by the introduction, or at
least the spread, of kharijism south of the Atlas. Samgu was not preaching
alone in the desert but he was instead part of a strong human settlement
with a powerful political organisation, whatever the nature of this organi-
sation was. It was this community that had to accept the changes of the
‘social consensus’. This struggle required a minority group to respect the
well-established structure in order to gradually instill new religious and
political aspirations without overthrowing the established order. This
would have been considered as an affront and would have represented
a potential source of violent response. Maintaining an already recognised

17
Capel and Fili 2018.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 602 Chloé Capel

local leader at the head of the state was part of this strategy, it ensured social
peace but did not prohibit progressive lifestyle changes or even a more
radical overthrow of the system at a later date. By stepping aside in favour
of ‘Isa, Abu al-Qasim Samgu only respected the authority ‘Isa already
enjoyed within the society. The final fall of ‘Isa, which occurred 15 years
later, probably resulted to a shift in the local power balance due to the
progressive arrival of khariji immigrants from the north, following the
setbacks of their leaders in major battles of the 140s AH. It probably
provoked a deep swing in attitudes and social habits. Therefore, Sijilmasa
was probably not born ex nihilo by structuring isolated and unattached
shepherd communities, but by the gradual transformation of a local
authority.

The Myth of Midrar: Rereading the Parable of the Blacksmith

in Light of the Filalian Context
The second version of the founding of Sijilmasa, involving the rebels from
Cordoba in the birth of the city, has been ruled out of the debate until now
because of its improbability. But rereading this text actually provides
information of major importance and would enlighten the history of the
city. To this day, this text has never been read in the light of its context of
production, though this probably represents the key to its interpretation.
Al-Bakri gets this myth, like the majority of his information about
Sijilmasa, from al-Warraq. But al-Warraq, who died in 363 AH/AD 973
in Cordoba, exercised his talent at the court of the Umayyads of al-
Andalus, which would not have been without any effect on his work. He
composed his own Kitab al-Masalik wa-l-mamalik at the end of his life on
behalf of the caliph al-Hakam.18 This intellectual and political context
alone would justify the omnipresence of Cordoba, and the real or imagined
involvement of Cordoba, in many historical events in the Islamic West, as
presented by al-Warraq. But in more concrete terms, it should be noted
that it is precisely during the reigns of al-Hakam and his father ‘Abd al-
Rahman, that the Umayyads significantly extended their authority in the
Maghrib, to the detriment of the Fatimids, their biggest political rivals.
They first took Fez and then Sijilmasa in 366 AH/AD 976, in both cases
through Maghrawa clans, opposed to the Fatimids, but loyal supporters of
the caliphs of Cordoba. In Sijilmasa, Maghrawa armies overthrew the
founding Midrarid dynasty. This Cordoban control on Sijilmasa and the

18
Gilliot 2002.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 603

will to fully integrate the Saharan site in the caliphate could justify the
production of a new foundation myth giving Cordoba a significant respon-
sibility, even if it was indirect, in the development of the city. One should
notice that Sijilmasa was not the only city to attach an important role to
Cordoba in its founding myths. For instance, the same process was adopted
in Fez, where the name of an Andalusian district was explicitly linked to the
migration which occurred in the aftermath of the revolt of the suburbs of
Cordoba. Although the first mention of this fassi (Fez) myth is quite late
(fourteenth century), the district’s name dates from the Umayyad period at
least.19 However, neither al-Warraq nor al-Bakri after him, reported the
Maghrawa conquest of Sijilmasa in their chronicles, simply because the
first of these Andalusian authors was already dead at this time. It thus
follows that the invention of the myth of the Andalusian foundation of
Sijilmasa post-dated the effective conquest of the city. Yet this does not
invalidate its origin since the expansionist policy of the Umayyads, and
their justifications, date back to the early days of the caliphate. Historical
and genealogical references to Corboda and al-Andalus are numerous in
literature from the western Islamic world, whatever the period. The prac-
tice goes beyond the case of Sijilmasa and therefore the passage from al-
Bakri’s text must not be retained for itself. But it must not be ruled out
either because if it does not actually inform on the period of the city’s
emergence, it is nonetheless a valuable document for understanding its
history in the tenth century.
If we leave aside the origin of this founding myth and its doubtful
chronology, it is the theme of the blacksmith that is of interest. The myth
of the blacksmith founder was considered unlikely by al-Bakri. He related
the story for documentary rigour, but then dismissed its historicity.
Historical literature of the twentieth century has followed his lead and
considered it to be an unclear anecdote, devoid of any real historical
substance.20 For one thing, the social function of the blacksmith was not
so innocuous as a trader or a peasant. It was a despised activity in Medieval
Islamic culture, especially in the Middle East, where it was associated with
impurity and dishonesty, relegating the blacksmiths to the bottom of the
social scale. Considering a blacksmith to be the founder of a dynasty
tarnished the prestige of a power seeking for political recognition.21
However, blacksmithing did not have the same connotations on the
African continent, especially in the Saharo-Sahelian area. Here, even if the

19
Lévi-Provençal 1938. 20
‘Abd al-Raziq 1985, 114; Mellouki 1985, 70; Pellat 1960.
21
Chelhod 1978.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 604 Chloé Capel

smithy is usually synonymous with some social constraints and social

isolation, the blacksmith could be esteemed for his know-how and his
technical superiority that allowed him to be closely associated with the
symbolic world in many Sahelian societies. This form of power gave him
the position of a well-advised and shrewd man whose social position could
be related to a privileged caste or an intellectual activity.22 Thus, the
blacksmith parentage has been invoked repeatedly in the genealogies of
major African dynasties, especially among Mandé populations, including
the case of the Soso Kingdom in the twelfth century.23 From this new
perspective, the association of the founder of Sijilmasa with a blacksmith is
understandable if the emergence of this political structure was in an
African, not an Arabic, cultural environment. This is an altogether logical
situation in the eighth century south of the Atlas. Al-Bakri related an
historical version that was obviously totally incomprehensible for an
Arabised author like himself. This text reinforces the hypothesis of
a ‘black’ power at the beginnings of Sijilmasa, strongly rooted in local
social stratification and indigenous political references, and not so much
on the eastern ideology of kharijism that was still little known in the region.
A historical detail about al-Warraq is also meaninful. Al-Warraq was born
in North Africa and travelled in the Maghrib in his youth, perhaps even to
Sijilmasa, which was the main topic of a monographic text he wrote in al-
Andalus (but which is now lost).24 Al-Warraq could have collected and
understood such a mythical tale in Tafilalt. Thus, the theme of the black-
smith certainly has a genuine ideological and social foundation and per-
haps it reveals a lot about the real context of the emergence of the filalian
state, in this non-orientalised social body, which was probably not even
Islamised.

Environmental Survey: The Hypothesis of a ‘Salt Cure’

in Tafilalt before Sijilmasa

As mentioned above, Medieval authors and contemporary historiography

have not considered the area where Sijilmasa emerged as having any pre-
existing form of political power, let alone being a proto-state, before the
foundation of the city. Tafilalt, in the account of al-Bakri, was described as
a vast plain with favourable pastures for nomadic activities, but hostile and
lacking any permanent settlement and territorial development. However,

22 23 24
Heusch 1956, 62. Dieterlen 1964, 13; Levtzion 1973, 51. Gilliot 2002.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 605

in the light of the previous arguments, the hypothesis of a well-established

and structured political power in Tafilalt before the foundation of Sijilmasa
must not be ruled out. Nor can we exclude the idea of the presence of
a significant, or even dense, population in the region. Several archaeologi-
cal arguments tend to support this option. Indeed, the large number of pre-
Islamic monuments known in the Tafilalt is a very visible sign of human
occupation before the rise of Sijilmasa.25 All around the current oasis
occupying the heart of the Tafilalt plain where stand the ruins of
Sijilmasa, are thousands of funeral structures, mostly pre-Islamic tumuli.
The area of Erfoud, about 20 km north of the archaeological tell of
Sijilmasa, is the only place where they have been studied archaeologically,
but there are many other spots with funeral monuments standing within
short distances of the Medieval site.26 Unfortunately, these graves are not
dated, but Jean Margat believes that many of them only slightly preceded
the emergence of Sijilmasa. He even proposes the possibility that such
funeral rites survived the early days of the introduction of Islam.27
Populations were necessarily associated with these burial practices and
thus were established in the region, at least seasonally.
This human concentration in Tafilalt can find several explanations. First,
although it is a desert region, it is favourable to life because of exceptional
hydrological conditions. Indeed, Tafilalt forms a vast depression of about
50 km in diameter, located between the foothills of the Atlas mountains and
the Algerian hamadas, into which many wadis naturally drained (Fig. 16.1).
The two most important of these wadis, the Ziz and the Gheris, flow from
north to south, cross the plain before drying up a short distance beyond the
main depression. These two rivers have exceptional flow and are active for
a substantial part of the year, but especially in spring time when they are
swollen up by snowmelt.28 The recurring humidity of the depression, added to
a rainy season in the autumn, promotes the development of pastures.
Furthermore, Tafilalt has a loamy soil that is highly favourable to the devel-
opment of pastures compared to the sterile rocky floor of the surrounding
hamadas. This would have encouraged the seasonal arrival of pastoral people
as evoked by al-Bakri and is why Samgu was one among numerous sheperds
in Tafilalt before the foundation of Sijilmasa. But this does not fully explain the
record of the annual trade gathering as related by al-Bakri in his explanation of
the emergence of Sijilmasa. The foundation myths in connection with sufrism
or with the blacksmith are superimposed in al-Bakri’s text onto his account of
25
See Sterry et al., Chapter 6, this volume.
26
Margat and Camus 1959; Ruhlmann 1939. See now, Bokbot 2019, 317–24 for a useful summary.
27
Margat and Camus 1959, 370. 28 Margat et al. 1962, 32.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 606 Chloé Capel

Figure 16.1. Map of Tafilalt today (C. Capel).

this gathering. The hypothesis of a seasonal fair in Tafilalt, with (commercial)

exchanges, where Midrar was supposed to have sold/exchanged his tools, must
be discussed in relation to the khariji rise and the pre-Islamic political
structure of the region. These three assumptions are not mutually exclusive.
Rather, it could be that pastoralists, who regularly frequented the place, already
knew Tafilalt as a stable and famous regional place of exchanges and perhaps
the location of a settlement before the rise of Islam. But if a favourable climate
and environment were the reason for nomads to gather in Tafilalt, it is not the
only place with such natural conditions south of the Atlas Mountains and it is
probably necessary to consider other arguments of why Tafilalt was a place of
preferential attraction.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 607

Certainly, the annual gathering of nomadic people was justified by

economic necessities, as al-Bakri related. As some nomads still proceed
nowadays, one could also invoke tribal, social and marriage requirements
to justify such fairs in order to maintain bonds between communities that
were isolated from one another for the rest of the year. But the reason why
it would have taken place precisely in Tafilalt could be because of its salt
deposit. The (re)discovery of a sabkha in the immediate neighbourhood of
Sijilmasa, which presents clues of human exploitation, brings a decisive
argument in the choice of populations to gather and even settle in
Tafilalt.29 It also sheds a new light on the way the region could have
developed economically. The commercial fair that al-Bakri describes
could have been primarily motivated by the completion of a ‘cure salée’
(salt cure) among shepherds, even if the antiquity of this activity cannot be
estimated. This salt deposit is certainly a key element in understanding the
seasonal aggregation phenomena in pre-Islamic times in Tafilalt, but also
the development of sedentary lifestyles at the time of emergence of
Sijilmasa. The metabolism of domestic ungulates in desert environments
requires nutritional supplements, particularly in terms of minerals; this
may be facilitated by organising stays in areas with brackish water and/or
salty pastures. The geomorphology of the Tafilalt plain is of a closed
hydrological basin from which groundwater cannot escape underground,
but only by evaporation. Consequently, the loamy soil of the plain has
a significant salt content, especially in the southern part of the basin, along
the Ziz river, where the waters are concentrated and blocked against the
surrounding mountains.30 From Cordoba, al-Bakri himself heard about
this information since he mentions it in his description of the vicinity of
Sijilmasa. On the south-eastern side of Sijilmasa, about 15 km from the
main archaeological mound, lies a wide area of salty land called the
Tamezguida sabkha. Moreover, at the end of the rainy season, a large
salty pond forms in the immediate vicinity of the sabkha. The brackish
water of this Daya Tamesguida makes it a valued watering point for wild
and domestic animals. Thus Tafilalt provides ideal and rare conditions for
a salt cure, either by brackish surface water, by salty pastures or by the
direct consumption of natural salty efflorescences forming at the surface of
the sabkha. This is a good reason for why pastoralists would have gathered
annually in Tafilalt.

29
The existence of this sabkha is not a real discovery as it is mentioned on old topographic maps,
especially the one produced in the 1930s (but not on the versions of the 1970s).
30
Margat et al. 1962, 120–28.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 608 Chloé Capel

Figure 16.2. Aerial view of a salt extraction site along Wadi Ziz at the place called
Tamellaht. This spot is not dated and may be modern but does not go back before the
ninth century AD (image: Google, DigitalGlobe).

The salt deposit of Tafilalt is a new aspect for understanding the eco-
nomic development of the region and its position on the caravan routes,
regardless of any Trans-Saharan economy prior to the appearance of
Sijilmasa. The salt, which was certaintly exploited in Tafilalt between the
twelfth and the eighteenth century (and later for a period of ten years),31
may have been mined much earlier.32 It is possible that salt exploitation
dates back as far as the pre-Islamic period when pastoralists would have
collected salt in Tafilalt for the needs of their animals and society, either
freely and informally, or according to community regulations.
Furthermore, salt may have been more than an informal resource for
pastoralists, but exploited for its economic potential with a ‘mining’ man-
agement earlier than the twelfth century. Clues of salt extraction in the
sabkha of Sijilmasa (Fig. 16.2) do not reach the scale and the density of
those in the salt deposits of Taghaza and Tawdenni, well known for their
intensive exploitation and traded at a continental scale. But even if the
saltwork of Sijilmasa was of a smaller scale to these Saharan deposits, its
role in the local economy could have been considerable and it may have

31
Unpublished personal doctoral research. The exploitation of salt in Tafilalt is made by leaching
and not by mine working.
32
Unfortunately, archaeological evidence of salt extraction in Tafilalt cannot be dated at the
moment. Historical sources confirm this activity since at least the twelfth century, but the
practice could be older.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 609

had important repercussions for the structuring for the state of Sijilmasa.
Indeed, the salt would have become a capitalised property of the state on
which it could have imposed control and therefore legitimated part of its
authority. This mineral resource may be key to explaining the basis of the
Sijilmasian authority, beyond the justification of the spirituality of khariji
teaching. Sijilmasa might also have created an economic monopoly on salt,
although to this day it remains impossible to establish its effectiveness or to
define its terms. If an ancient salt exploitation in Tafilalt is now proven, the
way this operation took place and the date of its beginnings remain
unknown.

Archaeological Survey: The Hypothesis of Dense Settlements in

Tafilalt before Sijilmasa
Could the annual gathering in Tafilalt have led to some people adopting
a sedentary lifestyle here even before the rise of Sijilmasa? The environmental
optimum of the Tafilalt plain may at first sight have favoured permanent
settlements and not just seasonal dwellings as in the context of pastoral
activities. As this volume makes clear, this phenomenon was observed else-
where in the Sahara, especially during episodes of climatic downturn in the
Neolithic period. On these occasions, people gathered in areas of privileged
environment and developed a light agriculture based on the exploitation of
ponds or springs by means of a moderate irrigation.33 As a matter of fact, the
Western Maghrib in general and the Tafilalt plain in particular may have faced
an episode of sharp climatic deterioration between the seventh and ninth
centuries, compared to a much wetter early millennium.34
This significant change in environmental conditions, unfavourable to
pastoral activity, may have forced population settlement and the develop-
ment of agriculture, although the environmental realities prohibit any dry
farming in the region. But in Tafilalt, whether these climatic variations
happened or not, if environmental conditions are ideal for pastoral life-
style, they seem to be less appropriate for agriculture. Indeed, without
irrigation, there is no means to develop long-term agriculture here. The
Wadis Ziz and the Gheris provide surface water during most of the year,
but are not well adapted to irrigation because of the deep downcutting of
their channels in the substratum of the plain – except in the downstream

33
Amblard-Pison 2011.
34
Publications on recent climate history in the Maghrib are scarce and this prohibits any
definitive statement. However, it seems that the rise of Sijilmasa more or less coincided with
a period of desiccation of the environment, Capel, forthcoming; cf. Schneider 2017.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 610 Chloé Capel

Figure 16.3. Bed and riverbanks of Wadi Ziz on the northern side of Tafilalt plain (C.
Capel). The power of floods has cut the riverbed into the level of bedrock and
continuously erodes the banks, which makes cultivation along the Ziz difficult.
Agriculture is most feasible downstream, where the riverbed is not so deep, in the south
of Tafilalt plain, or close to seasonal ponds.

area, south of the basin. Other factors that constrain irrigation include the
great instability of the river banks and the violence of the flow during the
flood season (Fig. 16.3). But cultivation was nonetheless certainly possible
on small short-lived plots, established in the beds of the rivers – especially
just after the flood season – and around seasonal water ponds left by the
rains, like the (large) Daya Tamesguida. Therefore, prior to the foundation
of Sijilmasa, Tafilalt may have not only been a focus for seasonal nomadic
people but also sedentary agricultural populations practicing a strategy of
itinerant exploitation along the main wadis of the plain. Margat and
Jacques-Meunié, with different arguments, both put forward the thesis of
sedentary and rural settlement in Tafilalt before the Islamic era.35 This type
of practice enables crops which require little water and have a short growth
cycle (less than 120 days). Cereals are the most suitable species (durum
wheat and barley especially), but the culture of a number of hardy species
of legumes (peas, beans, lentils, alfalfa), many of which are forage plants,
can also be made possible.36 The date of introduction of palm trees in
Tafilalt, and in the Maghrib in general, is not known and we must not rule

35
Jacques-Meunié 1982, 195–96; Margat and Camus 1959, 370.
36
Chiche 1984, 151; Pascon 1984, 17.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 611

out a pre-Islamic presence as date palm cultivation has been identified in

Kharga oasis (Egypt) as early as the fifth century BC and even earlier in
Fazzan.37 But the date palm cultivation requires a lot of water and, without
irrigation all the year long, it can only be possible in the immediate vicinity
of wadis, which seems unlikely in Tafilalt without a system of dykes and
canals being developed to mitigate the violence of the floods.
Consequently, the introduction of the date palm in Tafilalt seems relatively
late and could be contemporary with the huge irrigation system that was
dug contemporaneously with the beginning of Sijilmasa (see below).
Under these conditions, it would be possible to argue for the existence of
a pre-Islamic dense sedentary society in Tafilalt associated with agricul-
ture. Did not al-Bakri tell us that the city of Ziz gradually emptied of its
inhabitants for the ultimate benefit of Sijilmasa?38 The city of Ziz has never
been located, but following a proposal by Daniel Eustache in 1971, most
authors agree to place it far upstream from the Tafilalt, in the area of the
High Atlas plateau.39 A careful reading of the Arabic sources mentioning
the city of Ziz leads me to reconsider the location of this city. There is no
evidence in textual sources that Ziz would be within the Atlas Mountains
and it appeared that this suggested location is based on an incorrect, or
truncated, translation by Mohammed Hadj Sadok published in 1949.40
Contrary to what has been usually asserted since Eustache, Ziz cannot be
located in the Atlas Mountains because according to Ibn Khurradadhbih,
who is the only author to mention the city at the same period of its activity,
it was located south of Der’a, in the same direction as ‘the land of Habasha’,
that is the Bilad al-Sudan, ‘the land of the Blacks’.41 This topographical clue
requires us to locate Ziz south of the Atlas and because the wadi had
already taken its name by the late ninth century,42 this suggests that this
city was located not far from its banks. The crosschecking of this informa-
tion makes Tafilalt one of the best candidates for the location of the city of
Ziz as it is situated both south of the Atlas and along the river of the same
name. In addition, as discussed above, it enjoys a very favourable environ-
ment in this desert landscape where a light cultivation could be possible.
Moreover, if Sijilmasa had such an impact on the demography and the
economy of Ziz, by depopulating it as al-Bakri said, then Ziz should have
been located in the area of direct influence of the new city of Sijilmasa, and

37
Tenberg 2012, 146; Van der Veen 2010, 514; see also Mattingly et al., Chapter 2, this volume.
38
Al-Bakri 1911, 148; Al-Bakri 1913, 282. 39 Eustache 1971, 142.
40
Hadj-Sadok 1949, 32–33. 41 Ibn Khurradadhbih 1865, 209; 1889, 88.7.
42
Al-Ya‘qubi 1892, 358–60; 1937, 225–26.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 612 Chloé Capel

probably much closer than the High Atlas plateau, which is geographically,
economically and socially distinct of the Saharan oasis system.
Archaeology, if it does not bring (yet) decisive arguments in this debate,
provides evidence that urban-scale settlements were possible in Tafilalt before
the rise of Sijilmasa. The site of Sijilmasa has remained quite poorly docu-
mented until now from a strictly archaeological point of view. Nevertheless,
three major excavation programmes have taken place there and a fourth one is
currently underway. Due to a combination of political and nationalistic
reasons, since Tafilalt is the cradle of the dynasty reigning over Morocco
today, Sijilmasa became a major archaeological focus in the early 1970s. The
Moroccan authorities appealed to an Italian team to start a programme of
research in Sijilmasa in 1971 and 1972. These two campaigns led to a short
publication of limited archaeological interest, but with valuable information
about oral traditions.43 At the end of this programme, scientific and political
disputes led the Moroccan authorities to ban any foreign teams from new
excavations in Sijilmasa. A second Moroccan project made a short campaign
in 1974, but without any follow-up or publication. It was ultimately in the
1980s that Morocco consolidated its archaeological institutional basis – with
the creation of an archaeological department at the Ministry of Culture and the
opening of a national institute for training in archeology and heritage
(INSAP). INSAP was then able to reconsider Sijilmasa within the framework
of an ambitious archaeological programme. In partnership with the ceramics
laboratory of Lyon (France), Morocco organised a survey and surface collec-
tions at Sijilmasa, which resulted in a major article about trade connections
between Tafilalt and Awdaghust (Tagdaoust) but which told little about
Sijilmasa itself.44 This Moroccan–French project had no follow-up, but the
renewed interest was beneficial to the team of Ronald Messier (Middle
Tennessee State University) that was finally allowed to open new excavations,
as the Moroccan–American Project at Sijilmasa (MAPS) in cooperation with
INSAP. MAPS started a work of greater magnitude with six successive cam-
paigns between 1988 and 1998, mostly focused on the main archaeological
mound of the site. The MAPS programme is today the main source of
archaeological information about Sijilmasa which has been published between
1993 and 2015 in a series of 20 articles and a recent monograph. Despite some
archaeological weaknesses, these documents are of paramount interest since
they are the first to offer a fairly accurate view of the archaeological reality of
the site.

43 44
Rachewiltz 1973. El-Hraiki et al. 1986.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 613

Since 2011, a new Moroccan–French programme, led from Toulouse,

has taken over the site of Sijilmasa. The first publications are expected
soon. Meanwhile, my doctoral research, initiated in 2009, built on the
remarkable but still under-exploited work of MAPS. Furthermore
I conducted extensive and fruitful survey of the Tafilalt plain between
2011 and 2013, in order to gain a broader historical view and a finer
understanding of the context of development of the city.
Two interesting sites are located in the vicinity of the Medieval city. Both
consist of the ruins of several dozen buildings or enclosures, constructed in
a stone architecture. Both sites are located on rocky hilltops (Jabal Afilal
and Jabal Oulad Zohra) directly overlooking the course of Wadi Ziz, being
enclosed by an enceinte, defining a perimeter of 7 and 3.5 ha respectively.
Respectively located 46 and 12 km from the archaeological tell of Sijilmasa,
they were previously reported in the archaeological literature, but have
never been studied.45 Surveys on these two sites, as part of my work in
Tafilalt, revealed that they were likely permanent dwellings (although with
less certainty for the second site which has more enclosures), each one
containing several structures of unusual design and dimensions which can
be identified as collective places or monuments of power (Figs 16.4–16.6).

Figure 16.4. Jabal Afilal and its pre-Islamic settlement on the top of a hill, view from
north-east (C. Capel).

45
Allain and Meunié 1956; Margat and Camus 1959. Jabal Afilal is the site described in Sterry
et al., Chapter 6, this volume, as Taouz 1, from which engravings of >200 chariots have been
recorded.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 614 Chloé Capel

Figure 16.5. Aerial view of Jabal Afilal, surrounded by cliffs and overlooking Wadi Ziz
(south-eastern corner). About 50 houses are enclosed in a wall with a separate ‘citadel’
(west side) where some unusual buildings (by size and plan) are concentrated, opposite
the site’s eastern gate (imagery: Google, DigitalGlobe).

Figure 16.6. The northern wall of the Jabal Afilal settlement. In the background, the
Wadi Ziz runs through the southern part of Tafilalt plain, where the riverbed is not deep
and where agriculture is feasible along its banks (C. Capel).

The closeness of these sites to Wadi Ziz would have allowed some agricul-
tural cultivation of the river bed. Consequently, all these elements could
allow us to identify these sites as villages, or as urban or proto-urban areas.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 615

The main issue remains the question of their dating. Without any excava-
tion, it is of course difficult to accurately identify and date these sites. The
surface collection provided very few artefacts, mainly ceramics which were
badly abraded and with almost no diagnostic shapes. Despite this, it seems
assured that these sherds are not Medieval or modern because their fabric,
with a lot of vegetal temper, has nothing in common with the ceramics
from Sijilmasa. All other Medieval or modern sites discovered in the
Tafilalt plain during my survey campaigns, even far from Sijilmasa,
revealed thin or domestic ceramics of the type found in the MAPS excava-
tions of the city. The ceramic artefacts collected on both sites can plausibly
be suggested to indicate a pre-Islamic dating.46

Conclusion: From the Emergence of the Filalian State to the

Building of the City of Sijilmasa

In Tafilalt, as elsewhere in the world, the emergence of a new political

authority, of a new power, does not necessarily mean the birth of a city and
the rise of an urban society, as Judith Scheele emphasises.47 The process
leading to the birth of a city and the close match between the state and its
capital city derives from another scheme, from a complementary develop-
ment. Social constraints, but also economic and technological factors are
involved in this separate process. However, the special case of Sijilmasa
could apparently reflect Bourdieu’s model of the birth of the state, although
this model was conceived for and from modern industrialised and capital-
ist societies. In Bourdieu’s model, the final stage completing the structuring
of a power is the production of arguments of its own legitimacy. Once the
state starts to think about itself in order to definitively establish its author-
ity, it gives birth to a wide range of instruments which not only justify the
existence of this power, but in turn, fully contribute to build and develop it
further. This initiates a vicious circle where the foundations of power and
its means of action are the same.48 It is during this final step that the
instruments and productions of the state appear and multiply, as issues of
currency or writing of historical vulgates, as creation of legal and legislative
rules or production of state symbols, as physical materialisation of dom-
ination through architectural programmes or the exercise of this
46
Mattingly et al. 2017 discuss comparable hilltop sites in the Draa valley to the west of Tafilalt,
which have similar ceramics and have been radiocarbon dated to the pre-Islamic period. See
also Sterry et al., Chapter 6, this volume.
47
Scheele, Chapter 18, this volume. 48 Lenoir 2013, 148.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 616 Chloé Capel

domination by the management of collective rules of life. All these ele-

ments ultimately define the state and its essence but, as mentioned in the
introduction to this chapter, they follow the birth of the state.
In the case of Sijilmasa, all these structural elements may indeed gradu-
ally have appeared after the emergence of a political power and a state in
Tafilalt. In the intellectual and symbolic field for example, the creation of
an Eastern descent for the Midraride dynasty, with references to the
preacher ‘Ikrima, was part of this legitimacy step.49 We propose to connect
the founding of the city of Sijilmasa to this precise step. It then follows that
the emergence of the new power in Tafilalt was in a temporality that would
not be exactly superimposed. Indeed, we propose to date the foundation of
Sijilmasa, as the city documented by the vast archaeological site located
west of the modern town of Risani in the heart of Tafilalt, from around 200
AH/AD 815, that is to say about half a century after the social and political
changes that occurred according to al-Bakri in Tafilalt in the middle of the
eighth century. My work, in the process of being published,50 has devel-
oped many historical, archaeological and environmental arguments, lead-
ing me to propose a new conclusion. Under the reign of the fourth khariji
king of Tafilalt, Abu al-Muntasir al-Yasa‘, the second son of Samgu and
father of Midrar, who ascended the throne in 174 AH/AD 790–791, a royal
city was built on the initiative of the prince at some distance from the
original nucleus of sedentary filalian population. The prince would have
moved his headquarters and residence there after the completion of a vast
architectural and land improvement programme. One of the aims, but not
the only one, was an ambitious assertion of his power, justified by the
contemporaneous threat of a growing Idrisid hegemony over the Maghrib
and the will to strengthen his own supremacy towards his own allies and
people. The symbolic value of such an initiative would be decisive. The
construction of this city was also coupled with a vast programme of
agricultural development in Tafilalt in order to support the population
growth of this new sedentary nucleus and the involvement of the city in
Trans-Saharan trade. It meant the establishment of intensive agriculture,
that is to say irrigated agriculture, thanks to the construction of a huge
artificial hydraulic system. This was probably constructed by a decision of
the state rather than a collective societal decision. It is this new city that
subsequently became the urban heart of Tafilalt and was named Sijilmasa.
This new city was therefore a relatively late foundation in connection with
the constitution of the filalian state. This mirrors the delay between

49 50
Capel and Fili 2018. Capel 2016; 2018.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 617

assuming power and starting new architectural programmes by other

Islamic dynasties such as the Umayyads of Damascus (who built the
Dome of the Rock, for instance, 30 years after taking power) and the
Abbasids (who built Baghdad as their capital 12 years after they overthrew
the Umayyad caliphs). Hence, the Midrarids of Tafilalt visually imposed
their authority on the filalian landscape one to two generations after their
effective political rise in the region. Finally, the date we propose for the
founding of Sijilmasa, based on historical, archaeological and environ-
mental arguments, roughly coincides with the uprising of Cordoba. It is
thus very likely that the Cordoban foundation myth of Sijilmasa is actually
the expression of the remembrance of the founding of this new city and one
that linked it to the discourse of the Umayyad caliphs.
The analysis of the process of emergence of Sijilmasa, first as a social and
political authority and then as a city, emphasises that Tafilalt was probably
a major meeting place and perhaps had several small rural settlements, with
quite strong political organisation, a long time before the rise of Trans-
Saharan trade. As David Mattingly and Martin Sterry remind us in this
volume, this pattern shows similarities with the history of Fazzan which was
a regular stop on Islamic-era trade routes across the Sahara because it had
previously developed an autonomous economic and political system. The case
of Tafilalt would underpin the assumption that Islamic trade routes that
formed across the Sahara from the eighth century, were determined according
to sites which were previously occupied and were not an ex nihilo creation.
This does not mean that all these sites were necessarily already involved in
a continental-scale exchange system, which remains essentially an Islamic
project. Some areas, such as Fazzan, were already maintaining long-distance
trade routes with the Mediterranean coast, but some others, like Tafilalt, were
merely local centres. Unlike Fazzan, the political and economic achievement
of Tafilalt was in the development of Islamic era Trans-Saharan trade.
Without it, and without the movement of gold with which it is closely
associated, Tafilalt would probably have only remained a major place of
nomadic gathering, a modest oasis zone, but not the huge city and the
powerful state it became in the Medieval period.

