The Archaeology of Afghanistan: From Earliest Times to the Timurid Period [2 ed.] 0748699171, 9780748699179

Afghanistan is at the cultural crossroads of Asia, where the great civilisations of Mesopotamia and Iran, South Asia and

531 53 119MB

English Pages 752 Year 2019

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Dedication
Contents
Analytical Table of Contents
List of Figures and Tables
Notes on Contributors
Acknowledgements
Note on Transliteration
Foreword • Mohammad Fahim Rahimi
Preface • Norman Hammond
Introduction to the First Edition • Norman Hammond and Raymond Allchin
Introduction to the New Edition • Warwick Ball
1 The Geographical Background • Sophia R. Bowlby and Kevin H. White
2 The Palaeolithic • Richard S. Davis
3 The Development of the Oxus Civilisation North of the Hindu Kush • Henri-Paul Francfort, Bertille Lyonnet, Cameron A. Petrie and Jim G. Shaffe
4 The Development of a ‘Helmand Civilisation’ South of the Hindu Kush • Cameron A. Petrie and Jim G. Shaffer
5 The Iron Age, Achaemenid and Hellenistic Periods • Warwick Ball, Simon Glenn, Bertille Lyonnet, David W. Mac Dowall and Maurizio Taddei
6 From the Kushans to the Shahis • Warwick Ball, Olivier Bordeaux, David W. Mac Dowall, Nicholas Sims-Williams and Maurizio Taddei
7 From the Rise of Islam to the Mongol Invasion • Warwick Ball and Klaus Fischer
8 From the Mongols to the Mughals • Warwick Ball and Klaus Fischer
9 Conclusion • Raymond Allchin and Norman Hammond
Notes
Bibliography
Copyright Acknowledgements
Index
Recommend Papers

The Archaeology of Afghanistan: From Earliest Times to the Timurid Period [2 ed.]
 0748699171, 9780748699179

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

THE ARCHAEOLOGY OF AFGHANISTAN From Earliest Times to the Timurid Period

Original edition by Raymond Allchin and Norman Hammond

Revised and updated by WARWICK BALL with NORMAN HAMMOND

THE ARCHAEOLOGY OF AFGHANISTAN

To the memory of our fellow-authors Raymond Allchin (1923–2010) Klaus Fischer (1919–1993) Maurizio Taddei (1936–2000)

THE ARCHAEOLOGY OF AFGHANISTAN FROM EARLIEST TIMES TO THE TIMURID PERIOD NEW EDITION

ORIGINAL EDITION EDITED BY RAYMOND ALLCHIN AND NORMAN HAMMOND

REVISED AND UPDATED EDITION EDITED BY WARWICK BALL WITH NORMAN HAMMOND

Edinburgh University Press is one of the leading university presses in the UK. We publish academic books and journals in our selected subject areas across the humanities and social sciences, combining cutting-edge scholarship with high editorial and production values to produce academic works of lasting importance. For more information visit our website: edinburghuniversitypress.com Original edition published 1978, edited by Raymond Allchin and Norman Hammond, © Academic Press Inc. (London) Ltd © editorial matter and organisation Warwick Ball and Norman Hammond, 2019 © the chapters their several authors, 2019 Edinburgh University Press Ltd The Tun – Holyrood Road 12 (2f) Jackson’s Entry Edinburgh EH8 8PJ Typeset in Trump Mediaeval by Servis Filmsetting Ltd, Stockport, Cheshire, and printed and bound in the EU by Fine Tone Ltd by arrangement with Associated Agencies Ltd, Oxford A CIP record for this book is available from the British Library ISBN 978 0 7486 9917 9 (hardback) ISBN 978 1 4744 5047 8 (webready PDF) ISBN 978 1 4744 5046 1 (epub) The rights of Raymond Allchin, Norman Hammond and Warwick Ball to be identified as Editor of this work have been asserted in accordance with the Copyright, Designs and Patent Act 1988, and the Copyright and Related Rights Regulations 2003 (SI No. 2498). Published with the support of the University of Edinburgh Scholarly Publishing Initiatives Fund.

Contents

Analytical Table of Contents vii List of Figures and Tables xvi Notes on Contributors xxx Acknowledgements xxxiii Note on Transliteration xxxv Foreword xxxvii Mohammad Fahim Rahimi Preface xlix Norman Hammond Introduction to the First Edition 1 Norman Hammond and Raymond Allchin Introduction to the New Edition 10 Warwick Ball CHAPTER 1 The Geographical Background 15 Sophia R. Bowlby and Kevin H. White CHAPTER 2 The Palaeolithic 61 Richard S. Davis CHAPTER 3 The Development of the Oxus Civilisation North of the Hindu Kush 99 Henri-Paul Francfort, Bertille Lyonnet, Cameron A. Petrie and Jim G. Shaffer CHAPTER 4 The Development of a ‘Helmand Civilisation’ South of the Hindu Kush 161 Cameron A. Petrie and Jim G. Shaffer

vi CONTENTS CHAPTER 5 The Iron Age, Achaemenid and Hellenistic Periods 260 Warwick Ball, Simon Glenn, Bertille Lyonnet, David W. Mac Dowall and Maurizio Taddei CHAPTER 6 From the Kushans to the Shahis 344 Warwick Ball, Olivier Bordeaux, David W. Mac Dowall, Nicholas Sims-Williams and Maurizio Taddei CHAPTER 7 From the Rise of Islam to the Mongol Invasion 460 Warwick Ball and Klaus Fischer CHAPTER 8 From the Mongols to the Mughals 546 Warwick Ball and Klaus Fischer CHAPTER 9 Conclusion 609 Raymond Allchin and Norman Hammond Notes 620 Bibliography 651 Copyright Acknowledgements 695 Index 699

Analytical Table of Contents

List of Figures and Tables xvi Notes on Contributors xxx Acknowledgements xxxiii Note on Transliteration xxxv Foreword xxxvii Mohammad Fahim Rahimi Preface xlix Norman Hammond Introduction to the First Edition 1 Norman Hammond and Raymond Allchin Introduction to the New Edition 10 Warwick Ball CHAPTER 1  The Geographical Background 15 Sophia R. Bowlby and Kevin H. White Climate and climatic change 18 Natural resources for subsistence 24 The high mountains 26 The mountains and foothills 31 The plains and lowlands 33 The Amu Darya, Oxus and Helmand-Sistan regions 34 The deserts 36 Patterns of subsistence in the 1950s–1960s 37 Agriculture 38 Nomads 50 Opportunities for trade 56 Summary 60

viii

ANALYTICAL TABLE OF CONTENTS

CHAPTER 2  The Palaeolithic 61 Richard S. Davis Problem areas 62 Brief history of palaeolithic research in Afghanistan 63 Initial occupation of Afghanistan – the Lower Palaeolithic 66 The Middle Palaeolithic 67 Darra-i Kur 67 Balkh Province 70 Hazar Sum 70 Kara Kamar 71 Discussion 72 The Late Palaeolithic 74 Kara Kamar Level III 75 The Epi-Palaeolithic 82 Summary 96 CHAPTER 3  The Development of the Oxus Civilisation North of the Hindu Kush 99 Henri-Paul Francfort, Bertille Lyonnet, Cameron A. Petrie and Jim G. Shaffer Introduction 99 The adoption of domesticates: the Neolithic and Chalcolithic periods 101 Early mobile pastoralists in northern Afghanistan 109 Early sedentary agriculturalists in northern Afghanistan 110 Developments in neighbouring regions and connections to northern Afghanistan 111 Central Asia 111 Eastern Iran 113 Pakistan 114 Development of stratified societies 116 The Eastern Bactria survey 117 The Late Chalcolithic/Early Bronze Age (c. 3500 to 2500 bc) 119 The end of the Early Bronze Age and Middle Bronze Age (c. 2500 to 1500 bc) 122 Shortughaï 125 Stratigraphy and architecture 127 Surface topography and magnetometry survey 127 Pottery typology and comparisons 132 Petrographic analyses of pottery 133 Small finds 133 Archaeobotany 135 Archaeozoology 137 Metals 137

analytical table of contents

Physical anthropology 138 Chronology 138 Ecology 139 Economy and society at Shortughaï 140 Ghar-i Mar 140 Darra-i Kur 141 Dashli sites 143 Architecture 143 Dashli 1 143 Dashli 3 144 Ceramics 149 Other artefacts 151 Other sites in northern Afghanistan 151 Conclusions about Bactria, Central Asia and the ‘Oxus Civilisation’ 157 CHAPTER 4  The Development of a ‘Helmand Civilisation’ South of the Hindu Kush 161 Cameron A. Petrie and Jim G. Shaffer Introduction 161 Geographical factors 162 Mundigak 163 Stratigraphy and architecture 166 Period I 166 Period II 168 Period III 170 Period IV 173 Period V 187 Chronology 189 Ceramics 192 Period III handmade pottery 197 Period III turned pottery, undecorated 197 Period III turned pottery, decorated 198 Period III turned pottery, intrusive or special function 199 Period IV handmade pottery 203 Period IV turned pottery, undecorated 204 Period IV turned pottery, decorated 206 Period IV turned pottery, intrusive or special function 209 Lithic artefacts 216 Bone artefacts 217 Metal artefacts 218 Miscellaneous artefacts 221 Spindle whorls 221 Stone vessels 221 Beads and pendants 222

ix

x

ANALYTICAL TABLE OF CONTENTS

Seals 222 Ceramic figurines 223 Faunal and floral remains 224 Said Qala Tepe 224 Stratigraphy and architecture 225 Chronology 228 Ceramics 230 Lithic artefacts 234 Bone artefacts 235 Metal artefacts 235 Small miscellaneous artefacts 235 Spindle whorls 235 Stone vessels 235 Beads and pendants 236 Seals 237 Ceramic figurines 237 Faunal and floral remains 237 Deh Morasi Ghundai 237 Stratigraphy and architecture 239 Chronology 239 Ceramics 239 Other artefacts 241 Miscellaneous artefacts 241 Spindle whorls 241 Ceramic figurines 241 Faunal and floral remains 242 Kandahar 242 Nad-i Ali 243 Helmand-Sistan surveys 245 External relationships with southern Afghanistan 246 Mundigak: Periods I–II 248 Mundigak: Period III 251 Mundigak: Period IV 256 Mundigak: Period V 259 CHAPTER 5  The Iron Age, Achaemenid and Hellenistic Periods 260 Warwick Ball, Simon Glenn, Bertille Lyonnet, David W. Mac Dowall and Maurizio Taddei Historical background 260 The question of Iron Age states 260 The Achaemenid empire 261 Alexander III of Macedon and his successors 263 The Graeco-Bactrians 264 The Parthians 265

analytical table of contents

Yuezhi and Saka invaders 265 The Indo-Parthians 267 Settlement, material culture, architecture and art 267 The Iron Age and Achaemenid periods 267 Bactria 267 The Helmand-Sistan region 276 The Herat region 278 Arachosia 279 The Kabul Valley and Gandhara 283 The Hellenistic period 285 Bactria 285 Begram 287 Kandahar and southern Afghanistan 288 Herat 291 The Greek city of Aï Khanoum 291 Sites outside Afghanistan 309 The nomad ‘invasions’ 311 Epigraphy 315 The Achaemenids 315 Elamite 317 Aramaic 317 Inscriptions of the Mauryan period 318 The Aramaic inscriptions from Laghman 318 The bilingual Greek and Aramaic inscription from Kandahar 320 The Ashokan inscription in Greek from Kandahar 320 The Indo-Aramaic inscription from Kandahar 321 The Sophytos inscription from Kandahar 322 Historical significance 323 Seleucid and Graeco-Bactrian inscriptions and epigraphy 324 The Aristonax inscription from Kandahar 324 Inscriptions from Aï Khanoum 324 The inscriptions of Klearchos 324 The Dedications in the Gymnasium 325 Finds from the necropolis 325 Economic and other documents 325 The Aramaic ostracon 326 Miscellaneous Greek documents from northern Afghanistan 326 Kuliab 326 Early Kharoshthi inscriptions 326 The inscription of Tiravharna the Satrap 327 The Bimaran vase 328 Numismatics 328 Evidence for currency and circulation 328 The Achaemenids 328

xi

xii

ANALYTICAL TABLE OF CONTENTS

Role of the siglos 328 The Oxus Treasure 329 The Miho Treasure 329 The al-Sabah collection 329 The 1966 Balkh hoard 330 The Chaman-i Hauzuri hoard 330 Other hoards of bent bar coins 330 Changes due to Alexander 331 Alexander’s currency reform 331 Later elements of the Oxus Treasure 331 After Alexander 332 The 1990 ‘Afghanistan’ hoard 332 Mauryan and Graeco-Bactrian coinage 333 The overall pattern 333 Mir Zakah I 335 Mir Zakah II 336 Finds from Begram 337 Excavated coins from Aï Khanoum 338 The Kuliab hoard 340 The Qunduz hoard 340 The currency of western Afghanistan 341 Yuezhi, Sakas and Indo-Parthians 342 The Yuezhi currency of Bactria 342 Coins of the Azes Dynasty and Su Hermaeus 342 The distribution of Indo-Parthian issues 343 CHAPTER 6  From the Kushans to the Shahis 344 Warwick Ball, Olivier Bordeaux, David W. Mac Dowall, Nicholas Sims-Williams and Maurizio Taddei Historical background 344 The Kushans 344 The Later Kushans, Sasanians and Kushano-Sasanians 345 Hunnic groups 349 The Western Turks 350 The progress of Islam 351 Turki and Hindu Shahis 351 Settlement, material culture, architecture and art 352 Begram 353 Kandahar 361 Wardak 365 Jaghatu 365 Kharwar 368 The Kabul sites 369 Mes Aynak 375

analytical table of contents

Surkh Kotal 376 Rabatak 379 Rag-i Bibi 381 Balkh 384 Zadiyan and the Balkh Oasis wall 386 Buddhist art and architecture 390 Cave monasteries 390 Stupa-monastery complexes 402 Sculpture 403 Classical connections and chronology 414 Pictorial art 419 Late Buddhist sculpture 423 Non-Buddhist art 424 Everyday life 432 Problems of dating 434 Epigraphy 436 Bactrian inscriptions and texts 436 The era of Kanishka 444 Inscriptions connected with the Kanishka era 444 Undated Kharoshthi inscriptions 445 The Jaghatu inscriptions 446 The inscriptions of Uruzgan 446 Indian inscriptions in Sharada script 446 Numismatics 447 The Kushan coinages 447 The Kushan monetary system 447 Foreign influence 452 Coins in Buddhist stupa deposits 453 Other hoards and site finds 453 The Late Kushan coinages 454 Sasanian influences 456 The Kushano-Sasanian coinages 456 The Tepe Maranjan hoard and other Sasanian finds 456 Coinage of the Huns 456 The coins of the Nezak 457 The issues of Vrahitigin 457 Gadhaiya paisa 458 Problems with the Shahi coinage 458 The pattern of the coinage 458 Silver coins of Spalapati Deva 458 Silver coins of Samanta Deva 458 The Shahi copper denominations 459 The billon currency 459 Hoards and site finds 459

xiii

xiv

ANALYTICAL TABLE OF CONTENTS

CHAPTER 7  From the Rise of Islam to the Mongol Invasion 460 Warwick Ball and Klaus Fischer Historical background 460 The Tahirids, Saffarids and Samanids 460 The Ghaznavids 462 The Ghurids 463 The Khwarazm-shahs 464 Architecture of the Early Islamic period 465 Background 465 The earliest period 465 Ghaznavid buildings 469 Lashkari Bazar 469 Bust 474 Ghazni 475 Baba Hatim Ziyarat 478 Charkh-i Logar 479 Daulatabad 484 Minor monuments 487 Sistan 487 Outside Afghanistan 506 Summary of Ghaznavid art and architecture 506 Palaces 509 Gardens 509 Mosques 509 Minarets 509 Decoration 509 Painting 510 Sculpture 511 Ceramics 511 Ghurid buildings 511 The minaret of Jam 513 The minaret of Sakhar 517 Shahr-i Arman and Naraiman 518 Ghur fortifications 519 Bamiyan fortifications 525 Chisht 527 Larwand 527 Herat 532 Shah-i Mashhad 534 Danestama 536 Ghazni 536 Bust 542

analytical table of contents

CHAPTER 8  From the Mongols to the Mughals 546 Warwick Ball and Klaus Fischer Historical background 546 Sistan 549 Herat 558 City plan and citadel 558 The ‘musalla’ complex 559 The Friday Mosque 565 Mosque of Hauz-i Karboz 568 Mausoleum of Shahzadeh Abdullah 572 Gazurgah 572 Herat region 578 Azadan 580 Ziyaratgah 580 Kuhsan 585 Khaniqah of Sard al-Din Armani 585 Kush Rabat 585 Zindajan 585 Ghazni 585 Mausoleum of Shah Shahid 585 Mazar-i Sharif 589 Shrine of Hazrat Ali 589 Balkh 591 Timurid works of art 592 Herat, bronze cauldron in the courtyard of the Great Friday Mosque 592 Metal ewer, British Museum 593 Summary of Timurid art 597 Architecture 597 Sculpture 602 Painting 602 Minor arts 603 Post-Timurid art and architecture 605 CHAPTER 9  Conclusion 609 Raymond Allchin and Norman Hammond The Palaeolithic 610 The appearance of sedentary settlements 611 The beginning of history in Afghanistan 613 The arrival of Islam 617 Notes 620 Bibliography 651 Copyright Acknowledgements 695 Index 699

xv

Figures and Tables

Figures 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 1.10 1.11 1.12 1.13 1.14 1.15 1.16 1.17 1.18 1.19 1.20 1.21 1.22 1.23 1.24 1.25 1.26 1.27

The location of Afghanistan within Eurasia 16 Afghanistan: relief and drainage, showing major rivers 17 Average precipitation and cultivated land 19 Average temperature by season 20 Climatic regions 21 The physiographic regions of Afghanistan 24 Natural vegetation 26 The Pamir Mountains in Wakhan 27 The north side of the Salang Pass 28 The Panjshir Valley 29 The Farah Rud, with the discolouration in the hills caused by iron ore deposits 30 The Ghorband Valley 30 The Kabul River Gorge 31 The southern foothills near Ghazni 32 The Balkh River in the mountains south of Balkh 33 The Herat-Farah lowlands between Adraskand and Farah Rud 34 The Balkh Plain near Balkh 35 The Amu Darya floodplain from Kampyr Tepe in Uzbekistan 35 The Helmand River 36 Giant sand-dunes of the Registan Desert on the edge of the Helmand Valley 37 Principal types of agricultural irrigation 39 Cross-section of a qanat system 40 Aerial view of a typical qanat system in southern Iran 40 Irrigation canal or jui in the Balkh Plain 41 Grape field in the Kandahar area 42 Traditional ploughing in the Bamiyan region 43 Winnowing in the Kandahar region 44

figures and tables

1.28 1.29 1.30 1.31 1.32 1.33 1.34 1.35 1.36 1.37 1.38 1.39 1.40 1.41 1.42 1.43 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 2.11 2.12 2.13 2.14 2.15 2.16 2.17 2.18 2.19 2.20 2.21 2.22 3.1 3.2

Abandoned cave dwellings in Ghur Province 45 Vaulted mud-brick domestic roofing near Kandahar 46 Flat timber and mud roofing at Istalif 47 A traditional qal’a in the Bamiyan Valley 47 Donkey transport in the Kandahar region 48 A camel caravan passing in front of the Iron Age citadel at Old Kandahar 49 Seasonal movement of nomads 51 Nomad market on the Wazmin Pass in the western mountains 52 Pashtun peaked black tent between Delaram and Gulistan 52 Pashtun vaulted black tent 53 Firuzkuhi yurts in three stages of construction in western Afghanistan 54 Firuzkuhi yurt in use 55 Black tents of the Taimani in western Afghanistan 55 Raw lapis lazuli in a trader’s shop in Kabul 57 Lapis bust of a Parthian prince in the Iran Bastan Museum 58 Lapis-bearing rock at the main lapis lazuli mines at Sar-i Sang in Badakhshan 59 Palaeolithic sites in Afghanistan 62 Satellite image of the Dash-i Nawur 65 Petroglyphs near Sarhad in Badakhshan 66 Darra-i Kur, Middle Palaeolithic 68 Darra-i Kur, Middle Palaeolithic 69 Kara Kamar 71 Kara Kamar, Level II, Late Palaeolithic 74 Kara Kamar, Level III, Upper Palaeolithic 77 Kara Kamar, Level III, Upper Palaeolithic 78 Kara Kamar, Level III, Upper Palaeolithic 81 Plan of Aq Kupruk showing the location of the archaeological sites 83 View of Aq Kupruk from the west 83 The Shelter of Aq Kupruk II 84 Flint nodules in situ near Aq Kupruk (Dar-i Archa) 85 Aq Kupruk, Epi-Palaeolithic 86 Aq Kupruk, microlithic component, Epi-Palaeolithic 87 Aq Kupruk, Epi-Palaeolithic 89 Aq Kupruk, Epi-Palaeolithic 91 Aq Kupruk, Epi-Palaeolithic 92 Plan of Aibak vicinity 93 Kok Jar Epi-Palaeolithic surface site 93 Kara Kamar, Level I, Epi-Palaeolithic 94 The Bronze Age in Afghanistan and surrounding regions 100 Neolithic material from northern Afghanistan 108

xvii

xviii

FIGURES AND TABLES

3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18 3.19 3.20 3.21 3.22 3.23 3.24 3.25 3.26 3.27 3.28 3.29 3.30 3.31 3.32 3.33 3.34 3.35 3.36 3.37 3.38 3.39 3.40 3.41 4.1

The Eastern Bactria survey 118 Abandoned canal, east Bactria 119 Example of a river diversion 120 Chalcolithic sherds from Taluqan 121 Andronovo/Burguljuk/Jaz I related sherds from AÈ Khanoum 124 Plan of Shortughaï 126 Shortughaï Period I–II 128 Shortughaï Period I–II 129 Shortughaï Period I jar in situ 130 Shortughaï wall Period II and intrusive Period Ib burial 130 Shortughaï Period III 131 Shortughaï Period IV burial 131 Shortughaï Period I Harappan sherd 132 Shortughaï Period II pedestal dish 132 Shortughaï ‘Andronovo’ sherds 132 Shortughaï Period I–II etched carnelian beads 134 Shortughaï Period I long carnelian bead 134 Shortughaï Period III bronze pin 135 Shortughaï Harappan seal 135 Map showing a reconstruction of the Bronze Age environment of the Shortughaï Plain 136 Shortughaï Period I–II crucible fragments 137 Shortughaï: raw lapis lazuli 138 Map of the Dashli and associated sites in the north-west of Balkh 143 Dashli 1 fort 144 Dashli 3 ‘temple’ 145 Dashli 3 ‘palace’ 146 Dashli 3 ‘sceptre’ in the National Museum 147 Dashli 3 fragments of alabaster mosaic in the National Museum 148 Dashli ceramics in the National Museum 149 Dashli ceramics in the National Museum 150 Bronze Age sites north of Tashkurghan 152 Fullol Hoard: gold vessels 152 Fullol Hoard: silver vessels 153 Fullol Hoard: silver vessels 153 Fullol Hoard: motifs 154 Bronze Age objects from illicit excavations in Afghanistan displayed on a Kabul street in 1975–6 155 A Bactrian bronze seal in the Louvre 156 A Bactrian decorated chlorite flagon in the Louvre 156 A Bactrian bronze axe-head in the New York Metropolitan Museum 156 Map of the Kandahar area in the Bronze Age 162

figures and tables

4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 4.21 4.22 4.23 4.24 4.25 4.26 4.27 4.28 4.29 4.30 4.31 4.32 4.33 4.34 4.35 4.36 4.37 4.38 4.39 4.40 4.41

Mundigak, general site plan of excavation areas 164 Mundigak, view of mound A in 1956 165 Mundigak, view of mound A in 1966 165 Mundigak A: Period 1, 5 167 Mundigak A: Period II, 2 168 Mundigak A: Period II, 3b 168 Mundigak A: Period III, 2 170 Mundigak A: Period III, 6a 171 Mundigak, general plan of the Period IV structures 174 Mundigak A: plan and elevation of ‘colonnade’ 175 Mundigak A: façade of the Period IV monumental building 176 Mundigak A: façade of the Period IV monumental building 176 Mundigak A: ‘palace’ first reconstruction 177 Mundigak A: ‘palace’ final reconstruction 178 Mundigak B: angle of rampart and contemporary structures IV, 1 180 Mundigak D: bastion IV, 1 181 Mundigak G: plan and section of ‘temple’ 182 Mundigak G: view of the ‘temple’ from Mound A in 1966 183 Mundigak G: view of the ‘temple’ showing triangular ‘buttresses’ 184 Mundigak, conjectural reconstruction of the Period IV ramparts 185 Mundigak A: plan of the ‘massive monument’ 187 Mundigak, pottery from Period I 193 Mundigak, pottery from Period II 194 Mundigak, pottery from Period III 195 Mundigak, decorated pottery from Period III 198 Mundigak, decorated pottery from Period III, showing Quetta ‘solid’ style 200 Mundigak, decorated pottery from Period III, showing Quetta ‘linear’ style 201 Mundigak, pottery from Period III 202 Mundigak, wheelmade and decorated pottery from Period IV1 203 Mundigak, wheelmade and decorated pottery from Period IV1 204 Mundigak, wheelmade and decorated pottery from Period IV1 205 Mundigak, wheelmade and decorated pottery from Period IV1 206 Mundigak, decorated pottery of Quetta style, Period IV1 207 Mundigak, special function and/or intrusive pottery, Period IV2 208 Mundigak, wheelmade and decorated pottery from Period IV2 210 Mundigak, polychrome vessel from Period IV2 211 Mundigak, wheelmade and decorated pottery from Period IV3 212 Mundigak, wheelmade and decorated pottery from Period IV3 213 Mundigak, special function and/or intrusive style pottery from Period IV3 214 Mundigak, decorated pottery from Period V 215

xix

xx

FIGURES AND TABLES

4.42 4.43 4.44 4.45 4.46 4.47 4.48 4.49 4.50 4.51 4.52 4.53 4.54 4.55 4.56 4.57 4.58 4.59 4.60 4.61 4.62 4.63 4.64 4.65 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12

Mundigak, miscellaneous objects 216 Mundigak, stone bowls; copper/bronze objects 219 Mundigak, metal objects 220 Mundigak, stone vessels in the National Museum 221 Mundigak, seals, spindle whorls and Period IV ceramics in the National Museum 223 Said Qala, sondage profiles 226 Said Qala, example of Period II architecture 227 Said Qala: example of architecture, Period III; oven from above structure 229 Said Qala, intrusive or special function sherds 231 Said Qala, intrusive or special function sherds 232 Said Qala, intrusive or special function pottery 233 Said Qala, examples of stone hoes 234 Said Qala, metal objects 236 Said Qala, terracotta figurines 238 Deh Morasi Ghundai: shrine complex, Period Ila; used magnetite nodule, Period Ila; female figurines, Period IIa 240 Mundigak IV bowls from Kandahar and Mundigak 242 Nad-i Ali, Sorkh Dagh mound 244 Characteristic artefact styles of the Quetta valley 249 Characteristic artefacts at Damb Sadaat II 250 Namazga pottery from Geoksyur 252 Comparison of decorated pottery of Namazga III and Mundigak III 253 Comparison of decorated pottery, seals and figurines of Damb Sadaat II and Namazga III 254 Pottery of Damb Sadaat II 256 ‘Zhob mother goddess’ figurines 256 Map of the eastern Achaemenid Empire showing satrapies 262 Map of Hellenistic Afghanistan and adjacent regions, with the route of Alexander 264 Plan of Kohna Qal’a, the Iron Age ‘city’ adjacent to Aï Khanoum 269 The circular ramparts of Iron Age Kohna Qal’a 270 Satellite image of Qunduz Bala Hisar 271 Plan of Altin 10, Palace 1 272 Plan of Altin 10, Palace 2 273 Plan of Altin Dilyar Tepe 274 General view of the site of Chashma-i Shafa 275 The excavated fire temple and altar at Chashma-i Shafa 276 Painted Iron Age ceramics from Qala 350A in the Sar-o-Tar basin of Sistan 278 View of Kandahar from the ridge overlooking the site 279

figures and tables

5.13 5.14 5.15 5.16 5.17 5.18 5.19 5.20 5.21 5.22 5.23 5.24 5.25 5.26 5.27

5.28 5.29 5.30 5.31 5.32 5.33 5.34 5.35 5.36 5.37 5.38 5.39 5.40 5.41 5.42 5.43 5.44 5.45 5.46