References

‘Abd al-Raziq, M.I. 1985. Al-khawârij fî bilâd al-Maghrib hattâ muntasaf al-qarn
al-râbi’al-Hijrî. Casablanca: Maktabat al-Hurrîyah al-Hadîthah.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 618 Chloé Capel

Allain, J. and Meunié, J. 1956. Quelques gravures et monuments funéraires de

l’extrême sud-marocain. Hespéris 43: 51–86.
Amanis, S.A. 2009. Dictionnaire Tamazight-Français (Parlers du Maroc central).
Paris: Harmattan.
Amblard-Pison, S. 2011. Villages et territoires villageois néolithiques d’une zone
refuge au Sahara méridional. In G. Kourtessi-Philippakis and R. Treuil (eds),
Archéologie du territoire, de l’Égée au Sahara. Paris: Publications de la Sorbonne,
169–79.
Al-Bakri, 1911. Kitâb al-masâlik wa-‘l-mamâlik. 2nd edition (1st edition 1857).
Edited by W. Mac Guckin de Slane. Algiers: A. Jourdan.
Al-Bakri, 1913. Description de l’Afrique septentrionale, 2nd edition (1st edition
1859). Translated by W. Mac Guckin de Slane. Algiers: A. Jourdan.
Bokbot, Y. 2019. Protohistoric and pre-Islamic funerary archaeology in the
Moroccan pre-Sahara. In M. Gatto, D.J. Mattingly, N. Ray and M. Sterry (eds),
Burials, Migration and Identity in the Ancient Sahara and Beyond. Cambridge:
Cambridge University Press and The Society for Libyan Studies, 315–40.
Bouderbala, N., Chiche, J., Herzenni, A. and Pascon, P. (eds). 1984. La question
hydraulique. 1. Petite et moyenne hydraulique au Maroc. Rabat: Institut
Agronomique Hassan II.
Bourdieu, P. 2012. Sur l’Etat: Cours au Collège de France (1989–1992). Paris:
Points.
Capel, C. 2016. Une grande hydraulique saharienne à l’époque médiévale: L’oued
Ziz et Sijilmassa (Maroc). Mélanges de la Mélanges de la Casa de Velàzquez. 46.1:
139–65.
Capel, C. 2018. Sijilmasa in the footsteps of the Aghlabids: The hypothesis of
a ninth century new royal city in Tafilalt Plain (Morocco). In G. Anderson,
C. Fenwick and M. Rosser-Owen (eds), The Aghlabids and their Neighbours: Art
and Material Culture in the 9th-century North Africa. Leiden: Brill, 531–50.
Capel, C. Forthcoming. The impact of climate modulations on the economic, social
and political reorganisation of Maghreb al-Aqsa in the 8th-9th centuries. Al-
Masaq 32.
Capel, C. and Fili, A. 2018. Sijilmâsa au temps des Midrârides: Nouvelles approches
historiques et premier bilan archéologique. In C. Aillet (ed.), L’ibadisme dans les
sociétés de l’Islam médiéval. Modèles et interactions. Berlin/Boston: De Gruyter,
137–67.
Chelhod, J. 1978. Qayn. Encyclopedia of Islam, 2nd edition. Leiden: Brill, 2: 819.
Chiche, J. 1984. Description de l’hydraulique traditionnelle. In Bouderbala et al.
1984, 119–319.
Dieterlen, G. 1964. Contribution à l’étude des forgerons en Afrique occidentale.
Annuaire 1965–1966 de l’École Pratique des Hautes Etudes, Section des sciences
religieuses 73: 3–28.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 16 At the Dawn of Sijilmasa 619

El-Hraiki, R., Picon, M. and Robert, D. 1986. Ateliers et commerce transsaharien à

l’époque médiévale. In La ceramica medievale nel Mediterraneo occidentale.
Siena, Faenza, 8–13 ottobre 1984. Firenze: Edizioni all’Insegna del Giglio, 51–54.
Eustache, D. 1971. Corpus des dihrmas idrisites et contemporains: Collection de la
Banque du Maroc et autres collections mondiales, publiques et privies. Rabat:
Banque du Maroc.
Genêt, J.P. 2014. A propos de Pierre Bourdieu et de la genèse de l’Etat moderne.
Actes de la recherche en sciences sociales 201–202: 98–105.
Gilliot, C. 2002. Al-Warrâq. Encyclopedia of Islam, 2nd edition. Leiden: Brill,
11: 151.
Hadj-Sadok, M. 1949. Description du Maghreb et de l’Europe au iiie/ixe siècle:
extraits du Kitâb al-masâlik wa’l-mamâlik, du Kitâb al-buldân et du Kitâb al-
a‘lâq an-nafîsa. Alger: Jules Carbonel.
Heusch, L. 1956. Le symbolisme du forgeron en Afrique. Reflets du monde 10:
57–70.
Ibn Khurradadhbih. 1865. Le livre des routes et des provinces. French translation by
C. Barbier de Meynard. Paris: Imprimerie Impériale.
Ibn Khurradadhbih 1889. Kitâb al-masâlik wa-’l-mamâlik. Edition M.J. de Goeje.
Leiden: Brill.
Jacques-Meunié, D. 1982. Le Maroc saharien des origines à 1670. Paris: Klincksieck.
Lenoir, R. 2013. L’état selon Pierre Bourdieu. Sociétés contemporaines 87 (2012–
2013): 123–54.
Lévi-Provençal, E. 1938. La fondation de Fès. Annales de l’Institut d’études orien-
tales 4: 23–53.
Levtzion, N. 1973. Ancient Ghana and Mali. London: Methuen.
Margat, J., Destombes, J. and Hollard, H. 1962. Mémoire explicatif de la carte
hydrogéologique au 1:50000 de la plaine du Tafilalt, Notes et Mémoire du Service
Géologique. 150 bis. Rabat: Éditions du Service Géologique du Maroc.
Margat, J. and Camus, A. 1959. La nécropole de Bouïa au Tafilalt. Bulletin
d’Archéologie Marocaine 3: 63–74.
Al-Mas‘udî 1865. Murûj al-dhahab wa-ma‘âdin al-jawhar. Edition and French
translation, C. Barbier de Meynard and A. Pavet de Courteille, volume IV. Paris:
Imprimerie Nationale.
Mattingly, D.J., Bokbot, Y., Sterry, M., Cuénod, A., Fenwick, C., Gatto, M., Ray, N.,
Rayne, L., Janin, K., Lamb, A., Mugnai, N. and Nikolaus, J. 2017. Long-term
history in a Moroccan oasis zone: The Middle Draa Project 2015. Journal of
African Archaeology 15: 141–72.
Mellouki, M. 1985. Contribution à l’histoire des villes médiévales du Maroc:
Sijilmâsa des origines à 668/1269. Unpublished PhD dissertation, Université
Aix-Marseille.
Mezzine, L. 1984. Sur l’étymologie du toponyme Sijilmasa. Hespéris 22: 19–25.
Pascon, P. 1984. La petite et moyenne hydraulique. In Bouderbala et al. 1984, 7–44.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 620 Chloé Capel

Pellat, C. 1960. Les Banû Midrâr. Encyclopedia of Islam, 2nd edition, vol. 1. Leiden:
Brill.
Rachewiltz, B. de. 1973. Missione etno archeologica nel Sahara Maghrebino.
Quaderni delle rivisa Africa 2: 519–68.
Ruhlmann, A. 1939. Recherches de préhistoire dans l’extrême sud marocain, 5.
Rabat: Publications du Service des Antiquités du Maroc.
Schneider, A.W. 2017. The Medieval Climate Anomaly as a factor in the history of
Sijilmasa, southeastern Morocco. The Journal of North African Studies 1: 132–52.
Tenberg, M. 2012. Beginnings and early history of date palm garden cultivation in
the Middle East. Journal of Arid Environments 86: 139–47.
Van der Veen, M. 2010. Plant remains from Zinkekra. Early evidence for oasis
agriculture. In D.J. Mattingly (ed.), The Archaeology of Fazzan. Volume 3,
Excavations of C.M. Daniels. London: The Society of Libyan Studies, 489–519.
Al-Ya‘qubi 1892. Kitâb al-buldân. Edition, M.J. De Goeje. Leiden: Brill.
Al-Ya‘qubi 1937. Les Pays, French translation by G. Wiet. Cairo: Institut Français
d’Archéologie Orientale.

https://doi.org/10.1017/9781108637978.017 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market
Towns of West Africa
sam nixon

Introduction

In the early Islamic period, trade across the Sahara escalated to new levels
as the rich resources of West African gold, slaves and ivory were imported
on camel caravans to the markets of North Africa and the wider Islamic
world trade system, these goods being exchanged for products from North
Africa such as copper, textiles and glasswares.1 The immediate destinations
for Trans-Saharan camel caravans coming from North Africa were the
market towns which developed on the southern fringes of the Sahara, in the
most arid zones of the Sahel (Sahel meaning ‘shore’ in Arabic; see
Fig. 17.1). These centres became the principal locations for the Trans-
Saharan commercial exchange, meeting points between traders from
North Africa and the Sahara and those coming from further south in
West Africa. Certain trading centres directly involved in the Trans-
Saharan exchange also grew up slightly further south in the Sahel, close
to the centres of the West African states controlling the gold, slaves,
and ivory resources (see Fig. 17.1). In the earliest period of study of
Trans-Saharan commerce, these various market centres were understood
largely as transplantations of Islamic urban culture.2 This was a somewhat
simplistic model centred around the concept that Muslim traders from
North Africa and elsewhere in the Islamic world imported wholesale
models of town planning and architectural design to create towns resem-
bling those north of the Sahara and catering to expatriate Muslims. With
the increasing awareness of the pre-Islamic origins of urbanism in West

1
See Austen 2010; Devisse 1988; Mitchell 2005, Chapter 5; Nixon 2017b. Cf. also Mattingly et al.
2017a.
2
For example, Bovill 1958; Lhote 1951; and for discussion McIntosh, R. 1999. 621

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 622 Sam Nixon

Figure 17.1. Map showing Trans-Saharan trade routes to West Africa and localities
referred to in the text.

Africa3 – and indeed the existence of pre-Islamic urban models from the
Sahara itself4 – as well as a greater recognition of the cultural diversity of
early merchant populations,5 a more nuanced approach has been devel-
oped in relation to these towns. This has also been significantly informed
by more in-depth archaeological studies of several key sites.
Despite this reassessment, the earliest market towns at the southern
fringes of the Sahara – from Mauritania to Niger – and the early trading
enclaves established further south are still often commonly discussed
together as a distinctive urban phenomenon.6 This is due to the belief in
them having a broadly similar function, based on the following criteria:
1. Being clearly identifiable urban nodes directly associated with the func-
tion of Trans-Saharan exchange in the early stages of this commerce;
3
See McIntosh, S. 1995; Chapter 14, this volume; McIntosh, S. and McIntosh, R. 1993.
4
See Mattingly and Sterry 2013; Chapter 19, this volume. 5 See Moraes Farias 2003.
6
For example, Devisse 1988; Insoll 2003, Chapter 5; Kea 2004; Monroe 2017.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 623

2. Their identification as centres with elevated levels of contact with the

central Islamic world and associated influence from those regions;
3. The presence there of significant immigrant communities of long
distance traders, many of whom were Muslim.
In line with previous scholars, this chapter maintains the utility of consider-
ing these towns side by side as a distinctive urban phenomenon whose
discussion can help us illuminate wider patterns of urban development in
West Africa.
Three localities are identifiable within the Arabic records as key early
Islamic centres of exchange at the immediate southern edge of the Sahara.
These are Awdaghust, Tadmakka (or Tadmekka) and Kawar,7 situated on
three different routes across the Sahara. In addition to being the earliest
entrepôts, these centres also continued to hold this role for centuries after
their foundation. All three centres can be associated with archaeological
sites, though only Awdaghust and Tadmakka have seen in depth investiga-
tion. The Arabic records also provide clear descriptions of the trading
centres located slightly further south in the Sahel, in close proximity to
the West African state centres. The most significant of these more south-
erly trading centres – celebrated both by the early Arabic historians and by
modern scholars – is that described as located close to the capital of the
Empire of Ghana (see below for historical descriptions). The town as
described in the Arabic records has never been unequivocally identified
with an archaeological site. The archaeological ruins of Kumbi Saleh,8
which is broadly within the zone of Ghana and its successor states,
potentially relates to a chronologically later entrepôt within the orbit of
late-period Ghana or its successors. The significance of this site and the
relatively extensive investigation undertaken there has established it as an
important site for considering the evolution of early merchant centres in
West Africa. The other important market centre described by the sources
further south in the Sahel lies in close proximity to the royal city of Gao, on
the Niger Bend. This is clearly identifiable within the archaeological com-
plex around and under the modern settlement of Gao. Despite the presence
of the modern town, the archaeological remains have seen quite significant
investigation.9
Together this grouping of market centres – Awdaghust, Tadmakka, Gao
and Ghana/Kumbi Saleh – forms the core set of historical and archaeolo-
gical data to allow the exploration of the nature of early Islamic urban
7
See respectively, Devisse 1983; Nixon 2017a; and Lange and Berthoud 1977.
8
Berthier 1997. 9 Cissé 2017; Cissé et al. 2013.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 624 Sam Nixon

centres of Trans-Saharan commerce in West Africa, as well as the evolution

of these centres with the growth of this commercial system. As a distinctive
grouping widely considered to be fundamental for understanding the
development of urbanism in the Trans-Saharan network, it is not surpris-
ing that various syntheses have been undertaken, both in terms of the wider
study of the archaeology of Islam in Sub-Saharan Africa,10 as well as studies
of West African urbanism.11 This chapter seeks to introduce these urban
centres through focused summary accounts of the Arabic historical
descriptions of them and the archaeological investigations which have
taken place, and within the context of this volume to provide a broad
discussion that can direct further debate and investigation of the Trans-
Saharan centres. While the review will focus principally on these key
market centres, this study will also provide a brief overview of some of
the other early market towns that we know of from Arabic texts and
archaeology, in order to collate the current body of knowledge relating to
them, and to address four key aspects within the broad debate that form the
latter part of this chapter:
• Pre-Islamic occupations at the sites of early Trans-Saharan market
towns;
• Relationships between the early Islamic Trans-Saharan market towns
and surrounding settlements;
• Urban landscapes and settlement planning at the early Islamic Trans-
Saharan market towns;
• Architectural construction and building types.

The Trans-Saharan Market Towns: Historical and

Archaeological Overview

Awdaghust/Tagdaoust
On the far western Trans-Saharan route running from Morocco through
Mauritania the first major settlement south of the Sahara recorded within
Arabic records was Awdaghust. This location is first mentioned in AD 872/
3 by al-Yaqubi who described it as ‘a town called “Ghust”, an inhabited
valley with dwellings . . . It is the residence of their king who has no religion
or law’.12 Within al-Yaqubi’s text the description of Awdaghust clearly

10
Insoll 2003, Chapter 5. 11 For example, Kea 2004; Monroe 2017.
12
Levtzion and Hopkins 2000, 22.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 625

shows it to be the first significant location after the desert on the way to the
kingdom of Ghana, though the description of its ruler appears to indicate
he was not Muslim. At the end of the tenth century (c.AD 967–988) the
town of Awdaghust was again singled out by Ibn Hawqal as the pre-
eminent settlement at the southern fringes of the Sahara on this western
route, the writer describing Awdaghust as ‘a pleasant town, and of all God’s
lands it most resembles Mecca . . . ’ – this description is clearly designed to
identify this as a locality with a strong Muslim presence and recognisable
features of an Islamic town.13 Ibn Hawqal’s description also provides
a compelling account of the high value of trade conducted at Awdaghust,
indicating this to be a flourishing market location:
I saw at Awdaghust a warrant in which was the statement of a debt owed to one of
them [the people of Sijilmasa] by one of the merchants of Awdaghust, who was
[himself] one of the people of Sijilmasa, in the sum of 42,000 dinars. I have never
heard anything comparable to this story in the East [i.e. in Asia].14

Seemingly confirming Awdaghust’s status at this period as the major town

in this region and a place of flourishing trade, al-Muhallabi (c.AD 990),
quoted in Yaqut (c.AD 1229), stated that:
The king of Awdaghust maintains relations with the ruler of Ghana . . . there are
excellent markets in Awdaghust, and it is one of the most important metropolises,
and there is a continuous flow of traffic towards it from every land.15

In addition to indicating its commercial importance, al-Muhallabi also

seems to identify it as an important Islamic centre: ‘its people are
Muslims, who recite the Koran, study Islamic jurisprudence and possess
mosques and oratories’.16 Further references appear to indicate certain
cultural practices imported from further north, including the cultivation
of wheat.17 In the mid-eleventh century the documents indicate that
Awdaghust was attacked by the Almoravids, an Islamic reform movement
that developed in the western regions of the Sahara.18 They went on to
subdue certain other urban centres in the Saharan network,19 before going
on to conquer and found a dynasty in the western Maghrib and southern
Spain. During this episode, many of the inhabitants of the town were
reportedly massacred and the town was said to have been sacked.20
Reports of Awdaghust after this eleventh-century destruction are less easy

13
Levtzion and Hopkins 2000, 46. 14 Levtzion and Hopkins 2000, 47.
15
Levtzion and Hopkins 2000, 168. 16 Levtzion and Hopkins 2000, 168.
17
Lewicki 1974, 22–26. 18 See Sterry et al., Chapter 6, this volume.
19
See also Capel, Chapter 16, this volume. 20 Levtzion and Hopkins 2000, 73–74.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 626 Sam Nixon

to interpret, it being unclear from the reports whether it revived again or

lost its status.21 It is difficult to distinguish in fourteenth-century references
whether the name was being used in connection with a permanent settle-
ment or simply in relation to a pastoral locality.22 This is the last documen-
tary reference to Awdaghust.
From these descriptions it is quite clear that by the ninth century
Awdaghust was a permanently inhabited centre, and by the end of the
tenth century a location of flourishing commerce with an (at least partly)
Islamic population.
The site of Tagdaoust (Tegdaoust) in central Mauritania has been
identified with the early Islamic centre of Awdaghust described in the
Arabic records (see Fig. 17.2).23 The equation of Tagdaoust with
Awdaghust rests in part on locational information within the early
Arabic documents that placed it south of the Tagant in Mauritania.24
The identification of Tagdaoust as a potential site of Awdaghust’s ruins
was made initially due to the fact that the site was larger than other sites
in the region – though it should be pointed out there are indeed
a number of other sites in the surrounding region – something which
appeared to correspond with Awdaghust’s major early Islamic role in
this zone of the Sahel.25 Another factor which suggested Tagdaoust and
Awdaghust were one and the same place was that the specific geogra-
phical description of Awdaghust appeared to show strong correspon-
dences with the contemporary environs of Tagdaoust.26 Although the
linguistic correspondence between ‘Tagdaoust’ and ‘Awdaghust’ can be
seen to be negligible,27 the contemporary groups possessing the name
‘Tagdaoust’ have oral traditions which correspond well with the known
written history of the town of Awdaghust.28 When these factors are
combined with the findings of archaeology the initial identification is
given considerable further weight,29 so much so that I am not aware of
any scholars who contest the assertion that Tagdaoust and Awdaghust
are one and the same.
The site of Tagdaoust is situated in a very arid part of the northern Sahel,
yet the immediate locality still retains a certain sparse vegetation cover. The
site comprises a central tell of c.10 ha over which stone architectural ruins
of the terminal occupation of the site can be clearly defined. The majority

21
See Robert et al. 1970, 15–27. 22 Levtzion and Hopkins 2000, 274.
23
Devisse 1983, 5–29. 24 Robert et al. 1970, 33–38.
25
Devisse 1983, 5–29; Robert et al. 1970, 39–67. 26 Devisse 1983, 539–49.
27
Robert et al. 1970, 29–30. 28 Robert et al. 1970, 97–107. 29 Devisse 1983, 5–29.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 627

Figure 17.2. Aerial photograph of the site of Tagdaoust (from Vanacker 1979, 2).

of buildings appear to be compounds constructed on a broadly rectilinear

plan in dry stone architecture, having rooms arranged around a central
courtyard (see Fig. 17.2). There is no evidence that the urban area was
defined by a surrounding wall or that it featured other elements of for-
tification. Amongst the building remains seen on the surface at least one
mosque was discovered. In addition to the central tell, there were a number
of associated structures surrounding the site – including a musalla (open

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 628 Sam Nixon

Figure 17.3. Plan of the excavated area at Tagdaoust, Mauritania, Building Phase 6
(after Holl 2006, fig. 13).

prayer ground) as well as a large cemetery.30 Fieldwork at Tagdaoust was

conducted between 1960 and 1965 by a team of French archaeologists.
During this period all zones of the site saw sub-surface investigation and
the area which was designated as the principal Medieval tell saw numerous
large-exposure excavations (see Fig. 17.2 where extensive excavations can
be identified in the northern section of the ruins; see also Fig. 17.3 for
a reconstruction of an area of buildings based upon excavation results).
The results of these excavations formed a number of theses and have been
published in four separate volumes.31
The results of the Tagdaoust excavations provided a good stratified
occupational and material culture sequence for the site that was dated
both by radiocarbon samples and trade goods of known age.32 The earliest
levels of Tagdaoust were dated to the eighth/ninth century AD. These
levels were not associated with clear evidence for a full-scale urban settle-
ment, but they did show significant occupation and evidence of mudbrick

30
For a more detailed description of the site see Devisse 1983, 5–29.
31
Devisse 1983; Polet 1985; Robert-Chaleix 1989; Vanacker 1979.
32
See Devisse 1983; Polet 1985, 231–43; Robert-Chaleix 1989.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 629

architecture. No significant quantities of material culture imports

from North Africa were found in the very earliest deposits and,
indeed, significant quantities of these are not seen until the tenth century.
The tenth century deposits of Tagdaoust appear to provide good evidence
for major settlement expansion and a change in architectural construction.
Following the early experimentation with mudbrick architecture, there was
a shift to mostly stone-built construction, although still combined with
mud, in the form of bricks and banco coating.33 Building and urban
planning appear to have followed a model which has strong parallels
with Islamic architectural construction and town planning. Specifically,
we see structures with an enclosure wall containing rooms arranged
around a courtyard (see, for example, Fig. 17.3), a prevalent type through-
out North Africa during this period,34 as well as in the Central and
Northern Sahara.35 At Tagdaoust this building type is identified as
one of the important markers of the arrival of North African trading
communities.36 Extensive excavations have enabled the reconstruc-
tion of entire urban quarters from this period, the period incidentally
when Tagdaoust was described as a flourishing location in the Arabic
records.
Coinciding with this tenth century expansion and redevelopment, there
was also a large increase in evidence for imported, North-African material
culture, such as vessel glass, glazed ceramics, and glass beads.37 The next
significant shift in the occupation sequence took place in the eleventh
century when the site appears to have suffered a major setback associated
with destruction,38 including the industrial quarter on the perimeter of the
town.39 This closely corresponds with the date of the Almoravid destruc-
tion recorded by the Arabic texts. The rebuilding following this rupture
was more limited, with certain areas of the site abandoned and never re-
occupied. However, some of the rebuilt buildings are of a noticeably high
quality – one feature of these is the use of red ochre plaster, a feature only
encountered in the late phase occupations.40 Following this rebuilding
phase, North African imports were less evident and local metallurgy
(which previously had been important) seems to have disappeared.41 In
the thirteenth century the site declined further in size and in the fourteenth
century it was abandoned.42

33
Polet 1985, 233–34. 34 Fentress 1987; Insoll 1999, Chapter 3.
35
Mattingly 2013, Chapter 12. 36 Polet 1985. 37 Devisse 1983; 1988.
38
Devisse 1983; Polet 1985. 39 Vanacker 1979.
40
Polet 1985, 239; Robert-Chaleix 1989, 182. 41 Vanacker 1979.
42
Devisse 1983; Polet 1985.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 630 Sam Nixon

The Merchant Town of ‘Ghana’/Kumbi Saleh

For traders travelling south of Awdaghust, the next major trading centre on
the far western route was the Empire of Ghana itself. While ‘Ghana’ is first
referred to in Arabic documentary sources in the first half of the ninth
century,43 it was not until the eleventh century that a distinct ‘merchant
town’ of Ghana is described. This description is as follows:
The city of Ghana consists of two towns situated on a plain. One of these towns,
which is inhabited by Muslims, is large and possesses twelve mosques, in one of
which they assemble for the Friday prayer. There are salaried imams and muezzins,
as well as jurists and scholars . . . . The king’s town is six miles distant from this one
and bears the name al-Ghaba. Between these two towns there are continuous
habitations.44

Shortly after this description was written, Ghana like Awdaghust came
under the influence of the Almoravids and ‘turned Muslim’ (that is,
converted to orthodox Islam) in the late eleventh century.45 The texts are
not clear as to the consequences of this for Ghana and its merchant town.
Part of the difficulty of interpreting Ghana’s history after the development
of relations with the Almoravids stems from the potential for the name
‘Ghana’ to have been used to describe multiple different states or centres
across the region.46 Our understanding of the ‘entrepôt of Ghana’ from the
sources therefore focuses essentially on the eleventh-century report of al-
Bakri. The origins of this eleventh-century town and its subsequent role are
unclear. Additionally, we cannot clearly gauge the possible histories and
natures of other potential merchant towns of Ghana.
Central to discussions of the merchant town of Ghana has been the site
of Kumbi Saleh, situated in the south of Mauritania close to the Malian
border. At one time the equation between the site of Kumbi Saleh and the
capital of the empire of Ghana was generally accepted, but in more recent
years there has been increasing scepticism.47 To begin with, the historical
description of Ghana located the stone-built merchant town with mosques
some six miles from the royal town. The failure to find the royal town, or
any other major settlements in a wide radius from the site, makes the
identification of Kumbi Saleh with the merchant town of al-Bakri some-
what problematic. More importantly, the fact that Kumbi Saleh is an
isolated centre without a hinterland presents further difficulties; it is

43
See Levtzion and Hopkins 2000, 5–10. 44 Al-Bakri in Levtzion and Hopkins 2000, 79–80.
45
Levtzion and Hopkins 2000, 99. 46 Cf. Masonen 2000; McIntosh, R. 1998.
47
See Lange 1996; Masonen 2000; McIntosh, R. 1998, 256–58.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 631

more than 100 km from the hundreds of major, contemporary and earlier
Middle Niger urban centres in the Méma, Middle Niger, and Lakes
Region.48 Finally, the publication of excavation results from the site has
revealed once and for all that Kumbi Saleh did not exist as an urban site
until the eleventh century – virtually the end of the historical polity of
Ghana of the Arabic sources, which notionally dates from sometime before
the eighth century to the late eleventh century AD.49 Based upon the
various evidence at our disposal, the most widely held opinion now is
that the archaeological site of Kumbi Saleh was a major entrepôt of the final
period of Ghana, and perhaps even served as its ultimate capital, continu-
ing as a trade entrepôt through the time of the Malian Empire.
Kumbi Saleh is truly vast, the central tell covering an area of some
100 ha (Fig. 17.4). In addition, there are multiple necropoleis surround-
ing it. From the surface of the main site, it is evident that at its
abandonment Kumbi Saleh was a well-defined urban complex, with
large blocks of stone architecture and carefully laid out streets. The
town does not appear to have had a surrounding wall or other elements
of fortification. Various fieldwork projects have been carried out at
Kumbi Saleh in the last 100 years.50 However, in terms of publication
the site has been poorly represented, with only one substantive mono-
graph, that of Berthier,51 emerging from 12 seasons of excavation – the
1975–1981 excavations of Serge and Denise Robert being almost
entirely unpublished. In more recent years renewed work has been
attempted, but due to insecurity this has not progressed. Archival
work on the Kumbi Saleh material has produced some important
results, including the work of Capel et al. on the human remains
found at the site, and a recent study of the pottery.52
Through the combination of Berthier’s study and the more limited
publications produced by others it is possible to build a fairly good idea
of the sub-surface deposits of Kumbi Saleh.53 The sector of the site
excavated by Berthier (see Fig. 17.5) was first occupied at the end of
the ninth century, with the first evidence for an urban structure and
permanent architecture dateable to the eleventh century, seemingly just
prior to the advent of the Almoravid movement.54 The apogee of the site
began in the twelfth century and continued through the fourteenth

48
Lange 1994; 1996; Togola 1996; and MacDonald, personal communication.
49
Berthier 1997; Levtzion 1980.
50
Berthier 1997; Bonnel de Mézières 1923; Thomassey and Mauny 1951; 1956.
51
Berthier 1997. 52 Capel et al. 2015; Van Doosselaere 2014. 53 Berthier 1997.
54
See Berthier 1997, 103.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 632 Sam Nixon

Figure 17.4. Plan of the central area of the urban ruins of Kumbi Saleh (after Berthier
1997; from Monroe 2017).

century. In the twelfth century a monumental columned mosque was

constructed at the site,55 coupled with well-made stone domestic struc-
tures. These structures consistently follow a rectilinear format and
can be broadly grouped into a Berber or Arabo-Berber tradition. The
site reached its largest extent by the thirteenth century. The end of
the fourteenth century through the fifteenth centuries saw a dramatic
decline in the portions of the site being occupied, with ultimate aban-
donment sometime in the fifteenth century. In addition to the main tell

55
See Nixon 2014 for reproduction of B. Nantet image of the mosque under excavation.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 633

Figure 17.5. Excavated buildings at Kumbi Saleh (Berthier 1997, 127).

site, there were also two large necropoleis to the west and east of the
site,56 the western necropolis featuring a monumental ‘columned
tomb’.57
Trans-Saharan trade goods and other material culture indicators of
a Trans-Saharan trade function are noticeably limited at Kumbi Saleh,
especially in view of the large-scale excavations which have taken place
there. For example, Berthier’s extensive excavations found only 125 glass
beads, scattered relatively evenly across the entire sequence, with a minor
quantitative peak in the twelfth century.58 Glass gold-weights (seven in
all) – a rare but excellent indicator of commerce – were also recovered from
twelfth- and fourteenth-century contexts.

Tadmakka/Essouk
On the central Trans-Saharan route, leading from the region of Tiaret in
Algeria in the direction of Gao on the Niger Bend, the first major centre

56 57
See Capel et al. 2015, including Figure 1. See Capel et al. 2015 for extensive discussion.
58
Berthier 1997.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 634 Sam Nixon

recorded in Arabic records south of the Sahara is Tadmakka (or Tadmekka).

For the tenth century Ibn Hawqal provided a description of some form of
cultural centre at Tadmakka, describing an ethnically and culturally mixed
Berber/black population whose rulers were apparently Muslim:
The supreme kings of Tadmakka in our time are Fusahr b. Alfara and Inaw
b. Sabanzak. They are the rulers, who combine leadership with learning, jurispru-
dence, and political skill, as well as some knowledge of biographies and they are
versed in traditions and history. They are Banu Tanamak.59

While crucially Ibn Hawqal’s report does not specifically describe

a permanent settlement at Tadmakka, certain oral traditions recounting
tenth-century events at Tadmakka, but recorded in the eleventh/twelfth
century, suggest there was at that time a place of thriving commerce with
‘treasure houses’, implying therefore a permanent settlement:
Three shaykhs, Abu ‘Amr [‘Uthman b. Khalifa al-Sufi], Abu Nuh [Salih
b. Ibrahim b. Yusuf al-Mazati] and Abu ’l-Rabi [Sulayman b. ‘Abd al-Salam
al-Wisyani] (may God have mercy upon them), recount that a man of the
B. Wisyan named Tamli was mean (muqill) in his early years . . . He belonged
to the people of the Qusur . . . He began to travel to Tadmakkat where he attained
great wealth . . . He began to send from Tadmakkat every year 16 bags, each bag
containing 500 dinars. These bags were of cowhide and on each of them was
written ‘This is God’s property’. The shaykh returned. He used to recount of his
treasure houses: ‘I saw there marked bags which I can only liken to puppies piled
one on another, as full of gold. On each bag was written: This is God’s property.
Praise be to God the Lord of the Worlds’.60

These oral traditions recounted above related to Ibadi Muslims, a minority

Islamic group who were however of great importance within early Trans-
Saharan trade,61 seemingly indicating the importance of this group at early
Tadmakka. Likewise, Yaqut, writing in the early thirteenth century but
considered to be borrowing from the tenth-century text of al-Muhallabi,
refers to the ‘Kingdom of Tadmak’ and its capital ‘Zakram’,62 Lewicki
maintaining this latter name relates to the Berber meaning of ‘town’ or
‘built up locality’.
Following the tenth century, the next mention we have of Tadmakka is
al-Bakri’s description of 1068, where it was described as a thriving large
town with a strong market component and merchants, and a burgeoning

59
Levtzion and Hopkins 2000, 50–51. 60 Levtzion and Hopkins 2000, 90–91.
61
See for example, Savage 1992; Van Berchem 1960; see also Capel, Chapter 16, this volume.
62
Lewicki 1981, 441–42.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 635

gold trade, the latter described as supported by a locally produced ‘coin-

age’. Like Awdaghust, the town was also described as ‘resembling Mecca’,
but in the case of Tadmakka (as al-Bakri states) its name can actually be
translated as ‘the Mecca like’:
From Bughrat you go to Tiraqqa and from there across the desert plain to
Tadmakka, which of all the towns in the world is the one that resembles Mecca
the most. Its name means ‘the Mecca like’. It is a large town amidst mountains and
ravines and is better built than Ghana or Kawkaw. The inhabitants of Tadmakka
are Muslim Berbers who veil themselves as the Berbers of the desert do. They live
on meat and milk as well as on a grain which the earth produces without being
tilled. Sorghum and other grains are imported for them from the land of the Sudan.
They wear clothes of cotton, nuli, and other robes dyed red. Their king wears a red
turban, yellow shirt, and blue trousers. Their dinars are called ‘bald’ because they
are of pure gold without any stamp. Their women are of perfect beauty, unequalled
among people of any other country, but adultery is allowed among them.63

Another important aspect of this description is that the inhabitants of the

town appear to have had a culture based on strong relationships to the
nomadic pastoralists of the surrounding region. While al-Bakri’s text dates
from just after the mid-eleventh century, in fact his account is widely seen
to have drawn on the tenth-century work of al-Waraq – it is quite possible
therefore that this description of Tadmakka dated from that time, copied
from this earlier text.64
Towards the end of the eleventh century Tadmakka, like Awdaghust and
Ghana, came into contact with the Almoravid movement and is said to
have ‘turned Muslim’ (converted to Sunni Malikism) under Almoravid
influence.65 Whereas Tadmakka was not specifically referred to as having
been sacked like Awdaghust (see above), the reference to it as having
‘turned Muslim’ does appear to be associated with some form of conflict.
Due to the vagueness of the source which refers to Tadmakka’s involve-
ment with the Almoravids we cannot gain a clear sense of the effects this
might have had on the town, either in a socio-cultural or physical sense.
However, there is no clear description of the town being sacked, as is the
case with Awdaghust. The sources for Tadmakka from the late-eleventh
century up until the early-fourteenth century are generally either very
sparse or obviously copies of earlier sources, and the significance of the
few new references to Tadmakka during this period is unclear. In the early
fourteenth century however, there is another clear reference to Tadmakka,

63 64
Levtzion and Hopkins 2000, 85. See Levtzion and Hopkins 2000, 62 for explanation.
65
Levtzion and Hopkins 2000, 99.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 636 Sam Nixon

by al-Umari, it being described as associated with a nomadic cultural group

who can clearly be related to ancestral Tuareg groups.66 There is no specific
reference to Tadmakka as a market centre nor as a permanent settlement
and therefore it is difficult to judge whether we are seeing a reference to
a flourishing trade centre or some other form of cultural centre, perhaps
not even urban. After this fairly detailed early-fourteenth-century descrip-
tion there are no clear records of Tadmakka, with the last reference in the
early fifteenth century being just to the name of Tadmakka as a locality,
potentially even a copying of earlier sources.67
The ruins of Tadmakka, today called Essouk (Figs 17.6–17.8), are situ-
ated in the mountains of the semi-autonomous Tuareg region of north-east
Mali, at the most arid extremes of the Sahel.68 Specifically, they are located
45 km north-west of Kidal at a point where the Wadi Essouk runs through
a dense range of cliffs. Essouk consists of c.50 ha of stone structural remains
in the central town area, within a valley defined by the surrounding rocky
cliffs. Beyond the cliffs, the town is surrounded by six Islamic cemeteries
bearing Arabic epigraphy from the eleventh to fourteenth centuries.69
A wadi, which experiences flooding in the summer months, runs through
the middle of the site. The ruins of Essouk are the most extensive in the
wider region where the Arabic records place Tadmakka, and oral traditions
also clearly identify these ruins as those of Tadmakka.70 Additionally, in
the 1980s the epigrapher Paulo Moraes Farias discovered an Arabic
inscription at the ruins reading ‘and there shall remain a market in con-
formity to (or with a longing for) Mecca (or Bakka)’,71 providing confir-
mation of the site’s identification with Tadmakka, fitting as this does with
the translation of Tadmakka’s name quoted by al-Bakri (‘resemblance to
Mecca’). This is one of a series of Arabic and Tifinagh (indigenous Berber
script) inscriptions carved in the cliffs overlooking the town. These Arabic
cliff inscriptions also include the earliest known internally dated writing in
West Africa, from the early eleventh century AD.72 On the Essouk cliffs
there are also petroglyphs of animals, people and objects, some clearly pre-
Islamic in age.73 While the wider landscape around Essouk is devoid of
other known settlement evidence, at a distance of c.40 km is the much
smaller site of Junhan, which also features some early Islamic Arabic
inscribed gravemarkers.74

66
Levtzion and Hopkins 2000, 274. 67 See Levtzion and Hopkins 2000, 354.
68
See Nixon 2017a, Chapters 3–4. 69 Moraes Farias 2003; 2017.
70
See Nixon 2017a, Chapter 2. 71 Moraes Farias 2003; 2017.
72
Moraes Farias 2003, 87–89. 73 Dupuy 1999. 74 Moraes Farias 2003, 151–52.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 637

Figure 17.6. Map showing the Essouk-Tadmakka ruins in relation to the Wadi Essouk
and surrounding cliffs, and also illustrating excavation locations (adapted from Mauny
1961; Moraes Farias 2003).

The central town ruins of Essouk comprise c.50 ha covered with stone
structures, and the excellent preservation means that these can be easily
traced both on the ground and with aerial photography.75 The various
districts of the ruins are composed of clusters of buildings with streets and
alleyways weaving around and through these, with no clear overall govern-
ing structure to layout. The level of preservation and visibility of surface
75
Nixon 2017a.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 638 Sam Nixon

Figure 17.7. East-west view across a portion of the central Essouk-Tadmakka ruins
showing the various stone structural remains distributed across the surface (the wadi
and the island within it are seen in the middle-distance).

remains allows one to trace individual roadways and alleyways, the dimen-
sions of individual buildings, and even individual room spaces. The
majority of structures were rectilinear buildings having an enclosure wall
and a courtyard with rooms arranged around it, most likely commercial or
residential buildings. There were two mosques, one a pillared building
c.25 m long with attached courtyard, principally constructed of stone, but
also having some preserved traces of mud construction. The other mosque
was a single room structure (un-pillared) c.8.5 m in length, again con-
structed of stone. In the south-eastern area of the ruins there was also
a series of large open spaces defined by low walls, possibly for the corralling
of animals. There is no evidence of a surrounding wall for the town (though
a long wall is traceable on the west side of the ruins, seemingly to prevent
flooding from the wadi). Beyond the area of the central town and ceme-
teries several isolated structures can be observed, most importantly
a musalla (open prayer ground) to the north of the town.
In 2005 excavations were undertaken by the author and colleagues
to achieve a better understanding of this previously unexcavated town
and its trade.76 Three areas of the site were excavated (see Fig. 17.6).
The excavations recorded and radiometrically dated a 5-m occupation

76
Nixon 2009; 2017a.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 639

Figure 17.8. Plan of selected structures present on the surface of Essouk-Tadmakka,

providing representative examples of the nature of the buildings seen across the surface
at the site (developed by Benoit Suzanne from EOM aerial photos taken by Raymond
Mauny). For full site plan, see Nixon 2017a.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 640 Sam Nixon

Figure 17.9. Essouk-Tadmakka: looking down into excavation unit Ek-A from the
surface – the image shows digging at depth of c.5 m, the depth of the earliest walls
revealed within the unit (note: stone structural remains seen in this image represent
a palimpsest of different building constructions built one on top of the other, using
previous walls as the foundations for new constructions).

sequence associated with permanent architecture, dated c.AD

900–1400, and a further 1.5 m of undated cultural deposits below
this. Essouk’s occupation was seemingly uninterrupted, despite major
rebuilding phases. In the earliest levels (those below the first walls
recorded) post-holes were recorded within relatively rich deposits,
probably indicating semi-permanent structures. No Trans-Saharan or
other long-distance trade evidence was found in association with
these. The earliest architecture recorded, dated to c.AD 900 and at
c.5 m below the site surface, is a well-built structure in a combination
of dry-stone and rammed earth (Fig. 17.9). Throughout the sequence,
varying combinations of dry stone and mud-architecture existed. The
later phases of construction saw more rudimentary dry-stone con-
struction, though associated also with the use of red ochre plaster.
From the survey results it can be seen that the majority of the
structures on the surface dated to the fourteenth century – thereby
covering the earlier buildings – though certain sectors of the ruins
were abandoned in the eleventh century.
Evidence for long-distance trade is seen throughout the sequence,
though significant North African imports were not present until c.AD

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 641

1000. Before evidence for North African imports there is a significant

marker for the gold trade, in the form of gold-coin moulds.77 The portion
of the sequence c.AD 1000–1100 saw the richest evidence of North African
imports, in the form of glass, glass beads, and glazed ceramics. Amongst
the botanical remains we also recorded wheat grains, providing a strong
parallel with the Arabic references to wheat cultivation at Awdaghust.

Gao
For traders continuing south of Tadmakka the next major market centre
was that of Gao, c.300 km distant. While Gao (Kawkaw) was first men-
tioned by name in the ninth century,78 it was not until the end of the tenth
century that there was reference to a merchant town of Gao. This is the
record of al-Muhallabi, supposedly dating to the tenth century, but
recounted in the twelfth century by Yaqut:
Their king pretends before his subjects to be a Muslim and most of them pretend to
be Muslims too. He has a town on the Nile [i.e. the Niger], on the eastern bank,
which is called Sarnah, where there are markets and trading houses and to which
there is continuous traffic from all parts. He has another town to the west of the
Nile where he and his men and those who have his confidence live.79

This text therefore describes a market town as one element of a larger

settlement, and seemingly separated from the indigenous royal town by the
Niger river, though this is complicated by the statement that the ruler
himself purported to be Muslim and to also be associated with the market
town.
Al-Bakri also recorded the presence of two settlements at Gao/
Kawkaw.80 One of the towns was evidently inhabited by Muslims and the
king of the Sudanic polity of Gao was also described by al-Bakri as being
Muslim. In the twelfth century al-Idrisi referred to Gao as a major town,
but there is no reference to a separate merchant centre.81 The twelfth-
century documents do provide an account of the merchants of Gao, though
they do not obviously describe North African merchants, and more likely
referred to local merchants who inhabited the Tuareg/Songhay world
between and including Tadmakka and Gao:

77
Nixon et al. 2011.
78
On the identification with ‘Kawkaw’, see Levtzion and Hopkins 2000, 7.
79
Levtzion and Hopkins 2000, 174. 80 Levtzion and Hopkins 2000, 87.
81
Levtzion and Hopkins 2000, 113, 186–87.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 642 Sam Nixon

Their merchants wear chemises and mantles, and woollen bands rolled around
their heads. Their ornaments are of gold. The nobles and eminent persons among
them wear waist-wrappers. They mix with the merchants, sit in their company, and
take shares in their wares [participating in the profit] by way of muqarada.82

From the thirteenth century onwards we do not have particularly clear

descriptions of the merchant centre of Gao but, it is assumed on the basis of
the Arabic sources that the Sudanic polity of Gao as a whole continued to
thrive in terms of commerce.83 Indeed, one can only assume that a strong
commercial function was retained in the town right up to the end of the
early Islamic era, since it continued to be a major centre and in the fifteenth
century became the capital of the Songhay empire.
It can be seen therefore that we can gain useful ideas of Gao and its
merchant population from the sources regarding its earlier period espe-
cially, and the key motif relates to some form of division between the
merchant town and the royal town at that time. It is unclear what we
should make of the lack of new references to this merchant settlement in
the twelfth century and the documents after that point are not terribly clear
on the nature of merchant activities at Gao nor on the nature of the urban
complex as a whole.
Modern Gao is situated at the Niger Bend in Mali, on the left bank (or
northern side) of the river. In and around it there are archaeological
components of varying ages, including Gao Ancien, Gao-Saney, and the
Gao-Saney cemetery (see Fig. 14.7).84 Gao Ancien has been identified with
the Sudanic capital of Arabic records, and Gao-Saney with the merchant
centre. Due to the relationship between the names of the site of Saney and
the reference in the historical records to the merchant site of ‘Sarnah’ (see
above) this seems a reasonable identification and this appears further
confirmed by the epigraphy at Gao-Saney cemetery, which contains within
it references to the name ‘Saney’.85 It is important to point out, however,
that these two settlements are not separated by the Niger River but rather
by a small river which runs into the Niger (see Fig. 14.7).
The site of Gao Ancien is difficult to access as the modern city of Gao
rests atop it. This ‘urban archaeology’ was first seriously investigated by
Insoll who opened a small exposure in an open square in the city’s old
quarter at the northern end of the town.86 There he found a sequence of
occupation that, on the basis of dateable trade goods rather than 14C
82
Al-Idrisi in Levtzion and Hopkins 2000, 113.
83
See index entries in Levtzion and Hopkins 2000.
84 85
Cissé 2017; Cissé et al. 2013; Insoll 1996; 1997; 2000. Moraes Farias 2003.
86
Insoll 1996; 1997; 2000.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 643

determinations, started in the ninth century. However, structural remains

do not appear to be present until the late ninth or early tenth century when
mud architecture began. In eleventh-century deposits Insoll found
a substantial cache of hippopotamus ivory, no doubt destined for the
Trans-Saharan trade.87 Directly overlying this cache was more mudbrick
architecture, which Insoll believed to date to the eleventh/twelfth centuries.
It was in these deposits that Insoll found the bulk of his Trans-Saharan
trade goods (glass beads, glass fragments, and glazed pottery). Overlying
this horizon was a very substantial structure made of fired brick, with fired
tile floors (Figs 17.10–17.11). This structure was seized upon by the Malian
media as a portion of the ruins of the Mosque of Mansa (Kankou) Moussa,
rumoured to have existed in the vicinity of these excavations. The name has
stuck to the site, although it is now known with certainty that Insoll’s
structure pre-dates the fourteenth century reign of this illustrious Malian
monarch. Insoll believed the structure to date to the twelfth or thirteenth
century on the basis of associated trade items. However, much more
extensive excavations of this area from 2003 to 2005 by Téréba Togola,
the late director of the Malian Direction National du Patriomoine
Culturelle,88 and subsequently Mamadou Cissé’s continued work at Gao
have shed significant further light on the structures here. The dating of
these buildings now looks to favour the eleventh century, including a large
monumental pillared building made of schist and with extensive evidence
for trade goods, dating most probably to the eleventh century
(Fig. 17.10).89 The more recent deposits at Gao Ancien are much disturbed
and do not present a sequence of any great certainty.
The tell and cemetery of Gao-Saney have seen a number of investiga-
tions over the years, but have also suffered from heavy looting – largely to
get at the quantities of valuable glass trade beads which occur in the tell, or
to remove epigraphic stone stele from the cemetery.90 The site was inves-
tigated, but never fully published, by Flight in the 1970s.91 Moraes Farias
undertook an extensive investigation of the Arabic inscriptions from the
cemetery,92 recording a number of very important inscriptions, dating
from c.AD 1100 onwards, including a number that refer to royal connec-
tions at the site. Certain of these were marble stelae imported from south-
ern Spain. Some of the inscriptions refer to individuals with Berber names
and point to a population with mixed Berber-Songhay ancestry. The royal
presence at the supposed merchant site of Gao-Saney raises questions as to

87 88 89 90
Insoll 1995. Sanogo et al. 2006. Cissé 2017; Cissé et al. 2013. Cf. Insoll 1993.
91 92
Flight 1975. Moraes Farias 2003.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 644 Sam Nixon

Figure 17.10. Structural remains detected during large-scale exposure at Gao Ancien,
including pillared schist building (after Cissé et al. 2013).

the clear division between an indigenous royal site at Gao Ancien and
a merchant site at Gao-Saney, a question in fact already indeed raised by
a careful reading of the Arabic descriptions themselves. A single radio-
carbon date for the tell 1000 ± 70 BP (calAD 891–1204) was obtained by
Raimbault and Sanogo as part of the Malian national site inventory pro-
gramme in the 1980s.93 Flight suggested that the site was abandoned in the
thirteenth century, as there are no dated inscriptions from after that time.
However, Insoll has suggested that occupation may have continued until
much more recent times, given the date of a seventeenth-century AD
93
Raimbault and Sanogo 1991, 520.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 645

Figure 17.11. Close-up of example of architectural construction in early Islamic Gao

Ancien – this illustrates the use of multiple building materials in this structure,
including fired brick, stone, and air dried mud brick (image courtesy of Timothy
Insoll) – this is part of the larger complex illustrated within Figure 17.10.

Chinese ceramic fragment found on the surface of the site.94 New investi-
gations are required on this heavily damaged site before the nature of its
sequence of occupation can be fully understood.

Other Early Islamic Market Centres in West Africa

and Their Archaeology
While the localities reviewed above are the most prominently referred to in
discussions of early Islamic entrepôts at the Southern Saharan fringes in
West Africa, it is important to clarify that there were a range of other
trading localities closely associated with the Trans-Saharan trade networks
in West Africa. I am here referring to sites potentially identifiable as
primarily centres of Trans-Saharan commerce – rather than other centres
of trade further south in the Sahel such as Jenné-jeno and Dia, as well as
other Medieval sites, including those associated with the capitals of the
kingdom of Mali.95

94 95
Insoll 1996, 12. Bedaux et al. 2005; MacDonald et al. 2011; McIntosh, S. 1995.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 646 Sam Nixon

Working my way west to east across the Sahel, firstly, there is a group of
sites within the wider sub-region around Tagdaoust. Located to the desert
to the north of Awdaghust, Azugi has been identified as an important early
centre of the Almoravids (Fig. 17.1).96 Aerial photography has provided
a useful record of the stone-built ruins found on the surface, and limited
excavation enabled a sample of these to be dated to the Almoravid era. Not
far from Azugi is Chinguetti (see Fig. 17.1), a locality which became very
important for caravan routes from at least the fifteenth century, and was
particularly associated with the Hajj journeys across the Sahara – there is
though strong potential its trading history precedes this period and the lack
of archaeological research there needs to be rectified.97 Located just south
of Tagdaoust is the town of Walata, clearly an important part of the Trans-
Saharan system by the mid-fourteenth century, and specifically recorded as
a trading centre associated with the Empire of Mali.98 Walata is another
centre which has seen no archaeological research.
The Niger Bend is the next area where a series of further trade localities
are located where future research has great potential. Excavations close to
Timbuktu, at the site of Tombouze, have revealed a settlement dating back
to the mid-first millennium AD, and whose finds include long distance
trade goods, with the site seemingly abandoned by the tenth century.99
Given its proximity to Timbuktu this site has generated great interest and
highlights the need to explore the earlier levels of the city of Timbuktu
itself. Timbuktu is referred to in the Arabic sources from the mid-
fourteenth century.100 Excavations at Timbuktu have been attempted,
but did not manage to reach Medieval levels due to the great depths of
more recent deposits.101
Located between Timbuktu and Gao the early Arabic records refer to the
market centre of Tireqqa, at a point on the River Niger where merchants
from Ghana and Tadmakka met.102 Bonnel de Mézières claimed to have
identified this site 35 km east of Timbuktu,103 though this identification
has been much disputed and a clear resolution or further investigation is
still awaited. Downriver of the Niger Bend is the complex of Bentyia-
Kukiya (see Fig. 17.1), a place which was part of the same Songhay culture
sphere as Gao, and which also features cemeteries with significant quan-
tities of early Arabic inscriptions.104 The exact nature of this site and its
trading population is unclear, but given the significant presence of Arabic
96 97 98
Saison 1981. Nixon 2013; Norris 1962. Levtzion and Hopkins 2000, 284–86.
99 100
Park 2010. Levtzion and Hopkins 2000, 287. 101 Insoll 2002.
102
Al-Bakri, eleventh century AD – see Levtzion and Hopkins 2000, 84–85.
103
Bonnel de Mézières 1914. 104 Arazi 1999.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 647

inscriptions (and the rarity of these in West Africa) this is an important

place to flag up in this debate.
Further to the east, in Niger, is the site of Azelik-Takadda, related to the
town of Takadda, an important trade centre referred to in the Arabic sources
from the mid-fourteenth century.105 There have been some important
studies of the stone ruins here, albeit mainly restricted to surface
studies.106 To the south of the modern city of Agadez in Niger is the site of
Marandet, a locality which has presented some of the earliest Trans-Saharan
trade evidence, though it appears to have been principally associated with
copper production, rather than being an entrepôt centre as such.107
The most crucial missing ‘piece of the puzzle’ of the southern end of
Trans-Saharan trade is perhaps the Kawar oases sites.108 This complex is
a series of mud-built fortified villages on the route from Fazzan to the West
African state of Kanim, broadly relating to the forts of Kawar first referred
to in ninth-century accounts of the raid of Uqba bin Nafi.109 While
sustained archaeological investigations have not taken place here, a series
of sites identified by satellite remote sensing appear to provide some
support for a pre-Islamic permanent occupation at these sites.110

Discussion

Pre-Islamic Occupations at the Sites of Early

Trans-Saharan Market Towns
Since the beginning of research into early Trans-Saharan trade, a crucial
question has been whether the localities mentioned in the early Islamic
sources had pre-Islamic levels, and if so what was their nature. In parti-
cular, there has naturally been a desire to discover whether or not the early
Islamic urban centres overlay pre-Islamic urban sites, especially given the
failure to find any obvious pre-Islamic Trans-Saharan entrepôt sites else-
where in the western Sahel. There is then a great importance in under-
standing any pre-Islamic levels below the early Islamic sites, both in order
to contextualise the evolution of the specific Trans-Saharan market towns,

105
Levtzion and Hopkins 2000, 301–3. 106 Bernus and Cressier 1991.
107
Magnavita et al. 2007. 108 Lange and Berthoud 1977; Vikør 1987.
109
Levtzion and Hopkins 2000, 13.
110
See Sterry and Mattingly, Chapter 7, this volume; whether this hypothesised permanent
settlement can be associated with Trans-Saharan trade evidence would be the next crucial
question if and when the early dating can be confirmed.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 648 Sam Nixon

and also what this might tell us more widely about the interface between
early Islamic and pre-Islamic trade and settlement.
Looking firstly at the settlements of Awdaghust/Tagdaoust and
Tadmakka/Essouk – the best explored Trans-Saharan market towns in
the far northern Sahel – there is little positive evidence for substantial pre-
Islamic permanent settlement, either below or adjacent to these sites. The
extensive excavations at Tagdaoust produced an earliest date for perma-
nent occupation around the eighth/ninth century,111 while at Essouk the
earliest evidence recorded was dated to c.AD 900.112 These dates would
appear to correlate very well with the nascent Islamic trade described in the
early Arabic sources at these two locations. While clearly there is still
potential for recovering evidence of a pre-Islamic urban centre at Essouk,
this does seem unlikely. In addition to the absence of evidence from these
sites, there is also no clear evidence for pre-Islamic permanent settlement
in the wider catchment areas around these sites. Aside from the archae-
ological evidence, it is important to highlight the early Arabic historical
evidence discussed by Lewicki for the name of the capital of Tadmakka,
‘Zakram’ (‘town’ or ‘built up locality’), suggesting permanent settlement
was a concept that had only recently arrived in the region.113
While there is no compelling evidence for pre-Islamic urban settlements
at either Tagdaoust or Essouk, the evidence from these sites does indicate
some form of cultural activity at these locations prior to the urban settle-
ments, likely relating at least in part to the pre-Islamic period. At Essouk
for instance the c.1.5 m of deposits associated with multiple levels of post-
holes below the earliest recorded permanent architecture would appear to
indicate some form of semi-permanent settlement of tented or mat-and-
pole structures.114 In addition there is the extensive rock art at the site
testifying to Essouk’s long-term role as an important point in the land-
scape, and the obvious significance that all locations with good water
sources like Essouk and Tagdaoust would have had as meeting places
before the Islamic era. The existence of some form of pre-Islamic Trans-
Saharan trade network before the early Islamic era is unquestionable –
based upon such finds as those from Kissi in Burkina Faso – and the nature
of Tagdaoust or Essouk as obvious nodal points close to the southern edge
of the desert plausibly links them into these networks in some form.115
Accordingly, it does suggest the Islamic occupation at both of these sites

111
Devisse 1983; Polet 1985, 231–43; Robert-Chaleix 1989. 112 Nixon 2017a.
113
Lewicki 1981. 114 Nixon 2017a. 115 Fenn et al. 2009; Magnavita 2009; 2013.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 649

was grafted onto pre-existing settlements of some form, rather than start-
ing on ‘virgin’ ground.
We move next to discuss the trading settlements that lay south of
Awdaghust/Tagdaoust and Tadmakka/Essouk, in the realms of the states
of Ghana and Gao. The Arabic sources give the impression these West
African states were already in existence at the beginning of the Islamic era
(see above), therefore raising the distinct possibility that the early Islamic
Trans-Saharan market centres were built at pre-existing settlement com-
plexes. There is though as yet little compelling evidence from the archae-
ology for a true urban centre preceding the Islamic era at Ghana116 or
Gao.117 At the same time, it also has to be said that the absence of evidence
so far should not necessarily be taken as evidence of absence. At Gao the
sequence is not entirely clear, due to much of Gao Ancien being obscured by
the modern town and Gao-Saney being a hugely problematic site due to the
extensive looting that has taken place there. Likewise, the uncertainty over
the location of early ‘Ghana’118 has made the nature of this pre-Islamic state
and its settlement network an even more problematic question. The site of
Kumbi Saleh does not really illuminate the debate here as the site was a later
establishment to the ‘Ghana’ of the earliest Arabic sources. While it is clear
that urban settlements existed in the Sahel in the pre-Islamic period,119 we
are still a long way from defining the urban component of Ghana and Gao at
the start of early Islamic Trans-Saharan trade. This includes not being sure of
the exact nature of the ‘royal towns’ described in relation to these two polities
by the early Arabic sources – one must indeed keep open the idea that these
were settlements created for trade at the beginning of the early Islamic era,
rather than the capitals of the states themselves.
In addition to the four key localities discussed above, two other locations
are particularly crucial to this debate. Firstly, the site of Tombouze,120
immediately next to Timbuktu, which appears to show a pre-Islamic
settlement of some form connected with long-distance trade, and one
that then also continued into the earliest period of the Islamic era. While
this was clearly an important site, crucially we do not know if it was
a dedicated Trans-Saharan centre as such, and there exists no Arabic
records or solid archaeological proof to indicate the presence of an
Islamic trading population there.

116
Berthier 1997; Bonnel de Mézières 1923; Thomassey and Mauny 1951; 1956.
117
Cissé 2017; Cissé et al. 2013; Insoll 1996; 1997; 2000.
118
Lange 1996; Masonen 2000; McIntosh, R. 1998, 256–58.
119
McIntosh, S. 1995; Chapter 14, this volume; McIntosh, S. and McIntosh, R. 1993.
120
Park 2010.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 650 Sam Nixon

Away from the trading routes leading to Ghana and Gao and the western
Sahel, the route through Kawar to Kanim presents another key element to
this debate. The ‘urban’ nature of this is still highly debatable and clear
identification of a pre-Islamic urban presence here must await excavation
of this important site complex. The possibility that the Kawar oases were
plugged into Trans-Saharan trade networks at so early a date is also
unsubstantiated.

Trans-Saharan Settlement in the Early Islamic Era: Local

and Regional Dynamics
The broad model for understanding the development of the earliest for-
mation of Islamic entrepôts in West Africa is of a settlement patterning
related to two cultural and environmental zones, associated with two stages
of the trade routes at the western end of the Trans-Saharan exchange
network.121 Awdaghust and Tadmakka represent entrepôts established at
the most northerly habitable point of the Sahel, where the Berber world of
the Sahara has its most southerly extension. Situated in a precarious arid
environment, these towns represented isolated permanent centres in
a wider nomadic world, along the lines of desert oases. Further south,
there were then the merchant towns of Ghana and Gao, established in
proximity to the centres of the most powerful Sudanic states supplying the
key commodities of gold, slaves, and ivory. These centres were situated in
a relatively less arid environment and more easily able to support wider
networks of permanent settlement. Here I explore this broad model –
examining how the market towns were positioned in relation to contem-
poraneous settlements in their immediate environment, and the wider
landscape of settlement in their wider catchment area – as well as high-
lighting a number of other questions in relation to local and regional
settlement dynamics.
I first look at the desert edge towns of Awdaghust/Tagdaoust and
Tadmakka/Essouk.122 Exploration of the immediate periphery of these
settlements appears to confirm that they were isolated urban establish-
ments in a wider settlement landscape. However, we do need to consider
the question of temporary or semi-permanent occupations in the immedi-
ate environs of these sites that have left less obvious archaeological traces.
Given the functioning of these sites as exchange centres for at times
extremely large camel caravans – early Arabic authors referring in some

121 122
See Insoll 2003, Chapter 5. Devisse 1983; Nixon 2017a.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 651

cases to up to 12,000 camels in a single caravan123 – it is quite possible that

a series of semi-permanent settlements of some size might have existed on
their peripheries. While no systematic archaeological soundings were
made of the peripheries of these sites, one possibly indicator of activity in
their immediate vicinity are the musalla which are found c.1 km outside
both Tagdaoust and Essouk, religious structures that would have served the
short-term needs of visiting camel caravans.
Another question relates to the occupation of the wider area around
these sites by desert nomads, groups who were clearly the political powers
of the regions within which the market towns were located. An early Arabic
reference to Ibn Battuta’s visit to the trading town of Takadda (in modern
Niger) recounted the meeting at which he paid his respects to the nomadic
leader under whose authority Takadda fell. This took place at a location
one day distant from the town.124 Importantly, Ibn Battuta referred to the
existence of ‘houses’ at this location, though it does not appear that this
leader himself lived in a house. Almost certainly the same dynamic would
have existed at Awdaghust and Tadmakka. While archaeologically there is
not a huge visibility of nomadic or semi-permanent settlements in the
wider region, we should not ignore the well-defined interaction networks
and political relationships that would have existed between these desert
edge urban centres and nomadic groups in the surrounding region.
On a regional level, also, we need to be aware of the existence of other
permanent settlements contemporaneous to these major desert-edge cen-
tres. The existence of a range of smaller permanent settlements in the wider
region around Tagdaoust is seldom commented on.125 Likewise, the exis-
tence of the permanent settlement of Junhan in the wider region around
Essouk seldom features in relation to discussion of this town.126 There
needs to be further consideration of how such smaller permanent centres
functioned within the orbit of the larger Trans-Saharan towns, including
the possibility that they fulfilled a function as semi-permanent centres of
the ‘nomadic’ political authorities of the region.
While Awdaghust and Tadmakka evidently continued for some time as
important centres in the region, it appears that their roles changed, poten-
tially following the rise to power of the Almoravids in the mid-eleventh
century. Certainly, by the fourteenth century they were no longer the major
locations, with Timbuktu, Walata, and Takadda described as of
importance.127 Effectively then the urban legacy of the two principal
123
Levtzion and Hopkins 2000, 336–38. 124 Levtzion and Hopkins 2000, 302–3.
125
Robert et al. 1970. 126 Moraes Farias 2003.
127
See index entries in Levtzion and Hopkins 2000.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 652 Sam Nixon

early Trans-Saharan market centres in the Sahel/Sahara borderlands was

taken up by these other localities. While considering these large-scale
regional shifts within the urban network of Trans-Saharan trade in the
relatively abstract terms of larger economic, geographical and political
shifts in the networks, we must also recognise the role of population
transfers between these centres. For example, in the immediate post-
Almoravid period at Awdaghust it seems very likely that merchants and
skilled craftspeople were either forcibly moved to new localities within the
Almoravid network – potentially even Kumbi Saleh for instance – or that
they migrated of their own will to new centres. In the case of Tadmakka,
the oral historical accounts and the archaeology demonstrate a transfer in
the fourteenth century of the population of Tadmakka towards the region
of Azelik-Takadda and the new urban centres which grew up there.128
These strands of movement and influence between urban centres are
difficult to explore, but they nonetheless form an important element of
the story of these urban centres in the Sahel/Sahara borderlands.
Further south, the market towns of Ghana and Gao are most commonly
understood in terms of their positioning within a ‘dual town’ relationship,
in proximity to an indigenous ‘royal town’ (even though the pre-Islamic
background to settlement in the areas of these polities is unclear). Due to
the confusion concerning the identification of archaeological sites in rela-
tion to Ghana, we are unfortunately unable to make a clear assessment of
the statement in the Arabic sources that the market town of Ghana was
located six miles distant from the royal town.129 However, we do have
unequivocal archaeological evidence for the dual town complex described
by the Arabic sources at Gao.130 The existence of two distinct early Islamic
settlements at a distance of c.2 km, and the clear correspondence between
the name ‘Sarnah’ given for the market town in the Arabic sources and the
name ‘Saney’ within the epigraphy of the cemetery of the site known as
Gao-Saney, allows us complete confidence in the existence of such a dual
early town.131
While accepting the existence of this ‘dual town’ paradigm, the Arabic
historical records and archaeology provide important data to help nuance
this model. The Arabic historical records describe the attempt by the ruler
of Gao to negotiate a dual identity in relation to both the royal and the
merchant town, appearing alternately as both non-Muslim and Muslim
respectively in the royal and the merchant town. Unsurprisingly, these
128
See Nixon 2017a, Chapter 12. 129 Levtzion and Hopkins 2000, 79–80.
130
Cissé 2017; Cisse et al. 2013; Insoll 1996; 1997; 2000; Levtzion and Hopkins 2000, 174.
131
Moraes Farias 2003.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 653

records describe not only cultural exchange, but a culturally mixed mer-
chant population, seemingly linked to and merged with the indigenous
population. The archaeology provides concrete evidence of such
a culturally mixed and complex picture, in the form of the Arabic inscrip-
tions bearing royal titles in the merchant town.132
Given that Gao-Saney is an abandoned settlement with no apparent
modern-era occupation, while Gao Ancien is still occupied, it seems clear
that at some point the ‘dual town’ model was abandoned, though the
archaeology is not clear enough for us to know exactly when and why.
One assumes this was at a point when the increasing Muslim population
within the indigenous town made the existence of separate royal/indigen-
ous and merchant/Muslim settlements no longer necessary or desirable.
Perhaps this might have been as early as the twelfth century, as by that time
there is no mention of a ‘dual town’ in the Arabic records and the archae-
ology could accord with this too.133 While it remains at the level of pure
hypothesis, one can also imagine that a similar process of abandonment of
a ‘dual town’ model took place in late period Ghana or with its successors
in the region.
Given the very limited nature of the archaeology conducted on the Kawar-
Kanem Trans-Saharan route it would be pure speculation to examine such
questions in that region. The regional patterning of a significant series of
settlements in the Kawar oases area may suggest a different regional dynamic
to that occurring in relation to the settlements further west.134