Map of the site of Kandahar showing excavated areas 280 Section through the east rampart at Kandahar 281 The cut through the rampart at Kandahar 281 Plan of the Site H building at Kandahar 282 Achaemenid ceramics from Site D at Kandahar 284 View of the excavations at Tepe Zargaran, Balkh, with fragmentary Classical column remains 286 Plan of the circular Hellenistic town of Emshi Tepe 287 Plan of part of the excavated Hellenistic temple of Jiga Tepe 288 The lower town and ramparts of Sikandarabad Shahr-i Kohna with the citadel in the foreground 290 Plan of Aï Khanoum 292 Aï Khanoum, view from the citadel 294 Aï Khanoum, photo of the site revealing extensive damage caused by robbing 294 Aï Khanoum, reconstructed view towards the south-west from the theatre 295 Aï Khanoum, reconstructed view to the south-east: the propylaeum and courtyard of the palace 295 Aï Khanoum, reconstructed view to the west: the main street, the temenos of the temple with indented niches and the courtyard of the palace 296 Aï Khanoum, plan of the administrative complex 297 Aï Khanoum, colonnade of the palace 298 Aï Khanoum, Corinthian capital from the south portico of the palace courtyard 299 Capitals from Aï Khanoum re-used in a nearby tea-house 300 Aï Khanoum, mosaic floor in the administrative building 301 Aï Khanoum, plan of the temple with indented niches 303 Aï Khanoum, the temple with indented niches 304 Aï Khanoum, a herm of an elderly man from the gymnasium 306 Aï Khanoum, funerary relief from the necropolis 307 Aï Khanoum, silver medallion with Cybele on a chariot 308 Aï Khanoum, bone figurine representing a standing goddess 308 Aï Khanoum, bronze statuette of a beardless Heracles 309 Aï Khanoum, headless female statuette from the sanctuary of the temple with indented niches 310 Tillya Tepe, pendant with seal of Athene 312 Tillya Tepe, tree 313 Tillya Tepe, ‘Bactrian Aphrodite’ 314 Tillya Tepe, clasp 315 Tillya Tepe, dagger and sheath 315 Tillya Tepe, ‘dragon king’ 316

xxi

xxii

FIGURES AND TABLES

5.47 Aramaic document, probably from Balkh, concerning the campaign against Alexander 318 5.48 Aramaic inscription from Darunta 319 5.49 The Graeco-Aramaic bilingual inscription of Aªoka from Kandahar 321 5.50 The Greek building inscription from Kandahar 322 5.51 The Sophytos inscription, purportedly from Kandahar 323 5.52 The Kharoshthi inscription of Tiravharna from the neighbourhood of Jalalabad 327 5.53 Sophytos silver coin of unknown denomination 332 5.54 Gold stater of the Graeco-Bactrian king Diodotus 333 5.55 Silver Attic tetradrachm of the Graeco-Bactrian king Antimachus 334 5.56 Square bronze coin of the Graeco-Bactrian king Antimachus 334 5.57 Silver Attic standard tetradrachm of Eucratides I 334 5.58 Square silver drachm of reduced Indian weight struck by Apollodotus I 335 5.59 Square copper bilingual coin of Pantaleon 335 6.1 Map to illustrate main sites discussed in Chapter 6 344 6.2 Map of sites associated with Begram on the Kuh-i Daman plain north of Kabul 354 6.3 The excavations at Begram 355 6.4 Plan of the site of Begram 356 6.5 Ivory from Begram in the Musée Guimet 357 6.6 Ivory from Begram in the Musée Guimet 357 6.7 Ivory from Begram in the Musée Guimet 358 6.8 Glass from Begram in the Musée Guimet 359 6.9 Glass from Begram in the Musée Guimet 359 6.10 Glass from Begram in the Musée Guimet 360 6.11 Plaster plaques from Begram in the Musée Guimet 361 6.12 The stupa-monastery at Kandahar 362 6.13 Spiral-burnished pottery from Kandahar 363 6.14 Plan of the site of Wardak 366 6.15 Stupa 1 at Wardak 367 6.16 Remains of paintings in the monastery at Wardak 367 6.17 Photo of the site of Kharwar 368 6.18 Plan of the remains at Kharwar 369 6.19 Map of the sites in and around Kabul 370 6.20 The city walls of Kabul 371 6.21 Isometric view of the Tepe Narnenj remains 372 6.22 Photo of the excavated stupa of Qol-e Tut 373 6.23 The stupa at Shiwaki 374 6.24 The Minar-i Chakri 374 6.25 Aerial view of Mes Aynak 375

figures and tables

6.26 6.27 6.28 6.29 6.30 6.31 6.32 6.33 6.34 6.35 6.36 6.37 6.38 6.39 6.40 6.41 6.42 6.43 6.44 6.45 6.46 6.47 6.48 6.49 6.50 6.51 6.52 6.53 6.54 6.55 6.56 6.57 6.58 6.59 6.60 6.61 6.62

A stupa at Mes Aynak 376 The excavations at Surkh Kotal 377 Plan of Surkh Kotal 378 The ‘altar’ at Surkh Kotal 379 One of the merlons from Surkh Kotal in the Musée Guimet 380 Pseudo-classical pilaster and cornice from Surkh Kotal in the Musée Guimet 381 Statue of Kanishka from Surkh Kotal in the National Museum 382 The sanctuary site of Rabatak 383 The Sasanian rock relief of Rag-i Bibi 383 The ramparts of Balkh 385 Tepe Rustam at Balkh 385 Takht-i Rustam at Balkh 386 A digital elevation model of Zadiyan from satellite imagery 387 The citadel of Zadiyan 387 The Balkh oasis wall 388 Approximate alignment of the Balkh oasis wall in relation to other Kushan fortifications in the Balkh oasis 389 Elevation, section and plan of the Balkh oasis wall 390 General view of the main cliffs at Bamiyan with the two colossal Buddhas 391 The 53-metre Buddha at Bamiyan 392 The 38-metre Buddha at Bamiyan 393 The standing Buddha at Kakrak 394 Seated Buddha at Bamiyan 395 Wooden architecture imitated in stucco in one of the caves at Bamiyan 396 Painting from Bamiyan in the Musée Guimet 396 Wall painting in the niche of the 38-metre Buddha 397 Plan of the caves at the foot of the 54-metre Buddha and 38-metre Buddha at Bamiyan 398 Plan of the fortifications at Shahr-i Zohak 399 The fortress of Shahr-i Zohak 400 The fortress of Chehel Burj 400 Detail of the decoration and arrow slits at Chehel Burj 401 The rock-cut stupa at Haibak 401 The stupa at Guldarra 402 Plan of the stupa-monastery at Tepe Sardar 404 The main stupa at Tepe Sardar 404 Tapa Sardar: male head from earlier period, Bodhisattva (?) 405 Tapa Sardar: bearded head from earlier period, Bodhisattva Vajrapani (?) 405 Tepe Sardar: colossal gilded foot from Room 100, Antique Period 2 406

xxiii

xxiv

FIGURES AND TABLES

6.63 Schist Buddha statue from Paitava in the Musée Guimet 407 6.64 Schist relief from Shotorak in the Musée Guimet 408 6.65 Schist reliefs from Chakhil-i Ghundi, Hadda, in the Musée Guimet 408 6.66 Relief from Qunduz in the National Museum 409 6.67 Map of the sites in the Hadda-Jalalabad area 410 6.68 Hadda: plan and front view of Stupa 121 at Tapa Kalan 411 6.69 Plan of Tepe Shotor at Hadda 412 6.70 Aerial view of Tepe Shotor 412 6.71 The ‘Heracles’ relief from Tepe Shotor at Hadda 413 6.72 The ‘fish porch’ from Tepe Shotor at Hadda 414 6.73 Stucco bust from Tepe Kalan at Hadda in the Musée Guimet 415 6.74 Stucco head from Tepe Kalan at Hadda in the Musée Guimet 416 6.75 Composite image of the gold reliquary of Bimaran in the British Museum 417 6.76 Painting from Bagha Gai at Hadda in the Musée Guimet 420 6.77 The Sasanian painting at Ghulbiyan 421 6.78 Interpretive drawing of the Ghulbiyan painting 421 6.79 Dilbarjin Tepe: wall painting in the temple depicting Shiva and Parvati 422 6.80 The so-called ‘royal couple’ from Fondukistan in the Musée Guimet 424 6.81 The bejewelled deity from Fondukistan in the Musée Guimet 425 6.82 The bejewelled Buddha from Fondukistan in the Musée Guimet 426 6.83 Tepe Sardar: row of clay stupas and thrones on the eastern side of the main stupa 427 6.84 Tepe Sardar: detail of the unbaked-clay sculpture in chapel 37 428 6.85 Tepe Sardar: detail of the unbaked-clay Parinirvana Buddha in shrine 63 429 6.86 Tepe Sardar: multiple mould used to obtain decorative plaques for clay stupas, thrones, etc. 429 6.87 Gudul-i-Ahangaran; Ghazni: an inscribed clay tablet from inside of miniature stupa; Buddhist profession of faith 430 6.88 Marble statue of Surya from Khair Khana in the Musée Guimet 431 6.89 The barrow cemetery at Kandahar 434 6.90 The Rabatak Inscription in the National Museum 437 6.91 The well inscription from Surkh Kotal in the National Museum 438 6.92 A polyandric text from the Bactrian documents in the Hirayama Collection 440 6.93 A mention of Shahanshah Peroz in a text from the Bactrian documents in the Nasser D. Khalili Collection of Islamic Art 441 6.94 The oldest mention of ‘Afghans’ in a text from the Bactrian documents in the Hirayama Collection 442

figures and tables

6.95 A translation of the Islamic invocation ‘bismillah’ in a text from the Bactrian documents in the Nasser D. Khalili Collection of Islamic Art 443 6.96 Copper didrachm of Soter Megas, with the rayed head of Mithra and the reverse type of the king mounted on horseback (late first–early second century ad) 448 6.97 Kushan gold dinar of Vima Kadphises, with the bearded and diademed king emerging from rocks or cloud, and Shiva naked and standing on the reverse (early second century ad) 449 6.98 Kushan gold dinar of Kanishka, with the king standing at an altar and the reverse type of Nana (mid-second century ad) 450 6.99 Kushan copper tetradrachm of Kanishka, with the reverse type of Mioro (Mithra) radiate 451 6.100 Kushan gold dinar of Huvishka, with the reverse type of Miiro (late second century ad) 451 6.101 Kushan copper tetradrachm of the reduced standard of Huvishka, showing the king riding an elephant and a figure of Mao 452 6.102 Kushan copper unit of Vasishka, with the king standing at an altar, and Oesho in front of a bull on the reverse (mid-third century ad) 455 6.103 Kushan copper unit of Kanishka II, with the king standing at an altar, and Ardochsho on the reverse (first half of the third century ad) 455 6.104 Kushan copper unit of Shaka, with the king standing at an altar, and Ardochsho on the reverse (early fourth century ad) 455 7.1 Map showing early Islamic sites described in the text 461 7.2 Plan of the Mosque of No Gunbad at Balkh 466 7.3 Balkh, mosque of No Gunbad, view of interior 467 7.4 Balkh, mosque of No Gunbad, detail of stucco decoration 468 7.5 Lashkari Bazar, general plan 470 7.6 Lashkari Bazar, ground plan of the south palace 471 7.7 Lashkari Bazar, forecourt of the central palace 472 7.8 Lashkari Bazar, entrance to the central palace 472 7.9 Lashkari Bazar, south palace, wall decoration 473 7.10 Lashkari Bazar, south palace, squinches 473 7.11 Lashkari Bazar, fragment of wall painting from pillar 474 7.12 Ghazni, palace of Mas‘ud III, ground plan 475 7.13 Ghazni, polylobated arch with inscriptions bearing the name of Mas‘ud III 476 7.14 Ghazni, palace of Mas‘ud III, fragments of marble decoration. a: dado with benedictory inscription. b: panel with octagon pattern 476 7.15 Ghaznavid copper dish. National Museum 477 7.16 Ghaznavid glazed bowl with lustre decoration. National Museum 478 7.17 Ghaznavid copper and gold dish. National Museum 479 7.18 Ghazni, minaret of Mas‘ud III 480 7.19 Ghazni, minaret of Mas‘ud III, detail of the carved brick-work 481

xxv

xxvi

FIGURES AND TABLES

7.20 Ghazni, marble cenotaph on the tomb of Mahmud 482 7.21 Baba Hatim Ziyarat, façade of monument to the memory of Salar Khalil 483 7.22 Charkh-i Logar, wooden mihrab in local mosque 483 7.23 Charkh-i Logar, zone of transition in mihrab 484 7.24 Daulatabad, minaret 485 7.25 Daulatabad, minaret detail 486 7.26 Sar-i Pul, stucco decoration in the Ziyarat-i Imam-i Kalan 488 7.27 Sar-i Pul, stucco inscription in the Ziyarat-i Imam-i Kalan 489 7.28 Darra-i Shakh oratory, stucco decoration 489 7.29 Plan and elevation of the Baba Rushnai mausoleum 490 7.30 Plan and elevation of the Abu Hurayra mausoleum 491 7.31 Plans of the two mosques at Shahr-i Gholghola, Bamiyan 492 7.32 Map of early Islamic sites in Sistan 493 7.33 Plan of Nad-i Ali 494 7.34 The minaret of Khwaja Siah Push 496 7.35 The stellate plans of the minarets at Ghazni, Nad-i Ali, Khwaja Siah Push and the Qutb Minar in Delhi 497 7.36 Gul-i Safid, ground plan of a courtyard house in the centre of the city 498 7.37 Gul-i Safid, decorated wall of ivan in the same courtyard house as Fig. 7.36 499 7.38 Gul-i Safid, the same courtyard house as Fig. 7.36 499 7.39 Detail of the decoration at Gul-i Safid 500 7.40 Gul-i Safid, double-storeyed mud-brick tower to the north of the ruin-field 501 7.41 The citadel at Peshwaran 502 7.42 Peshwaran, façade of the mosque 503 7.43 House façade at Chigini II 504 7.44 Citadel and walls of Shahr-i Gholghola in Sar-o Tar 504 7.45 Plan of Sar-o Tar 505 7.46 Plan and elevation of the fortress of Qal’a-i Nau 506 7.47 The fortress of Qal’a-i Nau 507 7.48 Jam, the minaret, with fortifications in the background 514 7.49 Jam, the minaret: view showing horseman at the foot 515 7.50 Jam, detail of the tile decoration on the minaret 515 7.51 Jam, detail of the brick decoration on the minaret 516 7.52 Sakhar, the minaret 518 7.53 Qal’a-i Chahar Baradar, line of square and round towers 519 7.54 Comparative plans of towers in Ghur 520 7.55 Nili, mud-brick tower on stone base 521 7.56 Qal’a-i Chahar Baradar 4, detail of the decoration 521 7.57 Muna ‘Ala tower decoration 522 7.58 Qal’a-i Qaysar fortress 523

xxvii

figures and tables

7.59 7.60 7.61 7.62 7.63 7.64 7.65 7.66 7.67 7.68 7.69 7.70 7.71 7.72 7.73 7.74

Qal’a-i Qaysar fortress Qal’a-i Qaysar fortress detail Bamiyan, tower decoration Chisht, view of the two dome chambers Chisht, plans and sections of the two dome chambers Chisht, details of the brick decoration on the east dome chamber Larwand, façade and entrance of mosque Larwand, plan and elevation Herat, Friday Mosque, the Ghurid portal Herat, Friday Mosque, detail of the Ghurid portal Shah-i Mashad, ground plan Shah-i Mashad, south façade Shah-i Mashad, detail of decoration Shah-i Mashad, detail of decoration Danestama, ground plan of ruin Ghazni, moulded terracotta from the Ghurid phase of the palace of Mas‘ud III 7.75 Tala Begum, plan 7.76 Bust, so-called Ziyarat of Ghiyath al-Din 7.77 Bust, Ziyarat of Ghiyath al-Din, interior 7.78 Bust, Ziyarat of Ghiyath al-Din, inscribed tombstone 7.79 Bust, the arch 7.80 Bust, the arch detail 7.81 Bust, the arch detail 7.82 Bust, façade of mud-brick palace 8.1 Map showing main sites discussed in Chapter 8 8.2 Map of late Islamic sites in Sistan 8.3 Characteristic fortifications in Sistan 8.4 Characteristic zone of transition used in Sistan 8.5 Plan of Diwal-i Khudaydad 8.6 Diwal-i Khudaydad, courtyard house 8.7 Nishk, castle with corner bastions 8.8 Nishk, interior of castle 8.9 Nishk, fortified gate of city wall 8.10 One of the ruin fields of Sistan with the University of Bonn mission in the background 8.11 Typical ivan house in Sistan, showing characteristic vaulting, squinches and niches 8.12 The scouring effect of the ‘wind of 120 days’ on the ruins of Sistan 8.13 Herat, map showing main monuments 8.14 Herat, distant view with the two minarets and dome of the Gawharshad complex and the four minarets of the Husain-i Baiqara complex clearly visible

524 525 526 527 528 529 530 531 532 533 535 536 537 538 539 540 540 541 541 542 543 543 544 545 546 550 551 552 553 553 555 555 556 556 557 557 559

560

xxviii

FIGURES AND TABLES

8.15 8.16 8.17 8.18 8.19 8.20 8.21 8.22 8.23 8.24

Herat, distant view of the city in 1974 560 Herat, the city walls photographed in 1885 561 Herat, plan of Qal’a-i Ikhtiyaruddin 561 Herat, tile decoration on the northern tower of Qal’a-i Ikhtiyaruddin 562 Herat, reconstructed plan of the Gawhar Shad complex 563 Herat, sketch of the Gawhar Shad mosque by Durand 564 Herat, sketch of the Gawhar Shad madrasa by Durand 564 Herat, surviving minaret of the Gawhar Shad madrasa 566 Herat, surviving minaret of the Gawhar Shad mosque 567 Herat, detail of the tile-work inscription on the minaret of the Gawhar Shad mosque 568 8.25 Herat, plan of the Gawhar Shad mausoleum 569 8.26 Herat, interior decoration of the Gawhar Shad mausoleum 570 8.27 Herat, interior dome of the Gawhar Shad mausoleum 571 8.28 Herat, exterior dome of the Gawhar Shad mausoleum 571 8.29 Herat, the minarets of the Sultan Husain madrasa 572 8.30 Herat, tile detail on the Sultan Husain madrasa 573 8.31 Herat, plan of the Friday Mosque 574 8.32 Herat, Friday Mosque 574 8.33 Herat, mosque of Hauz-i Karboz, mihrab 575 8.34 Herat, plan and section of the mausoleum of Shahzadeh Abdullah 576 8.35 Herat, plan of the Gazurgah shrine 577 8.36 Herat, shrine complex of Gazurgah 577 8.37 Herat, tile decoration on the façade of Gazurgah 578 8.38 Herat, detail of tile decoration inside the main ivan at Gazurgah 579 8.39 Herat Province, map showing the number of sites in the Hari Rud catchment 580 8.40 Ziyaratgah, plan of the Friday Mosque 581 8.41 Ziyaratgah, ivan in front of the prayer hall of the Friday Mosque 582 8.42 Ziyaratgah, courtyard and side ivan of the Friday Mosque from the main ivan 583 8.43 Ziyaratgah, plan of the Khaniqah of Mulla Kalan 584 8.44 Ziyaratgah, the Khaniqah of Mulla Kalan 584 8.45 Kuhsan, plan and section of the madrasa 586 8.46 Kuhsan, tile decoration on the madrasa 587 8.47 Kuhsan, detail of the tile decoration on the dome 587 8.48 Deh-i Minar, plan and section of the Khaniqah 588 8.49 Kush Rabat, plan of the caravanserai 589 8.50 Ghazni, plan and section of the mausoleum of Shah Shahid 590 8.51 Ghazni, the mausoleum of Shah Shahid 590 8.52 Mazar-i Sharif, the Shrine of Ali 591 8.53 Balkh, plan and section of the Shrine of Khwaja Abu Nasr Parsa 593 8.54 Balkh, the Shrine of Khwaja Abu Nasr Parsa 594

xxix

figures and tables

8.55 Balkh, tile-work detail on the Shrine 8.56 Balkh, tile-work detail on the Shrine 8.57 Herat, bronze cauldron in the Friday Mosque 8.58 Metal ewer, signed and dated by Muhammad ibn Ibrahim al-Ghuri 8.59 Herat, Chaharsuq, covered reservoir 8.60 Herat, terraced garden overlooking the city 8.61 Herat, axonometric reconstruction of the Bagh-i Nazargah 8.62 Herat, decorated marble grave cover near Gazurgah 8.63 Kabul, the Bagh-i Babur 8.64 Kabul, the mosque of Shah Jahan in the Bagh-i Babur 8.65 Balkh, the arch of the madrasa of Sayyid Subhan Quli Khan 8.66 A hauz or cistern between Herat and Kuhsan

595 596 597 598 599 600 601 602 605 606 607 608

Tables 1.1 2.1 2.2 2.3 3.1 3.2 4.1 4.2 6.1 6.2 6.3

Natural vegetation in the Central mountain zone and Nuristan Late Palaeolithic flint industries of northern Afghanistan Calibrated radiocarbon dates for the Palaeolithic in Afghanistan Faunal remains from the Palaeolithic levels of Aq Kupruk II Radiocarbon dates from sites mentioned in the text Calibrated radiocarbon dates for the sites of Aq Kupruk II, Aq Kupruk I, Darra-I Kur and Shortugaï Radiocarbon dates from sites mentioned in the text Calibrated radiocarbon dates for the Neolithic and Bronze Age in southern Afghanistan Kushan king list Approximate framework for rulers of Gandhara and adjacent region Kushan, Kushanshah and Sasanian links

29 73 76 90 102 104 190 191 346 347 348

Notes on Contributors

Raymond Allchin (1923–2010) was one of the foremost British archaeologists in South Asian archaeology, elected a Fellow of the Society of Antiquaries (FSA) in 1957 and of the British Academy (FBA) in 1981. He did fieldwork in Afghanistan, at Shahr-i Zohak, and for many years across South Asia in collaboration with his wife Bridget. They founded the biennial South Asian Archaeology symposia held across Europe since 1971, and the Ancient India and Iran Trust in Cambridge. He was a member of the Faculty of Oriental Studies, University of Cambridge, 1963–89, then Emeritus Reader in South Asian Archaeology there. Warwick Ball is a Near Eastern archaeologist who has carried out excavations, architectural studies and monumental restoration in Afghanistan (where he was Acting Director of the British Institute of Afghan Studies), Iran, Iraq (where he was Director of Excavations of the British School of Archaeology in Iraq), Jordan, Libya and Ethiopia. He is currently Editor in Chief of Afghanistan, the journal of the American Institute of Afghanistan Studies, and also author of many books and articles on the history and archaeology of the region, including: Archaeological Gazetteer of Afghanistan (2 vols, 1982; new edition, 2019); with Anthony McNicoll, Excavations at Kandahar 1974 and 1975 (1996); and Monuments of Afghanistan (2008). Olivier Bordeaux studied Central Asia numismatics under the supervision of Professor Osmund Bopearachchi at the Université de Paris Panthéon-Sorbonne. He is currently the Deputy Director of the Délégation Archéologique Française en Afghanistan (DAFA). He is also working on the publication of the Kushan coin collection at the Bibliothèque National de France (BNF). Sophia Bowlby has retired from the Geography and Environmental Studies Department of the University of Reading, where she specialised in feminist social geography. She continues her research as an Honorary Research Fellow at Reading and Visiting Professor at Loughborough University.

notes on contributors

Richard S. Davis is a prehistorian with particular focus on northern Afghanistan, southern Tajikistan, eastern Turkey and central Siberia. Recently, his work has shifted to maritime cultures of the Aleutians and Bering Sea coast. His basic research interests centre on the study of human adaptations to the changing environments of the Pleistocene and Holocene, and he is now Professor Emeritus in Anthropology at Bryn Mawr College, Pennsylvania. Klaus Fischer (1919–93) conducted archeological field research on pre-Islamic and Islamic art and archaeology in Afghanistan between 1959 and 1962, and between 1966 and 1974 carried out survey work on mainly Islamic remains in Sistan. From 1966 to 1985 he taught at the Faculty of Arts at the University of Bonn. Henri-Paul Francfort is a Member of the Institut de France (Académie des Inscriptions et Belles-Lettres), and senior researcher emeritus at Centre Nationale de Recherche Scientifique. He is Director of the Archaeology of Central Asia team of CNRS (1984–2004), and the French Archaeological Mission in Central Asia (1989–2012). He has carried out fieldwork in Afghanistan (Bamiyan, Aï Khanoum, Shortughaï), India, Tajikistan, Kazakhstan, Uzbekistan, Kirghizstan and Siberia, and is the author of more than 160 publications. Simon Glenn is Research Fellow in the Heberden Coin Room at the Ashmolean Museum, University of Oxford. He wrote his doctoral thesis on the coins of the early Graeco-Bactrian kings. Norman Hammond is a Senior Fellow of the McDonald Institute for Archaeological Research at Cambridge University and Professor Emeritus of Archaeology at Boston University. He directed the first archaeological survey of the Helmand Valley in southern Afghanistan in 1966, was founding editor of Afghan Studies and South Asian Archaeology, and has also worked in North Africa and Latin America. He was elected FSA in 1974, FBA in 1998 and a Member of the Academia Europaea in 2018. Bertille Lyonnet is an archaeologist, and Directrice de Recherche Emerita at the CNRS, Paris. She has worked in Central Asia (Afghanistan, Tajikistan and Uzbekistan) since 1972, and since 1990 in north-eastern Syria, the northern Caucasus and Azerbaijan. A specialist in ceramics of different periods, she has always shown a special interest in the interrelations between the different worlds where she worked. She is the author of several book, including (with Nadezhda Dubova) The World of the Oxus Civilisation (forthcoming), and over 150 articles. David W. Mac Dowall (FSA 1960) was formerly a numismatist at the British Museum, a civil servant, Master of University College, Durham, and Director of the North London Polytechnic (now University of North London). A specialist in

xxxi

xxxii

NOTES ON CONTRIBUTORS

the pre-Islamic numismatics of Afghanistan, he has written widely on the coinage and epigraphy of Afghanistan and South Asia generally. Cameron A. Petrie is Reader in South Asian and Iranian Archaeology at the University of Cambridge, and a Fellow of Trinity College. He has conducted fieldwork in India, Iran and Pakistan, which has involved the investigation of archaeological landscapes and settlement sites dating to the Neolithic, Chalcolithic, Bronze Age, Iron Age and historic periods. Jim G. Shaffer is Associate Professor of Anthropology at Case Western Reserve University, Cleveland. Since the late 1960s he has conducted archaeological field research throughout Afghanistan, Pakistan and India, focusing on the Bronze through Iron Ages, and has published widely on those subjects. Nicholas Sims-Williams is Emeritus Professor of Iranian and Central Asian Studies at SOAS, University of London, and was elected FBA in 1988. His research centres on the medieval Iranian languages of Afghanistan and Central Asia, in particular Sogdian and Bactrian, and his publications include a three-volume edition of Bactrian documents from northern Afghanistan (2001–12). Maurizio Taddei (1936–2000) was an officer of the Museo Nazionale d’Arte Orientale in Rome from 1964 to 1974 and from 1968 a lecturer in Indian Art and Archaeology at the Istituto Universitario Orientale, Naples, becoming Professor of Indian Art History from 1976 and Vice Chancellor from 1981 to 1984. He was Director of the Italian excavations at Tapa Sardar from 1967, returning there in 1999. Kevin White is Associate Professor in Environmental Remote Sensing in the School of Archaeology, Geography and Environmental Science at the University of Reading, UK. He has thirty years’ experience of fieldwork, mostly focused on soils, geology and vegetation in dryland regions of Africa and Asia.