Urban Structure and Urban Features

When we speak of Trans-Saharan market towns, we need to establish
crucially what these looked like, how they were structured and what
features they possessed. These data allow us to gain a sense not only of
their physical appearance, but also what facilities and services they pro-
vided as urban centres. In approaching this question obviously one must be
clear that these towns were not single-phase constructions, but had sig-
nificant occupations evolving over periods of 500 years plus, with phases of
destruction and rebuilding, and with urban stratigraphies of up to 5 m
depth. The complexity of understanding these urban forms is obviously
rendered more difficult for the earlier occupations, most of which are
covered by later occupations. This said, the combination of surface ruins

132
Moraes Farias 2003. 133 Levtzion and Hopkins 2000, 113.
134
Lange and Berthoud 1977; Sterry and Mattingly, Chapter 7, this volume; Vikør 1987.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 654 Sam Nixon

and excavated deposits enables us to build up a good picture of the urban

structure and features of the desert-edge entrepôts, as well as to make
certain statements about other towns. The Arabic sources also provide
important additional information. I shall explore how these towns might
have looked during the period of their initial urban structures and the
apparent period of their apogee, as well as trying to examine their nature in
their later phases, examining first the more easily approachable desert edge
towns of Awdaghust/Tagdaoust and Tadmakka/Essouk, followed by a brief
consideration of the available evidence for other towns.
The desert edge towns of Awdaghust/Tagdaoust and Tadmakka/Essouk
with their stone ruins, excellent preservation, and lack of overlying modern
structures enable us to understand a significant amount about the nature of
these key urban complexes, including their overall settlement distribution
and their individual attributes. Firstly, neither of these towns appears to
have been fortified, nor have contemporaneous refuge sites been identified
nearby, in the manner of Sedrata or Sijilmasa.135 Essouk lay within
a relatively protected valley with good lookout positions, and the terrain
around Tagdaoust likewise provides a good outlook. However, both places
relied for their security largely on being embedded within powerful noma-
dic confederations whose people occupied the surrounding landscape.
Both sites also had a symbolic definition as places of Islam. Most obviously
at both sites this is communicated by the presence of Muslim cemeteries on
the periphery of the town. In the case of Essouk practically every entry
point into the town passed a cemetery. The use of Arabic epigraphy at one
of the crucial entry-points into Essouk defined it as a place of Islam by
stating its identity as ‘a market resembling Mecca’.136 While there is no
concrete evidence of settlements in the immediate periphery of these sites,
these urban spaces surely existed within a surrounding semi-permanent
landscape of mobile settlement, associated both with visiting caravans and
with groups offering services to the town, including craft production. It is
also interesting that both towns possessed a musalla, positioned to the
north of the settlement in the direction of the Trans-Saharan routes. These
were no doubt used by large caravans arriving and departing the town, as
well as for larger communal prayer on feast days, for example. The final
thing we need to consider in terms of the periphery of the town is the
potential for market places to have existed there, in the same manner as is
hypothesised for Sijilmasa.137 Clearly the urban infrastructures of these

135
Aillet et al. 2017; Capel, Chapter 16, this volume; Messier and Miller 2015; Van Berchem 1960.
136
Moraes Farias 2003. 137 Messier and Miller 2015.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 655

towns were not designed to support the influx of very large caravans and it
would have made sense to have certain market functions take place outside
the town.
In terms of the internal structure of the urban centres of Awdaghust/
Tagdaoust and Tadmakka/Essouk the combined textual and archaeological
evidence enables us to build up a reasonable picture. In the earliest period,
both Tagdaoust and Essouk possessed multiple complex buildings across
different areas of the site. From the Tagdaoust evidence this seems some-
what random and unplanned, something which would make sense within
the initial phase of settlement. With the arrival of diverse merchant groups,
different family, cultural, and religious units, it is likely that occupation
took place at several locations within the valley,138 rather than a planned
urban centre developing outwards from a central core. The findings from
Tagdaoust however indicate that after this initial phase of occupation
a more clearly planned settlement developed from around the tenth cen-
tury, and the excavations of various wards from this period clearly show
a very solid structure of linked building compounds and associated lanes
weaving between them (see Fig. 17.3). The surface traces and excavated
buildings at Essouk suggest a comparable situation. Corresponding with
the dated urban deposits, epigraphy from Essouk clearly shows that at least
three of the six cemeteries were in use by the early eleventh century. This
gives us a sense of the scale of the town already by this point, as well as
potentially indicating the division of the town into wards relating to
different groups. Crucially, one of the cemeteries at Essouk features no
inscriptions, and it has been suggested that this might accordingly relate to
an early Ibadi group. One assumes this division seen in death would also
have been represented in the urban structure by a putative Ibadi popula-
tion enclave as well.
In terms of specific features of the urban environment which these
towns possessed, the Arabic records make mention of the following:
markets (though without specifying whether inside or outside the town),
mosques, ‘oratories’, and ‘treasure houses’. The archaeological evidence for
structures appears to indicate unsurprisingly the presence of a range of
domestic and commercial buildings. Amongst these are simple large
structures with walls enclosing a large courtyard space, featuring only a
few rooms. There is a strong chance that these were either the ‘treasure
houses’ noted in relation to Tadmakka,139 or potentially something
138
As is the case at the modern settlement developing near to Essouk, observed by the author
during fieldwork.
139
Note also the recording of what appears to be a door-key at Essouk-Tadmekka: Nixon 2017a.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 656 Sam Nixon

approaching a rudimentary caravanserai. Some further structures may also

potentially equate to caravanserai, while the layout of other buildings
appears to resemble a market-type street. Early mosques have been identi-
fied at both these town sites, at Tagdaoust a larger ‘Friday’ mosque and at
Essouk what appears to be a single-roomed mosque, potentially relating to
the Ibadi faith.140 A further important feature to note is the presence of
industrial areas and workshops. At Tagdaoust a developed industrial area
was located on the perimeter of the town,141 while at Essouk evidence
across the town indicates that metal working and other crafts took place
within the urban fabric.142 One important further feature of the towns was
the presence of possible patch gardens – wheat is referred to as being grown
at Awdaghust and it is also present in the excavated material at Tadmakka.
Consideration must also be given to features that appear to be absent.
There is nothing distinguishable from other clusters of buildings that can
be clearly equated with a palace, and likewise certain key features of Islamic
towns have not been recorded, including hammams (bathhouses).
Nonetheless, the evidence we have available clearly shows the develop-
ment of towns that to the eyes of arriving merchants coming from North
Africa would have appeared as recognisably Muslim towns – regardless of
the ‘non-Muslim’ actions of some of the residents, as commented on for
example by al-Bakri. Accordingly, it is not surprising that al-Bakri and
others compared these towns to Mecca, as well as alluding to Tadmakka as
being ‘better built than Ghana or Kawkaw’,143 thereby identifying it as
a recognisable Islamic town. We must however be careful not to over-
characterise these places as classic Islamic ‘urban’ centres of trade. Again,
returning to al-Bakri, we should note his description of the inhabitants of
Tadmakka living like the Berbers of the desert. This would also surely have
been reflected in their ‘urban practice’. The contemporary practice of
erecting a tent within the courtyard of buildings in Kidal is perhaps such
a desert-dwelling practice that might have occurred in these towns.144
Following the early period of both Awdaghust/Tagdaoust and Tadmakka/
Essouk many aspects of the urban structure of these towns continued
through to their decline in the fourteenth century. A measure of continuity
can be seen in the practice at Essouk of using earlier house structures as the
foundations for new buildings throughout the entire sequence
(see Fig. 17.9). However, there were clearly significant changes within the
140
Elizabeth Fentress, personal communication; the pillared mosque seen on the surface at
Essouk relates to the terminal occupation of the site.
141
Vanacker 1979. 142 Nixon 2017a. 143 Levtzion and Hopkins 2000, 85.
144
As noted by the author.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 657

urban structure. The abandonment of whole areas of both towns half-way

through the sequence is striking, clearly showing that for long portions of
their occupations these towns had ruins in some sectors. The entire indus-
trial area at Tagdaoust was abandoned from the eleventh century. One can
obviously over-dramatise this period of decline and both centres retained
key features of Islamic towns. For example, Essouk’s Muslim cemeteries
continued into the fourteenth century and there was a fourteenth-century
pillared mosque built, with attached courtyard. The final phase structures of
both sites provide a clear picture of a complex urban establishment, with
multiple wards and a variety of built features. The lack of compelling
accounts of these towns by the Arab authors at this time has, however,
been seen as a sign of their declining status as Muslim urban centres.
Compared to Awdaghust and Tadmakka, we have far less evidence of the
urban structure and features of the merchant towns of Ghana and Gao. The
Arabic records for Ghana paint a picture of a flourishing town, mentioning
numerous mosques, as well as the religious, judicial, and scholarly autho-
rities of the town.145 The inability to link the Ghana of the Arabic histories
to an archaeological site prevents us saying more. For Gao, the Arabic
records are less informative about the merchant town as a physical space.
Archaeology has provided evidence of Muslim cemeteries associated with
the merchant town, but little else besides. The large, complex series of
structures from Gao Ancien does at least provide us with evidence of some
clear urban features from the Gao archaeological complex, including what
is potentially a pillared mosque.146
While Kumbi Saleh is obviously a later phenomenon, it does however
provide us with a good example of an urban locality in the semi-arid Sahel
which was built in the eleventh century. Given the limited publications on
this site, it is difficult to say much about its possible planned character.
What one can say is that certain features recorded from the town indicate
a developed urban formation. The complex domestic or commercial
structures that have been recorded over multiple periods (see Fig. 17.5),
were accompanied by a very significant pillared mosque, as well as an
intriguing pillared tomb within the cemetery.147
While again we cannot speak with any certainty about early Islamic
occupation in the Kawar oases, one fundamental difference can be posited.
The archaeological sites in the Kawar oases complex appear to have been
strikingly different from those further west, principally in relation to

145 146
Levtzion and Hopkins 2000, 79–80. Cissé 2017; Cissé et al. 2013.
147
Capel et al. 2015.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 658 Sam Nixon

having a fortified aspect to them, far more in keeping with the Central
Sahara and North African pre-desert settlements.

Architecture: Materials, Construction Techniques, and Form

The nature and evolution of architecture in West Africa has been a crucial
area of discussion for both historians and archaeologists, including in
debates on construction materials, form, and techniques.148 Beyond the
physical appearance of these Trans-Saharan centres, because of the impor-
tant role they played in bringing new ideas and cultural influences to West
Africa, these localities also provide insight into an important area of
experimentation and expression within the ‘arts’ of pre-modern West
Africa. Of particular interest is the extent of transfer of architectural style
from Islamic traditions in North Africa,149 as well as the background to the
evolution of regional styles.
In terms of materials, a key feature of the discussion of these towns
relates to the relative use of stone and mud. Stone has come to be identified
with the merchant towns – the surface remains of the key towns of
Tagdaoust and Essouk are of stone, the merchant town of Ghana was
described as built of stone, and likewise the ruins of Kumbi Saleh are of
stone. Only Gao amongst the earliest merchant towns in the western Sahel
appears not to strongly feature stone. However, this strong identification
between the merchant towns and stone is deceptive, given the sub-surface
deposits of Tagdaoust and Essouk. At Tagdaoust the initial occupation
featured primarily mud-architecture; and the later move away from solely
mud-architecture may have reflected its unsuitability in an environment
where it was subject to rapid water erosion. In the second phase of the site,
building construction tended to use both stone and mud to create more
durable structures. At Essouk likewise, the sub-surface deposits reveal far
more in the way of mud-architecture than is displayed by the surface ruins.
This includes a mixed stone/mud construction in the earliest building
excavated. The desire to build in mud was certainly influenced by cultural
factors; many of the towns like Sijilmasa or Sedrata, where merchants were
coming from, made extensive use of mudbrick and rammed earth (pisé).
Also, the load-bearing and insulating properties of mud made it an ideal
building material. However, the plentifully available local stone was also
turned to as an expedient measure.

148
Aradeon 1989; Insoll 1996; 1999; Prussin 1986; and see Nixon 2014 for a brief summary.
149
Cf. Aradeon 1989.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 659

The mud-architecture at Tadmakka was mainly rammed-earth, although

some mudbrick is also seen. At Tagdaoust mudbrick appears to have been
far more plentifully used. Whether this was just a cultural preference is
unclear, but rammed-earth does have different load bearing potential, and
this may have contributed to the differences of construction between the two
localities. The earliest building recorded at Tadmakka features either
a massive pillar, or a staircase to a roof.150 Regardless of which of these
techniques was being applied, they are evidence for the presence of skilled
artisans in mud-architecture in these desert-edge locations. While there are
only a few excavated or visible archaeological structures from Gao, there is
clear evidence of the use of mudbrick at the site of Gao Ancien.151 Given the
difficulty of defining buildings in excavations at Gao-Saney and elsewhere at
Gao, it is most likely that much of the construction was in the piled or
rammed earth technique. Indeed, the mudbrick building recorded at Gao
Ancien appears to be an exceptional structure.
One particularly important technique to point out is the extensive use of
mud-plaster, covering up stone wall cores, used both at Tagdaoust and
Essouk. The use of mud plaster both within and on the outside of buildings
clearly related to an aesthetic choice, and this appears to have been con-
nected to the relatively rough finish provided by the stone in this region
(see below). Accordingly, even where we see extensive stone building in the
sequence, the outer appearance of these buildings would have been one
related to mud-architecture. In this respect it is interesting to observe that
none of the early Arabic commentators refer to Awdaghust or Tadmakka
as ‘stone towns’, even though the market town of Ghana is described
exactly in this way (see below for further discussion of Ghana). One
particularly distinctive aspect of the mud plastering to note is the use of
thick red ochre plaster in the final phase at both Essouk and Tagdaoust
(where white paint was also used in association with this) - clearly these are
early examples of the ochre and white painted plaster techniques that
became a strong aesthetic tradition in the Southern Sahara.152
There are a number of important observations to make concerning the
nature of the stone used in the Trans-Saharan market towns. At both
Tagdaoust and Essouk, though the locally available stone made for easy if
rudimentary constructions, it required great skill to create refined stonework
(only one example of this is seen at Essouk). As inferred above, the difficulty
of dressing this stone probably led to it often being coated with mud plaster

150
See Nixon 2017a. 151 Cissé et al. 2013.
152
Prussin 1986, including for example in the modern town of Walata.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 660 Sam Nixon

to create a more aesthetic finish. While these two towns have not produced
much good quality stonework, regular stonework was more prominent at
Kumbi Saleh, indeed being the dominant feature of the architecture. Unlike
the stone found around the other two towns discussed here, the geology of
Kumbi Saleh features a particular type of stone that enables the easy creation
of thin, flat, almost slate-like blocks. These were used to build a very refined
style of stone architecture. Clearly this stone finish became a feature to be
displayed, including in the creation of triangular niches which are a feature
of the site.153 This refined stonework is also seen in the large mosque
excavated at the site. The fact that Ghana’s merchant town is referred to as
‘of stone’ within the early Arabic documents is important, suggesting that
stone was not just a core material to be covered with mud, but was the visible
aesthetic of the town. This tradition of refined stone construction is seen
more widely in the region, including for example at Chinguetti. One final
type of stone to comment on is the schist seen in the pillared building at Gao
Ancien154 and in the pillared tomb at Kumbi Saleh155 – the restriction of this
evidence to what one might term exceptional or elite structures so far does
suggest this was a highly restricted use.
There are two further specific materials and techniques to discuss.
Firstly, one of the most distinctive architectural traditions in the early
Islamic towns at the northern fringes of the Sahara is highly distinctive
decorative carved plaster work, for example seen at the Ibadi site of Sedrata
in Algeria.156 Given the supposed very strong early Ibadi influence at
Tadmakka and other West African trading sites one might perhaps expect
to see an experimentation with such styles – it is possible though that the
social and material technology for such a transfer did not exist. Secondly,
there is some evidence of the technology of fired-brick work, as at Gao
within a mihrab and portions of the large and complex structure uncovered
at Gao Ancien.157 While one sees this at North African towns linked to the
Trans-Saharan networks, including Aghmat and the Medieval site at
Zagora,158 no evidence of this has been discovered at either Tagdaoust or
Essouk. Other places where fired-brick has been found in significant
quantities in West Africa include the Kanim-Bornu zone (as at Garu
Kime). Potentially then, fired brick was restricted to elite structures and
away from the Southern Saharan edge towns.
In terms of the form of building seen in the Trans-Saharan trading towns,
the key type to highlight, and the most common seen at the sites of
153
Berthier 1997. 154 Cissé et al. 2013. 155 Capel et al. 2015.
156
Aillet et al. 2017; Van Berchem 1960. 157 Cissé et al. 2013; Insoll 1996.
158
Fili et al. 2013; 2014; Mattingly et al. 2017b.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 661

Tagdaoust and Essouk, features an enclosure wall containing rooms

arranged around a courtyard. This relates to a very common tradition in
the Islamic world that would have been the prevalent type throughout North
Africa during this period,159 as well as in the Central and Northern
Sahara.160 At Tagdaoust this building type is identified as one of the
important markers of the arrival of North African trading communities.161
One important thing to note is that in addition to the private internal rooms,
the courtyard also provides a self-contained space to undertake a range of
domestic activities removed from the public world of the town; something
which is often discussed as one of the classic hallmarks of Islamic
architecture.162 This differs greatly for instance from the small structures
built today at the settlement near Essouk-Tadmakka which lack a courtyard
and where consequently a greater amount of activities take place in the
public areas outside the structure.
The other key building types to highlight are the mosques. The
recorded mosques from the Trans-Saharan trading towns provide
a useful illustration of the wider development of Southern Saharan
mosque design. At Tadmakka the earliest mosque recovered was
a simple single-cell structure, likely a ward mosque. When the Arabic
geographers described the Southern Saharan towns as having multiple
mosques and ‘oratories’ they were most likely referring to this sort of
building. The grander form of ‘Friday mosques’ would appear to post-
date the eleventh century. That from Kumbi Saleh, dated seemingly to
around the twelfth century, has already been noted as providing a very
good example of a highly developed mosque form in well-worked stone.
The largest mosque observable amongst the ruins at Tadmakka (measur-
ing 23.5 × 15.5 m) dates most likely to the late fourteenth century in the
form we see it. This mosque had five aisles parallel to the qibla, identifi-
able by the remains of pillars. Study by Cressier indicated that it had no
minaret or moveable minbar,163 this being standard within the Southern
Saharan mosque traditions. Cressier has already highlighted the potential
regional importance of this mosque in terms of influencing later mosque
design in the wider region,164 and certainly we should consider this
important religious building within the context of Tadmakka’s acknowl-
edged role as an origin point for Islamic religious movements in Niger in
the late Medieval and early modern era.165

159
Fentress 1987; Insoll 1999, Chapter 3. 160 Mattingly 2013, Chapter 12. 161
Polet 1985.
162
Insoll 1999, Chapter 3. 163 Cressier 1992. 164 Cressier 1992.
165
Cressier 1992; Moraes Farias 2003; Norris 1975.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 662 Sam Nixon

References

Aillet, C., Cressier, P. and Gilotte, S. (ed.). 2017. Sedrata. Histoire et archéologie d’un
carrefour du Sahara médiéval à la lumière des archives inédites de Marguerite van
Berchem. Madrid: Casa de Velázquez.
Aradeon, S. 1989. Al-Saheli. The historian’s myth of architectural technology
transfer from West Africa. Journal des Africanistes 59: 99–131.
Arazi, N. 1999. An archaeological survey in the Songhay heartland of Mali. Nyame
Akuma 52: 25–43.
Austen, R. 2010. Trans-Saharan Africa in World History. New York: Oxford
University Press.
Bedaux, R., Polet, J., Sanogo, K. and Schmidt, A. 2005. Recherches archéologiques à
Dia dans le Delta intérieur du Niger, Mali: Bilan des saisons de fouilles 1998–2002.
Leiden: CNWS.
Bernus, S. and Cressier, P. 1991. La région d’in Gall – Tegidda N Tesemt (Niger).
Programme archéologique d’urgence 1977–1981. Azelik-Takadda et l’implantation
sédentaire médiévale (Études Nigériennes No.51). Niamey: Institut de Recherches
en Sciences Humaines.
Berthier, S. 1997. Recherches archéologiques sur la capitale de l’empire de Ghana:
Etude d’un secteur d’habitat à Koumbi Saleh, Mauritanie. Campagnes II–III–IV–
V (1975–1976)-(1980–1981). (Cambridge Monographs in African Archaeology
41/BAR International Series 680). Oxford: Archaeopress.
Bonnel de Mézières, A. 1914. Découverte de l’emplacement de Tirekka. Bulletin du
Comite des travaux historiques et scientifiques de l’Afrique occidentale française
29: 132–35.
Bonnel de Mézières, A. 1923. Recherche de l’emplacement de Ghana (fouilles à
Koumbi et Settah). Mémoires présentés par divers savants à l’Académie de
Inscriptions et Belles Lettres 13.1: 227–64.
Bovill, E. 1958. The Golden Trade of the Moors. London: Oxford University Press.
Capel, C., Zazzo, A., Saliège, J-F. and Polet, J. 2015. The end of a hundred-year-old
archaeological riddle: First dating of the columns tomb of Kumbi Saleh
(Mauritania). Radiocarbon 57.1: 65–75.
Cissé, M. 2017. The Trans-Saharan trade connection with Gao (Mali) during the
first millennium. In Mattingly et al. 2017a, 101–30.
Cissé, M., McIntosh, S.K., Dussubieux, L., Fenn, T., Gallager, D. and Smith, A.C.
2013. Excavations at Gao Saney: New evidence for settlement growth, trade, and
interaction on the Niger Bend in the first millennium CE. Journal of African
Archaeology 11.1: 9–37.
Cressier, P. 1992. Archéologie de la devotion soufi. Journal des africanistes 62.2:
69–90.
Devisse, J. (ed.). 1983. Tegdaoust III. Recherches sur Aoudaghost (campagnes 1960–
1965, enquêtes générales). Paris: Éditions Recherche sur les Civilisations
(ADPF).

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 663

Devisse, J. 1988. Trade and trade routes in West Africa. In M. El Fasi (ed.), General
History of Africa, Volume 3: Africa from the Seventh to the Eleventh Century.
Berkeley: University of California Press, 367–435.
Dupuy, C. 1999. L’art rupestre à gravures naturalistes de l’Adrar des Iforas (Mali).
Sahara 11: 69–86.
Fenn, T., Killick, D., Chesley, J., Magnavita, S. and Ruiz, J. 2009. Contacts between
West Africa and Roman North Africa: archaeometallurgical results from Kissi,
northeastern Burkina Faso. In Magnavita et al. 2009, 119–46.
Fentress, E. 1987. The house of the prophet: North African Islamic housing.
Archeologia Medievale 14: 47–68.
Fili, A., Messier, R., Capel, C., Naji, S. and Fili, L. 2013. Archaeology, Conservation,
Development. Progress and Prospects of the Moroccan-American Project.
Casablanca: Aghmat Foundation.
Fili, A., Messier, R., Capel, C. and Héritier-Salama, V. 2014. Les palais mérinides
dévoilés: le cas d’Aghmat. In Y. Lintz, C. Déléry and B. Tuil Leonetti (eds),
Maroc Médiéval. Un empire de l’Afrique à l’Espagne. Paris: Hazan, 446–50.
Flight, C. 1975. Excavations at Gao (Republic of Mali) in 1974. Nyame Akuma
7: 28–9.
Holl, A. 2006. West African Early Towns: Archaeology of Households in Urban
Landscapes. Ann Arbor, MI: Museum of Anthropology, University of Michigan.
Insoll, T. 1993. Looting the antiquities of Mali: The story continues at Gao.
Antiquity 67: 628–32.
Insoll, T. 1995. A cache of hippopotamus ivory at Gao, Mali and a hypothesis of its
use. Antiquity 69: 327–36.
Insoll, T. 1996. Islam, Archaeology and History. Gao Region (Mali) ca. AD 900–
1250. (Cambridge Monographs in African Archaeology 39/BAR International
Series 647.) Oxford: Tempus Reparatum.
Insoll, T. 1997. Iron Age Gao: An archaeological contribution. Journal of African
History 38: 1–30.
Insoll, T. 1999. The Archaeology of Islam. Oxford: Blackwell.
Insoll. T. 2000. Urbanism, Archaeology and Trade: Further Observations on the Gao
Region (Mali) – The 1996 Fieldseason Results (BAR International Series 829).
Oxford: BAR.
Insoll, T. 2002. The archaeology of post-Medieval Timbuktu. Sahara 13: 7–22.
Insoll, T. 2003. The Archaeology of Islam in Sub-Saharan Africa. Cambridge:
Cambridge University Press.
Kea, R. 2004. Expansions and contractions: World-historical change and the
western Sudan World-System (1200/1000 B.C. – 1200/1250 AD). Journal of
World-Systems Research 10.3: 723–816.
Lange, D. 1994. From Mande to Songhay: Towards a political and ethnic history of
Medieval Gao. Journal of African History 35: 275–301.
Lange, D. 1996. The Almoravid expansion and the downfall of Ghana. Der Islam
73: 122–59.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 664 Sam Nixon

Lange, D. and Berthoud, S. 1977. Al-Qasaba et d’autres villes de la route centrale du

Sahara. Paideuma 23: 19–40.
Levtzion, N. 1980. Ancient Ghana and Mali (Revised Edition). New York: Africana.
Levtzion, N. and Hopkins, J.F.P. (eds). 2000. Corpus of Early Arabic Sources for
West African History. Princeton: Markus Wiener Publishers.
Lewicki, T. 1974. West African Food in the Middle Ages According to Arabic
Sources. Cambridge: Cambridge University Press.
Lewicki, T. 1981. Les origines et l’islamization de la ville de Tadmakka d’aprés les
sources arabes. In J. Devisse, C.H. Perrot, Y. Person and J.P. Chrétien (eds), Le
sol, la parole et l’écrit. Mélanges en hommage à Raymond Mauny, vol.1. Paris:
Societé française d’histoire d’outremer, 439–44.
Lhote, H. 1951. Sur l’emplacement de la ville de Tademekka, antique capitale des
berbères soudanais. Notes Africaines 51: 65–69.
McIntosh, R. 1998. The Peoples of the Middle Niger. Malden, MA: Blackwell
Publishers.
McIntosh, R. 1999. Western representations of urbanism and invisible African
towns. In S.K. McIntosh (ed.), Beyond Chiefdoms: Pathways to Complexity in
Africa. Cambridge: Cambridge University Press, 56–65.
McIntosh, S. 1995. Excavations at Jenne-Jeno, Hambarketolo and Kaniana (Inland
Niger Delta, Mali), the 1981 Season. Berkeley and Los Angeles: University of
California Press.
McIntosh, S. and McIntosh, R. 1993. Cities without citadels: Understanding urban
origins along the middle Niger. In T. Shaw, P. Sinclair, B. Andah and A. Okpoko
(eds), The Archaeology of Africa: Food, Metals and Towns. London: Routledge,
622–41.
MacDonald, K., Camara, S., Canós, S., Gestrich, N. and Keita, D. 2011. Sorotomo:
A forgotten Malian capital? Archaeology International 13: 52–64.
Magnavita, S. 2009. Sahelian crossroads: Some aspects on the Iron Age sites of
Kissi, Burkina Faso. In Magnavita et al. 2009, 79–104.
Magnavita, S. 2013. Initial encounters: Seeking traces of ancient trade connections
between West Africa and the wider world. Afriques 4 [online only]. Available at:
http://afriques.revues.org/1145 [last accessed 16 September, 2019].
Magnavita, S., Maga, A., Magnavita, C. and Idé, O. 2007. New studies on Marandet
(central Niger) and its trade connections: an interim report. Die Zeitschrift für
Archäologie außereuropäischer Kulturen 2: 147–65.
Magnavita, S. Koté, L. Breunig, P. and Idé, O. (eds). 2009. Crossroads: Cultural and
Technological Developments in First Millennium BC/AD West Africa. Frankfurt
am Main: Africa Magna Verlag.
Masonen, P. 2000. The Negroland Revisited: Discovery and Invention of the
Sudanese Middle Ages. Helsinki: The Finnish Academy of Science and Letters.
Mattingly, D.J. (ed.). 2013. The Archaeology of Fazzan. Volume 4, Survey and
Excavations at Old Jarma (Ancient Garama) Carried out by C.M. Daniels
(1962–69) and the Fazzan Project (1997–2001). London: Society of Libyan Studies.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 17 The Early Islamic Trans-Saharan Market Towns of West Africa 665

Mattingly, D.J. and Sterry, M. 2013. The first towns in the Central Sahara.
Antiquity 87.366: 503–18.
Mattingly, D.J., Leitch, V., Duckworth, C., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017a. Trade in the Ancient Sahara and Beyond. Cambridge: Cambridge
University Press and The Society for Libyan Studies.
Mattingly, D.J., Bokbot, Y., Sterry, M., Cuénod, A., Fenwick, C., Gatto, M., Ray, N.,
Rayne, L., Janin, K., Lamb, A., Mugnai, N. and Nikolaus, J. 2017b. Long-term
history in a Moroccan oasis zone: The Middle Draa Project 2015. Journal of
African Archaeology 15: 141–72.
Mauny, R. 1961. Tableau géographique de l’Ouest Africain au Moyen Âge d’après les
sources écrites, la tradition et l’archéologie. Dakar: IFAN.
Messier, R. and Miller, J. 2015. The Last Civilized Place. Sijilmasa and Its Saharan
Destiny. Austin: University of Texas Press.
Mitchell, P. 2005. African Connections. Archaeological Perspectives on Africa and
the Wider World. Walnut Creek: Altamira.
Monroe, J.C. 2017. ‘Elephants for want of towns’: Archaeological perspectives on
West African cities and their hinterlands. Journal of Archaeological Research.
Doi:10.1007/s10814-017–9114-2.
Moraes Farias, P.F. de. 2003. Arabic Medieval Inscriptions from the Republic of
Mali. Epigraphy, Chronicles and Songhay-Tuāreg History. Oxford: Oxford
University Press (British Academy).
Moraes Farias, P.F. de. 2017. Arabic and Tifinagh inscriptions. In Nixon 2017a,
41–51.
Nixon, S. 2009. Excavating Essouk-Tadmakka (Mali): New archaeological investi-
gations of early Islamic trans-Saharan trade. Azania: Archaeological Research in
Africa 44: 217–55.
Nixon, S. 2013. ‘A longing for Mecca’: The trans-Saharan Hajj and the caravan
towns of West Africa. In V. Porter and L. Saif (eds), Hajj: Collected Essays.
London: British Museum Press, 65–73.
Nixon, S. 2014. West Africa: Islamic archaeology. In C. Smith (ed.), Encyclopedia of
Global Archaeology. New York: Springer, 7720–33.
Nixon, S. (ed.). 2017a. Essouk-Tadmekka: An Early Islamic Trans-Saharan Market
Town. Leiden: Brill.
Nixon, S. 2017b. Trans-Saharan gold trade in pre-modern times: Available evi-
dence and research agendas. In Mattingly et al. 2017a, 156–88.
Nixon, S., Filomena Guerra, M. and Rehren, T. 2011. New light on the early Islamic
West African gold trade: Coin moulds from Tadmekka, Mali. Antiquity 85:
1353–68.
Norris, H. 1962. The history of Shinqit according to the Idaw ‘Ali tradition. Bulletin
de l’institut fondamental d’Afrique noire 24: 39–413.
Norris, H. 1975. The Tuaregs: Their Islamic Legacy and its Diffusion in the Sahel.
Warminster: Aris and Philips.
Park, D.P. 2010. Prehistoric Timbuktu and its hinterland. Antiquity 84: 1076–88.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 666 Sam Nixon

Polet, J. 1985. Tegdaoust IV, Fouille d’un quartier de Tegdaoust (Mauritanie

Orientale): Urbanisation, architecture, utilisation de l’espace construit. Paris:
Éditions Recherche sur les Civilisations (ADPF).
Prussin, L. 1986. Hatumere: Islamic Design in West Africa. Berkeley: University of
California Press.
Raimbault, M. and Sanogo, K. (eds). 1991. Recherches archéologiques au Mali.
Paris: Éditions Karthala.
Robert, D., Robert, S.D. and Devisse, J. 1970. Tegdaoust I: Recherches sur
Aoudaghost. Paris: Arts et Métiers Graphiques.
Robert-Chaleix, D. 1989. Tegdaoust V, Une concession médiévale à Tegdaoust:
Implantation, évolution d’une unité d’habitation. Paris: Éditions Recherche sur
les Civilisations (ADPF).
Saison, B. 1981. Azugi; Archéologie et histoire en Adlar mauritanien. Recherche,
pedagogie et culture 5: 66–74.
Sanogo, K., Cissé, M. and Fane, Y. 2006. Site de la mosquée Kankou Moussa à Gao:
Homage au docteur Téréba Togola. Bamako: Ministère de la Culture/ Direction
Nationale du Patrimoine Culturel.
Savage, E. 1992. Berbers and blacks: Ibadi slave traffic in eighth-century
North-Africa. Journal of African History 33.3: 351–68.
Thomassey, P. and Mauny, R. 1951. Campagne de fouilles à Koumbi Saleh. Bulletin
de l’Institut Français [later Fondamental] d’Afrique Noire (Dakar) 13: 438–62.
Thomassey, P. and Mauny, R. 1956. Campagne de fouilles de 1950 à Koumbi Saleh
(Ghana?). Bulletin de l’Institut Français [later Fondamental] d’Afrique Noire
(Dakar) 18: 117–40.
Togola, T. 1996. Iron Age occupation in the Méma region, Mali. African
Archaeological Review 13.2: 91–110.
Vanacker, C. 1979. Tegdaoust II, Fouille d’un quartier artisanal. Paris: Memoire de
l’I.M.R.S.
Van Berchem, M. 1960. Sedrata et les anciennes villes berbères du Sahara dans les
récits des explorateurs du XIXe siècle. Bulletin d’Institut Français d’Archéologie
Orientale 59: 289–308.
Van Doosselaere, B. 2014. Le roi et le potier. Etude technologique de l’assemblage
céramique de Koumbi Saleh, Mauritanie (5e/6e–17e siècles). Frankfurt: Africa
Magna Verlag.
Vikør, K. 1987. The early history of the Kawar oasis: A southern border of the
Maghrib or a northern border of the Sudan? The Maghreb Review 12.3–4: 78–83.

https://doi.org/10.1017/9781108637978.018 Published online by Cambridge University Press

 part iv

Concluding Discussion

Published online by Cambridge University Press

Published online by Cambridge University Press
 19 State Formation in the Sahara and Beyond
david j. mattingly and martin sterry

Introduction

In this concluding chapter we focus on debates about state formation and

to what extent it is possible to recognise historical instances of this key
process among Saharan societies. Here we make the case for a dynamic
Saharan history with frequent episodes of urbanisation and state formation
in the past. However, we also emphasise the ephemeral nature of these
polities, which challenges more evolutionary models of the rise of towns
and states. Towns and states have existed among a range of forms of
organisation utilised by Saharan societies, with the balance of power
swinging back and forth between sedentary and more mobile lifestyles.
Our exploration of these factors commences with general questions
about the definitions and types of towns and states that are detectable in
the Trans-Saharan world, coupled with a discussion of the sorts of models
that can explain their rise and fall.
A crucial debate, and one of the reasons why the material presented in this
volume is important, concerns the identification and interpretation of pre-
colonial states in Africa.1 As Monroe observes, for much of the twentieth
century the emphasis on external factors and influences was perniciously
pervasive, ignoring endogenous pathways to complexity.2 As part of post-
colonial theoretical backlash, there has sometimes been a tendency for
external contacts to be entirely airbrushed out of the discussion. This
volume emphasises the central importance of local actors, who in articulat-
ing ‘both local and exotic sources of power . . . shaped the political contours
of early states in Africa’.3 While the existence of early states has been
recognised for many areas bordering the Sahara, hitherto there has been
1 2
Monroe 2013, 703–5. On African pathways to complexity, see inter alia McIntosh, S. 1999.
3
Monroe 2013, 703. 695

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 696 David J. Mattingly and Martin Sterry

a reluctance to accord similar status to Saharan societal developments.4 If we

are correct in identifying the existence of Saharan polities and early inter-
connectedness of the Trans-Saharan world, this creates a new paradigm and
opens the possibility of parallel development in Trans-Saharan regions that
were most likely already in contact with each other in the pre-Islamic era.5
History reveals a great flaw in evolutionary models – social change is not
unilinear and states have often proved unstable and ephemeral. ‘Social pro-
gress’ may in some cases be rejected and communities revert to less centralised
forms of organisation. Nonetheless, it is also the case that knowledge of states
has been a powerful source of information that cannot be uninvented within
connected networks. Pristine states and empires will almost invariably have
significant impact and influence on their neighbours, who cannot remain
ignorant of their innovative socio-political organisation for long.6
Such approaches call into question traditional and simple evolutionary
models of social change, which emphasise progressive steps from primitive
to civilised, from tribe to chiefdom to state.7 Nonetheless, as the recent work
of Flannery and Marcus reminds us, there are common characteristics of
early states across space and time and one of their key arguments is that
rising levels of social inequality was fundamental in the emergence of states.8
We conclude the chapter with a final review of our work on the
Garamantes of Fazzan, as this is the ancient people and the oasis region
with the largest concentration of evidence. In Chapters 2 and 8 above, we
presented the factors that persuade us that Garamantian society was urban
and set out some ideas about the characteristics of Saharan urbanisation,
reflecting on how this differed from Mediterranean or Sub-Saharan urban-
ism. In this final discussion, we broaden our analysis to consider the
evidence from the wider Trans-Saharan zone, in relation to the vexed
issue of how we might define state formation in a Saharan milieu. Given
the Saharan focus of this volume, a key issue concerns the viability of oasis
states. From another perspective if we are correct to identity the Garamantes
as a polity, we also need to ask what was the contribution made by oasis
centres to the evolution and governance of such a desert state?

4
See Gronenborn 2013 (Lake Chad basin); MacDonald 2013 (Western Sahel); Moussa 2013
(Maghrib); Phillipson 2013 (Aksum); Shaw 2013 (Pharaonic Egypt); Welsby 2013 (Kerma and
Meroe).
5
The analysis offered here and in other books in this series demands a repudiation of
a conventional narrative that is outdated and outmoded in terms of a lack of engagement with
recent archaeological and theoretical advances, see for example Fauvelle 2018.
6
Flannery and Marcus 2012, 503–44 on how new empires learn from old. See also Whitehead 1992.
7
Trigger 2003; Yoffee 2005.
8
Flannery and Marcus 2012, especially 341–544 on the rise of kingdoms and empires.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 697

There are several questions to consider relating to semantics, ancient and

modern. We have tended to use the terms ‘state’ and ‘polity’ interchange-
ably, recognising that there is considerable variance in what this may have
meant even in relation to a single case study traced across several centuries.
Some of the other authors in this collection seem to discriminate between
states and polities, but without it being clear where the dividing line is to be
drawn. There is also a question of the extent to which states were part of
broader cultural groupings, as opposed to independent civilisations. It may
be a characteristic of more mature state formations that they reshape
material culture and behaviours in ways that make the society synonymous
with the political entity. We would argue that the Garamantes were not
merely a state, but a civilisation, in Trigger’s terms.9 Although we are not
advocating a checklist approach, the Garamantes meet the key criteria of
civilisation as defined by ancient Mediterranean societies: being based
around agriculture10 and urbanisation,11 having a writing system,12 using
monumental architecture,13 employing advanced technologies of produc-
tion (metallurgy, textiles, etc.),14 possessing a pronounced social hierarchy
(including the practice of slavery),15 and the development of wide trading
relations.16 Garama was also an important centre of religion, with
a substantial stone-footed temple building and freestanding statuary.17
Some of the elements of Garamantian culture appear to be shared with
other early Saharan societies and may be considered formative components
of oasis societies more generally. However, there are distinctive aspects of
the Garamantian material record, certainly as regards sites like Jarma at the
centre of Garamantian power, that seem to diverge markedly from other
settlements that lay more at the periphery, like Aghram Nadharif and Fewet.

What Type of Towns?

Throughout this volume there has been an underlying tension as to what

constitutes an urban centre and what constitutes a state. Scholarly defini-
tions of urbanism are highly varied, as discussed in Chapters 1 and 8
above.18 Our starting point in this final discussion is that there are distinctive

9
Trigger 2003, 43–48. 10 Pelling 2013a; 2013b. 11 Mattingly and Sterry 2013.
12
Mattingly 2013a, 517–18. 13 Mattingly 2013a, 73–125, 291–93.
14
Duckworth et al. forthcoming; Mattingly 2013a, 462–63; Schrüfer-Kolb 2007.
15
See arguments in Fentress 2011; Mattingly 2013a, 508.
16
Mattingly 2013b; Mattingly et al. 2017; Wilson 2012. 17 Mattingly 2013a, 289–93.
18
See Fernández-Götz and Krausse 2017.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 698 David J. Mattingly and Martin Sterry

and contingent aspects of Saharan urbanism that make it difficult to apply in

a neat way models developed for other regions or time periods in a checklist
approach.19 A similar stance that seeks to differentiate Late Iron Age central
European urbanism from Mediterranean models has proved equally helpful
in developing new paradigms.20
The previous chapters have presented examples of ancient towns and
states from across the Trans-Saharan zone. A starting point in defining
urbanism in a Saharan context is to suggest that towns were different to
other settlements – whether in scale, architecture and monumentality,
function and occupation of their population. It is also worth considering
that, regardless of modern definitions, a number of Saharan sites were
explicitly recognised as urban by Roman and Arabic sources. The
Garamantian capital at Jarma was specifically referred to by Greco-
Roman sources in the late first and second century AD as a clarissimum
oppidum (top-level centre) and a metropolis – marking it out as something
out of the ordinary for the Central Sahara.21 Archaeology fully supports
this judgement with the evidence of monumental stone architecture,
deploying Classical elements, and abundant Roman material culture con-
sumed there. While the site reveals some influences from external sources
(a Roman style bath was built there), many other aspects are not so easily
recognisable from Mediterranean models of urbanism.
The Roman towns described by Wilson with their monumental centres,
aqueducts and water amenities and conspicuous elite competition fit
a classical, Mediterranean view of the ancient city.22 Yet, these cities are
peculiar to a relatively small, if populous, part of Africa and existed in this
idealised form for only a few centuries. Though they had an important role
to play in the Mediterranean, their influence on forms of urbanism in other
Trans-Saharan regions seems slight. In Roman Egypt, the towns of the
Western Desert follow some Mediterranean precedents, with densely-
packed urban areas of mixed usage (houses and workshops alongside
each other), but the main civic buildings were temples and these could be
argued to follow pre-Roman traditions.23 The significant aspect of the

19
Smith 2017. 20 See Fernández-Götz and Krausse 2013.
21
Pliny, Natural History 5.35; Ptolemy, Geography 4.6.12. Ptolemy records two other sites as
‘metropolis’ in inland Africa, Nigira (near the Nigir river – not certainly the Niger in this
context) and Gira (associated with the Gir river). This might be a clue to the existence of at least
two other state-level societies recognised by Rome. For some discussion on Pliny’s African
knowledge, see Desanges 1980; for Ptolemy’s world view, see Berggren and Jones 2000.
22
Wilson, Chapter 10, this volume. For a range of views on the Classical Mediterranean city, see
Bowman and Wilson 2011; Clark 2013; Marcus and Sabloff 2008; Osborne and Cunliffe 2005.
23
Boozer, Chapter 4, this volume.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 699

Roman period sites in the Egyptian oases is the relative size of the largest of
these desert towns, with Trimithis estimated by Boozer to have had
a population of 10–15,000 and an occupied area of perhaps 40–60 ha.24
As we have seen, that is much larger than other pre-Islamic oasis towns of
the Central and Western Sahara.
Roman urbanism in North Africa followed a model based on agricultural
hinterlands, monumentality and institutionalised inequality. These are com-
monly assumed to have originated due to external influences on North
African society. However, Sanmartí et al. and Bokbot identify pre-Roman
origins of these characteristics in the Numidian and Mauretanian centres of
modern-day Tunisia and Morocco.25 That suggests an earlier and more
active engagement with urbanism in African societies and that this may
have contributed to the precise form of towns that emerged. There are
indeed traces of this – for instance, in the evidence of Roman towns built
over Numidian fortified promontory sites like Althiburos and Cuicul.26 The
oft remarked successes of urbanisation and agriculture in Roman Africa
have rarely been associated with endogenous factors, but contrary to the
myth of ‘wandering pastoralists’ we now know that North Africa in the first
millennium BC was a landscape already punctuated by numerous nucleated
sedentary farming settlements. The Roman ‘urban success’ story and ‘agri-
cultural boom’ built on these pre-existing conditions.27
Although urbanism was central to society and economy to the north of
the Sahara, Edwards suggests that the early urban centres of Nubia and
Meroe in the Middle and Upper Nile were parasitic.28 Like Roman towns of
the north, they were centres of craft production, but these were linked to
militaristic royal households that drew their wealth from the exploitation
of external environments rather than productive hinterlands. The relation-
ship to surrounding lands was quite different, with seemingly few social
ties between groups of craftspeople (who held no land rights) and groups of
agriculturalists and pastoralists. For Edwards, these early towns were
negative places in which the population had to be compelled to live.
Leaving aside the chronological difference, there is to some extent
a parallel between these towns and the oasis forts of the Roman frontier –
where there was also a large concentration of military power, but little
direct connection to surrounding lands or flocks other than the extraction
of tax and tribute.29

24
Cf. Mattingly and Sterry 2013. 25 Chapters 11 and 12, this volume.
26
Mattingly 2016, 15–16. 27 Mattingly 2016, 21–22; Hobson 2015.
28
Edwards, Chapter 9, this volume; Welsby 2013.
29
Mattingly et al., Chapter 5, this volume; Wilson, Chapter 10, this volume.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 700 David J. Mattingly and Martin Sterry

The otherness of the urban is also seen in the construction of complexes

such as the dual towns of Gao, Ghana and Zuwila in which some urban
zones were mercantile and some were royal.30 This zoning within urban
environments can also be seen in the accounts of Takadda where the town
had trading and manufacturing functions, but the ruler was located some
distance away.
For West Africa a dominant trend in the last decades has been the
argument that many early urban sites in the Middle Niger region arose
out of heterachical rather than hierarchical societies, being ‘self-generated’
and not dependent on states.31 However, as Kevin MacDonald has
observed, even in Mali there may have been distinctive types of early
urban sites present alongside one another. He distinguishes between
Fadugu (settlements of state power) and Markadugu (towns of commer-
cial/spiritual function).32 A crucial distinction is that the former were often
ephemeral settlements due to the transitory nature of the power behind
them and the latter much longer lasting.
Craft specialisation plays an important role in many of the case studies of
this volume. In the Middle Niger and Senegal regions it reached an industrial
scale in the large mound accretions that form the urban landscape. While the
evidence is sparser in the walled enclosures around Lake Chad, metalworking
appears to have been particularly prominent at these sites. Edwards considers
the towns of Nubia and Meroe to be essentially craft centres and Wilson
stresses the importance of urban production in his chapter. Saharan towns
echo this pattern, as is evident at Jarma, the Garamantian capital.33 At
Sijilmasa, Capel identifies salt production as a factor in the siting of the
town, while copper and silver mining were associated with the three early
Medieval towns of southern Morocco – Tamdult, Dar’a (Zagora) and
Sijilmasa.34 Other Saharan centres such as Tadmakka and Zuwila may not
have nearby natural resources, but both were involved in processing the
goods that passed through them, notably gold. An important but less archae-
ologically visible aspect of oasis manufacturing across history has been textile
and garment production. These goods have also been staples of trade in all
periods within the Trans-Saharan zone and again are much under-estimated
in many archaeologically focused accounts.35

30
Mattingly et al. 2015; Chapter 2, this volume; McIntosh, Chapter 13, this volume; Nixon,
Chapter 17, this volume.
31
McIntosh, R. 2005; 2015. 32 MacDonald 2013, 837; Mattingly and MacDonald 2013.
33
Mattingly 2013a, 511–17.
34
Capel, Chapter 16, this volume; Sterry et al., Chapter 6, this volume.
35
Mattingly et al. 2017, especially 211–30.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 701

Caravan towns have played an important part in the history of the

Sahara, though beyond the surviving literary sources, the detail is surpris-
ingly meagre. This is more a result of absence of archaeological evidence,
rather than evidence of absence. The case studies of Sijilmasa, Awdaghust,
Kumbi Saleh, Gao and Tadmakka are complemented by sites like Jarma,
Marandet and Ghadamis. Sam Nixon’s chapter in particular stresses the
different Islamic traditions of urbanisation that need to be recognised in
the Medieval trading entrepôts.36 Nonetheless, a key contribution of this
book is the observation that those Islamic trading towns were not entirely
externally produced, since, in some places at least, they can be shown to
have built on existing Saharan traditions of oasis centres and commercial
networks.
Fortification and defence are recurring themes for many, but not all
centres. The defining and enduring feature of the Lake Chad settlements
were the walls that protected the population and perhaps also livestock.37
In Fazzan, and large parts of the Sahara in historic times, fortifications
have been the norm for a wide range of settlements. However, there is
much variation. At Sijilmasa and Ghadamis, the walls enclosed not only
the core settlement area, but the oasis palmery; at the other end of the
spectrum, the towns of Zuwila and Qasr ash-Sharraba possessed walled
citadels rather than walls protecting the whole settlement and Tadmakka
had no defences at all.
There is thus a good deal of variation in the physical appearance of
towns in the Trans-Saharan zone. Whereas in the past most study has
focused on the Islamic-era context of urbanisation, it is now apparent that
there were also pre-Islamic initiatives and examples that may have influ-
enced later developments. Some desert towns have been extraordinarily
long-lived, like Ghadamis, but a recurrent pattern in the case studies
reviewed in this volume is that desert towns have often been quite ephem-
eral and sensitive to broader shifts in political power, which have periodi-
cally re-routed trade across different parts of the Saharan network. In
Fazzan, we can trace the successive shifts of the main capital from Jarma,
to Zuwila, to Traghan, to Murzuq and finally to Sabha.38 Each of those
relocations can be linked to the emergence of a new polity or over-riding
political change. It is to the connection between towns and incipient states
that we now turn.

36 37
Nixon, Chapter 17, this volume. Magnavita, Chapter 15, this volume.
38
Mattingly 2013a, 530–44.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 702 David J. Mattingly and Martin Sterry

What Type of States?

Scheele has made important points in Chapter 18 about the unsuitability of

Saharan societies for developing into recognisable states. In responding to
her arguments, we would first agree that any model of Saharan state
formation must take account of the peculiar geographical and social con-
ditions of the region, which since the first oases were created have always
interlinked the lives of sedentary communities and pastoral groups.
However, we diverge from her perspective in asserting that we believe
that state formation has been achieved a number of times in the Sahara and
that by and large it has been more commonly associated with the major
oasis clusters, where large sedentary populations lived alongside pastoral
groups, rather than in areas of the Sahara where pastoral groups have been
more dominant and oasis communities fewer and more thinly spread.
There has been considerable recent debate about the relationship
between urbanisation and state formation, with many commentators ques-
tioning evolutionary assumptions about progressive levels of social
complexity.39 There is a strong consensus among contributors to this
volume that simplistic progressive assumptions about rising levels of social
complexity from tribe, to chiefdom, to state are unhelpful in exploring state
formation in the Trans-Saharan world. The ephemeral nature of many
polities clearly indicates that states are not a natural high-end social out-
come in this sort of environment. Periods of state level organisation have
often been followed by periods of non-state structures.
The extent to which towns were a necessary precondition for the emer-
gence of states has also been questioned and indeed Jennings and Earle
have suggested that the initial development of towns may actually ‘have
served as a brake rather than an accelerant to central power’.40 There is
nonetheless a strong correlation between early towns and precocious state
formation, especially once hierarchy started to emerge among the
nucleated settlement agglomerations.41 This was also often accompanied
by a restructuring of the rural landscape – what Yoffee terms ‘an equally
important counterpart: “ruralisation” . . . the creation of the countryside’.42
While there are dangers of imposing western notions about state for-
mation in a region and social milieu they are ill adapted for, we think there

39
See Yoffee 2005 for a sustained challenge to neo-evolutionary interpretation.
40
Jennings and Earle 2016, 475. But note also the critical responses to their paper, 485–89.
41
Flannery and Marcus 2012, 457–60 on the rise of Uruk, with the emergence of four levels of
urban/village settlements.
42
Yoffee 2005, 60.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 703

is also a serious danger of anachronism in retrojecting across all of time the

political structures and patterns of power recorded in early modern and
contemporary times in the Sahara. There have been significant changes
across the last three millennia – the intensification of camel raising and
what that meant for the size, range and importance of pastoral groups,
a gradual decline in water availability in some oasis groups (like Fazzan),
conquests and migrations in the Medieval period, etc. It is arguable that
these have ‘rewired’ the Saharan networks in new ways.
Some aspects of the Saharan milieu have tended to influence the way that
state power has operated there. Distances between places can be vast, cross-
ing terrain with limited potential for settled occupation. Conventional
notions of territorial states are thus potentially misplaced and territorial
control will have become increasingly difficult to exercise over distance. But
the Garamantes suggest that the territorial ambition of Saharan polities
could be considerable – bearing in mind the documented ability of
Garamantian kings to engage in active warfare between the Mediterranean
coast and the Lake Chad region c.3,000 km to the south.43 The core territory
of the Garamantes has been estimated as c.250,000 km2, though plainly it
would be absurd to assert they exercised direct control of most of this
landscape. Control of the network of water sources and the main oasis
clusters within this was the key for the Garamantes and this may have
been achieved by working through intermediaries in the main oasis settle-
ments in the different regions. We would argue that the Garamantes were
a state to the extent that they appear to have maintained dominance over
a network of distinct oasis farming communities from their principal heart-
lands in Wadi al-Ajal. It also follows from the widespread zone of
Garamantian activity that power over people and control of trade must
have involved pastoral groups in the connective spaces between the oasis
centres. We shall return later to the question of whether pastoralists or
sedentary groups were predominant in Garamantian society.
Other types of Saharan state may also be envisaged, notably the city state
and the slaving state. City states represent one of the earliest and smallest
forms of polity across large parts of the world. That is not to say that
Saharan city states, like Ghadamis or Wargla, are exactly akin to the early
Near Eastern autonomous settlements, the poleis of ancient Greece, or the
small-scale states of Medieval Italy.44 Given the relative isolation of oasis
groups one from the next it is not surprising to see that a level of local
43
For source evidence to this wide-range of Garamantian activity, see Ptolemy, Geography 1.8–11;
Tacitus, Histories 4.50 and Mattingly 2003, 76–86.
44
Hansen 2000; 2002; Yoffee 2005, 44–62.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 704 David J. Mattingly and Martin Sterry

autonomy and self-defence capability developed in many oases. In many

periods of history we can identify a pre-eminent centre in most of the main
oasis clusters for which we have good archaeological or historical data.
These sites tended to be significant nodal points in the caravan trade as well
as productive oases. Indeed in the Saharan context, the trade connections
and wider networks of oasis towns might be suggested as a characteristic of
the Saharan city state. Shifts in the caravan routes and geo-political changes
played a part in the resilience of such towns and the continuance of their
city-state status. From the intensive evidence for Jarma being a centre of
trade, a case could be advanced for their society having expanded out from
a city state basis.
The slaving state has become most visible in relatively recent times, where
documentary records are more abundant, as in the well-documented case of
West African Dahomey.45 Such states have developed in scale and wealth
not through local production, but through long-range slave raiding and
trading. Typical characteristics are the development of significant mobile
armed forces and a well defended frontier region. Given the probable long-
term significance of slaves in Trans-Saharan commerce, the possibility of
pre-Islamic slave states needs to be carefully considered and the Garamantes
are also potential contenders for such a role.46
Finally, leaving the Garamantes to one side for the moment, it seems
simply wrong to infer that states have not existed in the Medieval and early
modern period within the Sahara. Just as in the pre-Islamic era, the geo-
graphical and political context in which they occurred was quite particular.
On the one hand, Ghadamis is an example of a long-lived mercantile city
state. But territorial states have also existed. Fazzan, the heartlands of the
Garamantes kingdom, was later the centre of a series of states and when
external powers have attempted to control the Central Sahara, they have
regularly focused on this region, often taking over the power structures of
the Saharan states they supplanted. Thus the Ibadi regime of the Banu
Khattab at Zuwila could be viewed as an attempt to reinvent the
Garamantian kingdom of Fazzan and after it was reduced to the status of
a province of the kingdom of Kanim for a while, the dynasty of the Awlad
Muhamid functioned as a state, with a number of interludes when the
Ottoman and Qaramanli powers from Tripoli again annexed the kingdom
as a province.47 A number of other prominent Medieval Saharan centres
were established or flourished as the power bases of Khajarite/Ibadi persua-
sion (for example, Sedrata by Wargla). In southern Morocco, the main oasis

45 46 47
Law 1992. Fentress 2011. Mattingly 2003, 90–106; Mattingly et al. 2015.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 705

zones may have rarely been independent, but they were important in the rise
of a number of new dynasties that took over wider territories. The densest
oasis clusters were thus by no means unfamiliar with state structures and
state functions. Nonetheless, there are interesting questions to address
concerning why the idea of the state was not more widely diffused within
the Sahara and whether we may be confronted by different models of the
state. City states, territorial states, pastoral or slaving states all seem options
to consider, as well as why some parts of the Sahara seem to have managed
perfectly well without state structures. The processual paradigm linking
inequality, urbanism and statehood is apparent in the Sanmartí et al. chapter
on the origins of the Numidian state.48 For them, the territorial states that
covered much of the Maghrib in the last centuries BC were the apogée of
a long process from intensification, agglomeration and an increasingly rigid
hierarchy. Sanmartí et al. recognise that as much as this makes sense of the
evidence, Althiburos must always have been entwined in networks of other
agglomerations and other communities.
The role of outside influences is poorly understood – would the
Numidian or Mauretanian states have occurred in the way they did without
the presence of Carthage and the Phoenicians? For McIntosh, outsiders
played a greatly reduced role (if any at all) in the development of the earliest
states in West Africa.49 Contacts from across the Sahara were present, but
only at a later date, initially through the introduction of technologies of
construction (mudbrick), animal husbandry (camels, horses, chickens)
and later still through forms of religion (Islam), trade and governance.
Nixon’s view of a particular brand of urban site associated with the
emergence of Islamic trade networks needs to be balanced against this.50
But let us step aside from the discussion of the development of the first
states, for no-one would claim that these should lie within the Sahara. Any
and all states within the Sahara must be seen as secondary states, states
created within a world in which power was already held and articulated by
neighbouring polities.51 The earliest forays by states into the Sahara appear
to be in Egypt’s Western Desert. This had the form of a colonial project, an
expansion of the Nilotic kingdom. The evidence is somewhat slight, but it
would appear that the oasis settlements of Dakhla and Kharga were not
independent, but instead linked to pulses of power from the Nile – the New
Kingdom, the Persians, the Ptolemies and the Romans. Similar imperial
expansions of power can be seen in the string of Roman forts in the
48
Sanmartí et al., Chapter 11, this volume. 49 McIntosh, Chapter 14, this volume.
50
Nixon, Chapter 17, this volume.
51
Emberling 2015; Parkinson and Galaty 2007; Price 1978.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 706 David J. Mattingly and Martin Sterry

northern Sahara from Bu Nijim in Libya to Castellum Dimmidi in Algeria –

a large investment in people, agriculture and power in regions that were
populated, but with little evidence for states. In the south, territorial states of
Mali, Kanim and Songhay also expanded to include parts of the Sahara. Here
there is a different mechanism (although with the caveat that our source
material is largely from the ‘outsider’ Arab sources rather than ‘insiders’)
whereby the Sahelian states made little investment, but undertook cam-
paigns against key nodes of Saharan trade. Small independent cities such
as Awdaghust, Tadmakka, Takadda and Agadez appear to have been the
main targets along with the salt mines of northern Mali and Kawar. The
control of Fazzan by the Kanimi kingdom in the thirteenth and fourteenth
century is the most extreme and unusual example due to the 2,000 km
distance. Control over these oases on the southern edge of the Sahara was
precarious (as with all oases) and typically lasted only a century or two at
most. Threats elsewhere might lead to their abandonment or they might be
sources of rebellion themselves as with the Laguatan or the Kharajites, or the
Tuareg in the south. Of the territorial empires, the Almoravids were unusual
in seeking to control a Pan-Saharan territory, rather than just its edge. This
reflects their origins, emerging from the Sanhaja of the Western Sahara. The
Almoravid movement thus had strong Saharan roots and many of its initial
targets were against Saharan entrepôts – Dar’a, Sijilmasa and Awdaghust.
However, as the Almoravids extended themselves into Morocco, they split,
with the northern Maghribian part separated from the Sahara. The Saharan
faction led by Abu Bakr is not well-documented, but does not appear to have
survived long after his death in 1087 (although it may have conquered
Ghana in the interim).52
There are many historical examples of Saharan polities beyond the
Almoravids (Table 19.1). Regrettably, we know little about most of these
beyond a couple of fleeting references, but in aggregate we can build
a picture suggesting some common characteristics. As with Saharan cities,
Saharan polities were uniformly small in terms of territory, population and
power. Most occupied only a single oasis, with one major trading town –
Sijilmasa is the clearest example, but we may note also Ghadamis, Waddan,
Wargla, Marandet, Tadmakka, Takkada and Awdaghust.
A second group occupied a string of neighbouring oases, for example the
Lawata Berber state in the Egyptian oases, the dominant group within the
oases of Kawar or the range of different polities that are mentioned as holding

52
On the Almoravids, see Lugan 2000, 63–86.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 707

Table 19.1 States and proto-states in the Sahara

Where Capital Who When Comments

Kharga, Dakhla, Al-Kalamun Lawata Berber – <941–942> Mixed population

Farafra and Al-kasr Abd al-Malik ibn and religions
Bahriya53 Marwan Several thousand
horsemen
Al-Jufra oases54 Waddan King 7th century Account is some-
what problematic
Sanhadja (zenaga) Tribal Led by Lamtuna 7th–11th
(also Anbiya) Azugi after 11th (list of kings in century?
century Ibn Khaldun)
Kawar55 Shifts, starts at al- 7th/9th century? Mixed Tubu,
Majority Berber
Qasaba (Gezabi?) Not beyond Berber, Arab
1220
Sijilmasa56 8th century
Air Maranda Marandet Tuareg 9th century Under sway of
kingdom of Gao
Air Al-Hazban57 Marquart identifies Tuareg 9th century Under sway of
with Air kingdom of Gao
Air Tikarkarin Somewhere between Tuareg 9th century Under sway of
Agadez and Tahua kingdom of Gao
Fazzan (ash- Zuwila Banu Khattab 10th–13th
Sharqiyat)58 century
Tadmakka Zakran Banu Tanmak 10th century Sanhadja, Zenata,
Saghmara Berber muslim and non-
muslim
Wargla Sedrata Ibadi, Zenata 10th–12th Mixed religion,
Berber centuries some Lawata,
some Christian
Collection of towns
in the oasis
Awdaghust Awdaghust59 Sanhadja 10–11th Mixed Arabs,
century? Berbers
Wadi Draa Dar’a/Tiymetin Banu Massufa 10–11th Fleeting reference
(Monteil thinks in century? only
region of
Tagounite)

53
Lewicki 1988, 281. 54 See Mattingly et al., Chapter 5, this volume.
55
See Sterry and Mattingly, Chapter 7, this volume. 56 See Capel, Chapter 16, this volume.
57
Marquart 1913, lxxviii, cix–cxvi. 58 Mattingly et al. 2015.
59
See Nixon, Chapter 17, this volume.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 708 David J. Mattingly and Martin Sterry

Table 19.1 (cont.)

Where Capital Who When Comments

Jarid and Jabal Gabes, Jarid Banu Ghaniya Late 12th– early Dynasty trying to
Nafusa (and Offshoot of the 13th century establish stability
parts outside the Almoravids in the eastern
Sahara inc. Maghrib
Balearics)
Air Agadez Tuareg? 14th century
DMWShH Awdaghust? ? 14th century
(awdaghust?)
Tadmakka Tadmakka60 Ibadi 14th century
Takadda Takadda61 Tuareg? 14th century
Karkari Kahir
Gobir Marandet (and mov- Hausa 15th century
ing south)
Air Agadez (1495)62 Tuareg 15th century Chronicle of
onwards Agadez
Timbuktu Timbuktu 16th century
Awlad Muhammad Murzuq Fazzani 16th century
Ader Tahoua area 17th century

sway over the Aïr massif. The Fazzani states of the Banu Khattab (at Zuwila)
and the Awlad Muhammad (at Murzuq) sit in between these models, both
appear to have tried to exert control over the entirety of Fazzan, but the
strength of this control seems to have been somewhat variable.
A third group comprises those polities that are described as more mobile –
the Tuareg polities of Aïr, the Sanhadja of the Western Sahara (but particularly
Adrar) although these may still have made use of towns. With populations
likely to have been in the 10,000s rather than 100,000s or higher, the claims to
statehood of all three groups is perhaps questionable to some readers. But these
forms of polity appear to have been recurrent in the Sahara.