Acknowledgements

This new edition has only been made possible with the assistance of a great many people. As editors we are grateful to all our co-authors, both those who contributed to the 1978 volume, who have in some cases made significant emendations to their original chapters, and those who have joined us for this revised and expanded second edition. We are grateful to Carole Hillenbrand, who persuaded EUP to take the book on. We, however, remain entirely responsible for all errors missed and advice ignored. Norman Hammond appreciates the encouragement given by our late co-editor, Raymond Allchin, and his wife Bridget, for the idea of a new edition of The Archaeology of Afghanistan, and to Sir Nicholas Barrington of the Ancient India and Iran Trust, which will benefit from the royalties. David Whitehouse allowed the use of his unpublished colour photographs of the 1974 Kandahar excavations, and a serendipitous encounter with Mitch Allen in Vancouver led to the inclusion of results from his and William Trousdale’s 1970s work in Sistan; Marc Abramiuk shared the results of his more recent work in the Helmand valley. Sophie Bowlby and Rick Davis went back four decades in their memories and files to revise our first two chapters, but NH is above all grateful to Warwick Ball for agreeing to join the project and taking charge of the bulk of the revision and augmentation: the extensive assistance he was able to recruit for both text and illustrations is reflected in his own acknowledgements and in the book itself. Warwick Ball is grateful to Norman Hammond for inviting him to take part in this exciting project in the first place. For advice, comments and discussions generally on parts of the text and many other matters relating to the revised version he would like to thank John Boardman, Osmund Bopearachchi, Nikolaus Boroffka, Robert Bracey, Pierre Briant, Joe Cribb, the late Nancy Dupree, Elizabeth Errington, Anna Filigenzi, Ute Franke, Simon Glenn, Frantz Grenet, Robert Hillenbrand, Jonathan Lee, Bertille Lyonnet, Rachel Mairs, Philippe Marquis, Cameron Petrie, Benjamin Mutin, William Trousdale and Edinburgh University Press’s anonymous reviewers. Mitchell Allen and William Trousdale allowed use of their unpublished material from the Helmand-Sistan Project and shared their presentations of that material. Paul Bucherer-Dietschi allowed access to the superb library and archives

xxxiv ACKNOWLEDGEMENTS of the Afghanistan Institut in Bubenthal, Switzerland. Leonard Harrow translated a passage from Farrukhi Sistani in Chapter 7. Mike Mantia shared his impressions and images of Kharwar while on active service there with the US Marines. WB spent a very profitable few days at the Oriental Institute, University of Chicago, at the invitation of its Director, Gil Stein, with their Afghan Heritage Mapping Partnership team, Emily Hammer, Anthony Lauricella, Rebecca Seifried, Kathryn Franklin and Emily Boak, with their unrivalled satellite images of Afghanistan. During the gestation period of this project he benefited greatly from conferences relating to Afghanistan in London, St Andrews, Chicago and St Petersburg and discussions with fellow participants. Although acknowledged as one of the authors, he owes a particular debt to Henri-Paul Francfort, who has been unstinting with his outstanding knowledge of the archaeology of Afghanistan in particular and Inner Asia in general for advice on many aspects of this revision. For procurement and use of new images, WB would like to thank: Flemming Aalund; David Adams; Mitchell Allen of the Helmand-Sistan Project for photos by Carol Ellick and Robert K. Vincent; Wendy Ball; Osmund Bopearachchi; Paul Bucherer-Dietschi of the Afghanistan Institut for photos by Klaus Fischer; Elizabeth Errington; Anna Filigenzi of the University of Naples ‘L’Orientale’ for IsMEO photos; Henri-Paul Francfort of the CNRS for his own photos and photos by Paul Bernard, Pierre Gentelle and others of the French Mission in Afghanistan; Simon Glenn; Norman Hammond for his own photos and photos by the 1966 Cambridge Expedition; Leonard Harrow for photos by Josephine Powell; Anthony Lauricella and Rebecca Seifried of the Chicago Oriental Institute for satellite images; Guy Lecuyot for the digital reconstructions of Aï Khanoum; Jonathan Lee; Bertille Lyonnet; Mike Mantia; Philippe Marquis; Tatsuhiko Maeda of the Hirayama Museum for photos of Bactrian documents; Nahla Nassar of the Khalili Collection for photos of Aramaic and Bactrian documents; Lolita Nehru; Bernard O’Kane; Zafar Paiman; Eric Schnittke of the University Museum, Pennsylvania, for the photograph by George Dales; Nicholas Sims-Williams; David Thomas. All other photos are either from the original authors or by Warwick Ball. It has been a real pleasure working with Edinburgh University Press, and both editors would like to thank their editorial, production and publicity staff, Michael Ayton, Ellie Bush, Ruth Campbell, Eddie Clark, Zuzana Ihnatova, Rebecca Mackenzie, Emma Rees, Rebecca Wojturska and Kirsty Woods. Their cartographer John Watson produced all colour maps for the revised edition. Above all, Head of Editorial Nicola Ramsey has been an enthusiastic supporter of this project from the beginning with unfailing patience, wisdom and humour. Note. Authors of chapters are listed in the book in alphabetical order. Warwick Ball and Norman Hammond

Note on Transliteration

Afghanistan is a transliteration minefield. Three main languages are spoken, not to mention several minor languages and many dialects, with different systems for all main ones. There is no single system that fits all, but for Persian/Dari names, the system that was recommended for the pre-1979 issues of journal Iran has been adopted here. For Pashto names, a standardised system has been formulated by the International Centre for Pashto Studies in Kabul, but to minimise the use of diacritical marks we have used the same system as for Persian. Post-independence Uzbekistan and Turkmenistan have developed their own systems for their variants of Turkish but in general we have transliterated from the version written in Arabic script using the same system as for Persian. Diacritical marks have been avoided for the sake of simplicity (apart from when they occur in bibliographical titles). In general the more modern accepted forms of names are used throughout instead of older forms (Ghur instead of Ghor, Sistan instead of Seistan, Amu Darya instead of Oxus) except when quoting (such as in the ‘Introduction to the Original Edition’). For Chinese, the modern Pinyin instead of the older Wade-Giles system is used: Yuezhi instead of Yueh-Chih, Xinjiang instead of Sinkiang, and so forth. For ancient Greek names the more familiar forms are used: Seleucus instead of Seloikos, Bactria instead of Baktria. In general, place names that are more familiar from their main publications rather than correctly transcribed forms are used: Begram instead of Bagram, Tillya Tepe instead of Tila Tapa, Aï Khanoum rather than Ai Khanum, Tamerlane rather than Taimur-i Lang. Although we have attempted to adhere to these principles as much as possible, many inconsistencies inevitably occur. In many cases – such as place names recorded by nineteenth century travellers – the original forms are not known.

Foreword Mohammad Fahim Rahimi (Director, National Museum of Afghanistan)

I read parts of the original edition of this book, edited by Raymond Allchin and Norman Hammond, for the first time when I was studying at the University of Pennsylvania in 2014. I discovered the book in the library and upon opening the first pages immediately recognised it as the most comprehensive and authoritative work on the archaeology of Afghanistan so far written. I still remember my first impression, seeing for the first time images of the carved wooden mihrab from the Charkh area of Logar Province, as well as the excellent illustrations throughout which helped me to further appreciate some of the artefacts in the collections of the National Museum of Afghanistan where I had previously worked. On subsequently reading the book in more detail, I found it completely up to date on the state of our knowledge on the archaeology of Afghanistan just before the recent wars, with information on most of the excavated archaeological sites of Afghanistan that had contributed to shaping the history of my country until that time. Moreover, I found it included an excellent description of the geographical characteristic of the country, providing a very positive image of Afghanistan for its readers. This new and updated edition edited by Warwick Ball and Norman Hammond is even more comprehensive, with new illustrations and new topics by well-known scholars. There are some excellent discussions about recently discovered cultural material, including those illegally recovered, as well as the new excavations since 2001 that have further enhanced the glorious history of this country. The chapter on ‘The Development of the Oxus Civilisation North of the Hindu Kush’ is the newest and most radically revised part of the book. We owe our updated knowledge about the Oxus Civilisation mainly to new discoveries of the Bactro-Margiana Archaeological Complex (BMAC) in excavations in Central Asia. Much of the information on this civilisation in Afghanistan regrettably comes from material found in illegal excavations over the past four decades. Information from studies of archaeological sites in Turkmenistan, Uzbekistan and Tajikistan, together with the studies of illegally excavated material from Afghanistan, has better shaped our understanding of the history of early communities in this region, now described in the new edition. Moreover, we now have a very complete study

xxxviii FOREWORD of the key excavations of Shortughaï that came too late for the previous edition, contributed to this edition by the site’s excavator. Gratifyingly, I see a new chapter here on the ‘Helmand Civilisation’, a term that I very much favoured using to describe the settlements that developed along the Helmand and Arghandab Rivers, distinct from the Oxus and Indus Civilisations. The material culture from the excavated sites of the Helmand Civilisation is fully described, and gives a very clear image to the reader of the history of this region. The chapter on the early historic period in the new edition, ‘The Iron Age, Achaemenid and Hellenistic Periods’, is completely updated by new authors. It includes discussions on settlements, material culture, architecture, art, epigraphy and numismatics, giving the reader access to the most recent studies on these subjects by the scholars who first described them. Included here are new discoveries – of the Mir Zakah 2 hoard, the Balkh documents and new excavations – together with other individual discoveries over the past four decades that have cast further light on the history of Afghanistan during these periods. In the chapter ‘From the Kushans to the Shahis’, new information about the Rag-i Bibi relief, the Balkh wall and the Bactrian documents, in addition to updated studies of previous discoveries, contribute new material to our knowledge of the history of this period in Afghanistan. One of the most important sites from this period is Mes Aynak, which has greatly contributed to a better understanding of the history of the Kabul region. Since 2009 this has been one of the biggest ongoing excavation projects in Afghanistan. Objects illustrating the material culture from Mes Aynak are currently at the National Museum of Afghanistan, emphasising the very diverse history of the Kabul Valley. The last two chapters concern the Islamic art and architecture of Afghanistan, and are also comprehensively updated. The descriptions of the history of major Islamic dynasties, their art and architecture – especially some new sites and monuments of the Ghurid dynasty – all complement the glorious history of this period. The metalwork and other Timurid period art of Herat make this chapter a particularly interesting one. I will always enjoy reading this book. I am sure that it will become as much a classic on the archaeology of Afghanistan as its first edition did. For those who wish to learn about the physical geography, historical geography, history and cultural history of Afghanistan, this book is the one to read. I am sure that it will become the most cited book on the subject, and I congratulate the editors and contributors on their achievement.

Preface Norman Hammond

This book was conceived in a tropical forest in Guatemala in 1968: I was working on a Harvard University excavation at the Maya city of Seibal, but still thinking about my research in the Helmand Valley of southern Afghanistan in 1966. The lack of a basic conspectus of the country’s archaeology was apparent and regrettable, and on a page of my Harvard site-book I sketched out a list of chapters. My expertise was strictly limited, but when I returned to England I approached Dr Raymond Allchin (1923–2010), who with his wife Bridget (1927–2017) had given me enormous support in planning my 1966 Helmand survey and analysing its results. He had worked in Afghanistan, at Shahr-i Zohak in the Bamiyan Valley in 1951, and maintained an interest in archaeology there as he went on to a distinguished career in India and Pakistan. Raymond agreed that a book was needed, and to be joint editor, and suggested potential contributors; others were proposed by Dr Louis Dupree (1925–89), the pioneer of American archaeology in Afghanistan and author of the most comprehensive book on the country’s history. Some of those we asked, such as Paul Bernard, excavating at Aï Khanoum, were too immersed in fieldwork, or, like Jean-Marie Casal, excavator of prehistoric Mundigak, too retired; but we assembled an international team from Britain, Germany, Italy and the USA. This revised edition has added contributors from Australia and France. Unsurprisingly, the book took some years to get together, although helped by the establishment in 1971 of the biennial South Asian Archaeology conference under the Allchins’ leadership and the successive volumes of South Asian Archaeology. We were lucky in that Anthony Watkinson, Editorial Director at Academic Press, swiftly agreed that this was a book worth publishing: with a contract in hand, Raymond and I pushed our contributors, and the book went to press in 1977, its bibliography updated to that year. Seven hundred hardback copies were printed. The Archaeology of Afghanistan from Earliest Times to the Timurid Period was well-received, well-reviewed (including substantial excerpts republished in the Kabul-based Afghanistan) and sold out: it now fetches flattering sums on the rare-book market. Publication coincided with the beginning of a harder, sadder period in Afghanistan’s history, and archaeology took a back seat to instability and war. Some work was still done, however, notably by Soviet scholars in the

xl PREFACE Oxus valley, and after the US-led intervention in 2001 it emerged that less had been lost than we had all feared: much of this was due to the heroic efforts of Afghans in the National Museum, who squirrelled away significant collections of artefacts, of which the Tillya Tepe golden treasure is only the best-known. As archaeological fieldwork again became possible, if still hazardous, in more of the country, so the utility of a second edition of this book became more obvious. Unfortunately, several of our co-authors had died since 1978, including Maurizio Taddei and Klaus Fischer, joined in 2010 by my co-editor, Raymond Allchin. He had approved the idea of a new edition, however, joined in this by his wife Bridget. Several of the surviving authors agreed to revise their chapters, or to assist those recruited to do so. And I had the great good luck to entice Warwick Ball to become co-editor and principal – in fact predominant – reviser and expander of the text, with the assistance of a range of his colleagues. Warwick had worked in many parts of western Asia, including a stint as acting director of the British Institute in Kabul while it still existed, and had with the collaboration of Jean-Claude Gardin compiled the two-volume Archaeological Gazetteer of Afghanistan (1982), ‘as impressive an archaeological gazetteer as has ever been compiled’, as one reviewer said. He was revising this for Oxford University Press at the same time as revising The Archaeology of Afghanistan, and had also published The Monuments of Afghanistan: History, Archaeology and Architecture (2008). Elected to the Society of Antiquaries of London in 1999 for his achievements, he was a fortunate and obvious choice to bring our book into the twenty-first century, as can been seen from his extensive contributions to the text and splendid photographs. We also welcome the significant contributions of Olivier Bordeaux, Henri-Paul Francfort, Simon Glenn, Bertille Lyonnet, Cameron Petrie, Nicholas Sims-Williams and Kevin White to the revised and expanded text. We mourn the deaths not only of several contributors to the first edition of this book, as noted above, but also of several colleagues whose work has informed both versions, including Paul Bernard, Roland Besenval, David Bivar, Louis and Nancy Dupree, Jean-Claude Gardin, Svend Helms, Anthony McNicoll, Ralph PinderWilson, Viktor Sarianidi, Zemaryalai Tarzi, Maurizio Tosi and David Whitehouse.

Introduction to the First Edition Norman Hammond and Raymond Allchin

Afghanistan is a political rather than a geographical entity, formed by the ­competing imperialisms of Russia and Britain in the latter part of the eighteenth and nineteenth centuries and by the far-sighted ambition of the Durrani dynasty, who ensured that Afghanistan survived as an entity and was not simply absorbed between the two larger empires (Dupree, 1973). Afghanistan has well been called the crossroads of Asia, in that it is bounded on the south and east by the Indian subcontinent, on the west by the deserts and plateaux of Iran, and on the north by the great inland drainage basins of Central Asia, of the Oxus (Amu Darya) and the Jaxartes (Syr-darya). Throughout the known past these three major cultural areas have nourished their own traditions, civilisations arising and declining at different times in each of them, and through Afghanistan has passed much of the ­cultural and commercial traffic between these areas and more distantly between China, India, the Persian Gulf and the Mediterranean world. Afghanistan has been both a meeting place and the melting-pot of cultural influences from which it has evolved throughout history its own highly distinctive cultural contribution to civilisation. The broad geographical and cultural region of which Afghanistan forms the heart stretches from the Oxus to the Indus, and yet within the bounds of the modern state there is marked diversity both of physical and of human geography, with mountains, deserts and fertile valleys closely juxtaposed. Surprisingly there has never been a book primarily concerned with the archaeology of Afghanistan, although two recent publications by Masson and Romodin (1964) and Dupree (1973) have utilised much archaeological evidence within more general historical accounts. The former of these at least is not easily available to an English reading audience and neither can be said to perform the function of a basic introduction to the archaeology of Afghanistan for the English reader, one which we hope that this book will fulfil. Since the Second World War there has been an increase in international interest in Afghanistan and this has found expression in the growing number of countries who have sent archaeological expeditions and missions to work in the country. All of this means that the time is now ripe to present a synthesis of both earlier and recent work; one might almost

2

THE ARCHAEOLOGY OF AFGHANISTAN

say that for the first time there is more cheese than holes! There has also been an increase in interest within Afghanistan itself and we hope that this book, although the product of foreign scholarship, will prove useful and acceptable to the Afghan people. With the increase in archaeological knowledge of the high cultures of India, Iran and Central Asia the region of Afghanistan, peripheral to each of these yet central to all three, has become more important and knowledge of its archaeology vital to a study of any of them. Choice of Approach and Structure of the Book This book is both collaborative and international in its authorship because archaeological work in Afghanistan over the past three decades has been truly international in scope. In view of the time span, the range of cultural complexity, and the newness of much of the research here reported, it seemed desirable to choose authors who were immediately acquainted with the various regions and periods of the archaeology of Afghanistan rather than to rely on a necessarily more secondhand synthesis by one or two people. The contributors to this book come from both Europe and America and all have carried out fieldwork in Afghanistan within the past few years. Several have major research programmes there in progress and all are outstanding authorities in their fields, which often range far beyond the boundaries of Afghanistan itself. By making this work collaborative we have been able to obtain a much greater degree of local expertise than would have been otherwise possible and each chapter places its period of Afghan archaeology within a broader South Asian and Central Asian context in a way that would be difficult for a single author to comprehend while using the most up-to-date information. Similarly the editors have both carried out archaeological field work in Afghanistan, although their major efforts have lain elsewhere, and both have been involved in the archaeology of Afghanistan for many years. As editors we have interfered as little as possible with the texts of our coauthors, apart from the modifications necessary to weld their separate contributions into a coherent whole. Nor have we sought uniformity of English style. The basic organisation of the book is chronological but naturally the subject matter varies very much from the prehistoric through into the historic period. The geographical background to Afghan prehistory and history is described in the first chapter in both physical and human terms by Sophia Bowlby. This is followed by chapters dealing chronologically with each of the major periods of prehistory and history. Chapter 2 by Richard S. Davis gives an account of the earliest known inhabitants, and deals with the last major geomorphological changes at the end of the final glaciation. There is thematic continuity between this chapter and the next by Jim G. Shaffer, who opens with a discussion of the final stage of the Stone Age and the prelude to settled agricultural production in this part of the world.a This chapter has been split into two, dealing with the areas north and south of the Hindu Kush.

a

introduction to the first edition

This chapter embraces the rest of Afghan prehistory down to the comparatively little known second millennium and early Iron Age. Recorded history begins with the Achaemenid conquests of the sixth century bc. Chapters 4 and 5b are the joint work of David Mac Dowall and Maurizio Taddei, the former dealing with the evidence of inscriptions and numismatics, and supplying a short historical introduction, and the latter dealing with the evidence of settlements, architecture and art history. Chapter 4c covers the centuries between the Achaemenid conquest and the arrival in the last two centuries bc of the nomadic Yueh-Chi in the north. It includes, therefore, the campaigns of Alexander the Great and the subsequent Greek kingdoms in Bactria and Kabul. Chapter 5d begins with the rise of the Kushan empire and continues through the successor dynasties to the Hephthalites and the Turks, and the Arab invasions which heralded the start of Muslim rule in Afghanistan. This chapter therefore includes reference to the caves at Bamiyan and allied monuments. The archaeology of the Muslim period is dealt with in Chapters 6 and 7 by Klaus Fischer. The first opens with the Arab invasions and covers the centuries up to the Mongol invasions, including the Ghaznavid and Ghorid dynasties; and the latter deals with the dynasties who succeeded the Mongols, down to the Timurids. Each lays stress on the discoveries and study of monumental remains which have hitherto formed a major aspect of the archaeology of this period in Afghanistan, but includes also notice of the other principal categories of evidence. The short historical introductions which preface these chapters were kindly supplied at short notice by Peter Jackson, Fellow of Churchill College Cambridge, during Professor Fischer’s temporary indisposition. Within the wide scope of the book we have tried to make the emphasis archaeological rather than historical. We are all too conscious of the resultant gap between the archaeological accounts of the historical period, and full historical and archaeological synthesis of the kind so brilliantly achieved by Ghirshman in his works on Iran and Afghanistan. Our justification, if indeed justification be required, must be that we are presenting so much new material that to have included this sort of synthesis must have greatly extended the length of this already considerable volume. For the reader who wishes to have further reference to the history we recommend him to consult the bibliographies of such works as Fraser-Tytler (1967), Masson and Romodin (1964), Dupree (1973), and more generally in the accounts of the history of Afghanistan in the Encyclopaedia of Islam and incidentally in the Cambridge History of Iran. As editors we have attempted, within reason, to adopt a single system of transliteration for all our authors, but we are aware that inconsistencies still occur. Where 3

Note that apart from updating this first edition Introduction, all notes are at the end of the book, starting on p. 620. b Now Chapters 5 and 6, also with new contributors. c Now Chapter 5. d Now Chapter 6.

3

4

THE ARCHAEOLOGY OF AFGHANISTAN

several spellings of a name or place name are current, we have sometimes allowed different forms, but we are conscious that in many instances where we have tried to be consistent we shall offend some by our choice of one form or another. As far as possible we have omitted diacritical marks, even if regretfully. We have evolved and applied our own system of conventions rather than any published schema, with the emphasis on simplicity and accessibility to the general reader as well as the specialist, who we hope will not be too critical. Radiocarbon dates are quoted either as raw dates (Libby halflife unless otherwise stated) followed by bc/b.p. and the laboratory number, or based upon bristlecone pine calibration as dates in calendar years. The calibration tables published by MASCA (1973) have been used.e Most dates are published in the journal Radiocarbon and the reliability and technicalities of the original dating may be ascertained there. For the historical period dates are quoted in bc/ad. 5

The History of Archaeological Research in Afghanistan If any fool this high samootch explore Know Charles Masson has been here before

Unlike her two great neighbours, India and Russia, both of whom developed state archaeological services during the nineteenth century, Afghanistan remained almost unknown to archaeology and without any department of its own until recent years. The earliest records of its monuments are in the accounts of travellers, first Chinese, then Muslim and finally European. The Chinese records are mainly those of Buddhist pilgrims who, on their way to visit the holy places in north India, traversed the often hostile and uncomfortable lands of Central Asia and Afghanistan. Although this traffic must have begun at least by Kushan times, the surviving records which from our point of view are important only start with Fa-Hsien (c. ad 400). The detailed account of the travels of Hiuen-Tsang in the middle of the seventh century contains a number of references to monuments, both living and ruined, just as it provides us with our earliest secondary source for many of the Asokan sites and monuments in India. Among the places mentioned by Hiuen-Tsang are the ‘New monastery’ (nava sangharama or nau bihar) at Balkh, and monasteries at Kapisa and Nagarahara. The description of the great Buddhas at Bamiyan, which in those days must have still been relatively new, is particularly interesting, and he informs us that the great Buddha shone golden in the sun (Beal, 1906). From the eighth century, Arab and other Muslim travellers and geographers begin to write about Afghanistan, and in some of their works too there are accounts of monuments. Thus the Hudud al-Alam (c. ad 982) mentions Balkh and its famous, For the new edition the OxCAL v.4.3.2 (2017) calibration has been used, with dates as bc/ad or bp (before present = ad 1950).

e

introduction to the first edition

painted nau bihar, as well as Bamiyan and Nagarahara, also commenting on the great Buddha figures (Minorsky, 1937). These attracted the attention of many subsequent writers, including Yaqut (ad 1461) who gives a description of the town and some of its remains. He mentions the wall paintings at Bamiyan, adding that they represented ‘all the birds created by God’. With European writings we encounter for the first time not only the accounts of travellers, but also coin collectors. As might be expected, much of this interest found its focus in India or Russia. It was the recognition of coins of Eukratides and Theodotus which suggested to Theophilus Bayer the plan of his history of the Greek kings of Bactria (Historia Regni Graecorum Bactriani) in 1738, and thereafter further Bactrian coins found their way to collectors in France, Britain and Italy. Indeed it seems that these coins led European scholars towards Bactria and its Greek rulers. This interest received a special stimulus, as did that in all other branches of oriental archaeology, in India and in Europe, from the foundation of the Asiatic Society of Bengal in Calcutta by Sir William Jones in 1784. The new spate of researches which immediately began to appear found their publication in the pages of the society’s Journal, while in the person of the society’s librarian, James Prinsep, the study of Indian epigraphy and numismatics found a notable proponent. Prinsep undertook to publish and illustrate many of the coins and antiquities which travellers and others recovered, both from the northwest of India itself and from Afghanistan. Among early travellers we may mention W. Moorcroft and G. Trebeck, whose Travels in the Himalayan Provinces (1819–25) were published in 1841, and H. W. Bellew’s Afghanistan and the Afghans (1839). They noticed numerous monuments on the road from Peshawar to Kabul and thence to Bamiyan and Balkh. Another early traveller was Sir Alexander Burnes whose acute eye and keen intellect made many valuable historical and archaeological observations on his way to Bukhara in 1831–3 (Burnes 1833, 1839). The 1830s saw a sudden outburst of archaeological work, fed no doubt by the growing British concern for Russian imperial expansion towards Central Asia. Foremost among those who contributed to knowledge of the archaeology of Afghanistan was Charles Masson who between 1834 and 1837 travelled widely there. To him we owe the first report of the old city of Begram, which he proposed to identify with the city founded by Alexander and known from classical sources as Alexandria ad Caucasum or Alexandria Paropamisadae (Masson, 1834, 1836a, b). Here he collected over the next few years some 30,000 coins, including many Greek and Kushan. It was this find more than any other which drew attention to the importance of Afghanistan for classical archaeology. Masson discovered other sites and antiquities around Kabul, and visited Bamiyan where he noticed the charcoal graffiti left by earlier visitors, including Moorcroft and Trebeck (Masson 1836c, d); and he left his own doggerel couplet (quoted at the head of this section) high up in an inaccessible place above the head of the great Buddha, where it was discovered a century later by the French mission. Another area which he surveyed was in the vicinity of Jalalabad, in the plain of the Kabul river valley, with s­ triking

5

6

THE ARCHAEOLOGY OF AFGHANISTAN

results, discovering dozens of Buddhist stupas, as well as many mounds. The results of his work at Begram and elsewhere were first published in the Journal of the Asiatic Society of Bengal (Vols 1–5) and later as a part of H. H. Wilson’s Ariana Antiqua (1840). Masson excavated several of the stupas and among the relics recovered were numerous coins and the famous Bimaran casket, now in the British Museum. At about the same time as Masson was visiting these places a Swiss physician (in the service of the Sikhs), Dr M. Honigberger, travelled through Afghanistan on the way home to Europe. He too visited many of the sites around Jalalabad and near Kabul, also doing some excavations and collecting coins. The results of his work were published in the Journal Asiatique, 1836–9. Another medical man, J. G. Gerard accompanied Burnes and left a memoir of these same sites (Gerard, 1834). The various researches of the third decade of the nineteenth century led to several important new publications. We have already mentioned Wilson’s Ariana Antiqua. Another writer to use the new material was Christian Lassen, a Norwegian who spent most of his working life at the University of Bonn. He wrote a monograph on the history of the Greek and Indo-Scythian kings of Bactria, Kabul and India, published in 1838. With the British invasion of Afghanistan in 1839, and the political uncertainties that ensued, the volume of original research dries up, and once more the stray traveller’s report is the only new material forthcoming. For instance, J. P. Ferrier’s Caravan Journeys, published in 1856, records his travels in 1846, with many references to archaeological remains at Karabagh, Kandahar, and elsewhere. He also included a tantalising reference to remains of an old city in a valley between Sar-ipul and Boodhi, where he noticed some great reliefs cut out of the rock. These have so far eluded the attempts of subsequent archaeologists to rediscover them. The collecting of coins continued as is witnessed by P. Gardner’s Catalogue of coins of the Greek and Scythic kings of Bactria in the British Museum (1886); while such geographical accounts as H. G. Raverty’s Notes on Afghanistan, 1878 and T. H. Holdich’s Indian Borderland (1901) and Gates of India (1910), contain much of value to the archaeologist, even if not primarily written from this point of view. Russian interest in Central Asia had already found expression in F. Nazarov’s work (1821), and developed an archaeological or historical bias in a number of other works, such as I. Bichurin’s study of the antiquities of Central Asia (1851), K. Ritter’s Kabulistan and Kafiristan (1867), and the writings of O. Tomaschek, particularly his Central Asiatic Studies (1877). The twentieth century produced a number of new names among whom we may mention three in particular, V. V. Bartold, 1869–1930, whose many works on the history of Central Asia contain much of relevance for Afghanistan; Sir Aurel Stein, 1862–1943, whose longstanding interest in archaeology took him on a series of expeditions in Central Asia, India and Iran, which although they largely skirted the geographical confines of Afghanistan nonetheless contributed much that relates to it; and A. Foucher, 1865–1952, whose consuming interest in the

introduction to the first edition

extensions of Greek and Hellenistic culture into the east led him to Afghanistan, where the studies of his later years were mainly concentrated. His interest early found expression in the monumental L’art greco-bouddhique du Gandhara (Vol. 1. Paris, 1905; Vol. 2. Paris, 1918); and in a whole series of supporting publications. By these works Foucher established a claim to be the leading art historian in this field. The collection and publication of basic data continued with work such as that of Tate (1912) on Sistan. A new era for archaeology in Afghanistan began in 1922 with the signing of the Franco-Afghan archaeological convention. The terms of this created a permanent Délégation Archéologique Française en Afghanistan (DAFA), giving it a virtual monopoly over research. This established for the first time a permanent base for archaeological research in the country. Its creation was in no small way the result of Foucher’s enthusiasm, and he was the natural person to become its first director. The fruits of its labours can be seen by reference to the pages of the series of Memoirs which it published. These include, among others, Foucher’s magnum opus on La vielle route de l’Inde, de Bactres à Taxila (2 vols, 1942 and 1947); excavations and studies at Hadda by J. Barthoux and others; a series of studies of Bamiyan, particularly by J. Hackin, who succeeded Foucher as director; excavations and studies at Begram, including the discovery and publication of its wonderful treasure, by Hackin, R. Ghirshman and others; numismatic studies of the coins of the Hephthalites and of important hoards of Greek coins; the discovery and excavation of a Kushan dynastic temple at Surkh Kotal, by D. Schlumberger, 1952–63; studies of the Ghaznavid palaces at Bust and Lashkari Bazar; and of the discovery of the minaret of Jam by A. Maricq; and finally two prehistoric excavations, of Nad-i-Ali in Sistan by Ghirshman, 1938, and at Mundigak near Kandahar by J.-M. Casal, 1951–8. The most recent discovery, which sets the seal upon Foucher’s great vision, has been of the Greek city of Ai-Khanum, in 1963, and its subsequent excavation by Schlumberger and his successor P. Bernard. In the years leading up to the Second World War other nationalities began to take an interest in the subject. In 1938 a small British expedition, consisting of E. Barger and P. Wright, surveyed sites in north Afghanistan, particularly around Qunduz and in Badakhshan. But it was not until the end of the war that international interest in Afghanistan and her archaeology took on a new form. As much of the work and many of the names which are connected with this period are dealt with in detail in subsequent chapters of this book, we shall not enumerate them here. Rather we shall indicate something of the breadth of this new interest. In 1946 the Indian Government sent a team headed by the Director General of Archaeology, Mortimer Wheeler, to visit Afghanistan and make a survey of sites, mainly in the north. Indian interest has continued and found expression in a number of expeditions aimed at offering assistance to the Government of Afghanistan in the conservation of Bamiyan; plans are also in hand for an excavation. We have already mentioned the continuing work of the French Delegation. An Italian Mission, affiliated to the Istituto Italiano per il Medeo ed Estremo Oriente (IsMEO), and

7

8

THE ARCHAEOLOGY OF AFGHANISTAN

inspired by G. Tucci, has been working since the 1950s. This mission has carried out extensive excavations and explorations around Ghazni, under A. Bombaci and U. Scerrato, and at the neighbouring Buddhist complex of Tapa Sardar, under M. Taddei. British work was mainly on an individual basis,f until 1972, when a British Institute of Afghan Studies was established in Kabul. This institute has undertaken excavations at Kandahar. German work has been mainly concerned with the survey of sites of the Muslim period in Seistan and the Herat area. There has been a considerable volume of American research. In 1949 W. Fairservis led the first expedition to Kandahar and Sistan areas, surveying a number of sites. In the following season Louis Dupree excavated at Deh Morasi Ghundai, while Fairservis made further explorations in Sistan: Dupree also excavated at Shamshir Ghar. In 1953 Rodney Young excavated part of the defences of the lower city of Balkh, obtaining important chronological data; and in 1954 Carleton Coon carried out pioneering excavations in Kara Kamar cave near Haibak. revealing Stone Age deposits (Coon, 1957). These were the first excavations of a prehistoric cave site of this period in Afghanistan. Since 1959 Dupree has made further surveys and excavations of prehistoric and later sites, notably at Aq Kupruk in western Badakshan, and elsewhere in north Afghanistan. In 1970 George Dales carried out a survey in Sistan and excavated at Nad-i-Ali.g The Soviet Union have recently established co-operation in the form of a Soviet– Afghan archaeological expedition which since 1971 has undertaken explorations and excavations at several sites in the neighbourhood of Akcha, including Dashli, Tillya Tepe and Altin Dilyar Tepe. These excavations have yielded very promising materials, to judge by the published results to date. A Japanese expedition from Kyoto University have been working in Afghanistan since 1960; initially led by S. Mizuno it made surveys in the Haibak region, and more recently in collaboration with Afghan archaeologists has been excavating and surveying Buddhist sites in the Hadda region. Finally some mention must be made of the initiation of fieldwork by the Afghan Government itself. Since 1965 Dr S. Mustamandi and subsequently Dr Z. Tarzi have been excavating at Tapa Shotor and other sites in the neighbourhood of Hadda, and thus laid the foundations for a local school of archaeological research. In view of the number of ancient sites already known and the magnitude of the work involved in their excavation, one can only applaud this development, and look forward to its future progress.h To conclude this brief survey of the history of archaeological research in 6

8

This included the 1951 recording of Shahr-i Zohak in the Bamiyan Valley by Allchin, now published as Baker and Allchin 1991. g Other work in south-western Afghanistan included William Trousdale’s work around the lower Helmand in the early 1970s, and Norman Hammond’s survey of the middle Helmand Valley in 1966 (Hammond 1970). h This chapter covers the history of archaeology in Afghanistan up to the mid-1970s: further ­developments are discussed in the Introduction to the new edition. f

introduction to the first edition

Afghanistan, we would like to point to two trends which have become apparent during the past three decades. The first is towards international participation. The Government of Afghanistan have permitted and encouraged the presence of foreign teams and archaeological missions, and these have been able to make a substantial contribution. The international authorship of this book is one of the fruits of this tendency. The second, in the long run more important, trend has seen the beginnings of an indigenous pool of trained archaeologists, undertaking their own researches. This was looked forward to by one of us at the beginning of the period (Allchin, 1957, 141): It is to be hoped that the time is not too far distant when the Government of Afghanistan will recognize the tremendous interest of its ancient sites and institute its own program of research. If the work is to succeed it will do so best when local interest supports local workers. Through the harmonious progress of these two trends Afghanistan has made great strides, and its past – once so tantalising because unknown and inaccessible – has begun to grow more distinct.