Nomads and Other Non-State Entities

There are two risks in focusing upon cities and states. First that we equate
one with the other without critique; second, that we exclude all other types of

60
See Nixon, Chapter 17, this volume. 61 See Nixon, Chapter 17, this volume.
62
See Nixon, Chapter 17, this volume; Scheele, Chapter 18, this volume.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 709

community and social organisation from debate. Hence we turn next to

consider these other groups (and by inference, what was distinctive about
Saharan states). Scheele demonstrates in her chapter how other types of
network have existed across the Sahara such as the religious connections of
the Senussi of the Eastern Sahara or the zawiya of Morocco.63 In many ways
these religious communities had much in common with what we might
regard as states, but they lacked the formal bureaucracies or the territorial
ambitions. Trading families provide a second major form of network. Letters
and accounts preserved from the last centuries provide a particularly rich
account of merchant families from Ghadamis in Libya, Timbuktu in Mali or
Ighil in Morocco. Finally, nomads have always been a key characteristic of
the Sahara, whether in the past or in more modern times able to build
alliances and confederacies, but often without a permanent base.
Honeychurch has outlined four ways in which mobile pastoralists have
coalesced into states:64 by constructing their own complex political system,
primarily involving nomadic populations as their constituents;65 by inte-
grating urban centres and trade into their networks and movements; by
conquering sedentary polities; or by being subsumed by an expanding non-
pastoralist state. He specifically suggests the Garamantes as an example of
the second category of state formation.66 While we can observe examples of
all four within and around the Sahara, it is the first of these that is the least
explored in the Sahara.
To identify such an entity we need to demonstrate the existence of
a polity that was internally complex. The archaeological evidence for
a wholly nomadic state is thin (although this is arguably due to a lack of
exploration than necessarily a lack of evidence per se). Although the burial
record of the Sahara points to a degree of hierarchy since c.3000 BC with
the construction of a wide array of complex monuments and commem-
orations, these can hardly be taken to be the evidence of states. The
ethnographies of the Tuareg indicate a great degree of social and political
complexity, with slave owning, land and water rights and so forth, but there
is again little evidence of more integrated political action. What might
define them as a polity, according to Honeychurch, was their capacity to
transform productive mobility into power and the control of space along
with their involvement in long-distance trade.67 We might conclude that,

63
Scheele, Chapter 18, this volume. 64 Honeychurch 2014, especially 292–99.
65
Cf. Emberling 2015 who proposes that Kush was a pastoral state – a state with significant
mobility of population and for which herd animals were a source of wealth and symbolic
capital. See also, Edwards, Chapter 9, this volume.
66
Honeychurch 2014, 292–93. 67 Honeychurch 2014, 296–97.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 710 David J. Mattingly and Martin Sterry

though there was a capacity among such groups for statehood, this has only
rarely occurred. The list of kings of the Sanhadja that appear to denote
a polity is perhaps the only compelling example.
The polities based in Air and the Azawagh are more convincing as states.
Here although there was a demonstrable urban centre for the different
states, they differed from more traditional city states in that the political
authority remained with a community that went to great lengths to express
their nomadic heritage and practices. It is in this sense that the meeting of
the Sultan of Takadda with Ibn Battuta should be viewed or the chronicles
of Agadez. This drawing on nomadic values is not unique to the Sahara but
can be found in many of the Medieval North African states in origin myths,
as at Sijilmasa, or dynasties – Almoravids, Marinids. This latter category
follows a pattern of a network of nomadic Berber groups becoming
a military force, then conquering a swath of the Maghrib before settling
and founding a new capital. It is striking that in the Medieval period the
elite emphasised their nomadic roots in direct contrast with the elites of the
classical period who focused upon farming and sedentism.68
In the opposite sense, though, a problem with the fixation on the reality
of the nomad state is that it also ignores the contribution of sedentary
communities in regions like the Sahara. The archaeological evidence from
the Sahara suggests that sedentary oasis communities co-existed with the
pastoral groups and that there were strong mutual interests in exchange
and trade development to promote patterns of symbiotic co-existence and
coalition, rather than continual competition and confrontation. Although
the mobile pastoral groups are commonly considered to have been the
more dominant militarily, this has not necessarily been the case through-
out history.

Abandonment Versus Resilience

We have noted already that a characteristic of some of the Saharan urban

centres has been their ephemeral nature. Although there are a few centres
that have been proved resilient and long-lived (such as Ghadamis in
western Libya), much more numerous are the instances where Saharan
towns have flourished for a few generations (or at most a few centuries)
and then been abandoned or shrunk back to a smaller size and functionally

68
Groupe de recherches sur l’Afrique antique 1993, on the funerary monument of the Flavii at
Kasserine, with its emphasis on the establishment of estates and viticulture close to the town.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 711

less diverse status. The reasons for the seeming lack of resilience of many
desert towns are varied, but readily comprehensible when their geographi-
cal, environmental and historical contexts are explored. Factors contribut-
ing to the abandonment of Saharan towns include:
• Collapse of political regime;
• Climate change or environmental deterioration (including hydrological
difficulties);
• Changing trade routes;
• Conquest by external group;
• Religious change.
The same factors that contributed to the lack of resilience of Saharan towns
have also impacted on early states in the region.
It also needs emphasizing that the different types of Saharan oasis and
their varied hydrological resources and supports, as highlighted in the first
section of this book, contributed to their different historical trajectories. At
the simplest level, those with the most abundant and readily exploitable
water sources had the potential to develop large populations. But equally,
difficult and precarious water reserves could constrain development and
precipitate decline. Even the existence of water did not guarantee success;
trajectories of urbanisation and state formation were also impacted by
factors like connectedness and trade.

What of the Garamantes?

Finally then, we return to the Garamantes. In a previous paper we suggested

that Saharan cities should not be measured up against cities of the
Mediterranean, Mesopotamia or Mesoamerica, but rather they should be
seen primarily in comparison to other settlements within the Sahara.69
Although these Saharan cities had a small population they were an order
of magnitude larger than other villages and fortified farms around them,
they contained craft specialists and had a particular economy and forms of
authority related to conducting and supporting long-distance trade.
A similar kind of argument could be made for Saharan states – these tended
to be territorially small, mostly equivalent to city states dominating indivi-
dual oasis groups, though sometimes with greater reach along networks of
routes. Typically, they exhibit evidence of craft production and trade

69
Mattingly and Sterry 2013.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 712 David J. Mattingly and Martin Sterry

connections, the organisation of labour and the promotion of individuals as

rulers. But the question we wish to end on is whether the Garamantes fit into
the different modes that we and our contributors have explored above and
throughout this volume or if they are a different kind of entity altogether?

Political Complexity
The Roman sources provide some indications that the Garamantes were ruled
by kings, not by chieftains.70 This is important as the Roman writers tended not
to overinflate the status of neighbouring rulers. When the term rex is used, it is
generally reserved for rulers of polities or for individuals with a formal status of
client king.71 There is no reason to dismiss these references to kings as fanciful
or meaningless. The conclusion is supported by the archaeological evidence
from Jarma, where there are tombs that were an order of magnitude grander
and more richly furnished than others in the area. The conclusion that these are
burials of a small high elite group, perhaps focused on a royal dynasty or
dynasties, is highly plausible.72 Ptolemy indicates that the Garamantian kings
on occasion campaigned far to the south of their core territory and that
Ethiopian people were subject to their authority.73 After supporting the ulti-
mately unsuccessful Tacfarinan revolt against Rome in the reign of Tiberius,
the Garamantes sent a deputation to Rome to plead their cause with the
emperor and embassies were also sent in the reign of Augustus.74 This shows
not just a Garamantian self-evaluation of their importance, but also Roman
recognition of the proper way to receive royal emissaries.
In terms of political complexity the Garamantes certainly have strong
evidence of hierarchy, reflected in things like funeral monuments and of
material consumption, but this inequality is also reflected in the differences
between the capital Jarma, its hinterland, more distant parts of the Wadi al-Ajal
and other parts of the Fazzan. Trade and conspicuous consumption of
imported goods was evidently highly centralised as was the experimentation
in building and funerary forms. Indeed this diversity in forms also seems to
reflect a diversity of people in and around Jarma itself. In this sense Jarma
appears very similar to many of the later Medieval trading towns of the
Sahara – structurally locked into long distance trade networks and highly
diverse.
70
Historical references: Pliny, Natural History 8.142; Ptolemy, Geography 1.8; 1.11; Tacitus, Annals
4.23–26; Lucian Dipsades 2; Isidorus of Seville 9.11.125. Poetic: Virgil, Aeneid 4.198; Silius Italicus,
Punica 2.58. For similar analytical conclusions, see also Liverani 2006a, 434; 2006b; 2007.
71
See Braund 1984 on client kings.
72
Ayoub 1967a; 1967b; Mattingly 2007, 140–44; 2010, 359–69; Mattingly et al. 2011.
73
Ptolemy, Geography 1.8. 74 Tacitus, Annals 4.26; Aurelius Victor, de Caesaribus 1.7.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 713

Scale of Population
Roman sources describe the Garamantes as a great and numerous people
(though the picture is partly obscured by the existence of strong stereo-
typical tropes in the ancient sources’ depiction of the ‘barbarian other’).75
Although individual settlements were quite small (typically 1–4 ha), this
largely reflects the limiting factors of water availability and ease of access to
land in excessive summer heat. In aggregate, Garamantian society appears
large when modelled on the basis of hundreds of thousands of tombs or
hundreds of settlements of village or larger scale. The extent of
Garamantian settlement across Fazzan is extraordinary. Conservative esti-
mates put the population of Fazzan in the region of 100–150,000 people in
c.AD 1–300 compared to the c.25,000–35,000 people in the 1850s–1950s
(just after the end of the Awlad Muhammad state and during the Ottoman
and Italian colonial regimes).76

Territorial Scale of Kingdom

The large extent of Garamantian territory and hegemony is indicated by
a number of factors. Tacitus records in the first century that Garamantian
authority extended far to the north and they intervened in the civil discord
between two of the coastal cities in AD 69, 1,000 km from Jarma.77
A Roman visitor to Fazzan in the late first century AD accompanied
a Garamantian king on a four-month journey (again arguably far in excess
of 2,000 km) to the south of Fazzan, to a region of ‘Aethiopia’ where the
rhinoceros was to be found.78 We may also note the intensive development
within Garamantian territory of hundreds of substantial settlements of
intrinsically similar layout and appearance and investment in a labour-
intensive irrigation regime.79 The import of raw carnelian and amazonite
for primary processing at Jarma from quarry sources located several
hundred kilometres away is another factor indicating wide territorial
control.80

75
Herodotus, Histories 4.183; Dionysius Periegetes, Description of the Known World 217; Iulius
Honorius, Cosmographia 47, 53, A48, 54 – all imply that Garamantes were one of major peoples
of inland Africa. Liverani 2006a, 431–34; Mattingly 2011, 34–36 for discussion of the topos of
the barbarian other.
76
See Mattingly 2013a, 538–42, for the population modelling from archaeological data; Mattingly
et al. 2015, for detailed summary relating to Zuwila.
77
Tacitus, Histories 4.50 (Garamantes attacked lands and the city of Lepcis Magna).
78
Ptolemy, Geography 1.8. 79 Mattingly 2013a, 530–34; Mattingly and Sterry 2013.
80
Mattingly 2013a, 456–60.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 714 David J. Mattingly and Martin Sterry

Contrast between Jarma and Peripheral Settlements

The discussion in this volume highlights significant differences between
the archaeology of the Garamantian heartlands and the peripheral zone
near Ghat where Mario Liverani, Lucia Mori and the Italian mission have
worked.81 If Jarma has the look of a primate centre evoking a strong sense
of social hierarchy, Fewet or Aghram Nadharif are much more standar-
dised and homogenous settlements with little differentiation also in the
burial monuments and grave good assemblages associated with them.82
This seems to suggest the existence in the Tanzzuft of much more egalitar-
ian communities. The difference between the Jarma area and what may
well have been a peripheral territory allied to or subject to the Garamantian
kingdom requires an explanation and for the moment the state model
really does this best. As has been stated already, sedentary oasis settlement
was more widely spread in the Sahara, but if we had more archaeological
data for the pre-Islamic period we suspect that the picture would look more
like the Tanzzuft than the Wadi al-Ajal. Again, we return to the exception-
ality of the Garamantes and this needs further discussion.

Territorial State or City-State?

By the standards of the Nile Valley or the northern Maghrib, the achieve-
ments of the Garamantes may seem relatively minor. But thinking about
this in emic terms highlights the extraordinary nature of their society in
its Saharan context at this early date. Certainly this was a complex society,
a polity or a state depending on the preferred nomenclature. But was it an
early form of the type of polities that would subsequently form in the
Medieval period? This question rests on the degree of integration across
Fazzan. Historical parallels from other parts of the Sahara such as the
Azawagh or the Wadi Righ would suggest that the urban centres of Jarma,
Qasr ash-Sharraba and Zuwila most likely competed for trade, making
Jarma the leading town at the time of Roman contact. In this model the
territorial control of kings ruling from Jarma may have been quite weak,
perhaps with more of a role as an arbitrator than a despot. Complexity
would be as much horizontal – in the different familial and tribal net-
works linking groups of fortified oasis settlements and pastoral groups
together – rather than hierarchical. If we follow this line of reasoning we

81
Liverani 2006a; Mori 2013.
82
The monumental grave of In Aghelachem is some 30 km distant and post-dates the main
occupation of these settlements.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 715

can see Jarma in a long line of trading towns with Sijilmassa, Tadmakka,
Tagdaoust, Marandet, etc. However, we have also seen how the Medieval
states of North Africa were structurally different to the Classical states.
The scale of settlement and population in Fazzan outstrips anything seen
anywhere in the Sahara, apart from the Egyptian Western Desert,
until the modern period (at which point Fazzan was constituted as
a province). The rugged geography and limited water sources of
Fazzan have historically made territorial control difficult, but perhaps
a larger mass of population helped to overcome these barriers. This would
make the Garamantes a unique form of state in the history of the Sahara,
but in archaeological terms they stand out dramatically. The evidence of
trade goods is not the few imported vessel fragments found at Medieval
trading towns, but that of thousands of amphora carried in by camel
every year. In this model the closest parallel is the manner in which the
oases of the Western Desert operated between the Persian and
Roman periods. But instead of being a provincial expansion of a larger
state beyond the Sahara, the Garamantes existed independently of any
major empire.
A recurrent question throughout this chapter has been whether the
Garamantes were a pastoralist dominated state – in the manner that
oases have often been subservient to nomadic groups like the Tuareg in
later times. The focus of wealth, power and material consumption at the
Garamantian oasis centres is a strong argument in favour of the signifi-
cance and autonomy of these sedentary populations. The widespread
erection of fortified buildings at the heart of oasis settlements suggests
a continued capacity for self-defence (but also perhaps a growing threat
from pastoral groups). But the strict dichotomy between oasis dwellers and
nomadic pastoralists may be misleading here. A key argument emerging
from our work is that the Garamantes were a close amalgam of sedentary
and pastoral groups, with their mutual interests sufficiently aligned to
promote oasis development, trade and raiding across a broad expanse of
the Central Sahara. Returning to Honeychurch’s four models of nomadic
state, this looks more akin to his third case of synthesis of pastoral and non-
nomadic systems ‘with long histories of integrated nomadic and sedentary
sectors’, rather than, as he argued, seeing the Garamantes as aligned with
his secondary model of state structures emerging in a pastoral society
through trade networks.83

83
Honeychurch 2014, 292–93.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 716 David J. Mattingly and Martin Sterry

Changes in Urbanisation and State Formation

We conclude that there are several different types of state that have existed
within and around the Sahara. Although there are many differences
between these in terms of political complexity (and the degree of heter-
archy or hierarchy), territoriality, demographic size, power and stability,
we can identify recurring themes.
Firstly, the environment, the Sahara itself, played a major role in the
establishment of states and their sustainability. At its most hostile it could
spell the end of a state as aquifers dried up, rains failed or groundwater
levels dropped to unreachable depths. But, we should be careful of
thinking that these dramatic environmental events were the major dri-
vers of change. Of the many different states discussed, it is notable how
the same places appear again and again – the Fazzan, the Azawagh,
southern Morocco, the Niger bend. It is the more subtle changes in the
environment then that have characterised its effects, the increase in rains
that brings a trading town within reach of a Sahelian state, the gradual
drying up of a well far from any major oasis that limits the numbers of
caravans that can pass in any year. In this sense, the Sahara desert is
a dynamic player, always in flux, changing the terms on which each state
is built from year to year. At all times, the states that we have been
examining were both affected by and in turn affected their environment.
Equilibrium is a characteristic that has generally been absent from the
states of the Sahara.
Resilience is the second major theme, or its lack thereof. Few of the states
that engaged with the Sahara were long-lived, many had a powerful leader
who established the state followed by a dynasty that desperately tried to
hang on to power as circumstances changed outside their control. The
most resilient states were often the most flexible and the smallest – city
states such as the Banu Khattab of Zuwila. But there were far more resilient
systems in the Sahara than state networks or even urban networks. The
mercantile ties of Ghadamis endured for centuries linking places as far
flung as Timbuktu, Bornu and Tripoli and resisting the imperial ambitions
of territorial states until the nineteenth century. In so doing, the
Ghadamensi remained flexible enough to reconfigure their networks so
that the city remained a key node in Trans-Saharan trade even as other
states grew and fell, while also managing their delicate but essential rela-
tionship with their Tuareg partners, so as to maintain the independence of
the sedentary population from would-be nomadic overlords.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 717

The third theme revolves around trade and networks. The continuing
success of a state relied on the success of states around it. A disruption in
one part of the network could have far-reaching consequences elsewhere in
the network. The notion of the western, central and eastern north–south
routes across the Sahara can cloud the complexities and interconnected-
ness across the Sahara.
A fourth theme of the volume has been the inter-relationships between
sedentary and pastoral populations. A key contribution of this volume has
been to open up debate on the extent of pre-Islamic sedentarism, urbanisa-
tion and state formation in the Saharan region. We hope to have demon-
strated impressive evidence of the spread of oasis cultivation and the
growth of agglomerated settlement, with the most spectacular instance
represented by the Garamantes meriting recognition as a precocious ter-
ritorial state. Pastoral groups undoubtedly played an important role in
Garamantian military and trading success, but what is most striking about
the archaeological record of this kingdom is the extent to which its wealth
seems to have been founded on and invested in and around oasis commu-
nities. In later times, it is more difficult to perceive sedentary communities
having the whip hand in the ‘symbiotic’ relations with mobile groups. The
camel may have been a key element in redefining the relationships and the
relative balance of power between sedentary communities and pastoralists
in the Sahara. Camels had already been around for several centuries, when
its importance started to increase in Late Antiquity. However, the enlarge-
ment of the long-range networks of routes into the Western Sahara and the
falling water table beneath much of the Sahara meant that longer stages
between major water points became more common, making the camel
increasingly indispensable, thus strengthening the hand of pastoral ele-
ments within Saharan communities.84
Finally, we want to highlight the chronological complexity that is
emerging from recent studies. The radiocarbon date lists we have
assembled demonstrate the potential to assemble more detailed historical
models. The methods for dating mudbrick and pisé architecture advo-
cated in this book have been shown to be effective and remarkably
consistent in the chronological indications. The challenge now will be
to augment the date list in a systematic manner, to include more of the
key, but as yet undated cities of the Sahara. There is also a clear need for
more stratified excavations that penetrate to the natural subsoil beneath
oasis towns, to assess the ultimate origins of such sites. This may look

84
Mattingly et al. forthcoming.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 718 David J. Mattingly and Martin Sterry

a forlorn prospect in the current political situation, with most of the

Sahara archaeologically off-limits for foreign missions and also difficult
to access for local archaeologists. We hope that the long-term legacy of
this book will be an expansion of work on Saharan historical settlement
and society whenever conditions allow.

References

Ayoub, M.S. 1967a. Excavations in Germa Between 1962 and 1966. Tripoli:
Ministry of Education.
Ayoub, M.S. 1967b. The Royal cemetery at Germa. A preliminary report. Libya
Antiqua 3–4: 213–19.
Berggren, J.A. and Jones, A. 2000. Ptolemy’s Geography. An Annotated Translation
of the Theoretical Chapters. Princeton: Princeton University Press.
Bowman, A. and Wilson, A. (eds). 2011. Settlement, Urbanization, and Population.
Oxford: Oxford University Press.
Braund, D. 1984. Rome and the Friendly King. London: Duckworth.
Clark, P. (ed.). 2013. The Oxford Handbook of Cities in World History. Oxford:
Oxford University Press.
Desanges, J. 1980. (Pline l’ancien) Histoire naturelle Livre V.1–46 (L’Afrique du
nord). Paris: Persée.
Duckworth, C., Cuénod, A., and Mattingly, D.J. (eds). Forthcoming. Mobile
Technologies in the Ancient Sahara and Beyond. Trans-Saharan Archaeology
Volume IV. Series editor D.J. Mattingly. Cambridge: Cambridge University Press
and The Society for Libyan Studies.
Emberling, G. 2015. Pastoral states: Toward a comparative archaeology of early
Kush. Origini XXXVI: Preistoria e protostoria delle civiltà antiche/Prehistory and
Protohistory of Ancient Civilizations: Rome: Sapienza Università di Roma,
125–56.
Fauvelle, F.-X. 2018. The Golden Rhinoceros. Histories of the African Middle Ages.
Princeton: Princeton University Press.
Fentress, E. 2011. Slavers on chariots. In A. Dowler and E.R. Galvin (eds), Money,
Trade and Trade Routes in Pre-Islamic North Africa. London: British Museum
Press, 64–71.
Ferguson, N. and Whitehead, N. (eds). 1992. War in the Tribal Zone: Expanding
States and Indigenous Warfare. Santa Fe: School for American Research.
Fernández-Götz, M. and Krausse, D. 2013. Rethinking early Iron Age urbanisation
in Central Europe: The Heuneberg site and its archaeological environment.
Antiquity 87.336: 473–87.
Fernández-Götz, M. and Krausse, D. 2017. Eurasia at the Dawn of History.
Urbanization and Social Change. Cambridge: Cambridge University Press.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 719

Flannery, K. and Marcus, J. 2012. The Creation of Inequality. How Our Prehistoric
Ancestors Set the Stage for Monarchy, Slavery and Empire. Cambridge, MA:
Harvard University Press.
Gronenborn, D. 2013. States and trade in the Central Sahel. In Mitchell and Lane
2013, 845–58.
Groupe de recherches sur l’Afrique antique. 1993. Les Flavii de Cillium.
Étude architecturale, épigraphique, historique et littéraire du mausolée de
Kasserine (CIL VIII, 211–216). Rome: Collection de l’École Française de
Rome, 169.
Hansen, M. (ed.). 2000. A Comparative Study of Thirty City-State Cultures.
Copenhagen: Reitzels.
Hansen, M. 2002. A Comparative Study of Six City-State Cultures. Copenhagen:
Reitzels.
Hobson, M. 2015. The North African Boom: Evaluating Economic Growth in the
Roman Province of Africa Proconsularis. Portsmouth, Rhode Island: Journal of
Roman Archaeology Supplementary.
Honeychurch, W. 2014. Alternative complexities: The archaeology of pastoral
nomadic states. Journal of Archaeological Research 22.4: 277–326.
Jennings, J. and Earle, T. 2016. Urbanization, state formation and co-operation.
A reappraisal. Current Anthropology 57.4: 474–93.
Law, R. 1992. Warfare on the West African Slave Coast 1650–1850. In Ferguson
and Whitehead 1992, 103–26.
Lewicki, T. 1988. The role of the Sahara and Saharians in relationships
between north and south. In M. El Fasi (ed.), General History of Africa
III: Africa from the Seventh to the Eleventh Century. London: James Currey/
UNESCO, 146–62.
Liverani, M. (ed.). 2006a. Aghram Nadharif. The Barkat Oasis (Sha‘abiya of Ghat,
Libyan Sahara) in Garamantian Times. Firenze: All’Insegna del Giglio.
Liverani, M. 2006b. Imperialismo, colonizzazione e progresso technico: Il caso del
Sahara libico in eta romana. Studi Storici 4: 1003–57.
Liverani, M. 2007. La struttura sociale dei Garamanti in base alle recenti scoperte
archaelogiche. Rendiconti dell’Accademia Nazionale dei Lincei Ser 9.18: 155–204.
Lugan, B. 2000. Histoire du Maroc des origenes à nos jours. Saint-Amand-
Montrond: Perrin.
MacDonald, K. 2013. Complex societies, urbanism and trade in the western Sahel.
In Mitchell and Lane 2013, 829–44.
McIntosh, R. 2005. Ancient Middle Niger: Urbanism and the Self-Organizing
Landscape. Cambridge: Cambridge University Press.
McIntosh, R. 2015. Different cities: Jenné-jeno and African urbanism. In The
Cambridge World History: Volume 3, Early Cities in Comparative Perspective,
4000 BCE–1200 CE. Cambridge: Cambridge University Press, 364–80.
McIntosh, S. 1999. Beyond Chiefdoms: Pathways to Complexity in Africa.
Cambridge: Cambridge University Press.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 720 David J. Mattingly and Martin Sterry

Marcus, J. and Sabloff, J.A. (eds). 2008. The Ancient City: New Perspectives on
Urbanism in the Old and New World. Santa Fe: School for Advanced Research
Press.
Marquart, M. 1913. Die Benin-Sammlung des Reichsmuseums für Völkerkunde in
Leiden. Leiden: Brill.
Mattingly, D.J. 2003. The Archaeology of Fazzan, Volume 1, Synthesis. London: The
Society for Libyan Studies.
Mattingly, D.J. 2007. The Archaeology of Fazzan, Volume 2. Site Gazetteer, Pottery
and Other Finds. London: The Society for Libyan Studies.
Mattingly, D.J. 2010. The Archaeology of Fazzan. Volume 3, Excavations Carried
out by C.M. Daniels. London: The Society for Libyan Studies.
Mattingly, D.J. 2011. Imperialism, Power and Identity. Experiencing the Roman
Empire. Princeton: Princeton University Press.
Mattingly, D.J. 2013a. The Archaeology of Fazzan, Volume 4. Survey and Excavations
at Old Jarma (Ancient Garama) Carried out by C.M. Daniels (1962–69) and the
Fazzan Project (1997–2001). London: The Society for Libyan Studies.
Mattingly, D.J. 2013b. To south and north: Saharan trade in antiquity. In
H. Eckardt and S. Rippon (eds), Living and Working in the Roman World.
Portsmouth, RI: JRA Supplementary, 169–90.
Mattingly, D.J. 2016. Who shaped Africa? The origins of agriculture and urbanism
in Maghreb and Sahara. In N. Mugnai, J. Nikolaus and N. Ray (eds), De Africa
Romaque. Merging Cultures across North Africa. London: The Society for Libyan
Studies, 11–25.
Mattingly, D.J. and MacDonald, K. 2013. Early cities: Africa. In Clark 2013, 66–82.
Mattingly, D.J. and Sterry, M. 2013. The first towns in the Central Sahara.
Antiquity 87.366: 503–18.
Mattingly, D.J., Abduli, H., Ahmed, M., Cole, F., Fenwick, C., Fothergill, B.T.,
Gonzalez Rodriguez, M., Hobson, M., Khalaf, N., Lahr, M., Moussa, F., Nikita, E.,
Nikolaus, J., Radini, A., Ray, N., Savage, T., Sterry, M. and Wilson, A. I. 2011. DMP
XII: excavations and survey of the so-called Garamantian Royal Cemetery
(GSC030–031). Libyan Studies 42: 89–102.
Mattingly, D.J., Sterry, M. and Edwards, D. 2015. The origins and development of
Zuwīla, Libyan Sahara: an archaeological and historical overview. Azania 50.1:
27–75.
Mattingly, D.J., Leitch, V., Duckworth, C.N., Cuénod, A., Sterry, M. and Cole, F.
(eds). 2017. Trade in the Ancient Sahara and Beyond. Trans-Saharan
Archaeology Volume I. Series editor D.J. Mattingly. Cambridge: Cambridge
University Press.
Mattingly, D.J., Sterry, M. and Fothergill, B.T. Forthcoming. Animal traffic in the
Sahara. In V. Blanc-Bijon (ed.), XIe Colloque international Histoire et
Archéologie de l’Afrique du Nord. Marseille: Presses universitaires de Provence.
Mitchell, P. and Lane, P. 2013. The Oxford Handbook of African Archaeology.
Oxford: Oxford University Press.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 19 State Formation in the Sahara and Beyond 721

Monroe, J.C. 2013. The archaeology of the precolonial state in Africa. In Mitchell
and Lane 2013, 702–22.
Mori L. (ed.). 2013. Life and Death of a Rural Village in Garamantian Times. The
Archaeological Research in the Oasis of Fewet (Libyan Sahara), AZA
Monographs 6. Firenze:All’Insegna del Giglio.
Moussa, F.K. 2013. Berber, Phoenicio-Punic and Greek North Africa. In Mitchell
and Lane 2013, 765–76.
Osborne, R. and Cunliffe, B. (eds). 2005. Mediterranean Urbanization 800–600 BC.
Oxford: Oxford University Press.
Parkinson, W.A. and Galaty, M.L., 2007. Secondary states in perspective: An
integrated approach to state formation in the prehistoric Aegean. American
Anthropologist 109.1: 113–29.
Pelling, R. 2013a. The archaeobotanical remains. In Mattingly 2013a, 473–94.
Pelling, R. 2013b. Botanical data appendices. In Mattingly 2013a, 841–52.
Phillipson, D.W. 2013. Complex societies of the Eritrean/Ethiopian highlands and
their neighbours. In Mitchell and Lane 2013, 799–815.
Price, B.J. 1978. Secondary state formation: an explanatory model. In R. Cohen and
E. Service (eds), Origins of the State: The Anthropology of Political Evolution.
Philadelphia: Institute for the Study of Human Issues, 161–86.
Schrüfer-Kolb, I. 2007. Metallurgical and non-metallurgical industrial activities. In
Mattingly 2007, 448–62.
Shaw, I. 2013. Pharaonic Egypt. In Mitchell and Lane 2013, 737–50.
Smith, M.E. 2017. How can archaeologists identify early cities: definitions, types,
and attributes. In Fernández-Götz and Krausse 2017, 153–68.
Trigger, B. 2003. Understanding Early Civilizations: A Comparative Study.
Cambridge: Cambridge University Press.
Welsby, D. 2013. Kerma and Kush and their neighbours. In Mitchell and Lane
2013, 751–64.
Whitehead, N. 1992. Tribes make states and states make tribes: Warfare and the
creation of colonial tribes and states in northeastern South America. In
Ferguson and Whitehead 1992, 127–50.
Wilson, A.I. 2012. Saharan trade: Short-, medium- and long-distance trade net-
works in the Roman period. Azania: Archaeological Research in Africa 47.4:
409–49.
Yoffee, N. 2005. Myths of the Archaic State: Evolution of the Earliest Cities, States
and Civilizations. Cambridge: Cambridge University Press.

https://doi.org/10.1017/9781108637978.020 Published online by Cambridge University Press

 Index

[Page numbers in italics are figures; with ‘t’ are tables; with ‘n’ are notes.]