9

Introduction to the New Edition Warwick Ball

When I was first asked by Norman Hammond to update this classic work for a new edition, my immediate thought was that it would be a relatively simple task: the almost continual state of upheaval that has plagued Afghanistan since the first edition was published would have allowed little fieldwork or any other developments in the archaeology to have taken place. I could not have been more mistaken. Not only has important fieldwork taken place, but there have been major new discoveries and an outpouring of new publications (witnessed by the bibliography of this edition, now almost three times the length of the original), a tribute both to the cultural tenacity of the country and to the enduring international scholarly interest. In addition, archaeological research has continued unabated in the regions immediately adjacent: this has had major – occasionally fundamental – bearings on the archaeology of Afghanistan. This applies particularly to the opening of the former Soviet Central Asian republics, which has made previous Soviet scholarship more available internationally and has enabled international teams to work in the area. Much else has changed too (and not least Afghanistan itself). New technology has had an immense impact on archaeology in such fields as satellite imagery, remote ground sensing and massive data crunching by computer. The internet has made available far more information online than it is possible for any one person to absorb. Much might not be worth absorbing, but digital resources have also made many properly peer reviewed academic publications available online, both new publications and old classics. The publication industry itself has changed with online publication, self-publication and print-on-demand options. Approaches to archaeology have also changed, with more emphasis now on model methodologies and theoretical approaches. On a practical level, nearly all the images in the original edition of this book have been lost. This posed few problems for most of the line drawings, which were re-scanned, but it did mean that almost all the photographs had to be replaced from different sources. The editors of the first edition observed that ‘there has been an increase in international interest in Afghanistan’ since the Second World War. This gained momentum, first, after the Soviet intervention in 1979, and then with interna-

introduction to the new edition

tional interest reinforced by the US-led intervention in 2001. The latter was, in part, a response to the destruction of the great Buddhas at Bamiyan, which had focused attention on the threats to Afghanistan’s cultural heritage, already highlighted by the looting and destruction of the National Museum during the previous decade. As a consequence, Afghanistan’s cultural wealth had been regarded as lost, but then in 2004 it was revealed that Museum staff had secretly hidden its greatest treasures away. The highlight of these was the Tillya Tepe gold artefacts, but they also included other treasures such as the Begram ivories and glass, the Fullol hoard, objects from Aï Khanoum and much else. The resulting exhibition showcasing Afghanistan’s cultural riches toured the world between 2007 and 2014, attracting huge public attention. In its wake came numerous scholarly publications, international conferences and cultural initiatives either focusing directly on Afghanistan or relating to it. The Balkh Art and Cultural Heritage Project initiated at Oxford University in 2011, the Oriental Institute–National Museum Partnership Project initiated in Kabul and Chicago in 2014, the European Society for Afghan Studies initiated in St Petersburg in 2017 and the establishment of a new international scholarly journal, Afghanistan, in 20181 is a small sample of the many and increasing number of such initiatives. Clearly, the time is ripe for a new edition of this book. In the original edition the editors also emphasised that ‘[s]urprisingly there has never been a book primarily concerned with the archaeology of Afghanistan’, although noting that Dupree’s pioneering 1973 work ‘had utilised much archaeological evidence’. There are still remarkably few such works, and even fewer covering all periods. The exception is the 1982 Archaeological Gazetteer of Afghanistan (with a revised and extended edition in press at the time of writing),2 although this is a sites and monuments record designed to be consulted rather than read as a narrative. However, there is now a gratifyingly much larger number of works which either incorporate archaeology into more general works or cover specific issues – and Dupree’s seminal work (updated to 1980) still remains valid.3 In addition, major fieldwork projects have been published, many of them of work carried out up to fifty years ago. These include the ongoing Aï Khanoum publications, the excavations at Kandahar, Shortughaï, Surkh Kotal, Lashkari Bazar and Tillya Tepe, and surveys in Eastern Bactria and the Hindu Kush – the list is by no means exhaustive. Important works of synthesis and discussion have also appeared on archaeology, art, architecture, numismatics, epigraphy, religion, prehistory, Graeco-Bactria, the Kushan, Ghaznavid, Ghurid and Timurid dynasties, and historical studies generally, to list just some of the main areas. The discovery in 1978 by Viktor Sarianidi of the Afghan–Soviet Mission of six burials containing some 20,000 gold objects was too late for inclusion in the first edition, which covered the history of archaeology in Afghanistan up to the mid1970s. Tillya Tepe was probably the most spectacular archaeological discovery of a gold treasure in the later twentieth century and has important ramifications not only for the art of Afghanistan but for Inner Asia more widely; the treasure formed

11

12

THE ARCHAEOLOGY OF AFGHANISTAN

the highlight of the touring exhibition from the National Museum mentioned above. Another major discovery too late for the first edition (apart from a brief report) was the Harappan site of Shortughaï on the Aï Khanoum plain, excavated between 1976 and 1979 by Henri-Paul Francfort. This, together with newly available data in former Soviet Central Asia, led Francfort to propose a major reappraisal of the Central Asian Bronze Age as a broader ‘Oxus Civilisation’, a term that has now gained wide acceptance. The discovery of Shortughaï was itself a result of a survey of the plain initiated by the late Jean-Claude Gardin to contextualise Aï Khanoum, which soon expanded into a broader survey of Eastern Bactria between 1974 and 1978. Francfort has written a summary of the Shortughaï excavations in Chapter 3 of the new edition, together with an overview of the Oxus Civilisation, and Bertille Lyonnet, Gardin’s co-investigator, has written a summary of the Eastern Bactria survey for the same chapter. Indeed, studies and interpretations of the Bronze Age in Afghanistan and surrounding regions have probably changed the most since this book was published. Hence, we decided to divide the original chapter that included the later prehistory into two, Chapters 3 and 4, north and south of the Hindu Kush, centred on the Oxus and Helmand Rivers respectively, albeit naturally with considerable overlap and interrelationships. Discussions of the external relationships of these chapters, in particular relating to new discoveries in Pakistan and eastern Iran, have been updated by Cameron Petrie. In the historic periods, too, more recent excavations have added considerably to our overall picture. The reopening of Afghanistan to international researchers in the early 2000s – particularly the reopening of the DAFA in 2002 together with the return of Italian, German, Japanese, US and other missions – initiated a new era in archaeological research. Between 2002 and 2004 the Istituto Italiano per l’Africa e l’Oriente (IsIAO, formerly IsMEO), originally under the late Maurizio Taddei, returned to Ghazni and Tepe Sardar for conservation, reappraisal and some limited soundings. In 2003 Zemaryalai Tarzi, as part of a joint French Ministry of Foreign Affairs and National Geographic mission, returned to Bamiyan to carry out new excavations through to 2008. From 2004 until 2008 a new Mission Archéologique Française de Bactriane Afghane (MAFBA, with the support of the DAFA) led by the late Roland Besenval and succeeded by Philippe Marquis carried out investigations on the Balkh Plain. These included: new excavations at Tepe Zargaran at Balkh revealing Classical remains there for the first time, confirming Balkh as the capital of Greek Bactria; preservation and new studies of the Mosque of Noh Gunbad carried out by the late Chahryar Adle; and new excavations upstream from Balkh at the major Achaemenid urban site of Chashma-i Shafa, remains of which included a fire altar that presents major implications for the interpretation of early Zoroastrianism. In 2005 the Deutsche Archäologische Institut (DAI) and the DAFA initiated new excavations and architectural studies at the Kuhandizh and citadel mounds of Herat, led jointly by Ute Franke of the DAI together with Besenval and Marquis,

introduction to the new edition

which for the first time confirmed an Iron Age date for Herat. At the same time the DAI carried out the first properly comprehensive survey of the Herat region (the results of which are still in preparation at the time of going to press). This comes as a welcome ‘reorientation’ of the archaeology of Afghanistan to the west, long neglected by studies focusing on the north and east. As part of this reorientation, in 2003 and 2005 David Thomas led an IsAIO–UNESCO team that carried out detailed investigations – including soundings – around the Minaret of Jam, confirming for the first time extensive urban remains, contradicting the isolated impression that the minaret had previously given. Dramatic new evidence for the extent and survival of Buddhism has been revealed by new and ongoing excavations in and around Kabul. Threats posed by copper mining at the site of Mes Aynak south-east of Kabul prompted international rescue efforts under the auspices of the Afghan Institute of Archaeology from 2009 to 2014 that uncovered extensive Buddhist monastic remains, substantial Buddhist sculpture and paintings, and gold and semi-precious objects. In the hills overlooking Kabul itself, excavations begun in 2004 by Zafar Paiman of the Afghan Institute of Archaeology at a number of Buddhist sites are ongoing. The main site excavated was at Tepe Narenj, an extensive monumental Buddhist site on a series of artificial terraces that recalled Surkh Kotal, but more monumental Buddhist remains were uncovered by Paiman at Qol-e Tut and elsewhere. The excavations both here and at Mes Aynak confirmed, first, that Kabul was a major Buddhist religious centre on a par with Bamiyan and other such centres, and second, that Buddhism continued to flourish in Afghanistan until well into the Islamic era, as late as the eleventh century. Extensive programmes of conservation and preservation have been carried out at monuments throughout the country by the Afghan Cultural Heritage Consulting Organization (ACHCO), the Aga Khan Trust for Culture (AKTC) and other international organisations. The evidence from new excavations has been complemented by a series of other dramatic discoveries, often accidental, since this book was published. In 1978, while searching for a Sasanian-style rock relief in north-western Afghanistan that was first reported in the nineteenth century, Jonathan Lee discovered a cave painting of the same period, one of the most important additions to the corpus of Central Asian painting in fifty years. The purported relief still remains elusive, but in 2003 Lee did discover a Sasanian rock relief depicting Shapur I at Rag-i Bibi near Pul-i Khumri in eastern Afghanistan, thousands of kilometres further east of where this art form is otherwise known. From the same region, the discovery of the Bactrian Rabatak inscription has shed new information on the genealogy of Kanishka, while a vast archive of Bactrian documents provides incomparable new historical information on much of the first millennium ad. The study of Bactrian epigraphy has thus been revolutionised, and Nicholas Sims-Williams, who has been to the forefront in publishing new Bactrian material, has written a new section on Bactrian epigraphy in Chapter 6. There have been discoveries since of more paintings at Chehel Burj in Bamiyan Province, another Greek

13

14

THE ARCHAEOLOGY OF AFGHANISTAN

inscription from Kandahar, and another Bactrian inscription at Tang-i Safidak; Aramaic documents, probably from Balkh, provide information on the last days of the Achaemenids and Alexander’s invasion; an extensive new Buddhist monastic ‘city’ at Kharwar north-east of Ghazni has been surveyed and documented; the Smithsonian Helmand-Sistan surveys undertaken in the 1970s are, at the time of going to press, being prepared for publication, detailing new information on hitherto unknown Hellenistic and Buddhist remains there, supplemented by a new US survey on the Helmand in 2011. An astonishing new treasure estimated to weigh up to four tonnes of statuettes, jewellery, gold strips and some 550,000 coins was discovered in the 1990s at Mir Zakah, adding to the approximately 11,000 coins discovered there in 1947. Perhaps no other field in the archaeology of Afghanistan has changed so much since the first edition than the field of numismatics, and new sections have been written for the new edition by Simon Glenn on Graeco-Bactrian coins in Chapter 5 and by Olivier Bordeaux on Kushan coins in Chapter 6. There have also been radical revisions to our understanding of Hunnic coins – previously lumped under the all-embracing ‘Hephthalite’ term.4 New discoveries, new discussions and new interpretations are an ongoing process. In the original introduction to this book the editors appropriately wrote: ‘one might almost say that for the first time there is more cheese than holes!’ To take the metaphor further, we now have even more cheese – but the cheese has grown considerably larger! The discoveries of the past forty years since the first edition merely foreshadow equally great discoveries in the future. Discoveries such as the Rag-i Bibi rock relief or the Bactrian documents or the Tillya Tepe treasure (to cite a few at random) merely hint at what still might lie hidden in the remote hills and underneath the ground of Afghanistan. Major publications still appearing as this book goes to press will undoubtedly soon outdate much that is new in these pages. This edition, therefore, like the first, can only be regarded as interim: the adventure continues.

CHAPTER 1

The Geographical Background Sophia R. Bowlby and Kevin H. White

Afghanistan lies at the heart of Eurasia within the vast belt of mountain range, steppe and semi-desert that stretches from the Mediterranean to China (Fig. 1.1). Routes from north to south, west to east, criss-cross the country and through the centuries have carried in a diversity of peoples – Mongols, Greeks, Indians and Iranians during the historic period alone. Many of these different groups can still be distinguished in the population today. This variety of peoples is easily matched by the variety of landscapes, which range from the icy pinnacles of the Hindu Kush to the blazing wastes of the Dasht-i Margo, from dry, dusty steppe-lands to oases and terraced valleys (e.g. Figs 1.8–1.17). Given this variety of environment and peoples, there are many forms of economy that the early inhabitants may have adopted. The purpose of this chapter is to reconstruct the nature and distribution of resources available to these settlers and to outline the range of economic alternatives open to them. Any such attempt at reconstruction must consider what resources were relevant to the peoples of the past. The archaeology of Afghanistan covers a vast span of time and a great variety of human groups, some with primitive and some with fairly advanced technology. However, most of them had to gain the bulk of their food and their materials for tools, shelter and clothing from their immediate locality, so that the distribution of natural resources must have been of great significance to their livelihood. Perhaps the major influence on the distribution of natural resources within an area is its climate. The climate of Afghanistan today certainly has a profound effect on its agricultural possibilities, natural vegetation and wild life. Thus, one of the first questions that needs to be answered is whether there have been significant changes in climate during the time period of interest here – namely from around 70000 bc to the sixteenth century ad. Before dealing with this question, however, it is pertinent to describe briefly the major physiographic features of Afghanistan, since these play an important part in its past and present climatic character. The whole country is part of the belt of recently folded mountains that stretches from the Pyrenees to the Himalayas and beyond.1 The Tethys Sea once lay between

16

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.1  The location of Afghanistan within Eurasia.

a large, stable continental block to the south (of which India and Africa were a part) and a similar stable block to the north. As time passed, the older rocks on the bottom of the Tethys Sea became covered over with sediments carried off the neighbouring land masses. Then, during the late Mesozoic and early Tertiary, the whole sea bed was folded upwards against the continental blocks to north and south, so forming a massive system of mountain chains. These chains tend to be composed of series of parallel ranges and their directions were partly determined by the outlines of the resistant continental blocks. The mountain-building a­ ctivity

the geographical background

Figure 1.2  Afghanistan: relief and drainage, showing major rivers.

was accompanied and followed by volcanic activity, rifting and fracturing. Rivers and weathering agents have further altered the landscape. In the Afghan area today, the Hindu Kush range sweeps south-west from the heights of the Pamir Knot, with Nuristan to the south and Badakshan to the north, to become the west–east spine of a complex, central mountainous area. This mountainous core of the country is divided north–south by the westward flow of the Hari Rud which flows 1,100 km from its source west of Kabul, through Herat, to lose itself in the deserts of Turkmenistan. To the south of the Hari Rud the mountains slope gradually downwards to the west, south-west and south until they vanish into the Helmand Basin (Fig. 1.2). To the north of the Hari Rud the mountains lose height northward towards the valley of the Amu Darya and westward towards the border. Within the highland area there are down-faulted basins, and the Helmand Valley forms a great intermontane basin between the Afghan mountains and those of Pakistan and Iran.

17

18

THE ARCHAEOLOGY OF AFGHANISTAN

The core of the mountainous area, the Hindu Kush around Kabul, is formed of a complex of older crystalline rocks with metamorphics.2 In Nuristan, to the northeast of Kabul, are found gneisses, schists and granites, and these crystalline rocks can be traced to the south-west. The part of the Hindu Kush running westwards from Kabul and ending near Herat is also geologically complex, with old crystalline and sedimentary rocks. The crystalline rocks of these two mountain branches are bordered by large areas of Mesozoic rocks, which comprise most of the remaining area of the northern mountain ranges. These rocks are largely sedimentary; for example, much of the northern foothills area is composed of limestone. The small basins within the highland area are often filled with late Tertiary sediments. Tertiary rocks are found to north and south of the Mesozoic highland area. In the north these do not cover a very extensive area, but in the south they form a large part of the high plateau between the southern Hindu Kush and the second main range of mountains, the Sulaimans. The Sulaimans are mostly outside the present boundaries of Afghanistan, but, although they are not as high as the main Hindu Kush, their strong folding has created a formidable mountain barrier between Afghanistan and the Indus valley that is only breached by a few principal passes. They are largely composed of Mesozoic and Tertiary rocks. The dry areas of the Helmand Basin and the lowlands along the Amu Darya are covered mostly with young Quaternary sediments and in the Helmand the Kuh-i Sultan is a reminder of the volcanic activity that is found along the lines of mountain building. Climate and Climatic Change Information on the present climate of Afghanistan has improved.3 Nevertheless, climatic information is still based on few weather stations relative to the size of the country, so local detail is limited. The climate is generally arid, but it is also strongly influenced by relief. As is shown in Figure 1.3, in much of the country, with the exception of the high mountains, the mean annual precipitation does not exceed 400 mm per year and in the south-western corner of the country true desert conditions prevail. The larger part of the lowlands fringing the mountains is semi-desert or steppe lands, with precipitation of 20–30 cm per year and scant vegetation except along the major rivers. These rivers have their origins in the mountains. In most areas the precipitation falls between October and April, with most rainfall in the spring – in the higher areas, much of it as snow. During the spring and early summer the melting snow feeds the rivers, so that these have their highest flow during periods with little rain; this seasonal peaking of flow often results in severe flooding and erosion, especially in the central high mountains and northern area.4 There are a few areas within Afghanistan with rather higher rainfall – notably the valleys on the south-east borders that open into the main Indus Valley, for the Indian monsoon usually reaches these areas, bringing some slight rainfall in early summer. The most famous of these valleys is the lower Kabul Valley around

the geographical background

Figure 1.3  Average precipitation and cultivated land.

Jalalabad, but even here irrigation is a vital part of agriculture, and Michel suggested that the best climatic analogue is the Imperial Valley in the south-western USA.5 The effectiveness of any rainfall that does arrive during the summer months is lessened by evaporation. Mean July temperatures are around 32˚C in the northern plains, 24˚C in the central highlands and 28˚C in the south-western lowlands.6 However, these hot summers are followed by very cold winters – for January, mean temperatures range from around +3˚C in the northern plains to –4˚C in the central highlands and +2˚C in the south-western lowlands (Fig 1.4).7 In the summer the snowline is around 4,500–5,000 metres, and in the winter 1,800 metres. Thus, Afghanistan has a dry continental climate, with altitude playing an important part in the spatial pattern of temperature and precipitation. Figure 1.5 shows that the majority of Afghanistan is classified as having hot or cold desert, semi-desert or steppe conditions. On the whole, winds are not abnormally strong except in the south-west and west where severe winter blizzards may occur and where, in the summer, a hot dry wind blows strongly from the north-west. Known locally as the bad-o sad-o bist ruz or ‘Wind of 120 Days’ it may reach over 100 knots and blows from July to September, whipping away loose topsoil and spreading and shifting the desert sands. In most parts of Afghanistan an important limit

19

20

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.4  Average temperature by season.

to the productivity of agriculture is the water supply available during the growing season, and high yields depend upon the possibility of irrigation (Fig 1.6). Moreover, the natural vegetation and fauna are also strongly influenced by the amount and incidence of precipitation. As regards the concerns of this book, it is important to establish whether these conditions prevailed through the archaeological period, and if so to what extent. There are few data collected within Afghanistan from which assessments of the earlier climatic conditions might be made. Most of the work on climatic change in prehistory in the Old World has been done on North Africa, Europe and the Middle East. These studies can be used to present a general picture of the possible changes in climate in the prehistoric period within Afghanistan, but further research, especially more locally specific research, is likely to modify this. Thus, only a broad sketch of the possibilities is presented here. Current knowledge suggests that the global climate trend during the Quaternary was characterised by an overall cooling, punctuated by dramatic and relatively rapid transitions from interglacial to glacial and interstadial to stadial conditions. These changes are recorded in Oxygen-isotope records in oceanic sediments and ice cores. The causes of these dramatic climate oscillations are complex. They are

Figure 1.5  Climatic regions.

22

THE ARCHAEOLOGY OF AFGHANISTAN

driven by variations in our orbit round the Sun, characterised by cycles of 100 ka, 40 ka and 23 ka, related to changes in eccentricity, obliquity and precession, respectively. Other millennial-scale oscillations result in slow cooling phases at the beginning of a stadial and fast temperature rises at the start of an interstadial. These features are termed Dansgaard–Oeschger Events8 and they have been attributed to deviations in ocean surface currents, surges of ice sheets, variations in sunspot activity or instabilities in the atmospheric carbon dioxide system. While the broad patterns may be identified, the exact spatial extent of these, and other similar climate perturbations, such as the Heinrich Events or the 8.2 ka event of rapid cooling,9 remains unclear. In particular, our understanding of Quaternary climate change in central Asia is very limited compared to our understanding regarding other parts of the globe, and many more climate records are required if we are to build up a coherent spatial pattern of these changes.10 What is currently known of past climate change in West Asia suggests that its climate system is governed by complex interactions between the mid-latitude Westerlies, the Siberian Anticyclone and the Indian Ocean Summer Monsoon.11 There is increasing evidence of an atmospheric teleconnection between the climate of the North Atlantic and the interior of West Asia via expansion of the Siberian Anticyclone.12 For example, arid climate phases recorded in peat sediments around Lake Neor in Iran correlate with Holocene ice-rafted debris events from North Atlantic sediments,13 which are also associated with intervals of reduced solar irradiance.14 These correlations indicate a climate coupling between West Asia and the North Atlantic.15 Evidence that allows us to tell a reliable story about changes in climate in West Asia during the Pleistocene is limited. However, data from Iran suggest that the Pliocene to Lower Pleistocene climate there was more humid than the presentday climate,16 resulting in highstands in lake basins such as Qom Playa and the Lut Basin.17 Lacustrine sediments deposited during these lake highstands have subsequently been eroded during more arid conditions. By the Middle Pleistocene, more arid conditions are thought to have formed thick accumulations of loess (a sedimentary deposit formed of wind-blown dust). These are usually interpreted as indicative of dry and cold conditions with sparse vegetation cover. However, these sedimentary sequences of loess are often punctuated by palaeosol horizons formed by weathering and pedogenesis under wetter and warmer conditions, suggesting fluctuations in the dry and cold conditions. On the basis of the extent of playas, Krinsley postulated that the Lower to Middle Pleistocene climate of the Iranian Plateau was probably cooler and, because of reduced evaporation, moister than today.18 The last Glacial/Interglacial cycle before the Holocene is recorded in a sediment core taken from Lake Urmia in Iran19 and the loess record of northern Iran.20 According to the pollen record from Lake Urmia, the last interglacial (Oxygen Isotope Stage 5e 124–119 ka) was slightly warmer and moister than the Holocene. As conditions deteriorated towards the Last Glacial Maximum (25 ka) evidence

the geographical background

from terrestrial sedimentary records indicates much drier, cooler conditions, which prevailed into the Lower Holocene.21 Kehl reports some estimates of the temperature depressions in Iran during the Last Glacial Maximum22 compared to present-day data, which range from 5°C in the Alborz and Zagros Mountains23 and 5–8°C in central Iran24 to up to 8–10°C for southern Iran.25 These cooler temperatures may help to explain the presence of lake highstands during the last glacial, due to reduced evaporation more than offsetting the lower precipitation amounts.26 Low temperatures would also result in lower vegetation cover, and hence lower evapotranspiration, leading to increased stream-flow into the lakes.27 In general, most recent palaeoenvironmental evidence from Iran indicates the alternation of dry and cold climatic conditions during the stadials, with wetter, warmer conditions during the interglacials and interstadials. This conflicts with earlier ideas, which assumed higher precipitation rates during the glacials than today.28 Vita-Finzi proposed increased cyclonic precipitation from 50 to 6 ka and the Little Ice Age (ad 1550–1850), on the basis of his interpretation of fluvial sedimentary records.29 In the Upper Pleistocene and Holocene, the Middle East and North Africa experienced periods of wetter climate.30 A Pleistocene wet phase may also have occurred in Afghanistan.31 These wet phases were probably caused by northward shifts of the Indian and African Monsoons.32 In contrast, Stevens et al. examined fossil and geochemical records from Lake Mirabad in western Iran and concluded that here the early Holocene was dry.33 Pollen and microfossil evidence from lakes in Turkey, Iran and Georgia also suggests that the early to middle Holocene was characterised by a dry climate.34 However, in the Holocene it becomes increasingly difficult to disentangle human versus climate impact, so that local variations may have occurred because of human activities.35 There is some evidence that periods of drier climate, which coincided with the 8.2 and 4.2 ka climate events, also coincided with the breakdown of agriculturebased societies in western Asia.36 Nearly all episodes of enhanced dust deposition in the Lake Neor peat deposits found by Sharifi et al. correspond with episodes of drought and famine in the region,37 and transitions in major Mesopotamian and Iranian civilisations, such as the fall of the Akkadian empire around 4200 ka (2150 bc). They conclude that ancient human societies in the region were probably susceptible to abrupt climate variability during the Holocene. The extent to which human activities contributed to Holocene environmental change is still very unclear due to the paucity of detailed, well-dated palaeoclimate climate records. Overall, the evidence points to several Quaternary climatic changes in the region, but there is considerable uncertainty over the detail and timing of these. Most evidence indicates that glacial stages correspond to drier and colder conditions, with wetter and warmer conditions returning in interglacials. The Upper Pleistocene and Lower Holocene experienced periodic periods of wetter conditions, but the data are still much too sparse to allow us to reconstruct the spatial extent of many of these climate changes.38

23

24

THE ARCHAEOLOGY OF AFGHANISTAN

Natural Resources for Subsistence The evidence on climatic change suggests that, at least since 500 bc, there have been no major variations in climate, although small fluctuations in aridity or temperature may have occurred. It is also probable that human activity impacted local climate conditions, but this can only be speculation in the face of very limited archaeological and climatic data. Thus we can gain a considerable understanding of the resources available in the recent past by examining the present situation. For the earlier periods we may perhaps take the drier, colder or wetter areas of Afghanistan as indications of conditions that may have been more widespread during drier, colder or wetter periods, respectively.

Figure 1.6  The physiographic regions of Afghanistan.

the geographical background

Afghanistan has seen a prolonged period of fighting and unrest since 1978. Four decades of war have meant that little research on contemporary Afghan patterns of life, especially in rural areas, is available. However, for the purposes of this chapter this lack is not especially important since there are three valuable sources of information on patterns of life in different rural areas of Afghanistan from the late 1950s to the 1970s which can be used to suggest the variety of viable livelihoods that are possible without extensive use of industrial products and methods. The Danish geographer Johannes Humlum produced a regional geography of Afghanistan in 1959 which provides an extremely thorough and comprehensive treatment of the economic and social geography of the country at that period.39 In 1973, Louis Dupree published a book exploring the history and society of Afghanistan (reissued and updated in 1980 and 1997) which also provides an extensive coverage of data on Afghanistan in the late 1960s.40 This work was later supplemented by the detailed Atlas of Indigenous Domestic Architecture by Albert Szabo and Thomas Barfield, published in 1991 but drawing on data collected between 1974 and 1976.41 Afghanistan was then, and still largely is, a society in which peasant agriculture, nomadism and traditional methods of providing shelter were dominant in rural areas. For those interested in the possible ways of life followed by people in prehistory in Afghanistan, Dupree, and especially Humlum and Szabo and Barfield, provide valuable information about the ways in which people in the 1950s and 1960s used the resources of different local environments to create viable local economies and societies without extensive use of new, industrial materials or techniques. Much of the following information is derived from their works. Before discussing the features of life in different regions of Afghanistan it is important to emphasise that each region is not isolated from others or from contacts with societies beyond its frontiers: Afghanistan lies on the confluence of several trade routes of great importance in the historic period and probably in the prehistoric period as well. Thus, there has been a long history of trade in goods, techniques and ideas. Both Humlum and Dupree suggested rather similar divisions of the country into regions based largely on criteria of climate and physiography. For the purposes of this discussion a modification of Dupree’s categorisation has been adopted. This is shown in Fig. 1.6 and a generalised map of natural vegetation is given in Fig. 1.7. The thirteen contrasting environmental areas shown in Fig. 1.6 have been grouped into five main types, distinguished on the map by shading. First are the high mountains of Badakhshan, Nuristan, the Wakhan Corridor and the central mountains. Next are the northern and southern mountains and foothills which border the central mountains. The Turkestan Plains, the Herat-Farah Lowlands and the two major river valleys follow, and last come the deserts to north and south of the Helmand.

25

26

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.7  Natural vegetation: 1. Glaciers or permanent snow. 2. High altitude steppes. 3. High altitude grassland. 4. Wooded steppes of North Afghanistan with gallery forest along watercourses. 5. Wooded steppes of South Afghanistan with gallery forest along watercourses. 6. Wooded sub-tropical steppes with gallery forest along watercourses. 7. Sub-tropical steppes with Salsolaceae. 8. Temperate steppes with Salsolaceae. 9. Temperate and sub-tropical steppes. 10. Tamarisk and other gallery forest. 11. Steppes with juniper thickets. 12. Deciduous forest. 13. Coniferous forest. 14. Irrigated areas.

The High Mountains Badakhshan and Nuristan are mountainous areas lying to the west of the main block of the Pamirs. Nowadays, international boundaries divide the Pamirs between Tajikistan, China, Pakistan and Afghanistan, but these boundaries do not follow any marked natural features, and the Wakan Corridor and Pamir Knot should be considered as part of a larger ‘Pamir region’ (Fig. 1.8).

the geographical background

Figure 1.8  The Pamir Mountains in Wakhan.

The Pamir area lies at very high elevations – most of it is above 3,000 m with individual peaks reaching up to 7,620 m. Perpetual snow covers land above 5,000 m and travel in the valleys is hindered by glaciers and steep and narrow gorges. The Wakhan, Badakhshan and Nuristan lie at slightly lower elevations and form a mountainous link between the Pamirs and the central mountains. Summer passes lead from what is now Pakistan and from the Inner Asian steppes through Badakhshan, across the Pamirs to the Tarim Basin and China. Throughout these areas the climate is extreme, with winter temperatures well below freezing, but daytime summer temperatures sometimes reach 26°C or so. The climate is greatly affected by elevation, so that there is considerable local variety in the habitat. The mountains are geologically complex with a great diversity of rocks which have been subjected to strong movements and pressures during the events of the orogeny that formed the Himalayas. Although there are local variations depending on rock type, the soils are predominantly thin, but in some valleys there is richer alluvial material originally deposited in the glacial lakes. This will support good pasture or crops in the lower valleys. The area of the central mountains is mostly at altitudes above 2,700 m with the higher peaks ranging from 4,270 to 4,180 m. The whole range is higher in the east than in the west. There are a number of passes over the mountains, the two most important being the Shibar and Salang (Fig. 1.9). Again, the local climate is

27

28

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.9  The north side of the Salang Pass.

strongly influenced by altitude and aspect, which affects not only temperature but also rainfall. Although this area gives rise to three major Afghan rivers – the Kabul, the Helmand-Arghandab and the Hari-Rud – it does not have a high rainfall and the natural vegetation is generally scant and low. Table 1.1, taken from Dupree, gives an indication of the natural vegetation zones within the central mountains and Nuristan.42 In general, this chart would also apply to Badakshan and the Pamirs. The forest zones offer food plants for collection and animals for hunting. In the Panjshir valley the forests were cut down during the pre-Islamic period (Fig. 1.10).43 At the lower elevations cleared land could be used for crop cultivation. The valley floors and riverbanks also offer good land for cultivation as well as opportunities for hunting and gathering. At the higher elevations there is good pasture during the summer months and this whole region is one where herding and transhumance are well-adapted to the natural conditions. Throughout the mountain zone there are places with mineral deposits of greater and lesser extent. Some of these were certainly of potential value to prehistoric humans. For example, iron ore is found in the Devonian and Permian rocks of the north-east, especially around Kabul, as well as the upper Farah Rud region in the west (Fig. 1.11); lead is also found in the north-east, as is copper. The exploitation of the copper deposits at Mes Aynak, one of the largest in the world, has achieved

the geographical background

Table 1.1  Natural vegetation in the Central mountain zone and Nuristan Altitude

Characteristics

Above 4270 m 3660–4270 m 3050–3660 m

None Mountain meadows of short grasses and seasonally flowering plants. Mountain scrub, grasses and seasonally flowering plants, small scattered bushes (juniper, dwarf, willow, rosebay, tragacanth, euphorbia). Dry scrub semidesert bushy plants (feather grass, wormwood, saltword, tragacanth, camel grass, tamarisk) and scattered clumps of pistachio trees. Conifer forests of pine, cedar, fir, larch and yew with a few broadleafed trees (willow, poplar); ivy found only in Nuristan.

Up to 3048 m Forest zones of Nuristan and Paktyaa 2438 or 2743 m to 3048–3352 m 1371–1528 m to 2438–2743 m Valley floors and river banks

Bushes and broadleafed forests of oak (including holly oak) with well developed undergrowth, and some walnut, alder, ash, juniper; above 1540–1828 m conifers included to form mixed forests. Plane trees, poplar, willow and mulberry thickets; much bush growth where land not cultivated.

Distinctive floral zones. Chart highly schematic. Because of the varied terrain within each zone, many local variations occur.

a

Figure 1.10  The Panjshir Valley.

29

30

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.11  The Farah Rud, with the discolouration in the hills caused by iron ore deposits.

Figure 1.12  The Ghorband Valley.

the geographical background

major publicity due to the consequent threats to the archaeological site there (see Chapter 6). Precious metals and stones – silver, jade, tourmaline and ruby – are available in many parts of the mountains, but the most notable resource that we know mined by prehistoric peoples is the lapis lazuli of Badakhshan (see below). The Mountains and Foothills The mountains and foothills to the south of the Hindu Kush slope gently from Kabul in the north-east towards the south-west (Fig. 1.14). They form a transitional zone of plateaux between the high mountains and the lower dry uplands bordering the Indus valley. In the Kabul area there are several broad mountain valleys – notably the Kabul valley, the Kohistan-Panjshir valley and the Ghorband Valley (Figs 1.10 and 1.12) – that have long been important inhabited areas. On the south-eastern boundary of the Kabul mountains, the Kabul River cuts a deep gorge through the mountains (Fig. 1.13) and flows into the open Jalalabad valley. Similar low-lying valleys, cut into the mountains bordering the Indus valley, are found further south. During the winter, snow lies at 1,800 m and above, and the temperatures throughout the region are generally below freezing, with high winds increasing

Figure 1.13  The Kabul River Gorge.

31

32

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.14  The southern foothills near Ghazni.

the chill factor. In the summer, daytime temperatures rise to 13–18°C below 2,100 m. Jalalabad is considerably warmer; average July temperatures are around 33°C and in winter they fall to around 7°C. Temperatures generally become higher as one moves south-westwards (Fig. 1.14), although winter temperatures may well be near freezing even at Kandahar. Rainfall also declines to the south-west as one moves away from the central mountains and the steppe vegetation merges into semi-desert. Where the rivers emerge from the Hindu Kush there are opportunities for irrigated agriculture and more abundant pasturage. The whole region is suitable for herding activities and the different altitudinal zones allow a variety of foods to be collected or produced. The broad mountain valleys of the north-east offer a favourable environment for dry or irrigated farming and quick access to the resources of a variety of natural environments (Fig. 1.15). The northern foothills and mountains form a rather desolate area of plateaux and rounded hills with low steppe vegetation. The rocks are generally limestone, sandstone or shale and support only thin, stony soils. However, in the lower valleys flood deposits and loess form richer soils. The Andarab and Surkhab Rivers combine in the Qunduz to cut through the main east–west mountains in narrow gorges but form wide, broad valleys to north and south and provide a routeway to the Shibar pass across the central mountains. As in the rest of Afghanistan, the

the geographical background

Figure 1.15  The Balkh River in the mountains south of Balkh.

climate is dry and temperatures show a great range between winter and summer. The most favourable habitats for humans are the river valleys, where the combination of better soils and the availability of water provides a richer flora and fauna and better prospects for cultivation (Fig. 1.15). The higher areas, however, still provide grazing, and a poor living could be made by dry farming and herding or hunting and gathering. In both the northern and southern mountains and foothills, and especially in the Kabul area, there are deposits of iron, copper and other minerals. Moreover, these areas are generally within reach of the minerals of the high mountains and the salt deposits and gypsum of the plains and lowlands. The Plains and Lowlands To the north and west the mountains and foothills give way to low, dry plains. The Turkestan plains are crossed by a number of rivers which eventually die out in the sands. However, loess blown off the Central Asian steppes is deposited in many areas of the plains and near the mountains and, with irrigation, forms a most fertile soil and easily worked building material. As one moves away from the mountains the stony soils give way to sandy desert and in some areas there are scattered dunes and dune fields. The climate is very variable, open as this region is

33

34

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.16 The Herat-Farah lowlands between Adraskand and Farah Rud.

to invasion by air masses from the north and west (Fig. 1.17). In the winter freezing weather is common but may be broken by interludes with temperatures around 18°C. Summer daytime temperatures can be extremely high (40°C) but at night fall abruptly. The rainfall is low and evaporation loss is high. The Herat-Farah Lowlands offer a slightly pleasanter environment. This area is a continuation of the Iranian plateau and the terrain is generally of low hills and broad basins and valleys (Fig. 1.16). Rainfall is low (about 18 cm per annum in Herat) and the temperatures range from night frosts in winter to fierce daytime temperatures in summer (sometimes 45°C). The hill soils are generally poor and stony and to the west there are areas of salt pans and mudflats. In both these areas there is pasture for animals, though it is hardly plentiful; only around the rivers is there more abundant vegetation and good conditions for cultivation with the aid of irrigation. The Amu Darya, Oxus and Helmand-Sistan Regions These two great rivers were used extensively for irrigation in the historic period and probably earlier. Both drain into inland seas and both are subject naturally to severe seasonal flooding. The floodplain along the lower course of the Amu Darya in Afghanistan is frequently marshy and separated from the desert by ill-drained alluvial terraces (Fig 1.18). In some areas there are salt flats. The Sistan Basin lies

the geographical background

Figure 1.17 The Balkh Plain near Balkh.

Figure 1.18  The Amu Darya floodplain from Kampyr Tepe in Uzbekistan. The river is in the distance and the northern edge of the Hindu Kush can be seen in the far distance.

35

36

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.19  The Helmand River.

at only 520 m above sea level and is an area of brackish marshes and fluctuating lakes. Much of the basin is covered with clay, silt, sand and gravel deposits from earlier lakes. The valley of the Helmand itself is separated from the deserts to north and south by low bluffs, and, where it is not cultivated, it is covered in low vegetation and tamarisk scrub (Fig. 1.19). The rivers and the lakes of the Sistan basin have fish and wildfowl which could have provided a useful source of food. The rivers themselves can obviously be used in irrigation and the frequent floods have laid down alluvial material that forms good soils. However, the hot dry climate leads to strong evaporation and careless irrigation can result in the formation of infertile saline deposits. The Deserts The stony western desert (Dasht-i Margo) and the sandy southern desert (Registan) lie on either bank of the Helmand. They are both true deserts with very little

the geographical background

Figure 1.20  Giant sand-dunes of the Registan Desert on the edge of the Helmand Valley.

natural vegetation (Fig. 1.20). Camel grass is the most frequent plant found. There are great diurnal ranges of temperature and despite cold nights the daytime summer temperatures reach 45°C or above. In the south there are dunes, which constantly shift over the stony pebbles underneath. In the north there is little sand but vast expanses of black pebbles. At present neither of these deserts offers much sustenance to humans or their livestock, although new investigations based upon satellite imagery have revealed a surprisingly high level of dams, irrigation canals and related hydrological features in the pre-modern period.44 Patterns of Subsistence in the 1950s–1960s Patterns of subsistence naturally differ not only between groups in different environments but between groups with different cultural histories. No attempt will be made here to describe in detail the ethnology of modern Afghanistan. Immediately prior to the Soviet invasion the population was a complex mixture of peoples who had migrated to Afghanistan at different times, and this still remains the case. The two largest population groups are the Pashtuns and Tajiks, who both speak Iranian languages. The Pashtuns are the dominant group south of the Hindu Kush; the Tajik are found mainly in the north-east. The other main Iranian language group are the Farsiwan around Herat and through the west of the country, who speak Persian. The other major linguistic group is Turkic, and Turkic languages

37

38

THE ARCHAEOLOGY OF AFGHANISTAN

are spoken largely by Uzbeks and Turkmen, who are found in the northern plains. However, the Hazara, a people of Mongol descent, from the central mountains, speak Persian. There are a number of other ethnic or linguistic groups, notably the Nuristani, Aimaq, Baluch and Dardic groups of Badakhshan. A detailed discussion of these groups can be found in Dupree.45 Despite this variety of peoples and customs, for the purposes of this chapter a number of major patterns of subsistence can be usefully identified without reference to specific ethnic groups. A high percentage of Afghanistan’s population is rural. Today the World Bank estimates it is about 73%. In the 1960s it was estimated to be considerably higher, at 90% of the population, with about 15% of people living nomadic or seminomadic lives.46 Fortunately for the archaeologist, in much of Humlum’s book, using data from the 1950s, he describes methods of subsistence that are probably comparable with those used in the late prehistoric period. Moreover, since there is a considerable diversity in the ways of life of the people, including among them nomads, hunters and gatherers, farmers and urban dwellers, it is possible to find many modern analogies for ways of life that may have existed in prehistory. The earlier section on climate made it clear than in Afghanistan the supply of water is the most critical limiting factor in the production of food. This applies not only to agriculture but also to the possibilities for gaining food by herding or hunting and gathering, since plants and animals also must have water. Agriculture As stated above, in the 1960s some 75% of Afghanistan’s population were sedentary rural dwellers and most of these gained their subsistence through cultivation. The major cultivated areas were and are largely in the plains and lowlands or mountain and foothill zones surrounding the mountainous core and are mostly situated along rivers or at oases (Figs 1.5 and 1.21). The conditions for agriculture in these areas are broadly similar – the most important condition being the need for irrigation. Most sedentary agriculturalists in Afghanistan use some method of irrigation and Humlum describes two principal traditional methods. The qanat or karez system is widespread to the south and west of the central mountains. Karez are long underground tunnels which tap underground water near the foothills of the mountains and channel it down to the oases.47 Some karez extend for more than 30 km although 8–16 km is more common. The tunnels are dug out by hand using simple tools and their lines are marked out on the surface by a series of holes surrounded by mounds of material thrown up from the excavation. Karez are found throughout Iran and also in Pakistan–Baluchistan (Figs 1.22 and 1.23). It has long been suggested that they were developed in Iran in the early first millennium bc at a time of increasing aridity and spread from there during the second half of that millennium.48 However, Magee has criticised the evidence for this dating and suggested that a more probable story is that they were developed towards the end of the second millennium bc in south-east Arabia and sub-

the geographical background

Figure 1.21  Principal types of agricultural irrigation.

sequently spread to south-eastern Iran.49 Dating their earliest use in Afghanistan would contribute to these debates. The technology and tools required to construct karez are simple, relying on gravity and human effort, and their efficacy in bringing water is considerable. Once the water has reached the oasis it is distributed to the fields by means of ditches which are breached and fed into smaller channels when water is required. The second major method of irrigation is to use the rivers. Water is diverted into canals (or juis) by means of small earth and stone dams placed upstream from the cultivated area; the water is then led by gravity to the fields (Fig. 1.24). In the mountains the fields are often elaborately terraced and the water is fed among

39

40

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.22  Cross-section of a qanat system.

Figure 1.23  Aerial view of a typical qanat system in southern Iran.

the geographical background

Figure 1.24  Irrigation canal or jui in the Balkh Plain.

41

42

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.25  Grape field in the Kandahar area. Note the slotted grape drying building in the distance.

them by a complex system of small ditches (Fig. 1.21). This method of irrigation is widespread not only in the major lowland river valleys but also in the smaller mountain valleys; it is found throughout the Himalayan foothills and extends far to the east. It would certainly have been possible for small groups of prehistoric peoples living in river valleys to irrigate in this way. However, in the great desert valley of the Helmand the successful large-scale irrigation that is thought to have once existed would have required co-operation among people living up and down the valley, both to maintain the irrigation works and to administer the distribution of the water. Maintenance would have been important because of the river’s seasonal flooding. In the north the Amu Darya and its tributaries are used extensively for irrigation and this is facilitated in areas with loess since it is a material from which it is easy to construct ditches and dams and when irrigated it forms a fertile soil. The principal crops that were grown on these irrigated lands in the south were wheat, barley, lentils, beans and maize. The first three were harvested in the spring and the last two in the autumn. A variety of other vegetables and some cotton and oil plants were cultivated along with vines, fruit trees and melons. The vines were, and still are, particularly important as grapes form one of Afghanistan’s principal exports. They are grown in trenches to protect them from the heat and to facilitate

the geographical background

Figure 1.26  Traditional ploughing in the Bamiyan region.

watering (Fig. 1.25). In the areas irrigated from the rivers some rice may be grown, and this crop was important in the northern irrigated lands along with cotton, wheat and sugar beet. Vegetables and fruit trees were and are grown in the north as in the south. Lentils, peas and beans were an important food crop throughout Afghanistan, but wheat is undoubtedly the most important single food. Although irrigation is such an important feature of agriculture, dry farming is also possible and is found especially in the mountains. Wheat is grown without irrigation here as a staple food crop and Humlum reports seeing it grown at altitudes of up to 3,000 metres.50 The farmers work with the aid of very simple tools. Humlum gives an excellent description of these along with line drawings and photographs.51 Before ­ploughing,

43

44

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.27  Winnowing in the Kandahar region.

the field could be prepared by pulling over it a heavy piece of wood drawn by oxen to break up lumps of soil. The ploughs were also pulled by oxen and were made of wood with a metal tip (Fig. 1.26). There are a variety of types; none of them turned the sod, instead they merely scratched a shallow furrow. Sowing was usually broadcast and generally a sickle was used for harvesting, while threshing was done by simply marching cattle round and round over the heaped-up grain or occasionally by beating it by hand with a flail. For winnowing the grain was tossed into the air with a wooden fork and then sifted through a simple sieve (Fig. 1.27). These tools, along with spades and shovels, completed the farmers’ most important mechanical equipment. Grain was sometimes milled by hand, but more often this was done in gravity-fed water mills. The fertility of the soil was maintained by leaving fields fallow for one or two years in three. When the fields were fallow, animals were sometimes put to graze on them, thus providing manure that was later ploughed in. The sedentary farmers in the oases, river valleys and mountains also kept livestock; these supplied not only food but also material for clothes, dung for fuel and motive power for transport and farm work. Oxen were kept for ploughing and threshing and were also used as beasts of burden over short distances. They were fed through the winter on lucerne grown for the purpose. Cattle supplied some milk, but this was also obtained from goats, which, along with sheep, fed on the natural scrub vegetation. Sheep were numerically the most important domestic animal in Afghanistan and, particularly in the north, supplied

the geographical background

Figure 1.28  Abandoned cave dwellings in Ghur Province. Note the human figure providing scale.

fine pelts for trading. Throughout the country poultry were found scratching for food around the houses and their eggs and meat supplied a valuable addition to the diet. Sedentary farmers need dwellings. There were and are a wide variety of dwelling types within Afghanistan, reflecting, on the one hand, the agricultural and social traditions of different cultural groups and exchanges of techniques and ideas between them, as well as, on the other hand, the nature of the local environment – most notably, the resources available for construction and the climatic conditions these dwellings need to withstand. Information on dwelling types and construction methods in Afghanistan in the late 1970s, prior to the Soviet invasion, is provided in a scholarly and detailed account by Szabo and Barfield.52 They record ten types of housing used by sedentary peoples in Afghanistan, ranging from caves to four types of curved roof construction and five types of flat roof construction. In their book they note that their maps of the geographical incidence of the different types of dwellings are ‘only approximations derived from previously published maps or descriptions, modified by our own observations and those of researchers we contacted’.53 The brief summaries of their findings given below thus further simplify descriptions that necessarily hide the local complexity of the distribution of dwelling types at this period. Szabo and Barfield report that cave dwellings were found in the Hazarajat and Pamir regions, where they were carved out either in sandstone rocks or in

45

46

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.29  Vaulted mud-brick domestic roofing near Kandahar.

c­ onglomerate rocks (Fig. 1.28). Such dwellings may have been more widespread in earlier historic and prehistoric periods. Indeed, archaeological finds suggest human occupation of rock-shelters during the Palaeolithic in more than one location (see Chapter 2). Curved roof constructions with domes or vaults were reported in the western desert, Herat-Farah lowlands and the Turkestan plain and parts of the northern foothills.54 Curved roofs are a useful form where wood is scarce (Fig. 1.29). Some used brick to create the domes with no use of timber, others used some timber beams, while in the western desert some people used tamarisk or bundled reeds to provide the frame for a vault. While these variations illustrate the importance of local materials they also evidence the significance of knowledge and cultural exchange of both techniques and ideas about the form of dwellings. Szabo and Barfield report flat-roofed dwellings throughout the central and northern mountains, the southern mountains and foothills and in Badakhshan (Fig. 1.30).55 The construction uses poplar poles or beams to support the roof, but the materials used for walls and roofing vary depending on local resources. The walls were usually made of pressed mud in rural areas or sun-dried brick in urban areas. In the south-east the qal’a was a common form – a fortified farm with massive walls of clay and straw enclosing a large compound (Figs 1.12, 1.31). In the high mountains stone construction is common with thick walls to insulate inhabitants from the severe winters. In Nuristan, two-storey buildings of stone and timber with no nails are found, each built above the other but stepped back from

the geographical background

Figure 1.30  Flat timber and mud roofing at Istalif. The main range of the Hindu Kush is in the background.

Figure 1.31  A traditional qal’a in the Bamiyan Valley.

47

48

THE ARCHAEOLOGY OF AFGHANISTAN

the building below – thus making effective use of constricted space for building. Finally, Szabo and Barfield describe the senj form of construction, which they suggest is more resistant to earthquakes than other methods. It uses brick walls with timber frames and is often used for upper floors. There was thus a similar pattern of subsistence farming throughout the country, varying in detail with altitude and the abundance of water but centred around wheat and barley as the staple grain crops, with pulses and vegetables as further important additions to the diet. Livestock were an integral part of this farming system, which also relies heavily on irrigation techniques. Crops were raised for exchange as well as for subsistence. Cotton, rice, sugar beet and oil crops were raised in the north and some cotton, oil crops and grapes in the south. Produce and livestock were traded long before more recent cash crops were introduced and allowed the sedentary farmer to acquire goods from far away. The nomads play an important part in this trade today and probably played a similar role in prehistory. Such trade is discussed further below. Here it is important to note the importance of animals for transport in the descriptions of Humlum, Dupree, and Szabo and Barfield. In addition to cattle, horses were used for transport in the north and in the Hindu Kush but in the south donkeys and mules were more common (Fig. 1.32). Camels were used especially for long-distance movement of heavy loads (Fig. 1.33); both the one- and two-humped varieties were described, the former predominantly in the south, the latter predominantly in the north.

Figure 1.32  Donkey transport in the Kandahar region.

the geographical background

Figure 1.33  A camel caravan passing in front of the Iron Age citadel at Old Kandahar.

It seems pertinent at this point to consider what aspects of the farming system of the past could be similar to the modern situation. Certainly many of the tools used in the 1960s were not more advanced than those that might have been used by an Iron Age farmer. Some of the crops grown today would have been unknown, but many probably grew as wild plants in Afghanistan in the past as they do today. If we accept that the farming system described by Humlum is not unlike that of the later prehistoric farmers then data on the yields that are obtained today or on the population supported by a given cultivated acreage would give some indication of the population that could have been supported in the past, but unfortunately there is little reliable data of this nature. However, Humlum, in studying the oasis of Pirzada which is irrigated by karez, found that samples from four fields showed a wheat yield varying from 469 to 1,440 kg/ha. A soil analysis suggested that these low yields were probably due to the primitive strains of wheat used and to disease rather than to a poor soil.56 It certainly seems that these yields may be of a similar order to those that could have been attained by prehistoric farmers. In the north, Michel quotes somewhat higher wheat yields in Kataghan Province in 1956 of 1,440 lbs/acre (approximately 1,615 kg/ha).57 These yields were gained on loess and alluvial soils using river irrigation and simple tools for cultivation but employing organic fertiliser and a good rotation. Lastly, where dry farming is used in the higher areas Humlum suggests that the wheat yields may be only two to

49

50

THE ARCHAEOLOGY OF AFGHANISTAN

three times the seed sown, although wheat grown by dry farming methods is said to have a higher protein content than that grown under irrigation. Nomads Not all the rural population of Afghanistan described by Humlum were sedentary farmers: there were groups showing all degrees of nomadism, from the pure ‘herding’ nomads to those who combined seasonal nomadism with the cultivation of crops. For example, in the mountains transhumance was common. The men took the flocks up to high pastures in the summer leaving the women and old men to cultivate the fields; the men and flocks then returned to the valleys with the colder weather. Many groups adjusted their degree of nomadism to fluctuating economic or climatic circumstances. Some groups moved over very short distances while others migrated up to 1,000 km between summer and winter pastures (Fig. 1.34). Ferdinand distinguished two main groupings of nomads – those of the south and west and those of the east.58 These divisions corresponded to broad cultural divisions as well as to contrasts in the environment. ‘Pure’ nomads and semi-nomads were found in both areas while in the east there were also groups who were essentially migratory farm workers with herds and groups who moved in a rather restricted area within the forested mountains of Nuristan. The ‘pure’ nomads’ principal means of livelihood were their herds of sheep and goats which supplied food, fuel, material for clothes and shelter, and goods for exchange with the farmers and traders in the towns. They moved from winter pastures in the uplands along the southern border to summer pastures in the central mountains (Fig. 1.34). Although their animals supplied them with so much, they were not self-sufficient and depended on exchange with the sedentary population to get many tools and manufactured objects in addition to items of diet. According to Ferdinand, the nomads of the south and west were less dependent on such trade than those of the east and south-east.59 Some nomads were primarily traders although they also kept herds. These groups were found mostly among the eastern nomads and they played an important part in the economy of central Afghanistan, bringing goods from Pakistan into the central mountains for trading with farmers and other nomads. Much of the trade was done at large ‘fairs’ in the Chahar-Aimaq area in Herat province and the north-west (Figs 1.35, 7.7). The traders brought in clothes, sugar, tea and metal wares and exchanged these for money or for wheat, butter and animals. In the Jalalabad area there were nomadic groups who spent the winter in the valley, sometimes living in houses, near, or in, villages where they performed odd jobs in return for living space, and who then moved off towards the uplands in the late spring to harvest the crops in other villages along their route. In return for this they got a share in the crop and freedom to put their herds in fallow fields, thus improving the soil fertility. They spent the summer with their flocks in villages

the geographical background

Figure 1.34  Seasonal movement of nomads.

in the uplands and returned to their winter quarters in the autumn. Their herds consisted largely of cows and donkeys with some sheep, goats and chickens. Yet another type of nomadism was found in the eastern mountains of Nuristan. Here were found herders of goats who lived in their winter quarters on the lower edge of the forest where they built huts of stones, earth, branches and straw to protect themselves against the bitter cold. In summer they took their tents and moved up to the treeline. They exchanged milk products, wool and animals for wheat and other foods and manufactured goods. Lastly, in the south and west were semi-nomads who cultivated their own land and lived during the winter in settled villages, but who moved away to the mountains with their herds during the summer months after planting their crops and returned in time for the harvest. In times of plentiful rain they might concentrate on farming, but during bad years

51

52

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.35  Nomad market on the Wazmin Pass in the western mountains.