abandonment (site), 510–13, 710–11, 716–17 al-Hamma see Aquae Tacapitanae

Awdaghust/Tagdaoust and Tadmakka/ al-Hayz oasis (Western Egypt), 124–25, 158
Essouk, 657 al-Jiarabub (Libya), 125–27
Jenné-jeno, 529 al-Jufra oases (Libya), 130–38, 131, 707–8t
Lake Chad, 567–68, 574 al-Quantara tomb (Morocco), 482, 483
Abdallah ibn Sa’ad ibn Abi-Sarh, 128 al-Qurayyat al-Gharbia (Tripolitania),
Abu Bakr, 706 189–90t, 192–93, 226, 425
Abu Ballas (Western Egypt), 120t, 123–24 Alexander the Great, 125
Abu al-Muntasir al-Yasa, 616 and Cyrenaica, 397
Ad Calceum Herculis (fort) (Algeria), 217 temple at Bawiti, 124, 158
Ad Maiores (Algeria), 212n116 Alexandria (Egypt), 167–68
Ad Medias (Tunisia), 212–13 Algeria, and the Andalusians, 488
ad Turres see Tamerza Almohad dynasty, 278
Adrar des Ilforas (Southern Sahara), Almoravid dynasty, 239, 250, 651–52, 706
279–97t, 316 and Azugi, 319
Aeneid (Virgil), 410 and Kumbi Saleh, 630
Aethiopes, 298 Southern Sahara, 278
Afunfun (Southern Sahara), 279–97t, 335 and Tadmakka, 635
Agadem (Southern Sahara), 307 and Tagdaoust, 625–26
Agadez (Niger, Southern Sahara), 277, 312, and Takrur, 551
313, 681, 707–8t Althiburos (Tunisia), 397, 406–7, 448, 467–71
Agarsel Nepte see Nafta Project on protohistory, 448–50
Agbe-Davies, A.S., 523, 524 environmental data, 449–50, 455–60, 456,
Aghram Nadharif (Fazzan), 68–69t, 94–95, 457, 458, 467–68
100, 337, 714 excavation results, 450–55, 451, 453, 454
agriculture funerary data, 461, 462, 464, 465, 466, 467,
Althiburos, 457–60 468, 470
Morocco, 478, 609–11, 616 Amazigh groups, 20, 20–21, 490
and Roman cities, 421, 424 amazonite, 302, 308, 315, 516, 573, 713
and sedentarisation, 332–35 Amblard-Pison, S., 501
and settlements, 334–36 Amheida see Trimithis
Western Desert, 138 Ammaedara (Haidra, Tunisia), 208, 415t, 431
Ahaggar Mountains, 311–12 Ammon see Siwa
Aïr (Southern Sahara), 279–97t, 312, amphitheatres, 418, 423
707–8t, 710 Carthage, 409, 410
Akjoujt (south-west Sahara), 279–97t, Amtal (Mauritania), 279–97t, 319, 322
317, 321 Andalusians (Algeria), 488
copper, 321, 526, 547, 554 animals
Akreijit (Mauritania), 279–97t, 502, 503, 573 Althiburos, 457–59, 458
Akumbu (Mali), 505 beasts of burden, 18, 174
al-Abid (Fazzan), 90 camels, 174, 703, 717
al-Barkat oasis, 68–69t, 91, 92 and caravans, 651
al-Bdayir (Fazzan), 83, 83–84 Middle Senegal Valley, 547, 554
al-Fuqha (Libya), 130 care for, 607
722 al-Guebba (Tunisia), 210 Euesperides, 406

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 Index 723

Lake Chad, 566, 574–75 Bâ, A., 553

Mauritania, 526 Badias (Bades) (Algeria), 213, 213, 226
Middle Senegal Valley, 544–45 Bagnall, Roger S., 160
Apollo, sanctuary and temple of, 403 Bahariya (Oasis Parva) (Western Egypt), 120t,
Apollonia (Cyrenaica), 396, 404 124–25, 158–59
Appian Bahr-al-Ghazal (Southern Sahara),
description of Carthage, 399 279–97t, 302
on Numidia, 477 Bailloud, G., 300–1
Aquae Tacapitanae (al-Hamma) (Tunisia), Bakri, al-, 130, 136, 256, 542
207–8, 210 on Awdaghust, 550
aqueducts, 416–18, 420 on Azugi, 319
Carthage, 410–11, 429 on Gao, 641
in Late Antiquity, 429 on Kumbi Saleh, 630n44
Arawan (Western Sahara), 317 on Sijilmasa, 596–98, 599–600, 602–4,
architecture 605–7, 611
Middle Senegal Valley, 544 on Tadmakka, 634–35, 656
Southern Sahara/Niger Bend area, 498–500, on Takrur, 550–51
499, 510–13, 512 Bani Walid (Libya), irrigation, 13
Middle Niger, 504–10, 507, 508 Baquates people, 251n25
and northern parallels, 514–16 Baradez, J., 217n139
Tichitt, 500, 502, 504 Barca (Ptolemais) (Cyrenaica), 396–97, 404,
West African, 658–61 415t
Arlit (Southern Sahara), 279–97t, 314 craft production, 422
army, Roman, 424–25 fortifications, 431
Arondo (Middle Senegal), 543, 542, 549, Barnett, T., 76
554 Barth, H., 53n1, 85
Arsinoë see Tocra baths (Roman), 208, 417–18
artesian springs/wells, 13, 14, 339–40 Capsa, 208
Darj, 198–99 Carthage, 411–12
Fazzan, 55, 84, 92 Late Antiquity, 426–27
Ghadamis, 195 Leptiminus, 429
Nafta, 211 Battiad dynasty, 397
Nefzaoua oases, 201, 206 Battos, King/Battiad dynasty, 397
Telmine oasis, 202–3 Bauer, A.A., 523, 524
The Jarid, 209 Bawiti (al-Qasr/Psobthis) (Western Egypt),
Tuggurt, 219 124–25, 158–59
Wargla, 221 beads, 540n69
Western Desert, 116, 150 beasts of burden
asnam, 195–96 camels, 174, 703, 717
Assodé (Southern Sahara), 313–14 and caravans, 651
Aswan see Syene/Aswan Middle Senegal Valley, 547, 554
Atlantic Coast, 321–22 Becker, C., 552
Atlas, Saharan, 217–18 Benghazi see Euesperides
Awdaghust see Tagdaoust Bentaher, Fuaad, 128
Awjila (Libya), 120t, 127–28, 137 Bentyia-Kukiya (Mali), 622, 646–47
Awlil salt mines, 550, 551 Berbers see Amazigh; Garamantes; Tuareg
axe, bronze, 482 Berenice see Euesperides
Ayoub, Mohammed, 53, 56 Berenice, Queen, 405
Azawagh region (Southern Sahara), 279–97t, Berenike (Egypt), 168
314–15, 710 Berthier, D., 631–33
Azelik-Takadda (Niger), 279–97t, 647 Besseriani (Nigrenses Maiores) (Algeria), 212
Azugi (south-western Sahara, Mauritania), Bilma sand sea area (Southern Sahara), 303–8
279–97t, 319, 317, 622, 646, 707–8t Biskra (Algeria), 213, 214–15, 225

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 724 Index

blacksmith, founder of Sijilmasa (Midrar), Middle Nile, Kerma, 363

596–97, 602–4, 606 Morocco, 490–94, 491, 493
blockhouses, 404, 430–31 Kach Kouch, 486–87
Bocoum, H., 552 Lixus, 481–82, 483
Bonnel de Mézières, A., 646 Roman North Africa, 419
Bonte, Pierre, 680–81 Sabratha, 428
Borku (Southern Sahara), 279–97t, 301–2 south-west Sahara
Bourdieu, Pierre, 594–95 Akjoujt, 321
brass, clochettes, 549–49 Atlantic coast area, 322
brick Kumbi Saleh, 631
fired, 539, 643, 645, 660 Wadan, 320
see also mudbrick Southern Sahara
bronze, Middle Senegal Valley, 546 Adrar des Ilforas, 316
Agadem, 307
Bu Nijim (Tripolitania), 188–92, 191, 226,
Aïr, 312–14
425, 426
Azugi, 319
Buda (Algeria), 267
Borku, 302
Bulla Regia (Numidia), 406–7, 415t, 445
Djado, 304
burges, 199 Ennedi, 301
Burgu (Punic town) (Tunisia), 401 Iwelen, 312–13
Burgus Speculatorum (al-Kantara) (fort) Kufra, 300
(Algeria), 217 Seguedine, 305
burials, 23 Tahoua, 312
Algeria Termit, 307
Andalusians, 488 Tibesti, 302–3
Wadi Jedi area, 215 Tadmakka/Essouk (Mali), 636, 654, 655
Carthage, 400 Tagdaoust (Mauritania), 628
central Southern Sahara Thugga, 456n47
Ahaggar Mountains, 311 Tripolitania
Tassili n’Ajjer, 309–10 Chawan, 199
Cyrenaica, 403–4 Darj, 198
Barca, 404 Ghadamis, 195–98
excavation techniques, 445–46 Kebili, 205
Fazzan, 100 Tripolitanian Jabal, 200
al-Barkat, 68–69t, 337 Tunisia, Althiburos, 450, 461, 462,
Aghram Nadharif, 95 464, 465, 466, 467,
Fewet, 68–69t, 93, 94 468, 470
Ghuddwa, 89 Wadan (south-western Sahara), 320
Murzuq basin, 80, 83 Western Egypt
Sabha, 90 Bahariya, 159
Tadrart Akakus, 96–97 Bawiti, 124, 159
Wadi al-Ajal, 76–77 Dakhla, 123, 160
Wadi ash-Shati, 87–88 Trimithis, 163
Wadi Tanzzuft, 92 Farfara, 124, 159
Gao Saney (Mali), 643 Kharga, 121
Late Antique, 428 Siwa, 125, 158
Libya Western Sahara
al-Jiarabub, 120t, 126–27 Azugi, 319
Awjila, 128 Foum Larjam, 262–63
eastern Libya, 138 Sijilmasa, 256, 604–5
Hun, 136 Taouz, 257
Tagrifet, 130 Tawdenni, 316
Waddan, 134, 136 Wadi Draa, 259, 262, 270
Meroitic kingdom, 385 Wadi Gir, 253
Middle Niger, Jenné-jeno, 532, 533 Wadi Noun, 263

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 Index 725

western Algeria, 268 Chirfa see Djado

Tabalbala, 268 Christianity, 427–29
see also monuments churches, 426
Burkina Faso, 509 Lepcis Magna, 427–28
Busi (al-Jufra group), 133 Sufetula, 427, 428
Byzantine period Cidamus see Ghadamis
Hodna, 217 Cintas, Pierre, 479, 482–83
Negrine, 212 circuses, 418–19, 423
Western Algeria, 267 Carthage, 409, 410
Cirta (Algeria), 217, 218, 397, 418, 445
Callimachus, on the siege of Euesperides, 405 see also El Kef
camels, 174, 547, 554, 651, 703, 717 Cissé, Mamadou, 505, 643
Camps, Gabriel, 446, 462, 489, 490, 491, 492 cisterns, 14, 417
Capel, C., 631 Althiburos, 452, 454, 468
Capitolium, 416 Carthage, 399, 400, 409, 410–11
Carthage, 410 Euesperides, 405
Thuburbo Maius, 429 citadels
Timgad, 412, 413, 416 Aghram Nadharif (Fazzan), 94, 95
Capot-Rey, R., 672n28 see also qsur
Capsa (Gafsa) (Tunisia), 208–9, 224, 347 city, definition, 522
caravans, number of camels, 651 city-states, 27–28, 586–89, 670n20, 703–5
Carthage civilisation, defined, 697
Late Antiquity, 430 civitas, 347
Phoenician, 396, 398–400, 399 Nefzaoua, 205–6
Roman, 398, 408–12, 409, 418, 429 Telmine, 202, 227
Cassius Dio, 249 Thiges, 209–10
Castellum Dimmidi (fort) (Algeria), 217 Clapperton, H., 136
castles see qsur Clark, Mary, 534
cattle, 301 climate, 8–12, 9, 334, 349–50, 716
Althiburos, 458–59 Lake Chad, 567–68
Mauritania, 526 Middle Niger, 11, 525–26, 527, 537
cemeteries see burials Middle Nile, Meroitic kingdom, 370–71
centralisation, 681–85 Middle Senegal Valley, 545, 546, 550
cereal cultivation, 332, 456, 478, 610 Tafilalt (Morocco), 609–10
Chami (Atlantic coast), 279–97t, 317, 322 Western Egypt, 115, 150–51
Champault, D., 268 clochettes, brass, 549–49
chapels, funerary, 253–54, 262, 318 collapse, urban, Western Egyptian Desert,
Chappelle, J., 303 148–49, 172, 174
chariot representations coloniae, 347, 423
Southern Sahara, 308 Tacapae, 207
Iwelen, 313 colonialism, 29, 228
Tahabort, 311 British, 677, 679n70
Timmissao, 311–12 French, 219, 250, 348, 541, 670, 675–76,
Taouz, 256–57 678–79
Wadi al-Ajal, 76n49 influence on scholarship, 446–47, 477
Wadi Noun, 263 see also Roman Empire
Chawan (Tripolitania), 191, 199 commodities, Western Desert, 171–72
Chebika (Speculum) (Tunisia), 211 Cooper, F., 679n70
Chemtou (Numidia), 406–7 copper
chickens, 516, 540, 547, 555 Jenné-jeno, 529, 531t, 535, 555
Childe, V. Gordon, 564 Middle Niger, Gao, 538
Chinguetti (Western Sahara), 317, 320, 622, Middle Senegal Valley, 545, 546,
646, 660 550, 554

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 726 Index

copper (Cont.) Western Desert, 120t

south-west Sahara, Akjoujt, 321, 526, Western Sahara, 248t
547, 554 see also individual sites
Southern Sahara Daumas, E., 214n130
Azawagh region, 314–15 Daza people, 298
Marandet, 315 defences see fortifications
Tamdult (Morocco), 700 defensive coalescence, 586–87, 588
Western Sahara, 248t Demeter, temples in Cyrenaica, 403, 404
Cornelius Balbus, Lucius, 188, 195, 213–14, 224 dendritic settlement systems, 25–26
cotton, 374, 379, 385–86, 549 desert, definition, 8
councils, tribal, 670–71 Desert Migrations Project (DMP), xx, 57
craft production, 347 Despois, J., 56, 670, 673
Gao, 541 Dhar el-Moudden (Morocco) see Kach Kouch
and identity, 389–90 Dhar Tichitt area (Mauritania), 278, 279–97t,
Lake Chad 318–18, 333, 565, 581
Dorota, 578 Di Vita, A., 129–30
Zilum, 572–73 Dia-Shoma (Mali), 504–5, 510, 513, 515
Zubo, 579, 580 trade, 516
Meroitic kingdom, 377, 385, 389 Diana Veteranorum (Algeria), 425
Mussawarat, 381 diggers, well, 219
Roman period, 421–24 Dio see Cassius Dio
see also copper; iron; textile production Djaba (Southern Sahara), 279–97t, 303–4
craft specialisation, 344, 553, 700–1 Djado (Chirfa) (Southern Sahara), 279–97t,
Cressier, P., 661 303–5
crocodiles, 257, 277, 318 Djemila see Cuicul
god, 116 Djenné see Jenné-jeno (Mali)
Crumley, C., 439 Dorota (Lake Chad), 574–78, 576, 587–88, 589
Cubalel (Middle Senegal Valley), 544, Doucen (Algeria), 215
546–47, 551 Douz (Tunisia), 202, 205
Cuicul (Djemila) (Algeria), 413–14 dual town relationships, 652–53, 700
Cyrenaica (north-eastern Libya), 396–97, 398 Duveyrier, H., 53
Cyrene (Cyrenaica), 396–97, 401–4, 403, dye production, 264, 422
405–6, 415t Euesperides, 406
Meninx, 401
Dakhla (Western Egypt), 122–23, 149, 149–50, Sabratha, 422
150, 707–8t
pre-Roman settlement, 155, 333 Earle, T., 439, 442, 702
Roman period, 159–60, 348 Eastern Libya, 112, 114, 127–37, 131, 133, 134,
Dakhlet et Atrous (Middle Niger), 335, 502n19 135
Daniels, Charles, 53, 56 see also Cyrenaica
Darj (Tripolitania), 191, 198–99 Eastern Sahara see eastern Libya; Western
date palms, 6, 7–8, 332 Desert (Egypt)
and groundwater catchments, 15 Echallier, J.-C., 267
iconography, 21, 224 Egypt
Morocco, 240–41, 610–11 New Kingdom colonialism in Nubia,
Tagdaoust, 550 365–68, 367
Wadi Draa, 257 Oxyrhynchus, 165–66
Wadi Tanzzuft, 76n47 see also Western Desert (Egypt)
dating, 37–39, 337, 717 El Kef (Tunisia), 397
Central Sahara (Fazzan), 68–69t El Ksour necropolis (Tunisia), 450, 461, 462,
Eastern Sahara, 120t 464, 465, 466, 467
Northern Sahara, 189–90t Emberling, G., 709n65
Southern Sahara, 279–97t Emi Tchouma (Southern Sahara), 306

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 Index 727

Ennedi (Southern Sahara), 279–97t, 300–1 Fewet (Fazzan), 68–69t, 91, 92, 94, 100, 714
environmental data Fez (Morocco), 602–3
Althiburos, 449–50, 455–60, 456, 457, 458, Flannery, K., 28, 696
467–68 Fletcher, R., 25, 349
Morocco, 478 Flight, C., 643–45
epigraphy floodwater farming, 14, 194, 199, 332–33
and Numidia, 446 foggaras, 13d, 15–16, 23
and Tadmakka/Essouk, 654 Algeria
Erfoud see Tafilalat/Tafilalt Ghardaia/Mzab, 223
Essouk see Tadmakka/Essouk Wargla, 221
euergetism, 420–21 eastern Libya, 138
Euesperides (Berenice) (Cyrenaica), 397, 401, Al-Fuqha, 130
402, 404–6, 418 Tagrifet, 130
craft production, 422 Waddan, 136
Eustache, Daniel, 611 Zala, 129
Euzennat, M., 200, 490 Fazzan, 55, 70, 73–75, 80, 99, 339
Evans-Pritchard, E.E., 676n48 al-Abid, 90
Umm-al-Aranib, 83–84
evolutionary theory, 439, 441–42, 444
excavation techniques of burials, 445–46 Tripolitania
Capsa (Tunisia), 209
exchange see trade and exchange
Chawan, 199
exotic goods, 170
Darj, 198–99
Middle Senegal and Niger rivers, 523–24,
Ghadamis, 13, 196–98, 341
530–31, 531t, 540, 550, 553–55
the Nefzaoua oases (Tunisia), 201, 206
Telmine, 204–5
Fachi (Southern Sahara), 306–7 western Nefzaoua, 205
fairs, Tafilalt, 596, 606–7 Western Egypt, 138, 153–54, 170
Fakhry, A., 124, 158 Bahariya, 125, 159
families, trading, 709 Dakhla, 155
Farfara (Farafra) (Western Egypt), 124, 159, Farafra, 159
333, 707–8t Kharga, 121, 154–55
Fatimids, 602 Western Sahara
the Fayyum (Western Egypt), 116–21 Tafilalat, 255
Karanis, 168 Wadi Draa, 258
Fazzan (south-western region of Libya), 58, Wadi Gir/Saoura, 252, 253
68–69t, 97–103, 701–2, 706, 707–8t Western Algeria
and the Garamantes, 53–55, 335–38 Gourara/Tuwat/ Tidikelt, 265, 267, 268
history of the research, 55, 58 Tabalbala, 268, 269
Murzuq/Hufra Basin and ash-Sharqiyat, fora, 415–16
68–69t, 77, 79, 80, 81, 82, 83, 85, Carthage, 408–10, 409
86–89, 98, 101 Sabratha, 413, 419
outlying oases, 89–90 Timgad, 412, 413
Wadi al-Ajal and Jarma, 58–77, 70, 72, 74, fortifications, 701
75, 98, 100–1, 102t, 103, 344–45 Althiburos, defensive wall, 454, 454, 455
Wadi ash-Shati, 84–89, 85, 86t, 98, 101, eastern Libya
102t, 103 Tagrifet, 130
Wadi Hikma, 58, 68–69t, 97, 98, 101 Waddan, 134–36
Wadi Tanzzuft/Tadrart Akakus, 68–69t, Mediterranean, late antique, 430–31
90–97, 91 see also qsur
water sources, 13, 15, 55, 58–70, 80, 84, fortified settlements, 336
338–40 Western Egypt
Fentress, E., 20, 336, 424–25 Dakhla, 123
Fenwick, C., 673, 683–84 Kharga, 121–22
see also Lake Chad; qsur
Février, Paul Albert, 489

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 728 Index

fortlets, Roman, 193–94, 198, 199, 201, 216, inequality/slavery/labour exploitation, 678–81
224, 226 oases, 22
fortresses, Western Egypt, Bahariya, 125 population, 713
forts (Roman), 347, 705–6 settlement, 332, 335–38, 347, 671–74
al-Qurayyat al-Gharbia, 189–90t, 192–93, and Southern Sahara/Sahel area, 514–16
226, 425 spiritual centres, 674–77
Bu Nijim, 188–92, 191, 226, 425, 426 as a state, 28–29, 667–68, 685–86, 703, 709,
Castellum Dimmidi, 217 711–15
Chawan, 199 centralisation and territoriality, 681–85
Doucen, 215 textile production, 225
Gemellae, 214, 215, 216, 225, 347 see also Fazzan; Jarma
Hodna, 216–18 gardens
Madress, 198 and market towns, 656
Mizda, 194 oasis, 13t, 16
Nigrenses Maiores, 212 see also individual sites
Northern Sahara, 223–24, 226 Garenne-Marot, L., 548
Remada (Tillibari), 200 Garrard, T., 516
in the Saharan Atlas, 218 garrison settlements (vici), 225–27, 347
Sidi Uqba, 214 al-Qurayyat al-Gharbia, 193, 425
Forty Days Road, 171 Bu Nijim, 188–92, 191, 226, 425, 426
fossatum (linear barrier), 216–17, 220 Gemellae, 215, 216
Foum Larjam (Wadi Draa) Mizda, 194
(Morocco), 262–63 Ras al-Ain, 201
fountains Gauthier, Yves and Christine, 256–57, 308
Carthage, 411 Gautier, E.-F., 267
Lepcis Magna, 417 Gawgaw see Gao
functional interdependence, 553 Gemellae (Algeria), 214, 215, 216, 226, 225, 347
and Jenné-jeno, 534, 535 Geoffroy, Auguste, 673n35
funerary chapels, 253–54, 262, 318 Gezabi (Southern Sahara), 305, 305, 707–8t
Fura Wells (Sudan), 369 Ghadamis (Cidamus) (Libya), 189–90t, 191,
348, 704
Gabes see Tacape/Tacapae irrigation regimes, 13, 341
Gabriel, B., 267 modern period, 18
Gaetuli, 20, 201, 224, 240, 269, 342 Roman period, 195–98, 197
Gafsa see Capsa Ghaki, M., 200
Gajiganna complex (Lake Chad region), Ghana (kingdom), 542, 630–33, 649–50
566–74, 569, 571 architecture, 658, 659–60
Gao (Kawkaw/Gawgaw) (Mali, Middle Niger), as a dual town, 652–53
515, 524–25, 537, 541, 555, 641–43, and Silla, 550
649–50 urban structure, 657
architecture, 658, 660 see also Kumbi Saleh
as a dual town, 652–53 Ghardaia (southern Algeria), 222–23
urban structure, 657 Ghat oasis (Libya), 91, 92, 98, 348n55, 348
Gao Ancien (Mali, Middle Niger), 279–97t, Ghuddwa (Fazzan), 89
537, 538, 540–41, 623–24, 642, 644, Giddy, L.L., 147n3, 170
645, 649, 657 Gill, J.C.R., 148n5
architecture, 659, 660 glass
Gao Saney (Mali, Middle Niger), 279–97t, 507, working
509, 537, 541, 642, 643–45, 649 Gao, 541
Gara Krima (Algeria), 222 Sabratha, 423
Garama see Jarma glass beads
Garamantes (ancient people), 21, 53–55, Dia, 516
335–36, 441 Jenné-jeno, 531t

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 Index 729

Kawar, 308 Western Sahara

Kumbi Saleh, 633 Sijilmasa, 256
Middle Niger, Gao, 538, 539, 554–55 Taouz, 257
Niger Bend, 540n69 Wadi Draa, 259–63, 260, 269–70
Godelier, M., 438, 442 Wadi Gir/Saoura, 253
gold, 157, 159 hinterlands, dispersed, 671, 674
Gao, 539 Hippodamean planning, 166, 167
Nubia, 366 Hitchner, R.B., 425
Tadmakka, 555, 700 Hodna (Algeria), 216–18
Takrur, 541 Honeychurch, W., 709
and trade, 153, 516, 623–24 Hope, Colin A., 160
Wadi Noun, 264 Horden, P., 671
Western Desert, 171 horses
Goulfei (Lake Chad region), 585 Middle Senegal Valley, 546n88, 549
Gourara (Algeria), 241, 266, 265–68, 674 representations
Gourma (Mali), 498, 499, 509, 513, 515, 539 Iwelen, 313
Gozalbes-Gravioto, Enrique, 482 Wadi Draa, 259, 260–61
Gracchus see Sempronius Gracchus and state formation in Numidia, 446
graffiti, Mussawarat, 381 Houlouf (Lake Chad region), 585
‘graffito’ pottery, 480, 481, 487 house types, Southern Sahara, 306–6
grain-beer, 385 Humera (Fazzan), 83, 83–84
grapes, Althiburos, 456–57 Hun (al-Jufra group) (Libya), 120t, 136–37
Grébénart, Danilo, 492 Hunwick, J.O., 537
Greece, and Cyrenaica, 396–97, 398, 401–6,
402, 403 Ibadi Muslims, 599, 634, 655–56, 660, 704,
groundwater catchments, 15, 321 707–8t
Gsell, Stéphane, 446 Ibn Abd al-Hakam, 307
Guédon, S., 225 Ibn Battuta, 267
Gutelius, David, 674 on Takadda, 651
Ibn Furtu, 305
Hadj-Sadok, Mohammed, 611 Ibn Hawqal
hafirs (water reservoirs), 373, 377, 381–82 on Awlil and salt, 550
Haidra see Ammaedara on Tadmakka/Essouk, 634
Hammoudi, Abdullah, 674 on Tagdaoust, 625
Hamy, E.-T., 215 Ibn Khaldun
Hardesty, D.L., 340–41 on Berber pastoralists, 673n35
Harris, M., 439 on qsur, 250
Hatté, C., 547 on Wargla, 221
Henchir Taddert (Algeria), 212–13 identity, craft-based, 389–90
Herodotus Idrisi, al-
on the Garamantes, 54 on Azugi, 319
on Libya, 477–78 on Gao, 642n82
heterarchy, 443, 523, 678 on Kawar, 305n44
Hibis (Western Egypt), 121, 153, 154 on Takrur, 551
and Roman period, 156 on Waddan, 134
hierarchical settlement ‘Ikrima, 601, 616
Garamantes, 671–74 Ile à Morphil see Middle Senegal
Lake Chad, 565–66, 569, 572, 581–83 Iligh (Morocco), 674–75, 676
hierarchy see social inequality Imazighen see Tuareg
hillforts, 335 In Teduq (Southern Sahara), 279–97t, 312, 315
Gara Krima (Algeria), 222 industry see craft production
near Remada, 200 Insoll, T., 642–43
Wadi al-Ajal, 72, 98 Inufan (Southern Sahara), 279–97t, 312

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 730 Index

Iol Caesarea (Algeria), 415t, 418, 423 Jews

iron Tabalbala, 268
Jenné-jeno, 529–31, 531t and Tuwat, 20n52, 267
mines Jodin, André, 483
Bahariya, 159 Johnson, A.W., 439, 442
Wadi ash-Shati, 84–85 Johnson, Douglas, 677
sickles, Morocco, 478 Jorf Torba (Wadi Gir, Algeria), 248t, 252, 254
and state formation, 446, 447 Julius Caesar, 408
working Junhan (Tadmakka), 636, 651
Althiburos, 450, 460–61, 470 Junker, L.L., 25–26
Jenné-jeno, 535 Justin, 469
Lake Chad region, 575, 578, 579, 580
Meroitic kingdom, 388 Kach Kouch (Dhar el-Moudden) (Morocco),
Hamadab, 378, 379, 388 484–87, 485, 486, 487
Meroe, 375 Kanim-Borno (Lake Chad region), 584–86, 660
Middle Senegal Valley, 552–53
Kanuri people, 303
Morocco, 484, 490
Kapteijns, L., 372
Southern Sahara, 302, 307
Karanis (the Fayyum), 168
irrigation, 16, 17
Kastiliya see Jarid
al-Hamma (Tunisia), 210
Kawa (Middle Nile), 366n23, 368
Ghardaia/Mzab (southern Algeria), 223
Kawar (Southern Sahara, Niger), 299, 303–8,
Ghat, 92
304, 305, 322, 623–24, 647, 650,
and slavery, 332
707–8t
Trimithis, 123
urban structure, 657–58
Western Sahara
Kawkaw see Gao
Tafilalat, 255
Wadi Draa, 13a,b, 258 Kebili (Tunisia), 202, 205
see also foggaras Keding, B., 300–1
‘Isa ibn Mazid, 598, 600–2 Kellis (Ismant al-Kharab) (Western Egypt),
Islamic period 120t, 122, 123, 160, 346
Ibadi Muslims, 599, 634, 655–56, 660, 704 Kerkouane (Punic town) (Tunisia), 400–1, 401
kharijism, 598–99, 600–2, 609 Kerma (Middle Nile), 360–61, 363, 364, 365
and trade, 350–51 Kharga City (Kharga), 174
zawaya (religious settlements), 674–77 Kharga (Western Egypt), 121–22, 149, 149–50,
Ismant al-Kharab see Kellis 150, 346, 707–8t
Iwelen (Southern Sahara), 279–97t, 312–13 pre-Roman settlement, 154–56, 333
Roman period, 156–57
Jaa-Ogo, 552 temple of Amun at Hibis, 153, 154
Jabal Afilal (Morocco) see Taouz and trade, 171
Jabal Bani (Middle Draa), 258–59 kharijism, 598–99, 600–2, 609
Jabal Oulad Zorha (Morocco), 613–15 Khartoum (Sudan), 362
Jacques-Meunié, D., 610 kingdoms, 28
the Jarid (Kastiliya) (Tunisia), 209–11, 213, Massyle, 469
707–8t Mauritanian, 469, 492–94
Jarma (Fazzan), 26, 58–77, 70, 72, 74, 75, Tadmak, 634
100–1, 344–45, 714–15 see also Ghana; Meroitic kingdom
and trade, 350, 713 kingship, Kerma, 363–65
urbanisation, 668–71, 698 Kolima-Sud-Est (Mali), 504–5, 509, 513
Javolenus Priscus, 209 Korandjé (language), 268
Jenné-jeno (Mali), 508–9, 510, 514, Koro-Toro (Southern Sahara), 279–97t, 302
521–22, 525 Kotoko (Lake Chad region), 584–89
trade, 516, 527–28, 530, 531t, 532, Kotula, Tadeusz, 446–47, 469
533, 555 Kriz (Tunisia), 210
Jennings, J., 702 Ksar al Barka (south-west Sahara), 317, 320

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 Index 731

Kufra (Southern Sahara), 300 Lull, V., 439

Kuhlmann, K.P., 153, 171, 174 Lyon, G.F., 137
Kumbi Saleh (Ghana kingdom) (Mauritania),
31, 279–97t, 318, 317, 623–24, 632, macella (markets), 416
633, 649, 657 Macina region see Dia-Shoma
architecture, 658, 660 McIntosh, Susan, 670
Kushites, 364, 365, 368–69, 387, 388 Mackensen, Michael, 193
Kusimba, C., 524 Maghrawa clans, 602–3
Kysis (Dush) (Western Egypt), 120t, 121, 154, Mahariat (Western Sahara), 279–97t, 317, 322
156–57 Marada (eastern Libya), 128–29
Marandet (Southern Sahara, Niger), 279–97t,
Lake Chad 315–16, 647, 707–8t
fortified settlements, 564–66, 575, 576, 579 marauders, Western Desert, 171, 174
Gajiganna complex, 566–74, 569, 571 Marcus, J., 28, 696
Iron Age sites, 574, 575 Margat, Jean, 605, 610
Zubo and Dorota, 575–81, 576, 579 Marius, Caius, 208, 224
and the origins of urbanisation, 581–83 market towns/trading centres, 318–21, 621–24,
and state formation, 583–88 622, 701
Lambaesis (legionary base) (Algeria), 425 architecture, 658–61
lamps, oil, Gao, 540 Awdaghust/Tagdaoust, 624–29, 627, 628
land survey, Roman, 201, 205–6 Azelik-Takadda, 279–97t, 647
languages Azugi, 279–97t, 317, 319, 622, 646
Korandjé, 268 Bentyia-Kukiya, 622, 646–47
Songhay, 268–69 Chinguetti (Western Sahara, Mauretania),
Lassère, J.-M., 342n34 317, 320, 622, 646
Late Antiquity, 166, 426–32 early Islamic era, 650–53
latrines, Roman, 417–18 Gao, 641–43, 644, 645
Lebeuf, J.P., 302 Kawar, 299, 303–8, 304, 305, 322, 623–24,
Lepcis Magna (Tripolitania), 413, 414, 647, 650
415t, 416 Kumbi Saleh (Ghana kingdom), 279–97t,
church, 427–28 318, 317, 623–24, 630–33, 632, 633,
circus, 418 649, 657
craft production, 422 Marandet, 279–97t, 315–16, 647, 707–8t
macellum, 416 pre-Islamic occupation, 647–50
olive presses, 429 Tadmakka/Essouk, 279–97t, 633–41, 637,
war with Oea, 423 638, 639, 640, 654–57
Leptiminus (Tunisia), 415t, 422, 423, 429, 430 Tagdaoust, 279–97t, 318, 317, 550, 623–29,
Leschi, L., 56, 95 627, 628, 648, 650–52, 654–57
Letheilleux, J., 56 Tireqqa, 646–47
Lévêque, Pierre, 447 Tombouze, 279–97t, 318, 506, 514–15, 646,
Lewicki, T., 634, 648 649–50
Lhote, H., 310, 311 urban structure/features, 633, 640, 653–58
Libya Martin, A.G.P., 266
eastern, 112, 114, 127–37, 131, 133, 134, 135 Masaesyli Numidians, 440
see also Cyrenaica; Fazzan; Syrtica; Massyli Numidians, 440, 448
Tripolitania kingdom of, 469
Libyan peoples, 115 Mauny, R., 305
Limagues (Tunisia), 209 Mauretania, 398
limes Tripolitanus, 200 Caesariensis, 250–51, 252
Liverani, Mario, 28, 57, 516, 667 Tingitana, 250–51
Lixus (Morocco), 480–82, 481, 482, 483 see also Akreijit; Azugi; Tichitt; Tipasa
Logone-Birni (Lake Chad region), 585 Mauritania see Akreijit; Azugi; Kumbi Saleh;
Lonsdale, J., 681–82 Tichitt