Figure 1.36  Pashtun peaked black tent between Delaram and Gulistan.

the geographical background

Figure 1.37  Pashtun vaulted black tent.

their flocks offered an alternative means of livelihood, and it may be that during dry periods such semi-nomadism increases.60 Szabo and Barfield describe the dwelling arrangements associated with different forms of non-sedentary lifestyles such as permanent dwellings in two locations, temporary and semi-permanent huts and mobile dwellings – most notably yurts and black tents – that were moved with the migrating owners.61 Temporary huts were found in northern Afghanistan and also in mountain areas and were made of a frame of semi-rigid wooden poles or reeds, which might be covered with a very wide variety of coverings such as woven reed mats, cloth of various types and even mudded plaster. Such huts are very light and easily portable and most were used seasonally, although some were semi-permanent and might be used as winter dwellings. Most huts were found only in north and central Afghanistan and were used as portable dwellings. In the south the black tent performed a similar role. Two main types of black tent are described, both used by Pashtun peoples – the Durrani and Baluch vaulted black tents and the Ghilzai and Brahui peaked black tents (Figs 1.36 and 1.37). Indeed, the distribution of black tent use mirrored the distribution of these ethnic groups within the country (a distribution that had been extended by political changes at the time Szabo and Barfield were writing). The tent covering was made of woven goat-hair palas supported by poles and the sides could be raised

53

54

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.38  Firuzkuhi yurts in three stages of construction in western Afghanistan.

for v ­ entilation. It could be transported on camel-back. Szabo and Barfield report that the black tent is particularly well-adapted to hot, dry conditions. Counterintuitively, the black material appears to cope with solar load more effectively than other tent types and facilitates air circulation inside the tent. Dupree reports a difference of temperature between the tent and outside air of 10–15 degrees centigrade.62 Black tents are not good at coping with cold and wet, in contrast to the yurt, which Szabo and Barfield found in northern Afghanistan and the Pamirs as well as among the Firuzkuhi of western Afghanistan (Figs 1.38 and 1.39). It was used by semi-nomadic pastoralists and semi-sedentary farmers. Yurts are highly sophisticated, circular, mobile dwellings which can be adapted to cold or heat.63 The frames were made of willow and the covers of woollen felt while the sides might be covered with reed matting. Two main types were described by Szabo and Barfield (the domical and smaller conical yurt), associated with different ethnic groups, patterns of land use and areas. The domical yurt needed transporting by camel because of its size and weight while the smaller conical yurt could be carried with donkeys, horses or oxen. Both types of yurt were expensive to construct. The Taimani black tent was an interesting hybrid structure used in the upper reaches of the Farah Rud (Fig. 1.40). It was a mixture of a yurt-style – but rectangular – frame of willow poles and woven reeds mats covered by a black tent covering of goat-hair palas. Szabo and Barfield note this as a good example of mixing cultural traditions to produce a new type of structure.

the geographical background

Figure 1.39  Firuzkuhi yurt in use.

Figure 1.40  Black tents of the Taimani in western Afghanistan.

55

56

THE ARCHAEOLOGY OF AFGHANISTAN

Nomadic groups may adjust their habits to economic circumstance and so change their patterns of life over time. It is clear that they formed (and still form) an important part of the economy of the sedentary farmers and that their trading activity was necessary to both parties. Study of their ways of life offers some ideas on possible relationships between nomads and farmers in the past and suggests what a wide range of adaptation to local environments and economic opportunities is possible. While the general patterns of subsistence in Afghanistan prior to the Soviet invasion have been outlined it is important to remember that there were many variations depending upon the local environment. For example, in Sistan the Sayyid population netted fish and migratory birds as the major part of their food supply; in Nuristan, gathering of wild plants had an important part in the economy. Again, the division between sedentary agriculture and nomadism is not firm and groups have shifted the emphasis from one to the other depending on changing local circumstances of climate or economy. Opportunities for Trade Afghanistan’s position gives it opportunities for trade with a variety of different areas. On the whole it does not possess any outstanding natural endowment or rare commodity, but in the 1950s and 1960s fresh and dried fruits, karakul, skins, raw wool and cotton, carpets, rugs and sheepskins were some of the principal items produced for trade abroad. Sugar and tea, textiles and manufactured goods were bought in return.64 Afghanistan’s position makes it a natural crossroads for routes between East and West and between India and Central Asia. In the past trade routes ran across the north of the country through Balkh – usually described as the Silk Routes. Routes also went up the Khyber Pass from the Indus valley towards the north and east. In the south, routes skirting the central mountains along the line of the present road led to Quetta and the Indus from the north-east via Ghazni and from the Iranian plateau via Herat and Kandahar. The Helmand Valley also provides a southward link from the central mountains to Sistan and on to the passes across the desert mountains to the Gulf of Oman and Straits of Hormuz; thus this route could have linked with coastal trading routes to Mesopotamia and the Indus. Many authors argue for the importance of the Silk Roads for east–west trade from about the end of the first millennium bc, although Ball disputes whether significant, organised, overland trading occurred at this time.65 Sherratt argues that later historic east–west exchanges were prefigured by the Trans-Eurasian Exchange – an episode of major west–east exchanges of knowledge of metallurgy and livestock herding – around the early second millennium bc.66 Jones et al. maintain that this was preceded by the gradual exchange of knowledge about the cultivation of important starchy food plants from about the sixth to the fourth millennium bc throughout Eurasia.67 Furthermore, Christian has argued that the

the geographical background

Figure 1.41  Raw lapis lazuli in a trader’s shop in Kabul.

focus on the east–west exchanges has diverted attention from the importance of exchanges across the ecological boundaries between the pastoralist and agrarian worlds of Eurasia.68 He argues that these north–south ecological exchanges were integral to east–west exchanges and allowed goods from the steppes and forests to be exchanged for goods from agrarian communities. In what ways people living in what is now Afghanistan were or were not involved in such exchanges is unknown, but they were certainly well-placed to be so. We have seen how important trade between the nomads and sedentary farmers has been in the recent past. Afghanistan’s location and varied ecological zones have facilitated both internal trade and trade south to Pakistan and India and north into Turkmenistan and the former Soviet Union. This pattern may also have occurred in the past. One rare commodity still produced in Afghanistan today that is known to have been traded in the past is the lapiz lazuli of Badakhshan (Fig. 1.41, 1.42, 3.24). Lapis lazuli artefacts are known from the Middle East, especially in the Syro-Mesopotamian area, dated to the third millennium bc. Lapis lazuli is not a common mineral and the only other probable known sources for supplying the Middle East are in the Pamirs, Tajikistan (Liajura-Dara), Lake Baikal and possible sources in the Swat Valley in Pakistan, in Azerbaijan and possibly in Kerman in Iran as well. The Lake Baikal source produces stones of an inferior quality to those of Badakhshan and was probably of far lesser importance in antiquity than the Afghan source. The Pamir deposit is at about 3,300 m and is extremely difficult of access, while the Iranian source, if it existed, has evidently been worked out. Herrmann concluded, after reviewing the evidence, that the Badakhshan lapiz lazuli was the principal source for an important trade with Mesopotamia that flourished from about 3500 bc.69 This view has continued to be dominant

57

58

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 1.42  Lapis bust of a Parthian prince in the Iran Bastan Museum.

among archaeologists, but there has been no other evidence to confirm it until recently. In 2008 Rea and colleagues began a study of the characteristics of lapis stones from Badakhshan, the Pamirs, Siberia and Chile using micro-PIXE – a noninvasive method which can be used without harming precious artefacts. They showed that it is possible to distinguish between stones from different sources.70 A recent study using this technique, of eleven carved lapis lazuli objects from the Egyptian Museum of Florence dated to the first millennium bc, has shown that the stone used in nine of the eleven came from Badakhshan.71 The remaining two objects are from either a Tajik or Badakhshan source and further analysis is needed before their provenance can be determined. As Lo Giudice concludes, ‘this is a further confirmation of trading of this stone up to 4,000 km from its extraction site in the first millennium bc’.72 Future micro-PIXE analysis of older lapis lazuli artefacts may provide further evidence that Afghan lapis lazuli was exchanged long distances in earlier periods. It is probable that this trade was not direct and that the Afghans used Shortughaï in north-eastern Afghanistan and Shahr-i Sokhta in south-eastern Iran as entrepôts where they exchanged the lapis lazuli for goods from nearer at hand than Mesopotamia (discussed further in Chapter 3). However, Mesopotamian or even Egyptian goods may have returned by the same route.

the geographical background

Figure 1.43  Lapis-bearing rock at the main lapis lazuli mines at Sar-i Sang in Badakhshan.

The Badakhshan mines were hardly easy of access. They lie above the KeranoMunjan valley at Sar-i Sang (Fig. 1.43), Stromby, Chilmak and Robat-i Paskaran. The valley is narrow and steep with scant vegetation and few permanent settlements. The mines can only be reached by precipitous paths along which all materials for mining must be carried. In the past the mining was done by fire-setting. Fuel and water were carried up to the mines, a fire was lit beneath the face to be quarried and then cold water was thrown onto the heated rock to make it crack. The lapis was then extracted by using picks, hammers and chisels. The provision of fuel and water for the mines must have involved great labour in amassing the fuel and transporting it, and this is suggestive of the value of the stone in the past. Once quarried the lapis could be taken south to Kabul and beyond via the Anjuman pass and Panjshir valley, to the north and west via Faizabad or to the east along the Wakhan corridor. The evidence on trade and cultural exchange suggests that the prehistoric peoples of Afghanistan are likely to have had significant contact with other peoples and had the opportunity to trade over considerable distances despite the problems of travel across desert and mountain barriers. Indeed, the country’s position would have encouraged such contacts and the development of trade.

59

60

THE ARCHAEOLOGY OF AFGHANISTAN

Summary A highland country surrounded by lowlands and lying at the crossing points of a number of important natural routeways, Afghanistan had considerable advantages in the past as a locale for trade and exchange. The products of the mountains and steppe lands could easily be exchanged for both manufactured and primary products from the Indus or from Iran and the west. The diversity of people at present inhabiting the country bears witness to its openness to migration and invasion in the past. Within the country itself the aridity of the climate means that cultivated areas are located either in the upland valleys or along rivers in the lowlands. Irrigation techniques are a particularly important aspect of agriculture. However, there is a great variety of natural habitats and a feature of Afghanistan in the period immediately before the Soviet invasion was the diversity of methods of subsistence including nomadism, hunting and gathering and settled agriculture. Because of the strong relief quite different environments can be found relatively close together. This feature would have been particularly helpful to Palaeolithic groups with their limited technology and small size. It must also have been important at times of climatic change, since groups could slowly shift from areas with deteriorating climatic conditions to more favourable ones, without the need to move far. This variety of habitats is a feature that is likely to have encouraged trade and exchange within the country. Lastly, it must be stressed again that data on the nature and distribution of resources in Afghanistan during the archaeological period are extremely scant. The picture presented above may be modified substantially as further research is undertaken.

CHAPTER 2

The Palaeolithic Richard S. Davis

Before 1951 the Palaeolithic period in Afghanistan was virtually unknown; Afghanistan was a blank space on the Palaeolithic map. It was never predicted, of course, that Afghanistan would prove to have been unknown to Palaeolithic hunters and gatherers, and probably most prehistorians of Asia in the early 1950s would have expected that evidence for at least Middle and Upper Palaeolithic occupations would eventually turn up. In the 1960s and 1970s several Palaeolithic expeditions began work, primarily in the northern part of the country, but following the Soviet invasion in 1979 Palaeolithic fieldwork came to a halt. This initial research clearly established the presence of Middle and Upper Palaeolithic sites and extremely tentative evidence for the Lower Palaeolithic. This chapter is intended to summarise the results of the fieldwork and also to include more recent interpretations of how the Afghan Palaeolithic can be seen in relation to the larger Eurasian Palaeolithic world. In this presentation the Palaeolithic is divided into four successive phases: Lower, Middle, Upper and Epi-Palaeolithic. The last two terms are collectively referred to as the Late Palaeolithic. It must be made clear from the outset that the present boundaries of Afghanistan have no special relevance to the distribution of Palaeolithic populations. Afghanistan is frequently referred to as a crossroads of Asia, and it is not unreasonable to think in those terms as regards peoples of the Palaeolithic. It is the case that portions of South-west Asia, Central Asia and South Asia meet in modernday Afghanistan, and the prehistorian, therefore, should be aware of Palaeolithic developments in all three areas. It should be remembered that Palaeolithic studies in Afghanistan were at a very early stage when they were brought to a close in 1979. It was a promising beginning, and it can only be hoped that future generations will be able to resume survey and excavation and thus bring the Afghan Palaeolithic record into a much wider world.

62

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 2.1  Palaeolithic sites in Afghanistan.

Problem Areas Why is Afghanistan an important region for Palaeolithic study? What problems can be profitably studied here? Archaeologists have time to reflect on these questions, and when secure conditions for fieldwork return, some of the areas indicated below should be considered: 1. In general, how does Afghanistan fit into the Palaeolithic world? Was Afghanistan witness to the early ‘out of Africa’ hominin radiation which brought early Homo into east Asia, possibly as early as 1.77 Ma?1 Are there traces of the early radiation of Anatomically Modern Sapiens (AMS) into Afghanistan? A first step is to build a firm cultural historical sequence stemming from a programme of survey and excavation. Particularly important is geological survey to identify appropriately aged land surfaces, stratigraphic deposits and depositional environments. 2. Which environmental zones were exploited during the Palaeolithic, and how did climatic fluctuations of the Pleistocene effect the settlement of particular areas? Afghanistan has a diversity of habitats including varieties of highland arid regions, steppe and true desert. Were there periods of significant population

the palaeolithic

decline or even abandonment during the Last Glacial? Afghanistan provides the opportunity to study a wide range of hunter-gatherer adaptations. 3. What part did the Late Palaeolithic hunters and gatherers play in the Neolithic Revolution? Is there evidence for subsistence practices during the EpiPalaeolithic which might signal steps towards early domestication of sheep and goats and in the cultivation of grains? These problems are all potentially answerable within certain constraints, and their solutions will have considerable relevance for understanding the dynamics of the hunting and gathering way of life. What follows is a review of all the important Palaeolithic data from Afghanistan, with some incomplete answers for the above questions. Brief History of Palaeolithic Research in Afghanistan The first discovery of Palaeolithic remains in Afghanistan was made in 1951 about ‘two miles southeast of Balkh’.2 It was a large flake struck from a prepared core which Raymond Allchin believed to show Mousterian characteristics. In March 1954, Carleton S. Coon, then of the University of Pennsylvania, motored north over the Hindu Kush and made significant survey and excavation discoveries. A narrative account of his travels and excavation is recorded in his book, Seven Caves.3 Coon reached Aibak, the capital of what is now Samangan province. His geologist had noted a large number of limestone outcrops in this area, and it appeared to be a good search area for Palaeolithic cave sites. Coon soon located a rock-shelter named Kara Kamar and began excavations. By mid-April Coon completed his work at Kara Kamar and returned to Kabul, where his Palaeolithic collection was divided with the National Museum. Kara Kamar had both an early Upper Palaeolithic occupation (the only one ever found in Afghanistan) and one from the Epi-Palaeolithic. In the autumn of 1959 Louis B. Dupree, a major figure specialising in Afghan archaeology from the Palaeolithic to the present and in Afghan studies generally, made a general archaeological reconnaissance of northern Afghanistan, and he was successful in identifying several Palaeolithic localities, which he later excavated.4 In Badakhshan Province he located a Middle Palaeolithic rock-shelter called Darra-i Kur, which he excavated in 1966. In Balkh Province he discovered three Epi-Palaeolithic localities near the town of Aq Kupruk. The sites consisted of two rock-shelters (Aq Kupruk I and II) and an open-air site located on a low river terrace of the Balkh (Aq Kupruk III). These sites were excavated in 1962 and 1965.5 Salvatore Puglisi, Director of the Istituto di Paleontologia, Sapienza University of Rome and part of the Italian archaeological mission to Afghanistan, conducted a survey in the vicinity of Aibak. Approximately 20 km south-west of Kara Kamar he located a rock-shelter in the dry wadi Dara-i Kalon, and he made a small

63

64

THE ARCHAEOLOGY OF AFGHANISTAN

test excavation in 1965 from which he recognised two distinct Epi-Palaeolithic assemblages.6 Dupree and I revisited the Aibak region in the summer of 1969 with a geologist, Laurence Lattman of the University of Cincinnati. We located one Epi-Palaeolithic surface site (Kok Jar) about 3 km from Puglisi’s test. During the same summer and in the following summer as well, I visited all of the above archaeological sites and made further reconnaissance in surrounding areas. Subsequently, I studied all of the Late Palaeolithic archaeological assemblages and subjected them to an extensive analysis.7 I also had the opportunity to study Late Palaeolithic collections in Samarkand, Tashkent and Dushanbe in what was then Soviet Uzbekistan and Tajikistan. In 1969 and 1970 Philippe Gouin of the Délégation Archéologique Française en Afghanistan made archaeological surveys north of the Hindu Kush on the very arid steppe just beyond the Tashkurghan Oasis. He located one Epi-Palaeolithic site, which he reported in preliminary form.8 For three seasons (1969, 1975, 1976) Alexander Vinogradov of the Soviet–Afghan Archaeological Expedition surveyed the uninhabited desert plain south of the Amu Darya from just west of Tashkurgan to Andkhoy. This survey sampled along its approximately 300 km extent, coarse alluvial deposits which derived from drainages flowing northward from the Hindu Kush, the desert leading up to the Amu Darya, and the margins of the south-eastern Kyzyl Kum desert. Vinogradov discovered a wide range of sites beginning with some clear elements of the Middle Palaeolithic, a large number of what he termed Mesolithic sites, and numerous Neolithic sites.9 His work was of great importance because it opened the possibility that in terminal Epi-Palaeolithic times there may have been two quite distinct cultural zones: one on the plain which typologically was similar to the eastern Caspian sites and the other in the foothills of the Hindu Kush – notably the Kuprukian – which has ties perhaps more closely with cultures of the Zagros. C. B. M. McBurney made an archaeological reconnaissance of known Palaeolithic sites in northern Afghanistan in July–August 1971.10 He concluded that the region was well worth continued investigation and noted the apparent distinction between the Kuprukian Epi-Palaeolithic and that of the Caspian area where he had worked extensively. Next, archaeological discoveries were made by Dupree south of the Hindu Kush in the Dasht-i Nawur region (Fig. 2.2). There, on ancient lake beaches surrounded by dormant volcanoes he collected surface concentrations of Middle and possibly Lower Palaeolithic artefacts.11 Finally, Dupree and I made a brief survey of the Dasht-i Nawur in 1976 and located surface sites which were typologically EpiPalaeolithic and had many obsidian artefacts.12 Mention might also be made of a number of petroglyphs, mainly in the mountainous Badakhshan region, which have been recorded by various expeditions (albeit not necessarily with archaeology as the objective) and tentatively dated to the Epi-Palaeolithic (Fig. 2.3).13 The dates of these petroglyphs are based upon

the palaeolithic

Figure 2.2  Satellite image of the Dash-i Nawur. (Image prepared by the Center for Ancient Middle Eastern Landscapes at the Oriental Institute of the University of Chicago. Imagery courtesy Esri and CNES/Airbus DS.)

stylistic comparisons with rock art elsewhere, mainly in neighbouring Central Asia and Northern Pakistan.14 Petroglyphs, however, are notoriously difficult to date unless associated directly with excavated strata, partly because there are no absolute dating criteria apart from stylistic ones, and partly because styles persisted for long periods of time. The approximate locations of all the above sites are indicated in Fig. 2.1. One may summarise this prolegomenon by stating that practically no one who has actually looked for Palaeolithic sites in Afghanistan has failed to find them, but with no exception has there been anything other than preliminary survey and excavation. We can hope that in the future the necessary combination of peace, expertise, recovery of Afghan institutions and funding can occur to make possible some long-term and intensive Palaeolithic research programmes.

65

66

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 2.3  Petroglyphs near Sarhad in Badakhshan; date is uncertain.

Initial Occupation of Afghanistan – the Lower Palaeolithic At present, the evidence for occupation of Afghanistan prior to the Middle Palaeolithic rests on the discoveries made by Dupree in the summer of 1974 on the ancient shores of the Dasht-i Nawur, a large and shallow brackish lake in the margins of the Hindu Kush in Ghazni province (Fig. 2.2).15 The Dasht itself is a large, flat volcanic–tectonic depression just over 3,000 m elevation. In a brief survey of the beaches to the east and north of the lake, Dupree found concentrations of quartzite tools from which he identified the following implement types: cleavers, large scrapers, choppers, chopping tools and pebble tools. According to Dupree, these finds represent a Lower Palaeolithic industry. To be sure, it is essential that careful geological work must be done in this area, and also it is hoped that further reconnaissance will reveal Lower Palaeolithic tools in association with a faunal assemblage. Nevertheless, Dupree’s discovery and typological assessment is exciting, and it is possible that the Dasht-i Nawur area will reveal more evidence for Lower Palaeolithic occupation. What is clear is that Afghanistan is surrounded by well-documented Lower Palaeolithic sites. The nearest are the loess sites in southern Tajikistan where pebble tool assemblages have been found in buried palaeosols below the Brunhes– Matuyama palaeomagnetic boundary with ages estimated at over 900,000 years.16

the palaeolithic

The Lower Palaeolithic has been recognised in the Indian subcontinent since the middle of the century. There is now a well-attested Acheulean presence earlier than 1.07 Ma, which is the earliest unequivocal archaeological evidence in South Asia.17 In Iran the evidence for Lower Palaeolithic is primarily from surface localities and the clearest traces are from north-west Iran near to the Caucasus.18 Given the early dates from Tajikistan and South Asia, it will not be very surprising if Afghanistan some day yields good evidence for Lower Palaeolithic occupation. Exactly what kind of stone tool industries eventually will be found is another question. Initial dispersals of hominins out of Africa utilised Oldowan core and flake technology as recognised at Dmanisi in Georgia at 1.8 Ma. Subsequently, perhaps 300,000 years later, bearers of Acheulean bifacial technology moved eastward into Asia.19 Industries with or without hand axes are certainly possible, but it will be even more interesting to learn how these suspected Lower Palaeolithic hunters and gatherers actually adapted and responded to the range of conditions in Afghanistan. The Middle Palaeolithic Although several localities in Afghanistan have been classified as Mousterian or Middle Palaeolithic, only one, Darra-i Kur, can confidently be placed in this category. My recommendation is that the term ‘Mousterian’ might best be avoided in Afghanistan in favour of ‘Middle Palaeolithic’ simply because there is not a big corpus of data to work with. The provisional definition for the Middle Palaeolithic in Afghanistan is: flake industries with discoidal and/or Levalloisian flaking technique. Darra-i Kur This is a stratified rock-shelter site located near the village of Chanar-i Gunjus Khan in Badakhshan Province. The shelter is located high up on the side of a valley and commands an excellent view of the surrounding countryside. There were major ancient rock-falls in the shelter which made excavation difficult. The Middle Palaeolithic implements (Figs 2.4, 2.5) were found in overbank deposits of silt and clay, which were laid down by an ancient stream which ran close to the shelter. There is one radiocarbon determination made on several small fragments of charcoal from a hearth disturbed by water action: GX-1122 30,300 + 1,900, – 1,200 bp. It is strongly suspected that some post-Middle Palaeolithic charcoal was mixed in the dated sample, thus rendering the determination too young. In fact the Geochron Lab analyst who ran the sample wrote: ‘I would not hang any major decisions or interpretations on this single date.’20 Undisturbed hearths and Middle Palaeolithic deposits probably exist buried beneath the massive roof fall. Dupree and I have already published a preliminary description of the lithic material,21 but several general observations can be made in addition. The Middle Palaeolithic

67

68

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 2.4  Darra-i Kur, Middle Palaeolithic: (1) flake core; (2) discoidal core; (3) retouched blade; (4) Levallois blade core.

layer produced approximately 800 lithic specimens. There is definite evidence of Levalloisian technique in the form of Levallois blade cores (Fig. 2.4, no. 4), but the predominant flaking technique was discoidal. Discoidal cores were numerous, and the majority of them were small in size and were well-struck (Fig. 2.4, no. 2). Multiple platform cores were also found (Fig. 2.4, no. 1). A prominent feature of this industry was the relatively high incidence of blades (Fig. 2.4, no. 3). Shaped

the palaeolithic

Figure 2.5  Darra-i Kur, Middle Palaeolithic: (1) Levallois point; (2) triangular point.

tools were infrequent. Irregularly or partially retouched artefacts were by far the more common. There appeared to be only rare side scrapers, no hand axes, and few ‘Upper Palaeolithic’ types. The material used was a variety of basalt which does not fracture well conchoidally. The lithic collection has not been completely analysed and it is, therefore, impossible to quantify the above observations. A hominin incomplete right temporal bone was found associated with the lithic material. This specimen was analysed by J. Lawrence Angel at the Smithsonian Institution,22 but more recent radiocarbon and DNA analyses re-dated this bone to the Neolithic.23 It represents the only known hominin material found in Palaeolithic context in Afghanistan. Angel compared the specimen to both Homo neanderthalensis material and modern Homo sapiens sapiens, and he concluded that it is nearer to modern sapiens than Neandertal in terms of the morphology of the tympanic bone, but that it would fit into a ‘partly’ Neandertal population like the Es-Skhul cave specimens from Mt Carmel in Israel as well as into a modern population.24 Trinkaus concurs that the temporal bone ‘may represent an early modern human’.25 Trinkaus makes the basic point that early modern humans older than 25,000 years outside of east Africa and South-west Asia exhibit ‘complex and varying mosaics of early modern, late archaic and regional anatomical features’.26 Glanz echoes this view as it applies to various fragmentary and more complete skeletal remains from Central Asia.27 In her view, the characteristic European Neandertal is not represented in this area by the Teshik Tash child found in Uzbekistan less than 150 km from the northern border of Afghanistan, and she does emphasise that it shares many traits with anatomically early modern humans. A recent morphological analysis of the Teshik Tash child,28 however, made a strong case for its Neanderthal character resulting from analysis of the frontal bone. In addition mtDNA extracted from the Teshik Tash child showed that it carried mtDNA of the Neanderthal type and was most likely related to western European populations.29 It is now well-established that Neandertals and Denisovans were in the Altai,30 so there is no reason not to expect to Neandertals