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 732 Index

Maysara, 599, 601 ‘mirror states’, 28–29

megalithic necropoleis, 446, 490 Mizda (Tripolitania), 191, 194
Althiburos, 450, 461, 462, 464, 465, 466, 467, Mlili (Milgis) (Algeria), 215
468, 470 Mogador (Morocco), 482, 484
Lixus, 481–82 Monod, T., 318
Méma region (Mali), 505–6, 509, 510, 511–13, Monroe, J.C., 695
512, 515 monumental architecture, 419–21, 431–32,
Meninx (Punic town) (Tunisia), 401, 418, 422 522
Mentes (Southern Sahara), 279–97t, 312 Althiburos, 406
Meroe (Middle Nile), 372–75, 374, 376, 699 Carthage, 409, 410
Meroitic kingdom, 369–72, 684 Garamantes, 72–73, 697
Meroe, 372–75, 374, 376 Jarma, 669
metropolis, Ptolemy on, 698n21 Lake Chad region, 584
Middle Kingdom Western Desert, 346
and Western Desert settlement, 116, see also Western Butana
124, 153 monuments (funerary), Western Sahara,
Middle Niger, 279–97t, 522 253–54
climate, 11 monuments (ritual), 303
see also Gourma; Méma; Tagant; Timbuktu; Moraes Farias, Paulo, 636, 643
Walata Mornou (Southern Sahara), 279–97t, 301
Middle Nile, 359–63 Morocco, 476–80, 477, 489–91
Kerma, 360–61, 363, 364, 365 archaeology programme, 612–13
New Kingdom period, 365–68, 367 Kach Kouch, 484–87, 485, 486, 487
Napatan period, 360, 368–69 Lixus, 480–82, 481, 482, 483
Meroitic kingdom, 369–72 Mogador, 482, 484
Hamadab, 378, 378, 385, 388–89 Sidi Driss, 487–88
Meroe, 372–75, 374, 376 storehouses, 673
other urban centres, 375 mosaics, Roman, 212
roles of towns, 383–87, 384 Mosque of Mansa (Kankou) Moussa, 643
and social identity, 387–90 mosques, 661
Western Butana, 379–83, 383 Agadez, 313
Middle Senegal, 522, 525 Awjila, 128
Takrur, 525, 541–53, 543, 545, 549 Chinguetti, 319
Midrar, as founder of Sijilmasa, 596–97, Ennedi, 301
602–4, 606 Hun, 136
Midrarids, 597, 617 Kawar oasis area, 305, 306
migration, 267, 336, 338, 340, 652, 703 Kumbi Saleh, 661
Jewish, 20n52 and market towns, 655–56
khariji, 599 Mosque of Mansa (Kankou) Moussa, 643
Miknasa tribe, 596–97, 598–99 Tadmakka, 638, 661
Milgis see Mlili Taghaza/Tawdenni, 316
Mills, A.J., 173n158 Tozeur, 210–11
mines/mining Wadan, 320
metal Wadi Draa, 248t
Akjoujt, 547n94 see also temples
Azawagh, 314
Mothis (Mut el Khareb) (Western Egypt), 122,
Bahariya, 159
153, 155, 174
Jaa-Ogo, 553
Tabalbala, 268 pre-Roman settlement, 155
Wadi ash-Shati, 84–85 Roman period, 160
salt Mouyssam II (Mali), 506
Azawagh, 315 mud, building material, 658–60
Kawar, 706 mud plaster, 659–60
Mali, 8, 706 mud-sealings, Meroe, 377

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 Index 733

mudbrick, 23, 336, 658–59 Niger (river), 12

Althiburos, 452 see also Middle Niger; Niger Bend area
Cyrenaica, 404–5 Nigrenses Maiores see Ad Maiores; Besseriani
and dating, 37–38 Nigrenses people, 212n116
Eastern Sahara, 120t, 122, 130 Nile (river), 12, 115–16
Garamantes, 72–73, 78, 79, 93, 102t, 335 see also Middle Nile
Middle Niger, 504, 505–10, 507, 508, 515 Nilotic peoples, 115, 138
Gao, 538, 540 ‘Noba’ (Nubians), 371–72
Jenné-jeno, 529 ‘nomad menace’, 3
Middle Nile, 363, 365, 370 nomads, 28–29, 228, 476, 708–10
Northern Sahara, 193, 199, 206, 211, 212 and camels, 174
Southern Sahara, 306 and disruption of trade, 172
Western Desert, 161 and market towns, 651, 654
Western Sahara, 248t, 262 and Tafilalt, 604, 606–7
Muhallabi, al-, 536 North-Western Sahara see Western Sahara
on Gao, 641 Northern Sahara, 187–88, 189–90t,
on Tagdaoust, 625 223–28, 226
mummies, Western Egypt, 120t, 124, 125, eastern Algeria, 211–18, 213, 216
126–27, 138, 159 southern Algeria, 218–23, 222
municipia, 227, 347 Tripolitania pre-desert/Hamada al-Hamra,
Capsa, 208–9 188–206, 191, 197, 202, 203, 204
Gemellae, 214, 215, 216, 225, 347 Tripolitanian coastal oases, 191, 206–7
Negrine, 212 Nubia, 366–68
Nigrenses Maiores, 347 ‘Noba’, 371–72
Telmine, 202–3 Numidia, 406–8
Munson, P.J., 501 political entities, 440
Murzuq/Hufra Basin and ash-Sharqiyat, Roman, 398
68–69t, 77, 79, 80, 81, 82, 83, 85, see also Althiburos
86–89, 98, 101, 336, 341, 348, 707–8t Nybgenii people, 201–2
foggaras, 341
musalla (open prayer grounds), 627–28, 638, oases, 6, 7, 338–43
651, 654–55 and climate, 11–12
Mussawarat (Aborepe) (Meroitic kingdom), networks, 17–19
379–81, 380 past, 19–24
Mut el Khareb see Mothis and refugia, 11–12
Muweis (Middle Nile), 376, 377 and sedentarisation, 333–34
Mzab (southern Algeria), 222–23 and settlement hierarchies, 671
Mzora (Morocco), 491, 492 term, 5–7
vegetation, 7–8
Nachtigal, G., 132, 137, 302, 361 water-sources, 12–17
Nafta (Nepte/Agarsel Nepte) (Tunisia), Oasis Magna (Great Oasis) see Kharga
210, 211 Oasis Parva see Bahariya
Napatan period, Middle Nile, 360, 368–69 ochre, red, 629, 640, 659
Naqa (Tolkte) (Meroitic kingdom), 381, 382 odeons, Carthage, 409, 410
Nasamones people, 128 Oea (Tripoli), 206
Nefzaoua oases (Tunisia), 201–6, 202 war with Lepcis Magna, 423
Negrine (Algeria), 212, 213, 226 Ogo (Middle Senegal), 542, 543, 548, 549
Nema (Mauritania), 318 Oka, R., 524
Nepte see Nafta Old Jarma (ancient Garama) (Fazzan), 68–69t,
New Kingdom and Western Desert settlement, 72–73, 79b
153, 159, 171 Old Kingdom
Niche Construction Theory (NCT), 12 and Western Desert settlement, 115, 173
Niger Bend area, 317 Abu Ballas, 123

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 734 Index

Old Kingdom (Cont.) Ponsich, Michel, 480–81

Dakhla, 122, 152–53 Pope, J.W., 368, 373
the Fayyum, 116 population
Kharga, 121 and state formation, 443
olives/olive oil Western Desert collapse, 172
Lepcis Magna, 429 see also individual sites
Roman cities, 421, 426 ports, Roman period, 423–24
Carthage, 408 pottery
Siwa, 158 Gao, 538–39, 539
Sufetula, 427, 428 Jenné-jeno, 531, 532, 555
Thuburbo Maius, 429 Lake Chad, 574
oratories, 625, 655, 661 Meroitic kingdom, 389
Oxyrhynchus (Egypt), 165–66 Middle Senegal Valley, 547, 547n97, 549, 554
Morocco
Pacho, J.R., 128 Kach Kouch, 487
palaces Lixus, 480, 481
Lake Chad, 585 Mogador, 482–83, 484
Middle Nile Sidi Driss, 487–88
Meroitic kingdom, 375–77, 376, Roman period, 423–24
383–86, 384 prestige goods, Althiburos, 470–70
Western Butana, 381, 382 Processual Archaeology, 564–65
Napatan period, 368 Protohistoric period, defined, 4
palms see date palms Prussin, L., 500, 509
Paris, F., 200 Pseudo-Scylax, 478, 489
pastoralism, 3, 269, 330–31 Psobthis see Bawiti
and sedentarisation, 331–33, 334–35, Ptolemais see Barca
341, 717 Ptolemy
and Sijilmasa/Tafilalt, 605–7 on ‘Aethiopians’, 298n11
Pauphillet, D., 56 on the campaign of early AD 40s, 249
Peopling the Desert Project, xx on Garamantian kings, 684, 712
Perret, R., 310 Geography, urban terminology, 345
Persians, 153–55, 173 on the Gir, 252–53
Pharusii, Strabo on, 218 on Sabha, 90
Phazania, 195 on sites as ‘metropolis’, 698n21
Phoenicians, 396, 398, 399, 401, 478–80 and Tuwat, 266
pilgrimage centres, 387 Ptolemy II, 397
Mussawarat (Aborepe) (Meroitic kingdom), Purcell, N., 671
379–81, 380 Purdue, L., 11
pisé (rammed earth technique), 419, 658–59
plague, 427 Qala Abu Ahmed (Sudan), 368–69
planning (town) qanats see foggaras
Hamadab (Meroitic kingdom), 378, 378 qasaba
Western Desert, 164, 166–69 Southern Sahara
plaster Djado, 303–4
decorative, 660 Kawar, 307
mud, 659–60 Seguedine, 305
red ochre, 629, 640, 659 Qasr ash-Sharraba (Fazzan) 79b, 78–79,
Pliez, O., 683 100
Pliny the Elder Qasr Mqdul (Ghadamis), 196
on Cornelius Balbus, 195, 213–14 qsur (fortified villages/settlements/castles), 20,
on Mlili (Milgis), 215 22, 102t, 670, 672–73
on Tacape/Tacapae oasis, 17, 207 Djaba, 303–4
on Wadi ash-Shati, 88, 98 Fazzan

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 Index 735

Murzuq/Hufra Basin and ash-Sharqiyat, Western Desert, 151–52, 170–71

78, 79, 80, 81, 82, 83, 101, 102t Robert, Serge and Denise, 631
Wadi al-Ajal/Jarma, 75, 76, 98 Robert-Chaleix, D., 552
Wadi ash-Shati, 85–87, 86t, 101 rock art, 23–24
Wadi Hikma, 68–69t, 97, 98, 101 Adrar des Ilforas, 546n88
Wadi Tanzzuft, 92, 102t Fazzan, 55, 76, 103
near Sabha, 90, 98 Tadrart Akakus, 96, 97
Old Jarma, 73 Wadi ash-Shati, 88
Qasr ash-Sharraba, 79 Morocco, 478
Sidi Uqba, 214 Southern Sahara
Southern Sahara Iwelen, 312–13
Abalessa, 311 Tahabort, 311
Ahaggar Mountains, 311 Tibesti, 302
Fachi, 306–7 Timmissao, 311–12
Kawar, 306 Tadmakka (Mali), 636, 648
Kufra, 300 Western Sahara, 240
Tadrart Akakus, 91, 94, 96–97 Taouz, 256–57
and trade, 672–73 Wadi Draa, 259–61
Ubari sand sea lake villages, 89 Wadi Noun, 263
Waddan (al-Jufra group), 132–33 western Algeria, 268
Wadi Rhir, 219 Rockman, M., 340n32
Western Sahara, 250 Roffo, P., 215
Wadi Draa, 262 Rohlfs, G., 129, 250, 255, 265, 268
Wadi Gir/Saoura, 252 Roman empire, 21–23, 398, 431
Western Algeria, Gourara/Tuwat/ architecture/buildings, 415–21,
Tidikelt, 265, 266, 267 417, 419, 500
Zuwila, 68–69t, 82–83 Carthage, 398, 408–12, 409, 418, 429
cities’ economies, 421–23
Rachgoun (Algeria), 488–89 city planning, 408–12, 413, 414, 415t
Rachid (south-west Sahara), 317, 320 Eastern Sahara, 121–23, 124–25, 129–30,
radiocarbon dating see dating 132–36
rammed earth technique (pisé) of architecture, see also below Western Desert
419, 658–59 and the Meroitic kingdom, 387
Ras al-Ain (Tunisia), 191, 201, 226 systems and networks, 423–25
Rebuffat, R., 129, 132–33, 192, 198, 199 and urbanisation, 342–43, 345–46, 698–99
refugia, 11–12 Western Desert, 113–15, 156–69, 173–74
religious communities (zawaya), 674–77, 709 collapse of urbanism, 148–49, 172, 174
Remada (Tunisia), 191, 200–1 economic links, 169–72, 173–74
Renfrew, Colin, 440–41, 523–24 and the Western Sahara, 241–49, 253
resilience see abandonment see also Northern Sahara
Retaillé, D., 683 Rønneseth, O., 303
Richardson, J., 199 Rouvillois-Brigol, M., 221nn162, 163
Rilly, C., 379 ‘Royal City’ see Meroe
rings, copper, 540n69, 550 ‘Royal tumulus’ of In-Aghelachem, 96
Risch, R., 439 run-off farming, 332
ritual
and burials, Morocco/Algeria, 488–89 Saadi kingdom, 239
and burials at Althiburos, 466 Sabha (Fazzan), 90–90, 98
monuments, 303 Sabratha (Libya), 166, 206, 413, 414, 415t,
rivers, as water-sources, 12, 13 419, 428
roads craft production, 421–22
Roman, 423 sacrifice, human, 400, 492
Tripolitania, ‘head of the rock road’, 188 Sáez Martín, Bernardo, 481

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 736 Index

Sahara see Northern Sahara; Southern Sahara; Sidi Driss (Morocco), 487–88
Western Egyptian Desert; Western Sidi Slimane (Morocco), 492, 493
Sahara Sidi Uqba (Algeria), 213, 214
Saharan towns, defined, 344 Sijilmasa (Morocco), 31, 35, 241, 242t, 615–17,
saints, Islamic, 674, 675 707–8t
Salama, P., 310 earlier Tafilalt settlements, 609–15, 610,
Sallust 613, 614
on Capsa, 208 founding of, 256, 595–98
on Numidian towns, 407–8 and kharijism, 598–99, 609
salt, 171, 672 and Midrar, 596–97, 602–4, 606
Awlil, 550, 551 mining, 700
and Sijilmasa, 604–9, 606, 608, 700 and salt, 604–9, 606, 608, 700
Southern Sahara and Samgu, 597, 598–602
Azawagh region, 315 Silla, 542, 550–51
Borku, 301 Sinawin (Tripolitania), 191, 199
Kawar, 303 Sincu Bara (Middle Senegal Valley), 548–49,
Seguedine, 304 549, 550, 554
Taghaza/Tawdenni, 316–17 Siwa (Ammon) (Western Egypt), 125, 137,
Ubari, 89 157–58, 174n163
Samgu (Abu al-Qasim Samgu ibn Wasul), 597, Siwré (Middle Senegal Valley), 544, 547–48
598–602 Skalník, P., 681–82
Samnu (Fazzan), 90 slavery and slave-trading, 26, 171–72, 278
Sanam (Middle Nile), 368 and Bu Nijim, 191
Sanhadja, 707–8t and Garamantian society, 678–81
Saniat Jibril (Wadi al-Ajal), 71, 73, 75c, 101 and irrigation, 332
Sanusiyya, 674–76, 683 and Kawar, 299, 322
saqia (waterwheels), 367–68 and Lake Chad region, 586
Sarnah see Gao Saney Meroitic kingdom, 371–72, 388
Scarin, E., 84, 129 and Samgu, 600
Scipio Aemilianus, 400 and zara’ib, 676–77
secondary states, 443–44, 705 slaving states, 703–5
sedentarisation, 5–7, 331–33 Smadja, Elisabeth, 446
Garamantian, 336–38 Smith, A.T., 359
and oasis formation, 333–34 social change, and evolutionary theory, 439,
and pastoralism, 331–33, 334–35, 341, 717 441–42
and settlements, 334–36 social inequality, 28, 442–43, 444, 470, 583,
Sedrata (southern Algeria), 221–22, 222, 660, 668, 696, 702
707–8t Garamantian society, 678–81
Seftimi (Tunisia), 209 Lake Chad, 584
Seguedine (Southern Sahara), 304 and Moroccan monumental tombs, 491,
Sempronius Gracchus, G., 408 493, 494
Senegal (river), 12 socio-cultural change, 442, 448–50
see also Middle Senegal Sognane, M., 552
Sesibi (temple-town), 366, 367 Songhay, 268–69
settlements, religious, 674–77 and Gao, 541, 642
Seyyid Ouinquil (Tichitt), 279–97t, 502–3 sorghum, Middle Senegal Valley, 544,
Shanks, M., 441n9 544–45, 553
sheep, 406, 459 Souag, L., 268–69
shell Southern Sahara, 277, 279–97t, 322–23
marine, 548 Akjoujt, 321
see also dye production Atlantic coast, 321–22
Sicca (Numidia), 407 Dhar Tichitt/Nema/Walata/Tagant,
sickles, iron, 478 279–97t, 318–18

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 Index 737

eastern, 298–308, 299, 304, 305 Stone, A.C., 534

Middle Niger oases, 317 stone (building material), 22–23, 658–60
Taghaza and Tawdenni, 316–17 Garamantes, 72–73, 86–86, 87, 93, 94–95, 96
trading centres in the south-west, oases of al-Jufra, 132
318–21, 317 Sukna, 137
Tuareg in the Central and Southern Sahara, see also Roman period
308–16, 309 Tagrifet, 130
see also architecture Zala, 128–29
Souville, Georges, 482 see also Roman period, architecture/
Spaulding, J., 372 buildings
Speculum see Chebika storehouses, Morocco, 673
Spencer, Herbert, 446 Strabo
springs, 13e, 14, 339–40 on Numidia, 476–77
Algeria on the Pharusii, 218
Biskra, 214–15 Sudani identity, 361–62
Saharan Atlas, 218 The Suf (Algeria), 15, 220
Tuggurt/Wadi Rhir, 219 Sufetula (Sbeïtla) (Tunisia), 415t, 427, 428, 431
Wargla, 221 sufrism, 598–99
Fazzan, 55 Sukna (al-Jufra group) (Libya), 132, 137
Wadi ash-Shati, 84 Sullecthum (Tunisia), 422, 430
Wadi Tanzzuft, 92 fortifications, 430
Northern Sahara, Al-Qurayyat al-Gharbia, 192 sultans, 681
Southern Sahara, 277 sword, bronze, 481–82, 482
Tripolitania Syene/Aswan (Egypt), 168–69
coastal oases, 206 symbols of state, 594
Darj, 198–99 Syrtica (Libya), 127, 137–38, 206
Ghadamis, 196, 341
Tripolitanian Jabal, 199
Tabalbala (Algeria), 268–69
Tunisia
Tacape/Tacapae oasis (Tunisia), 17, 22, 191,
Aquae Tacapitanae, 208
207, 224, 347, 707–8t
the Jarid, 209
Tacfarinan revolt, 201, 712
Kriz, 210
Nafta/Tozeur, 211 Tacitus, on the Garamantes, 684, 713
Nefzaoua oases, 201, 206 Tadeliza (Southern Sahara), 279–97t, 314
Telmine, 202–3, 203 Tadmakka/Essouk (Mali), 279–97t, 316, 322,
Western Egypt, 116, 150 633–41, 637, 638, 639, 640, 648–49,
‘stasis settlements’, 25 650–52, 707–8t
state formation, 27, 440, 695–97 architecture, 658, 659, 661
Althiburos, 469–71 gold, 555
definition, 27–28, 438–39, 499–500, 594–95 trade, 349
examples of states, 28–30 Tadrart Akakus (Fazzan), 68–69t, 96–97
history of research, 444–47 Tafilalat/Tafilalt (Erfoud) (Morocco), 248t,
Lake Chad, 583–88 254–56, 606
and social hierarchy, 680–81 irrigation, 13c
theory, 441–44 see also Sijilmasa
types of, 702, 707–8t Tagant (Mauritania), 318–18, 501, 515
see also Garamantes; Morocco site T150, 503, 504
stelae Tagdaoust (Awdaghust) (Mauritania),
Carthage, 400 279–97t, 318, 317, 550, 623–29, 627,
Gao, 643 628, 648, 650–52, 707–8t
Meroitic kingdom architecture, 658–61
Jebel Qeili, 382, 383 pottery, 548, 549, 554
Queen Amanirenas, 379 Taghaza (Western Sahara, Mali), 316–17
Western Sahara, 254 Tagrifet (Libya), 129–30

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 738 Index

Tahabort (Southern Sahara), 279–97t, 311 Trimithis, 155, 164

Takadda (Niger), 651, 707–8t Tutu at Kellis, 160
Takene Bawat (Southern Sahara), 279–97t, see also mosques
312 Termit (Southern Sahara), 279–97t, 307
Takezawa, S., 505 territorial states, 27–28, 681–85, 704–5
Takrur (Middle Senegal), 525, 541–53, 543, see also Garamantes
545, 549 textile production, 700
Tamdult (Morocco), 264 Meroe, 374, 385–86
mining, 700 Middle Senegal Valley, 549
Tamerza (ad Turres) (Tunisian-Algerian Sabratha, 422
border), 211 southern Algeria, 217n143, 220, 225
Tamgrout (Morocco), 674 Timgad, 425
Tanit (goddess), 400 see also dye production
Tanzzuft see Wadi Tanzzuft textiles, Middle Senegal Valley, 549
Taouz (Jabal Afilal) (Morocco), 256–57, 613, Thabudeos (Algeria), 214
614, 615 Thala (Tunisia), 407
Taqallit (Fazzan), 73, 74–75, 335n21 theatres, Roman, 418, 419
Tarfaya (Morocco), 322 Thelepte (Tunisia), 415t, 425
Tarradell, Miguel, 480, 482, 492 Thiges (Tunisia), 209–10
Tassili n’Ajjer (central Southern Sahara), thrones, daises, 384, 385
279–97t, 308–10 Thuburbo Maius (Tunisia), 415t, 427n77,
Taucheira see Tocra 429–29
Tawdenni (Western Sahara), 316–17 Thugga (Tunisia), 415t, 418, 445, 456n47
taxation, and the state, 594 Thusuros see Tozeur
technology, 334, 340, 341, 460–61 Thysdrus (Tunisia), 418
waterwheels (saqia), 367–68 Tibesti (Southern Sahara), 279–97t, 298, 302–3
Teda (Tubu) people, 278, 298–300, 303, 307–8, Tichitt (Mauritania), 279–97t, 320, 499, 510, 514
322–23 architecture, 500, 502, 504
tells, 147–48, 510–13 Dhar Tichitt, 278, 279–97t, 318–18, 333,
Telmine (Turris Tamalleni) (Tunisia), 202, 565, 581
203, 204–5, 227 Tichitt (people), 320, 333
temple-towns, Egyptian, 365–68, 367 Tiddis (Numidia), 418
temples Tidikelt (Algeria), 266, 265–68, 672
and Christianity, 428 Tifinagh, inscriptions, 24, 97, 636
Cyrenaica Tijika (south-west Sahara), 317, 320
Apollo, 403 Tilley, C., 441n9
Demeter, 403 Timbuktu (Mali), 31, 309, 319, 510, 646, 651,
Zeus 562.180, 403 707–8t
Garmantian, 73 trade, 349, 538
Middle Nile, 376–77 Timgad (Thamugadi) (Algeria), 412, 413, 414,
of Amun at Meroe, 373 415t, 430
of Isis at Meroe, 373 craft production, 421–22, 425
‘Sun temple’ at Meroe, 373 and the legionary base at Lambaesis, 425
Western Butana, 380, 380–82 macellum, 416
Roman, 416 theatre, 418
Carthage, 410 Timmissao (Southern Sahara), 311–12
Timgad, 413 Tin Hinan, tomb of (Abalessa, Ahaggar,
Zama, 445
Algeria), 279–97t, 311, 322
Western Desert, 154–56, 346
Tin Mahham culture, 322
Alexander the Great at Bawiti, 124, 158
Tintan (Atlantic coast), 279–97t, 317, 322
Ammon at Siwa, 125
Amun at Hibis, 153, 154, 156 Tipasa (Algeria), 415t, 430
Seth temple at Mothis, 155, 160 Tireqqa (Mali), 646–47

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 Index 739

Tisavar (Tunisia), 201 and Wargla, 221

Tmissa (Fazzan), 83, 83–84 see also Amazigh groups
Tocra (Taucheira) (Arsinoë) (Cyrenaica), 397, Tubnae (Algeria), 214
404, 415t Tubu people see Teda
Togola, Téréba, 505, 511–13, 643 Tuggurt (Algeria), 218–20
Tombouze (Mali), 279–97t, 318, 506, 514–15, Tuluk (Southern Sahara), 279–97t, 312
646, 649–50 Turris Tamalleni see Telmine
Tongo Maaré Diabel (Mali), 506, 507, 508, 516 Tuwat (Algeria), 266, 248t, 265–68, 672, 674
Tophet (Carthage), 400 French colonialism, 348
Toussaint, P.-M., 210
town planning see planning, town Ubari sand sea lake villages (Fazzan), 89
Tozeur (Thusuros) (Tunisia), 210–11 Umari, al-, 636
trade and exchange, 26, 336, 338, 347–48, Umayyad, 432, 599, 602–3, 617, 673, 683–84
523–24, 717 Umm al-Aranib (Fazzan), 83, 83–84
Cyrenaica, 406 United Nations Environment Programme, 8
Middle Niger/Middle Senegal river valleys, Uqba bin Nafi, 214, 219, 299, 307–8, 647
521, 522, 526, 527, 553–55 urbanism/urbanisation, 24–26, 342, 343–50,
Gao, 537, 541 361, 498–99
Jenné-jeno, 516, 527–28, 530, 531t, sustainability of, 350–51
532, 533 types of towns, 697–702
Takrur, 525, 541–53, 543, 545, 549 see also individual sites
Morocco/Western Mediterranean, 490 Utica, 396, 413, 415t, 422
Roman period in the Western Desert,
169–72 Vaga (Tunisia), 407–8
Southern Sahara, 516 Van der Veen, M., 99
and urban sustainability, 350 Vandals, 426, 430, 431–32
see also market towns/trading centres vegetation, oases, 7–8
Trans-SAHARA Project, xx, xxi–xxii, 4 Vernet, R., 322
treasure houses, 634, 655–56 Vescera (Algeria), 214
Treinen-Claustre, F., 301 vicus/vici see garrison settlements (vici)
tribal structures, 446–47, 469 Villard, François, 490
councils, 670–71 Volubilis (Morocco), 251, 413, 415t, 418
states, 681 Vuillemot, G., 488
and zawaya, 675
Trigger, B., 522 Wad ben Naqa (Middle Nile), 375–77, 376
Trimithis (Amheida) (Western Egypt), 122, Wadan (south-western Sahara), 320
123, 160–69, 162, 699 Waddan (al-Jufra oasis) (Libya), 132, 133, 134,
temple, 155, 346 135, 707–8t
Tripoli see Oea Wadi al-Ajal and Jarma (Fazzan), 54, 58–77,
Tripolitana (ancient province), 188 68–69t, 70, 72, 74, 75, 98, 100–1,
Tripolitania (north-west Libya), 188–206, 103, 334
189–90t, 191, 197, 202, 203, 204 water sources, 339
Tripolitanian Jabal, 199–200 Wadi ash-Shati (Fazzan), 84–89, 85, 86t, 98,
Trousset, P., 200 101, 102t, 103
Tuareg people, 3, 278, 322–23, 708, 709–10 Wadi Draa (Morocco), 240–41, 257–63, 260,
central Sahara, 308–12, 309 261, 269–70, 332n7, 707–8t
and Gourara/Tuwat/Tidikelt area, 267 irrigation methods, 12, 13a-b
Middle Senegal Valley, 547 Wadi Gheris see Wadi Rheris
in Morocco, 478 Wadi Gir/Saoura (Algeria), 13, 240, 248t,
and oasis settlement, 332 252–54
in the southern Sahara, 312–16 Wadi Hikma (Fazzan), 58, 68–69t,
and state formation, 29–30 97, 98, 101

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press

 740 Index

Wadi Jedi (Algeria), 213, 215–16 and the Nile Valley, 148
Wadi Noun (Morocco), 240, 248t, 263–64 pre-Roman settlement, 151–56
Wadi Rheris (Gheris) (Morocco), 12, 13d, 254, Roman period, 147–49, 156–69, 698–99
255, 605, 609–10 collapse of urbanism, 148–49,
Wadi Rhir (Algeria), 218–20 172, 174
Wadi Saoura see Wadi Gir economic links, 169–72, 173–74
Wadi Tanzzuft (Fazzan), 10, 68–69t, 76n47, 91, Western Sahara, 239–40, 241, 248t, 269
92, 714 north-west Algeria/southern Morocco,
Wadi Tifarti (Western Sahara), 317, 322 240–64, 260, 261
Wadi Ziz (Sis) (Morocco), 12–13, 240–41, 254, western Algeria, 265–69, 266
255, 605, 608, 609, 610 Wilson, A.I., 166, 267, 310, 336
wadis, seasonal, 12–14 Wright, E.O., 441
Wagadu see Ghana (kingdom) writing systems, 138
Walaldé (Middle Senegal Valley), 544, 545, 554 Demotic, 154
Walata (south-west Sahara/Mauritania),
318–18, 317, 320, 333, 514–16, Yaqubi, al-
646, 651 on Gao, 536
Ward, P., 129 on Kawar, 299
Warfajjuma people, 599 on Tagdaoust, 624–25
Wargla (southern Algeria), 220–21, 706 on Tamdult, 264
Warjabi b. Rabis, 550 Yaqut al-Hamawi, 306, 634, 641
Warraq, al-, 596, 602–3, 604 Yoffee, N., 348, 702
water sources (oasis), 10, 12–16, 13, 334
see also aqueducts; baths; foggaras; Zala (Libya), 129
fountains; individual sites; irrigation; Zama (Tunisia), 407, 445
rivers; springs; wells Zarai tariff, 225
waterwheels (saqia), 367–68, 387 zara‘ib (temporary camps), 676–77
wells, 13f, 14–15, 341 zawaya (religious settlements),
Algeria 674–77, 709
Tuggurt, 219 zenithal access dolmens, 462
Wargla, 221 Zeus (god), at Cyrene, 403
Cyrenaica, Euesperides, 405 the Ziban (Algeria), 213, 216
Fazzan, 55, 80, 81, 339 Ziegert, Helmut, 56–57
Wadi ash-Shati, 84 Zilum (Lake Chad), 515, 568, 569, 571, 574,
Southern Sahara, Djaba, 304 587–88, 589
Tripolitania, Bu Nijim, 192 Zinkekra (Wadi al-Ajal), 71–72, 100, 334, 335
Western Sahara, Wadi Draa, 258 Ziz (city) (Morocco), 611–12
Western Butana (Meroitic kingdom), see also Wadi Ziz
379–83, 383 Zubo (Lake Chad), 574–75, 578–81, 579,
Western Desert (Egypt), 112, 113, 115, 120t, 587–88, 589
137–38, 346, 705 Zuwila (Fazzan), 68–69t, 82, 82–83, 100–1, 348,
geography, 149, 150, 151 700, 704, 707–8t

https://doi.org/10.1017/9781108637978.021 Published online by Cambridge University Press