69

70

THE ARCHAEOLOGY OF AFGHANISTAN

and even Denisovans in Central Asia. There is much more to learn about the colonisation and evolution of early sapiens in Central Asia. The Middle Palaeolithic layer revealed some faunal material which has been identified by Dexter Perkins as sheep, goat and possibly a large bovid. This is a significant discovery because all known later Palaeolithic hunting and g­ athering groups also hunted these animals, particularly sheep and goat. It is interesting to note in this regard that at the Middle Palaeolithic site of Teshik Tash in Uzbekistan, Siberian Mountain Goat (Capra siberica) accounted for 96% of the identified large mammal bone.31 It appears, therefore, that a basic hunting adaptation to these particular species existed in the Middle Palaeolithic, and it continued on until the end of the Epi-Palaeolithic. Then, of course, in the early Neolithic sheep and goat were domesticated. There was, therefore, a long period of dependence on sheep and goat and different patterns of exploitation developed as humans developed an intimate knowledge of their behaviour and habitat. A very different situation was found at a large rock-shelter locally called Ghar-i Mordeh Gusfand (‘Cave of the Dead Sheep’) in north-west Afghanistan. Here, preliminary excavations by Dupree in 1969 revealed what appeared to be a Middle Palaeolithic industry made out of a poorly silicified limestone.32 The rock-shelter itself was enormous, measuring approximately 300 m across the dripline and approximately 100 m deep. There was a tremendous rockfall at the front of the shelter – presumably the one responsible for all the dead sheep! After the preliminary report was published, a large trench was excavated in the summer of 1970 in the rear of the shelter. The results of that excavation cast considerable doubt on the original interpretation of the lithic tools as belonging to a Middle Palaeolithic industry. Although the tools were artefacts, their stratigraphic position does not indicate sufficient antiquity for an early or mid- Upper Pleistocene occupation. For the present we can rule out Ghar-i Mordeh Gusfand as a confirmed Middle Palaeolithic locality. Balkh Province In Balkh Province close to the village of Aq Kupruk, McBurney reported the presence of ‘a middle Mousterian type of industry’.33 This potentially important find requires more investigation in order to allow us to understand the nature of this locality. Hazar Sum Further to the east, in the Hazar Sum valley near the town of Aibak, Puglisi has noted the presence of surface material which he has described as Mousterian.34 He found stone tools made from the locally abundant nodular flint which he interpreted as being manufactured according to Clactonian, Levalloisian and Mousterian techniques. None of the implements illustrated in his report, however, can be attributed to those Palaeolithic techno-complexes with any assurance, and

the palaeolithic

Figure 2.6  Kara Kamar. Arrow indicates location of the shelter.

these artefacts are in secondary context and not many show any real diagnostic characteristics. My own surveys in this area revealed no certain indications of Middle Palaeolithic industries, but there is a great abundance of surface material on the wadi terraces of the Hazar Sum valley. It consisted of the chipping debris of both Palaeolithic and post-Palaeolithic peoples who utilised the nodular flint outcrops on the sides of the wadi channels. No doubt Palaeolithic hunters and gatherers, Neolithic farmers and recent nomadic pastoralists have all made use of the excellent nodular flint from this area. Kara Kamar Adjoining the Hazar Sum valley is the site of Kara Kamar (Figs 2.6–2.10), excavated by Coon in 1954. Kara Kamar is a small rock-shelter located about 135 m above the valley floor (Fig. 2.6). The first and third cultural levels have been assigned to the Late Palaeolithic and will be described in some detail in a later portion of this chapter. The second and fourth levels have been identified by some prehistorians as Middle Palaeolithic. Coon, however, left the identity of the flint knappers of these two levels an open question. In an article on Upper Palaeolithic origins, Louis Pradel35 cites Kara Kamar for possible evidence of temporal overlap between Middle Palaeolithic and Upper Palaeolithic: ‘At Kara Kamar (Afghanistan) Upper Palaeolithic material may also exist, situated between two Mousterian layers … This information, however, lacks precision.’ Following an examination of the available material from levels two and four at the National Museum, it is clear to me that Pradel’s last sentence is quite an understatement. There simply is no good evidence that this material should be described

71

72

THE ARCHAEOLOGY OF AFGHANISTAN

as Middle Palaeolithic. The material available does not lend itself to a clear diagnosis. The cores from Level II were multi-platformed flake cores, and none of them could even remotely be described as discoidal or Levalloisian. There were no Levallois points, flakes or blades in the collection. The collection was too small and the modified pieces too undiagnostic for any definitive classification. Vinogradov came to a different interpretation following his examination of the material at the National Museum and made the case that it was a final Mousterian.36 It is interesting in this regard that many years ago G. P. Grigorev and V. A. Ranov37 posited a cultural historical picture of Central Asia wherein the Levalloiso-Mousterian and its Neandertal creators lived on until much more recently than in the Zagros or the Near East. Hence the paucity of Central Asian Upper Palaeolithic sites might be explained by a Middle Palaeolithic which lasted perhaps 10,000 years longer than elsewhere to the west. Does Kara Kamar reveal contemporaneity between Middle Palaeolithic and Upper Palaeolithic populations? There simply is not enough evidence to establish such a conclusion from Kara Kamar. The single Carbon-14 determination made from a contaminated sample in Level II has such a great standard deviation that it is of no use in the chronological placement of this industry. Only ten flint artefacts were recovered from Level IV. There was one edge retouched flake, one use retouched flake, two blades, five flakes and one core fragment in the collection. Coon38 has observed that Levels II and IV are similar in flint lithology and in debitage techniques, and my own inspection of the materials in the National Museum yielded a similar conclusion. The inventory of the existing components of the industries in question is given in Table 2.1, and Fig. 2.7 shows several implements from Level II. In addition to the Lower Palaeolithic finds already mentioned in the Dasht-i Nawur, Dupree also discovered some Middle Palaeolithic concentrations of tools on the ancient beaches of the lake. He noted similarities between the typology of the Dasht-i Nawur material and the industry from Darra-i Kur.39 Further survey and excavation is definitely indicated in this potentially extremely important region. Discussion The above descriptions of the various alleged Afghan Middle Palaeolithic sites must inevitably leave the reader with the impression that there is a high degree of uncertainty and lack of knowledge about this Palaeolithic phase. We may be quite certain, however, that the presence of Middle Palaeolithic populations is established; the materials from Darra-i Kur irrefutably demonstrate that. Of the other sites mentioned above, it is apparent that they are either definitely not Middle Palaeolithic (Ghar-i Mordeh Gusfand) or need much fuller investigation and description before they can be established as true Middle Palaeolithic occurrences (Dasht-i Nawur, Kara Kamar, Aq Kupruk and Hazar Sum). On the basis of discoveries in surrounding territories, it is not at all unlikely

73

the palaeolithic

SHAPED TOOLS

CORES AND DEBITAGE

Table 2.1  Late Palaeolithic flint industries of northern Afghanistan AK II

AK IIIA

AK IIIB

KK I

KK II

KK III

KJ

DK Ib

DK III

Cores (Flake/Blade)

5

5

1

2

3

25

0

2

0

Cores (Microblade)

18

7

1

2

0

0

1

18

5

Flakes

1597

552

70

5

5

15

111

6

7

Blades

48

21

10

8

0

27

0

42

46

Microblades

512

99

0

0

0

8

43

569

80

Retouched (Flake/Blade)

43

26

15

12

9

22

9

15

12

Endscrapers

11

8

6

6

1

13

1

34

27

Burins

24

15

5

2

0

0

1

3

0

Notched

18

9

4

4

3

3

0

1

2

Denticulate

1

1

3

0

1

2

3

2

1

Perçoir

2

0

0

0

0

3

0

7

5

0

0

1

1

25

0

2

0

743

115

42

23

143

169

701

185

Carinated Totals

2,279

AK = Aq Kupruk; DK = Dara-i Kalon; KK = Kara Kamar; KJ = Kok Jar.

that Afghanistan would have been occupied by Middle Palaeolithic peoples. The Zagros mountains in Iraq and Iran have revealed numerous Middle Palaeolithic sites (the earliest known occupations in that area). In Central Asia the presence of typologically Middle Palaeolithic assemblages is well-documented. Similarly in Pakistan there is no doubt of a Middle Palaeolithic presence. In a broad evolutionary perspective it appears that vast areas of Central Asia were thinly inhabited if at all until perhaps around the beginning of the Middle Pleistocene, when we find the earliest evidence of inhabitants in the Tajik loess deposits. At that point Lower Palaeolithic populations began to move and become established in increasingly interior areas of the continent. Exactly what combination of behaviour patterns and technology made this expansion possible is as yet unknown, and it remains an exciting area for future research. As yet there is no direct evidence of this process in Afghanistan. Unfortunately, knowledge of late Mid-Pleistocene environments in Afghanistan and surrounding regions is extremely limited, and hence the adaptive context of the Lower Palaeolithic expansion cannot be specified. It is known that the Middle Palaeolithic inhabitants of the Iranian Plateau, the Turan Lowlands and Afghanistan were successful hunters. They appeared to have concentrated on essentially modern ungulates such as the onager (Equus hemionus), aurochs (Bos primigenius), sheep (Ovis orientalis), goat (Capra hircus), red deer (Cervus elaphus) and gazelle (Gazella subgutturosa).

74

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 2.7  Kara Kamar, Level II, Late Palaeolithic (Scale 1/1): (1 and 2) flake cores; (3) core rejuvenation flake; (4) edge retouched primary flake; (5) edge retouched broken blade; (6) transverse scraper.

On a worldwide basis during the Middle Palaeolithic, adaptations to several hitherto sparsely inhabited environments evolved. These included the mid-latitude tundras, equatorial rainforests and, also, many mid-latitude arid forest-steppes and highlands. It is in the latter environmental zone that continued study in Afghan prehistory may be predicted to make a valuable contribution. The Late Palaeolithic The Late Palaeolithic is the best-known interval of hunting and gathering activity in Afghanistan. All of the sites are found north of the Hindu Kush in the semi-arid steppe zone where the present mean annual precipitation is less than 250 mm and there is an open steppe vegetation. All the known sites, with the exception of the Tashkurghan sites discovered by Gouin and surface localities discovered by Vinogradov north east of Tashkurghan, are within the northern foothills of the

the palaeolithic

Hindu Kush at elevations ranging from 700 to approximately 1100 m above sea level. (The Tashkurghan sites are located in the Turkestan plain at an elevation of approximately 370 m.) The present-day vegetation of the foothill zone, beyond the irrigated village gardens and shimmering poplar stands, is characterised by low shrubs and small and open stands of pistachio trees (Pistacia vera). Relief is considerable in the foothill zone. For example, within a 10 km horizontal distance of Aq Kupruk, the elevation ranges from 700 to 2,000 m. The landscape generally consists of faulted and folded ranges which are separated by river valleys. These valleys vary in form from narrow and deeply incised channels to open, terraced systems with flood plains. Topographically, this is ideal terrain for sheep and especially goat. The late Pleistocene and early Holocene environments are not well-known, because there has been little Pleistocene geological or palynological investigation in this area. In general, however, it is not unreasonable to conclude that the late Pleistocene environment was considerably colder than, and at least as arid as, today’s climate. Bar-Yosef summed up the Last Glacial Maximum climate in South-west Asia as cold and dry throughout the entire region.40 In Afghanistan the environment probably would have been even cooler and more arid, a combination producing a rather sparse steppe. Stephen Porter estimated an approximate 1,000 m depression for the full-glacial snowline of the Hindu Kush on the basis of the mapped extent of glaciation and its altitude distributions.41The source of precipitation in northern Afghanistan today is the same as it was during most of the Pleistocene. Storm tracks leading eastward from the Mediterranean bring the winter and spring rains. According to H. Bobek, the Siberian high pressure system (a semi-permanent winter anticyclone) was intensified during the Late Pleistocene and thus served to reduce the penetration of the Mediterranean low pressure storms into the Iranian Plateau and on into Afghanistan.42 General humidity (precipitation minus evaporation) may have been somewhat higher during cold phases of the Late Pleistocene, because lower temperatures would have had the effect of reducing evaporation and thereby increasing general humidity. This had the effect of raising lake levels in some cases. How this might have affected Late Palaeolithic settlement pattern is unknown. Palaeolithic studies will make few major advances until the Pleistocene environments in Afghanistan are further studied. Kara Kamar Level III Level III at Kara Kamar (Figs 2.8–2.10) is one of the most important Palaeolithic discoveries in Afghanistan. Typologically, the lithic industry from this layer is Upper Palaeolithic, and the chronological indications are that it is very early Upper Palaeolithic. Level III is composed of loess mixed in with occupational debris. Its lower boundary is marked by a contact with a ‘brown cave earth’ according to Coon, but its upper boundary has no definite stratigraphic indication. Instead, the boundary between Levels II and III was determined by Coon on the basis of lithic typology

75

76

THE ARCHAEOLOGY OF AFGHANISTAN

Table 2.2  Calibrated radiocarbon dates for the Palaeolithic in Afghanistan Lab sample #

Locality

14C age bp

1σ Calibrated bp

2σ Calibrated bp

Reference

P-53 W-224 W-225 W-226 P-42 P-49 P-50 P-51 R-274 HV-1358

Kara Kamar I Kara Kamar III Kara Kamar III Kara Kamar III Kara Kamar III Kara Kamar III Kara Kamar III Kara Kamar III Dara-i Kalon III AqKupruk II

10,580+720 34,000+3000 >32,000 >32,000 >25,000 >25,000 >25,000 >25,000 9475+100 16,615+215

11,290–13,122 35,355–42,038

10,281–14,159 33,775–47,773

10,581–11,069 19,770–20,325

10,501–11,144 19,540–20,578

Coon and Ralph 1955 Coon and Ralph 1955 Coon and Ralph 1955 Coon and Ralph 1955 Coon and Ralph 1955 Coon and Ralph 1955 Coon and Ralph 1955 Coon and Ralph 1955 Alessio et al. 1967 Dupree 1968

and lithology. According to L. Lattman, the geologist who inspected the Kara Kamar section in 1969, loess deposition in northern Afghanistan is not directly related to specific climatological causes. It is not the result of the deflation of glacial outwash plains as is the case in parts of Europe. Rather, the Afghan loess has its origin in the fine alluvium of the Amu Darya, and it is transported by the prevailing north wind. Loess deposition in northern Afghanistan may be a more or less continuous phenomenon and is found in both glacial and interglacial periods. The Kara Kamar Level III Carbon 14 determinations were made on charcoal fragments, and they consistently indicated an age greater than 25,000, and perhaps than 32,000, years (Table 2.2). How much older is not known, because the determinations were made in the middle 1950s with equipment the maximum range of which had been exceeded. In any case, the Kara Kamar Level III industry is at least broadly contemporary with the earliest phases of the Baradostian Upper Palaeolithic of the Zagros. The lithic industry of Kara Kamar Level III was manufactured from a homogeneous, locally abundant, nodular flint. The complete lithic industry excavated in 1954 is no longer available for study because a large number of unretouched flakes, blades and various forms of chipping waste were discarded at the time of excavation, and some of the retouched flakes and blades were subsequently lost. Nevertheless, the extant collection at the National Museum and the University of Pennsylvania Museum of Archaeology and Anthropology, as well as the drawings and photographs made at the time of the excavation, give a fairly clear picture of the general level of technology and of the major tool classes present. My tabulation and interpretation of the entire collection became possible after the Penn Museum collection was finally located in 2004.43 The retouched artefacts were made primarily on blades. This is evident from an analysis of the length and width dimensions of the retouched artefacts and the unretouched blanks. The mean length/width ratio for all retouched artefacts was 2.3 L × 1W. The lithic industry was classified into several components and major tool classes. The result of the classification is shown in Table 2.1.

the palaeolithic

Figure 2.8  Kara Kamar, Level III, Upper Palaeolithic (Scale 1/1): (1) bladelet core and bladelet; (2) retouched blade; (3) bladelet core and bladelet; (4) unretouched blade; (5 and 6) bladelet core; (7) retouched blade.

77

78

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 2.9  Kara Kamar, Level III, Upper Palaeolithic (Scale 1/1): (1–6) carinated scrapers/bladelet cores.

the palaeolithic

No blade cores were found in the collection, but bladelet cores of the type illustrated in Fig. 2.8 were present. There were a number of bladelets produced from these cores as well as a number of small curved bladelets – many of microblade dimension – which were the products of the carinated scraper manufacture. The bladelet cores and carinated scrapers formed an intergrading series. Figure 2.8, nos 1 and 3 show bladelets which actually fit on the core, shown alongside. None of these bladelets of either variety which I examined were retouched or appeared to have any noticeable use wear. The retouched artefacts have long been cited in the literature as containing ‘Aurignacian’ or Aurignacoid elements.44 The basis for this attribution is the morphology of the carinated scrapers. Indeed, there is a resemblance between the carinated scrapers from Kara Kamar and those from the Levantine Aurignacian. For example, from Mugharet el-Wad at Mt Carmel, Garrod illustrates several specimens which appear to resemble closely those from Kara Kamar Level III.45 Similar carinated scrapers, however, can be found in non-Aurignacian contexts. They appear in the Kebaran, an Epi-Palaeolithic manifestation in the Levant which is at least 10,000 years younger than the Aurignacian. In other words, carinated scrapers are a poor type-fossil or chronological marker because they have a wide distribution in time and in space. Several of the carinated scrapers from Kara Kamar Level III are illustrated in Fig. 2.9. It is the case that the bladelet cores mentioned above and the carinated scrapers form an intergrading series. This fact makes use of the carinated scrapers from Kara Kamar as diagnostics for ‘Aurignacian’ culture even more problematic because they are subject to a great degree of variation. There are no burins of any kind in the Kara Kamar Level III lithic industry. This is not an expected characteristic of an early Upper Palaeolithic industry, particularly an ‘Aurignacian-like’ one. Nor does it have such Aurignacian elements as Dufour bladelets, Font-Yves points or split based bone points. In fact Kara Kamar Level III is quite simple typologically. It consists of the carinated scrapers, endscrapers, marginally retouched and use retouched blades and a few notched pieces (Fig. 2.10 illustrates three additional endscrapers). No doubt the full typological repertoire of the Kara Kamar III people is not represented at this site; it is probable that the tools in Kara Kamar Level III are related to a few specific functions. Table 2.1 lists the frequencies of the major tool classes at Kara Kamar. Kara Kamar Level III remains a unique find. Since its discovery in 1954, nothing similar in date or typology has been found in Afghanistan. Further to the north in Uzbekistan and Tajikistan, however, discoveries and reinterpretations of already known sites have raised provocative new possibilities for the Upper Palaeolithic of Central Asia. In Uzbekistan just east of Tashkent, excavations at Kubulak have brought a view of a unified Upper Palaeolithic tradition which extends from a local origin in the Middle Palaeolithic up to the Mesolithic.46 The investigators introduced the term ‘Kubulakian’ to bring together a number of sites through technological and typological comparison, thus linking a series of Central Asian sites to a shared Upper Palaeolithic tradition. Included in this comparison is Kara

79

80

THE ARCHAEOLOGY OF AFGHANISTAN

Kamar Level III as well as the well-known sites of Shugnou in Tajikistan and the Samarkandskaya site in Uzbekistan. This Kulbulak tradition variant ‘was evidently an outcome of the gradual evolution of local terminal Middle Palaeolithic and transitional industries such as those of Khudji and Obi-Rakhmat’.47 Kolobova et al.’s typological comparison and interpretation relies heavily on carinated cores which are said to be highly similar throughout this large area, but their illustrated examples show considerable variation. As pointed out above, carinated scrapers or cores are exceedingly widespread and time transgressive. Other artefact similarities must be evaluated carefully. In addition, chronological control is very weak in Central Asia; there are exceedingly few published radiocarbon dates for the Upper Palaeolithic or Epi-Palaeolithic. Hence the unity and local origins of the Central Asian Upper Palaeolithic must remain a working hypothesis for the time being. Otte and Derevianko48 have conjectured a possible origin for the Aurignacian in Central Asia, sensu lato, on the basis of their interpretation of Anuy and UstKarakol as Aurignacian sites in the Altai. They identify Kara Kamar as Aurignacian as well as the Samarkandskaya site in Uzbekistan. Given that they do not find a gradual transition from the Middle Palaeolithic to the Aurignacian in the Levant or Turkey, they instead look to Central Asia as a likely hearth. Because of the paucity of sites and the limited basis for comparison with chronological control, their grand scheme of culture history has little support. It is important, however, that they do point out the Aurignacian character of the Altai sites and hence the mobility of these early Upper Palaeolithic settlers. Following the Kara Kamar Level III occupation there is a large temporal gap before the next sites are encountered in the Afghan archaeological record. A look at the C-14 determinations listed above shows a hiatus of approximately 15,000 radiocarbon years. It might be suggested that this gap is an artefact of the relatively little amount of Palaeolithic survey which has been conducted in this area. A gap of similar magnitude, however, apparently exists in the Zagros mountains, and lack of Upper Palaeolithic sites in Central Asia has already been mentioned. Ralph Solecki has attributed this scarcity of human population in the Zagros to the deterioration of climate during the main Würm stadial.49 This appears to be a very plausible inference which has garnered increasing support.50 Although Frank Hole and Kent Flannery have indicated that there is evidence that the Zarzian developed ‘directly out of the Baradostian’ at the site of Pa Sangar in the Khorramabad valley, Iran, the continuity of occupation is not well-documented.51 In fact there is a temporal gap of approximately 15,000 radiocarbon years between the Baradostian and Zarzian, a period which spans the Last Glacial Maximum (LGM). It is clear that the mere depression of temperature would not have dislodged hunting and gathering bands from the foothills of the Zagros and Hindu Kush. It is well-known in other Eurasian contexts that by Middle Palaeolithic times, midlatitude tundras were inhabited, and by Upper Palaeolithic times severe sub-arctic environments were exploited. The technology for cold temperature adaptation (fire, shelter, clothing and specialised hunting gear) was no doubt available to

the palaeolithic

Figure 2.10  Kara Kamar, Level III, Upper Palaeolithic (Scale 1/1): (1–3) endscrapers; (4) notched flake.

the Upper Palaeolithic denizens of highland South-west Asia and Afghanistan. The highland habitat under the cold and dry steppe conditions of the Last Glacial Maximum (LGM), however, perhaps did not support sufficient numbers of the gregarious ungulates, and, more critical to human populations, digestible vegetation was not at all abundant on the cold steppe. It seems highly probable, therefore,

81

82

THE ARCHAEOLOGY OF AFGHANISTAN

that human populations might have initially descended to lower elevations and perhaps settled near the Amu Darya in places where streams from the Hindu Kush joined the Amu Darya or formed deltas as can be seen today. The fragmentary traces of their camps have possibly been discovered by the surveys by Vinogradov described earlier. It is also quite possible that the entire region was uninhabited periodically during the LGM. Some investigators,52 however, strongly question an abandonment of Central Asia during the LGM, but the paucity of sites, lack of chronological control and an almost complete absence of climatic data gathered directly from Middle and Upper Palaeolithic localities put clear conclusions out of reach. A great deal of research will be required to solve this problem. The Epi-Palaeolithic The most abundant and well-known finds from the Afghan Palaeolithic are found approximately 15,000 years after the Kara Kamar Level III occurrence. Collectively, these sites may be classified as Epi-Palaeolithic, a designation which refers to the appearance of microlithic elements in the late Pleistocene and early Holocene. Indeed, in Afghanistan there appears an exceptionally fine microblade technique at this time which possibly might have developed from an Upper Palaeolithic carinated bladelet core technology long present in Central Asia. Near the town of Aq Kupruk, Balkh Province, several Epi-Palaeolithic localities were discovered and excavated by Dupree (Figs 2.11–2.19). Figure 2.11 shows the location of the town in a widened portion of the Balkh River valley, and Fig. 2.12 the valley from the west. Both north and south of Aq Kupruk the Balkh River is contained in a narrow gorge whose vertical limestone walls often exceed 100 metres in height. The Epi-Palaeolithic sites are found north of the town near the point where the Balkh River re-enters the steep-sided gorge. Aq Kupruk III is an open-air, stratified site found in river gravels of the lowest terrace of the Balkh. Aq Kupruk II (Fig. 2.13) is a large shelter located half a kilometre downstream from Aq Kupruk III and is within the gorge. The shelter measures approximately 60 m wide at the drip line and about 14 m deep. The back of the shelter is 68 m from the river, and at the surface the back wall is 12 m above the stream level. The stratigraphy of Aq Kupruk II was rather complicated, but may be summarised as follows. The deposit is composed of three elements: roof fall (limestone), loess and alluvium. The alluvium interfingers with the loess and roof fall on the slope outside of the shelter, but it does not extend beyond the drip line towards the back of the shelter. Alluvial deposits were only found lower in the section than the Epi-Palaeolithic level. The Epi-Palaeolithic assemblage was found in a single thin layer (25–40 cm thick) which contained charcoal, bone and flint artefacts. It sloped sharply towards the river. Dupree’s excavation trench sampled perhaps only 10% of the actual occupation surface of the cave and most of the flat occupation area under the drip line remains to be excavated. Further up the section are Neolithic layers and a historic Iron Age layer. No pollen analysis has been completed on the

the palaeolithic

Figure 2.11  Plan of Aq Kupruk showing the location of the archaeological sites.

Figure 2.12  View of Aq Kupruk from the west. The wadi, Gala Qudug, is in the foreground.

83

84

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 2.13  The Shelter of Aq Kupruk II. The Balkh River is in the foreground.

section and climatic interpretations of the stratigraphy are not well-developed at the present time. The Epi-Palaeolithic assemblage consisted almost entirely of flint artefacts made from a locally available nodular flint of high quality. Figure 2.14 shows two nodules in limestone bedrock which were found a short distance from the shelter. A single carbon-14 determination for the Palaeolithic layer of Aq Kupruk II was made on a solid piece of charcoal and the result was (Hv 1358) 16,615 ± 215 bp.

the palaeolithic

Figure 2.14  Flint nodules in situ near Aq Kupruk (Dar-i Archa).

This date is not unreasonable on typological grounds, but certainly, more determinations must be run before the age of the Epi-Palaeolithic layer can be known with more confidence. A determination of (HV 1355) 10,210 ± 234 bp has been made for the early Neolithic of Aq Kupruk II and this time range must be the upper limit for the Epi-Palaeolithic. The primary flaking techniques of the Aq Kupruk II lithic industry consisted of flake/blade production from simple single and multiple platform cores, and the manufacture of very fine microblades by pressure technique from exceptionally small prismatic bullet-shaped cores. Illustrations of the cores and their products are shown in Figs 2.14 and 2.15. No blade cores of the type which produced the kind of blades illustrated in Fig. 2.15, nos 3 and 4 were found at this or any other Epi-Palaeolithic site in Afghanistan. Figure 2.16 shows several microblade cores and microblades from Aq Kupruk II. The microcores show several common characteristics. The cores had distal preparation to make a pointed end. The striking platform (actually pressure platform) formed a 90° angle with the long axis of the core. This platform was invariably unfaceted and had a slightly concave contour. Several of the cores and microblades showed external signs of heat treating which was applied to facilitate the pressure removal of the microblades. Figure 2.16, nos 8 and 9 represent the kind of very fine marginal retouch present on a small percentage of the microblades. All in all, the microblade technology was very sophisticated, involving heat treating, pressure

85

86

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 2.15  Aq Kupruk, Eip-Palaeolithic: (1 and 2) flake cores; (3 and 4) use retouched blades.

removals, some kind of vice to hold the core steady and, obviously, considerable skill and experience on the part of the artisan. Although the technique of the microblade technology is fairly well-known, the function of these implements is not well-demonstrated at all. The most likely

the palaeolithic

Figure 2.16  Aq Kupruk, microlithic component, Epi-Palaeolithic: (1–6) microblade cores; (7–9) marginally retouched microblades; (10–15) unretouched microblades.

interpretation is that they were part of a projectile system, perhaps serving as barbs on a shaft. Both the microcores and blades are found in shelters as well as in the two known open sites of the Afghan Epi-Palaeolithic. It may be inferred from this contextual information that microblades were not only utilised in base camps, and that they are related to some extractive activity. The discovery of a microblade

87

88

THE ARCHAEOLOGY OF AFGHANISTAN

core at the Kok Jar surface site is difficult to interpret, because it would seem unlikely that microblades were being manufactured at small temporary camps. The microblade cores and microblades of the Epi-Palaeolithic are significant in many respects. First, they are found in all the Epi-Palaeolithic assemblages of northern Afghanistan. Similar prismatic microblade cores are found in southwestern Iran from the early Neolithic,53 and near Kerman,54 but nowhere are they found in Late Palaeolithic Zarzian contexts55 or in the Late Palaeolithic southern Caspian sites.56 In Central Asia similar microlithic techniques are known but only in a late ‘Mesolithic’ contexts and in early Neolithic sites from Uzbekistan.57 Several sites surveyed by Vinogradov along the southern margins of the Amu Darya have very fine microlithic technology including small prismatic microcores, but they are assigned to the ‘Mesolithic’ or Neolithic.58 A second feature of the microlithic industry at Aq Kupruk II, as well as all the other Afghan Epi-Palaeolithic sites, is that there are no geometric microliths or backed microblades, nor even obliquely truncated microblades. Of the 520 microblades in the Aq Kupruk II sample, only 31 (5·7%) were modified by secondary or use retouch. The secondary retouch consisted in all but one case of an extremely fine marginal retouch. The lack of geometrics and truncated elements is unusual in any microblade industry, and it perhaps represents an early phase in microblade technological evolution. In the Levant, for example, Ofer Bar-Yosef has characterised the evolution of the microblade technology in the Kebaran as a continuous development from simple microliths to industries including oblique-truncated bladelets and narrow micropoints. This is followed by the emergence of geometric manufacture which gradually increases in proficiency to the developed forms of trapeze-rectangles, reaching its final achievement in the production of lunates.59 This sequence in the Levant originates in the Palestinian Late Aurignacian and continues through the Geometric Kebaran, a span of approximately 10,000 years. It is tempting to suggest a similar sequence in Afghanistan beginning with the small bladelets struck from the carinated scraper/bladelet cores of Kara Kamar III and ending with the Afghan Epi-Palaeolithic microblades and cores of Tashkurghan. The problem is, of course, that the chronological hiatus discussed earlier between them is quite large, and it is consequently difficult to demonstrate any direct technological continuity between them. Small microblade cores and microblades are found in McBurnery’s site at Ali Tappeh on the southern shore of the Caspian sea.60 The microblade cores at Ali Tappeh, however, are quite distinct from the ones of northern Afghanistan. At Ali Tappeh the microblade cores were made on the margins of thick flakes and generally did not have microblade removals around the entire perimeter. Another major component of the Aq Kupruk II tool-kit was the steep endscraper (Fig. 2.17). Unlike the Kara Kamar Level III carinated scrapers/bladelet cores, the

the palaeolithic

Figure 2.17  Aq Kupruk, Epi-Palaeolithic: (1, 3) steep endscrapers; (2, 4) microblade core preparation forms/steep endscrapers.

examples from Aq Kupruk II have short bladelet removal scars and a steeper mean edge angle. There is some good evidence that the preform for the steep endscraper was highly similar to the preform for the microblade cores. In addition to the steep-ended scrapers, several standard endscrapers were present, and four examples are illustrated in Fig. 2.18. Dihedral burins, burins on retouched truncations and burins on snapped truncations formed the second-largest group of retouched stone artefacts. Three

89

90

THE ARCHAEOLOGY OF AFGHANISTAN

Table 2.3  Faunal remains from the Palaeolithic levels of Aq Kupruk II Species

Frequency

%

Ovis orientalis cycloceros Capra hircus aegagrus Ovis/Capra Cervus elaphus sp. Bos/Cervus Equus Canis aureus sp. Vulpes

 8  9 63  1  5  1  2  1

  9  10  70   1   6   1   2   1

Totals

90

100

­ ihedral burins and one burin on a snapped truncation are shown in Fig. 2.19. Edge d retouched blanks were the most frequent major tool class. This category consisted of laterally and distally trimmed flakes and blades exclusive of endscrapers. One of the most significant aspects of the discovery at Aq Kupruk II was the analysis of the faunal remains associated with the Palaeolithic deposit. Dexter Perkins61 identified the bone as to species with the results shown in Table 2.3. Clearly, the overwhelming number of species represented were sheep and goat (89% of the identifiable bones). It seems evident that these two species were the predominant meat resources for the Late Palaeolithic peoples in northern Afghanistan. As mentioned previously, hunter-gatherers of the foothills of the northern slopes of the Hindu Kush had had long familiarity with sheep and goat, their habitats and behaviour, since the Middle Palaeolithic. At Aq Kupruk III there are two distinct strata which have yielded EpiPalaeolithic assemblages. The upper one (AKIII–B) is nearly identical typologically and dimensionally with the Aq Kupruk II industry and it is inferred that they are roughly contemporary. The lower stratum (AKIII-A), however, presents a different picture. There, the microblade technique is not represented, nor are the steepended s­ crapers. The length and width dimensions of the retouched artefacts from AKIII-A are significantly larger than either AKIII-B or AK II. Aq Kupruk III-A is as yet undated by Carbon-14. This cultural stratigraphy is paralleled at the Dara i Kalon site discussed below. In the Aibak region the Epi-Palaeolithic is represented by the Kara Kamar level I, Kok Jar and Dara-i Kalon sites (Fig. 2.20). Technologically and chronologically all of these sites are very close to the assemblages from Aq Kupruk with some slight variations. None of them is so strikingly different in technology or typology that it should be included in a separate archaeological ‘culture’. Aibak is approximately 100 km from Aq Kupruk, is located in the same foothill zone, and is not isolated by any geographic barriers from Aq Kupruk. The location of the sites gives some indication of the settlement pattern. The Kok Jar surface site (Fig. 2.21) consisted of a single concentration of artefacts which were deflated from a thin layer of soil on a mesa top. The mesa was an erosional remnant preserved by a freshwater limestone cap-rock which was approximately 30 m above the valley floor. Figure

the palaeolithic

Figure 2.18  Aq Kupruk, Epi-Palaeolithic: (1–4) endscrapers.

2.21 shows the mesa and the arrow indicates the location of the site. A large area of the valley floor can be surveyed from this vantage point. It would have made an ideal observation post for a hunting party. The Dara-i Kalon rock-shelter, test excavated by Puglisi in 1965, is located in a dry wadi channel which has cut into the limestone bedrock to a depth of approximately 30 m. The shelter is narrow and long and probably was only used as a transitory station by small hunting groups. Mussi’s analysis of the lithics from the five levels of Dara-i Kalon clearly showed close similarity to the Aq Kupruk sites.62 Namely, in the lower levels of Dara-i Kalon there are no microliths, but they come in significant number beginning in Layer III. Mussi well demonstrates the typological/technological changes which occur in the Dara-i Kalon sequence, which may well, as she points out, be reflected in the Aq Kupruk III A and B succession.

91

92

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 2.19  Aq Kupruk, Epi-Palaeolithic: (1) burin on a snapped truncation; (2, 4) dihedral burins; (3) endscraper-burin.

Several artefacts from Kara Kamar, Level I are illustrated in Fig. 2.22. The microblade cores are similar to those from Kok Jar, Dara-i Kalon and Aq Kupruk. Like Kok Jar, Kara Kamar is a good observation post and may have been frequented by small hunting parties.

the palaeolithic

Figure 2.20  Plan of Aibak vicinity showing the location of the Palaeolithic sites.

Figure 2.21  Kok Jar Epi-Palaeolithic surface site.

The dating of Epi-Palaeolithic AqKupruk II, Dara-i Kalon and Kara Kamar level I relies on a single determination from each site. While Dara-i Kalon and Kara Kamar level I are in good agreement, Aq Kupruk II is some 6,000 radiocarbon years earlier. Given the close resemblance between Aq Kupruk II and Dara-i Kalon, this

93

94

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 2.22  Kara Kamar, Level I, Epi-Palaeolithic (Scale 1/1): (1–2) microblade cores; (3) endscraper; (4–5) retouched broken blades; (6) notched flake; (7) carinated scraper/bladelet core; (8) side scraper.

age disparity given by the radiocarbon determinations may be too great. More determinations are definitely needed. It seems clear that groups of hunters and gatherers exploited the broad interfluves adjacent to the Samangan River during the period of climatic amelioration at the beginning of Holocene times. Coon’s preliminary faunal analysis of the Kara Kamar I level indicated that gazelle was the primary species hunted, along with wild sheep. Mussi also reports the presence of gazelle at Dara-i Kalon as a result of J. Clutton-Brock’s faunal analysis of the site.63 The lack of gazelle at Aq Kupruk

the palaeolithic

may be explained by the qualitatively different terrain described earlier. It is simply unknown whether any riverine resources were utilised in the Aibak region, or whether the hunting and gathering groups seasonally moved further into the Hindu Kush during the summer and out to the Turkestan plains during the winter. In 1969, 1975 and 1976, Vinogradov of the Soviet–Afghan Archaeological Expedition located a large number of surface sites which he designated as Mesolithic and Early Neolithic.64 They were in a zone running east to west between Tashkurghan and Andkhoi on the lowland and arid Turkestan plain. The vast majority of these sites were discovered near the contact of the sandy desert and the alluvium deposited by the streams emanating from the Hindu Kush. The sites contain many geometric microliths and small blade tools. Vinogradov compared these sites to many known in the eastern Caspian region and southern [ex-Soviet] Central Asia. He emphasised their clear differences from the Kuprukian sites of the Hindu Kush foothills. Thus, northern Afghanistan may well have been a zone of contact between distinctive prehistoric populations at the end of the last glacial and on into the Holocene. Gouin and Dupree separately discovered some Epi-Palaeolithic surface sites near Tashkurghan. Gouin’s two sites are located on the flat deltaic plain just north of the city and the material was discussed as ‘Tashkurghan 40’ in his preliminary 1973 report, but his unpublished report, as well as his doctoral thesis, differentiates two Epi-Palaeolithic sites: Siah Rigan, 20 km north of Tashkurghan (his original ‘Tashkurghan 40’) and Shah Tepe, 23 km to the north-west.65 Dupree had also previously recognised possible Epi-Palaeolithic material on a surface site 3 km east of the modern town itself.66 Siah Rigan is a deflation site where the artefacts are exposed on top of a stabilised dune. Flint artefacts were collected off the surface over an area of more than several thousand square metres, as well as from Shah Tepe. No concentrations of artefacts were reported, and Gouin has interpreted the widely dispersed artefacts to have been locally transported by flooding of the Samangan River. The lithic industry from Gouin’s sites consists of over 3,000 pieces of flint of which about four hundred were tools. They all showed weathering from both wind and water. This industry is quite distinct from the Epi-Palaeolithic of Aq Kupruk and the Aibak vicinity and is probably from a later time period. This conclusion is based on the following points. First, the microblade component included backed elements, triangles, rectangles and lunates. Microburins were also present, which is also indicative of microlith manufacture. Second, the microblades were produced by a different method from Aq Kupruk or Aibak. At Tashkurghan, the microcores were made on a variety of forms, some of which were similar to the microcores at Ali Tappeh, Iran. None of the bullet-shaped prismatic microblade cores, however, were found. Third, true burins are almost entirely lacking from this industry. Gouin has interpreted this site to be representative of a seasonal campsite, and he predicted that other manifestations of this culture will be found in the northernmost foothills of the Hindu Kush.

95

96

THE ARCHAEOLOGY OF AFGHANISTAN

In 1976, Dupree and I made an archaeological reconnaissance in the Dasht-i Nawur, a highland volcanic-tectonic depression located approximately 60 km west of Ghazni in central Afghanistan (Fig. 2.2).67 The Dasht is a large, flat grassland plain with an average elevation of approximately 3100 m and is about 60 km long and 15 km wide. At the base of two low hills in the northern part of the Dasht, two surface concentrations of obsidian tools were found. The two sites were designated G.P.2 and G.P.4 (Ghazni Province). They were essentially lag deposits on a deflation surface. This is the first known occurrence of obsidian in Afghanistan in either a prehistoric or historic context.68 The source of the obsidian is almost certainly local, but the exact source location has not been identified by elemental composition analysis. The geochemical analysis of the obsidian tools clearly differentiated them from any previously known sources in the Near East. The Dasht-i Nawur, therefore, should be added to the short list of obsidian sources in South-west Asia. At present there are no reports of this obsidian source being utilised outside of Afghanistan. On the basis of technology and typology, G.P. 2 and G.P. 4 should be assigned to the Epi-Palaeolithic although it is possible to assign them to an early Neolithic. The collection can be described as a bladelet and microblade industry with rare backed elements, geometrics and burins. Slightly over 98% of the entire industry is obsidian, and the remaining fraction is flint. The microblades were produced from small cylindrical microblade cores (pencil or bullet cores), and only 7% of them were retouched or backed. Nevertheless, the presence of backed elements and geometrics argues for a later date than would be associated with the Kuprukian. G.P.2 and G.P.4 are the first Epi-Palaeolithic sites found south of the main divide of the Hindu Kush. They are the only Palaeolithic or Epi-Palaeolithic sites in Afghanistan with obsidian artefacts. Future survey and excavation in this area will undoubtedly reveal several more. Summary The above review of the Afghan Palaeolithic is more tantalising than satisfying for a number of reasons which should be obvious to the reader. Clearly, Afghanistan is an important Palaeolithic area, and continued research there will be significant for prehistoric studies. To date, however, the work has only been preliminary and has indicated fewer conclusions than unresolved problems. During the Lower Palaeolithic, Afghanistan south of the Hindu Kush may have been part of the South Asian world. To the north, it certainly would have fallen into the orbit of Central Asia. No well-documented Lower Palaeolithic localities, however, have yet been found in Afghanistan, but there has been little concentrated effort to find them. Search is made difficult because of the rapid alluviation and erosion caused by tectonic uplift and also by thick accumulation of loess in the north. Despite these obstacles, Afghanistan is a promising source for the Asian Lower Palaeolithic record in both cultural-historical and adaptational terms.

the palaeolithic

There is no question that by Middle Palaeolithic times Afghanistan was inhabited. During this period there was an expansion of human populations into higher latitudes, and the movement into the continental interiors of Asia is well-documented. Whether Afghanistan received populations from South Asia or South-western Asia during this period is unknown, and the typology of the Darra-i Kur artefacts in northern Afghanistan give no definite indications either way. The important fact, however, is that a population exhibiting anatomically modern sapiens characteristics69 was hunting sheep in the north during the Middle Palaeolithic, and this hunting adaptation continued in importance all the way into the Holocene. There is no evidence for a ‘smooth’ typological transition from the Middle Palaeolithic assemblage of Darra-i Kur to the early Upper Palaeolithic Kara Kamar Level III lithics. The distinctiveness of the two assemblages, however, is not adequate ground for concluding that there was no continuity in occupation during this time or that Afghanistan played no role in the evolution of Upper Palaeolithic technologies. The available C-14 determinations from the two sites, and the absence of additional sites, unfortunately do not provide enough information to help resolve this issue. The early Upper Palaeolithic from Kara Kamar Level III remains somewhat of an enigma: zagadochnyi or ‘enigmatic’, as Vinogradov termed it.70 It may well represent an eastern penetration of the early Upper Palaeolithic of South-west Asia which ultimately reached all the way to the Altai. It is clearly not related to anything in South Asia and bears only limited similarity to sites in the former Soviet Central Asian republics. It is roughly contemporary with the Baradostian of the Zagros but is technologically and typologically distinct. To describe Kara Kamar III with the epithet ‘Aurignacian’ or ‘Aurignacian-like’ because of the presence of carinated scrapers and no other typical Aurignacian traits is really to engage in word magic, and it does not increase our understanding of its origin or relationship to the early Upper Palaeolithic. Physically, Kara Kamar is a small shelter and only a few major tool types are present. This may indicate that the site was a transitory hunting station. Hence, it is not a good ‘type site’ for comparative typological purposes. The Epi-Palaeolithic is known from several sites, the most notable and welldocumented being Aq Kupruk II and III and Darra-i Kalon. The indications are that there was a heavy dependence on sheep and goat hunting within the northern folds of the Hindu Kush. In more open locales, as in the Aibak area, gazelle was also an important species. There is absolutely nothing known about the plant foods gathered in northern Afghanistan during the Epi-Palaeolithic. Wild barley is known from the region, and some simple grinding tools are known from the Aq Kupruk sites. It is possible, therefore, that this annual grass was exploited. Many years ago, Dupree stimulated the editors of Scientific American to recognise what he called the ‘2,500 mile revolution’.71 Namely, Dupree saw that the so-called Neolithic Revolution occurred not just in one place, but broadly contemporaneously from

97

98

THE ARCHAEOLOGY OF AFGHANISTAN

Macedonia to Afghanistan. That the process of plant and animal domestication was widespread and multi-centred was one of Dupree’s more important insights. As yet, none of the Epi-Palaeolithic sites gives any indication of a high degree of sedentarisation. The occupation levels are all thin and the areal extent of the sites are limited. It is probably safe to say that there was an increased overall population during the Epi-Palaeolithic compared to the early Upper Palaeolithic, but what the density was or what role it played in the early domestication of sheep at Aq Kupruk II (10,210 ± 235 bp) is unknown. A complex of interactions between human populations and wild sheep and goat in the Late Palaeolithic eventually led to their domestication, and northern Afghanistan was assuredly part of this process. Clearly, the long-term dependence on these species by the hunters and gatherers, changes in the species behaviour patterns, alterations in the early Holocene environment and the availability of plant foods are all variable phenomena which need to be measured more closely before the process of animal domestication can be better understood. We can hope that the time will soon come when Palaeolithic archaeologists and their colleagues can continue the initial work cut short some forty years ago. There is so much left to do.

CHAPTER 3

The Development of the Oxus Civilisation North of the Hindu Kush Henri-Paul Francfort, Bertille Lyonnet, Cameron A. Petrie and Jim G. Shaffer Introduction The later prehistoric periods of Afghanistan were essentially unknown until the French initiated excavations at Mundigak in the late 1950s. The importance of Afghanistan in later prehistoric times had been clearly demonstrated by the intensive prehistoric research conducted in former Soviet Central Asia, Baluchistan, the Indus valley and south-eastern Iran. However, it was only within the twenty-five years prior to the cessation of fieldwork in 1979 that knowledge about these periods in Afghanistan transcended the realm of mere speculation (Fig. 3.1). Since the first edition of this book there have been several major publications presenting research conducted in Afghanistan in the 1970s, and a considerable amount of new research in neighbouring areas. This research combines to increase our understanding of the later prehistory of Afghanistan significantly. The French surveys of eastern Bactria recorded large numbers of Chalcolithic to Iron Age sites, dramatically expanding our knowledge of the development and settlement of the region.1 An unexpected outcome of this survey was the discovery for the first time of an Indus Civilisation or Harappan settlement in north-eastern Afghanistan, Shortughaï, and its subsequent excavation,2 which led to a radical revision of our understanding of cultural connections across the region. Added to this have been considerable new developments in the regions immediately bordering Afghanistan. The collapse of the Iron Curtain made, first, a considerable body of former Soviet research available to international scholarship, and second, the former Soviet republics accessible to international researchers. Hitherto little-understood Bronze Age settlements were defined, initially by Viktor Sarianidi, as a single ‘Bactro-Margiana Cultural Complex’ (hence BMAC) delineated in the regions north of Afghanistan. The broader term ‘Oxus Civilisation’ was subsequently proposed by Henri-Paul Francfort, to encompass the BMAC and the evidence from northern Afghanistan and the neighbouring Central Asian countries. Connected to this development, new evidence and interpretations from the western borderlands of South Asia and from south-eastern Iran3 have placed the later prehistory of northern Afghanistan in a broader context. The information now available clearly indicates that the

100

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.1  The Bronze Age in Afghanistan and surrounding regions. Inset 1: Fig. 4.1; Inset 2: Fig. 3.25; Inset 3: Fig. 3.33; Inset 4: Fig. 3.3.

prehistoric populations which inhabited Afghanistan underwent fundamental socio-cultural changes and ecological adjustments that permitted the two most important transformations in later prehistory to occur: the development of an economy based upon domesticated plants and animals, and the development of stratified societies. In the first edition of this book, the matching chapter reviewed the evidence for the adoption of domesticates and early mobile pastoralist and sedentary behaviour in Afghanistan and the surrounding regions, followed by an overview of the development of stratified societies in Afghanistan. This second edition maintains a considerable amount of the original text, but revises, updates, expands and reorganises it. The most significant structural change comes from the fact that the overview of the development of stratified societies has been divided into two parts – with the discussion of the evidence from north of the Hindu Kush being presented in this chapter, and the evidence from south of the Hindu Kush in the following chapter. In both instances, the evidence from Afghanistan is contextualised in relation to that from the surrounding regions.

the oxus civilisation north of the hindu kush

The Adoption of Domesticates: the Neolithic and Chalcolithic Periods The process of domesticating plants and animals is considered to be one of the most important transitions in human technological and cultural development. It is arguably comparable to only three other such transitions: initial tool use by the earliest hominins, development of an urban way of life and development of an industrial economy. Each of these transitions represents fundamental alterations both in the relationships between humans and the physical environment and in their relationship with others. Although detailed discussion of the theoretical and methodological issues surrounding the development of domesticates is tangential to this summary of Afghan prehistory, the Afghan material does have some important theoretical implications that make it worthy of discussion. Even before the cessation of fieldwork in 1979 our knowledge concerning this problem from Afghanistan was limited to the results of excavations at only three sites: Ghar-i Mar (Snake Cave), Ghar-i Asp (Horse Cave) and Darra-i Kur (Valley of the Blind), as well as surveys to the north of Balkh.4 The late Roland Besenval and collaborators carried out a reappraisal of the Neolithic of the Balkh Ab and Amu Darya valleys in 2007–9,5 but there has been no new fieldwork on the Neolithic within Afghanistan after 1979. Ghar-i Mar, Ghar-i Asp and Darra-i Kur were excavated by Louis Dupree as a part of his overall research into the Palaeolithic period of Afghanistan,6 and provide all the known data about technological and sociological changes indigenous in this area. All three sites are located in north-eastern Afghanistan. On the basis of stratigraphic evidence and associated changes in the artefact inventory, Dupree defined Non-Ceramic Neolithic A and B, Ceramic Neolithic and ‘Goat Cult’ Neolithic (the latter since dated post-Neolithic: see below) periods of occupation.7 The NonCeramic and Ceramic Neolithic periods were identified only at Ghar-i Mar and Ghar-i Asp. At the time of the excavations and preliminary reports, research into Afghan prehistory was still in its infancy. Ceramic identifications were very provisional, there were few radiocarbon dates for the whole country, and AMS methods were not available. It is only in hindsight that both ‘Ceramic Neolithic’ and ‘Goat Cult Neolithic’ have been recognised as having later dates than initially proposed. The putative Neolithic ceramics have been re-dated to the fifth–sixth centuries ad,8 which is something that Dupree himself initially suspected,9 and the Goat Cult was recognised even in the first edition to be Bronze Age, and this has since been confirmed by later studies.10 Nevertheless, despite possible confusion arising from such initial misattributions, the results for the Non-Ceramic Neolithic are still valid, and remain the only Neolithic excavations so far for Afghanistan, with the rest of our knowledge coming from surveys. Because of the special nature of the finds, the ‘Goat Cult Neolithic’ will be discussed separately as part of the Bronze Age, but otherwise the initial terms can be discarded.11 Since the crucial technological innovation was the domestication of plants and animals, it would be appropriate to present the floral and faunal evidence.

101

III III III III IV IV

Shortughaï

Shortughaï

Shortughaï

Shortughaï

II

Shortughaï

Shortughaï

II

Shortughaï

Shortughaï

II II

I

Shortughaï

Shortughaï

I

Shortughaï

Shortughaï

I I

I

Shortughaï

Shortughaï

I

Shortughaï

Shortughaï

‘Goat Cult’ Neolithic ‘Goat Cult’ Neolithic

Chalcolithic

Aq Kupruk I Ghar-i-Mar

Darra-i-Kur

Chalcolithic

Aq Kupruk I Ghar-i-Mar

Darra-i-Kur

Ceramic Neolithic Ceramic Neolithic

Ceramic Neolithic

Aq Kupruk I Ghar-i-Mar

Aq Kupruk I Ghar-i-Mar

Non-Ceramic Neolithic

Aq Kupruk I Ghar-i-Mar

Aq Kupruk I Ghar-i-Mar

Non-Ceramic Neolithic Ceramic Neolithic

Aq Kupruk II Ghar-i-Asp

Kuprukian B/Upper Pal.

Aq Kupruk II Ghar-i-Asp

Aq Kupruk II Ghar-i-Asp

Period/phase

Site

MC-1730

Ny-421

Ny-424

Ny-422

MC-1729

MC-1728

Ny-427

Ny-428

Ny-429

MC-2445

MC-1727

Ny-430

Ny-425

MC-1726

MC-2446

MC-2447

GX-0910-carbon

GX-0910-collagen

Hv-428

Hv-429

Hv-1356

Hv-1354

Hv-1357

Hv-425

UCLA_1363F

Hv_1355

Hv_1358

Lab sample #

Table 3.1  Radiocarbon dates from sites mentioned in the text

3534

3432

3180

3050

3620

3975

3710

4190

4375

3890

3570

4075

4040

3875

3890

3725

3425

3780

7220

7030

6310

6765

6955

8650

4500

10,210

16,615

Date bp

92

160

335

250

105

90

100

125

160

80

95

95

100

95

80

80

125

130

100

110

70

85

75

100

60

235

215

Uncertainty

2135–1640

2200–1400

2465–760

1960–770

2290–1695

2860–2205

2460–1880

3310–2460

3510–2580

2575–2140

2200–1685

2895–2350

2880–2305

2620–2040

2575–2140

2435–1905

2110–1440

2580–1830

6355–5895

6100–5675

5470–5075

5840–5530

5990–5715

8170–7520

3370–2945

10,630–9286

18,630–17,590

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.4

95.3

95.5

95.4

95.4

Calibrated range bc %

the oxus civilisation north of the hindu kush

Unfortunately, the floral evidence, including carbonised grains, was never analysed, and is not available. However, the faunal data were processed by Perkins and initially indicated that some important revisions should be made concerning the natural range of domesticable species.12 To obtain a comprehensive picture of the situation it is necessary to review some of the evidence from the Palaeolithic periods. At Darra-i Kur, a Middle Palaeolithic (Mousterian) cultural complex was identified and initially dated to 30300 +1900/-1200 bp,13 but it has also been argued that this is probably a minimum age.14 A re-analysis of human bone material from these deposits produced a date of approximately ~4500 bp/~2500 bc (3989 ± 31 bp (OxA-31781; 4530 and 4410 cal. bp/2580–2460 cal. bc), with the authors suggesting that this particular sample was intrusive from overlying ‘Goat Cult Neolithic’ deposits, from which an almost identical date was initially obtained ([GX-0910]; 3780 ± 130 bp/2580–1830 cal. bc; Tables 3.1, 3.2).15 The presence of this bone in the Middle Palaeolithic/Mousterian layers suggests that there was contamination in the excavated deposits, but the Mousterian age is not questioned.16 The degree to which the Middle Palaeolithic (Mousterian) layers were contaminated is unclear, but the associated fauna included unidentified sheep/goat species and possibly cattle (Bos primigenius). Identification of these potentially domesticable animals at such an early date in Afghanistan in a context suggesting exploitation by humans indicates that this area was within the natural geographical range of these species before domestication. Also important are the faunal remains found at Ghar-i Mar and Ghar-i Asp, which were recovered in association with the Upper Palaeolithic (Kuprukian A and B) cultural remains.17 One radiocarbon date of 16,615 ± 215 bp (18,630–17,590 cal. bc) is available for the earliest phase, Kuprukian B (Table 3.2). Fauna identified from these cultural phases included: sheep (Ovis orientalia cycloceros); goat (Capra hircus aegagrus); sheep/goat (Ovis/ Capra); horse (Equus sp.); dog (Canis aureus sp.); fox (Vulpes sp.); red deer (Cervus elaphus); and cattle/deer (Bos/Cervus). It is quite clear even from this limited evidence that wild predecessors of several domesticable animals were present in this area and exploited by humans. Furthermore, the number of specimens for these domesticable species (80 or 89% of the total identifiable remains)18 indicates some degree of selective exploitation of these animals. Similar contemporaneous patterns have been noted at several sites in former Soviet Central Asia,19 which demonstrate that this is not an isolated phenomenon. Although the analysis was carried out in the 1960s–70s, fauna from the NonCeramic Neolithic strata at Ghar-i Mar and Ghar-i Asp are referred to as domesticated by both Perkins and Dupree: ‘definite domesticated sheep and goat’ from both Ceramic and Non-Ceramic Neolithic and Chalcolithic as well as ‘a possible onager specimen’ in the Chalcolithic and ‘probably domesticated cattle … as well as Red Deer (Cervus elaphus), gazelle (Gazella subgutturosa), and horse (Equus caballus)’ in the Non-Ceramic Neolithic.20 Unfortunately, a discrepancy exists between fauna listed for ‘Non-Ceramic Neolithic B’ by Dupree and Perkins.21 It would appear that the fauna associated with the ‘Ceramic Neolithic’ by Perkins

103

104

THE ARCHAEOLOGY OF AFGHANISTAN

Table 3.2  Calibrated radiocarbon dates for the sites of Aq Kupruk II, Aq Kupruk I, Darra-I Kur and Shortugaï

is listed as being associated with ‘Non-Ceramic Neolithic B’ by Dupree. However, the only domesticated fauna for which this discrepancy is relevant is cattle. Therefore, domesticated sheep (Ovis sp.), goat (Capra hircus and Capra hircus spp.) and unidentifiable sheep/goat (Ovis/Capra) were found in association with all strata referred to as Neolithic. The cattle (Bos sp.) bones are identified as being possibly domesticated. A complete listing of the non-domesticated fauna would include red deer (Cervus elaphus sp.), gazelle (Gazella subgutturosa spp.) and horse (Equus caballus spp. and Equus spp.). With wild predecessors of domesticates in an Upper Palaeolithic context, and what have been interpreted as fully domesticated forms in later strata (Neolithic), having been identified, attention should focus on s­ pecific chronology.

the oxus civilisation north of the hindu kush

Table 3.2 (continued)

Two radiocarbon dates are available from the ‘Non-Ceramic Neolithic’ deposits (Table 3.2), one of 10,210 ± 235 bp/10,630–9285 cal. bc from Ghar-i Asp, and another of 8650 ± 100 bp/8170–7520 cal. bc from Ghar-i Mar, which are both very early in comparison to other ‘aceramic’ Neolithic phases in Central and

105

106

THE ARCHAEOLOGY OF AFGHANISTAN

South Asia.22 Four dates are available for the ‘Ceramic Neolithic’: 4500 ± 85 bp/3370–2945 cal. bc from Ghar-i Asp, and 6765 ± 85 bp/5840–5530 cal. bc, 6310 ± 70 bp/5470–5075 cal. bc and 6955 ± 75 bp/5990–5715 cal. bc from Ghar-i Mar (Table 3.2). Thus, the cultural assemblages designated as representing Neolithic appear to span six millennia. The confusion of such an extended time span is further complicated by the dating of a Chalcolithic (metal using) cultural assemblage at Ghar-i Mar, which is stratigraphically later than but chronometrically earlier than (6095–5675 cal. bc and 6355–5895 cal. bc; Table 3.2; see below), or partly contemporary with, the Neolithic assemblages. Traditionally, the aberrant dates on both ends of the chronological spectrum might have been disregarded as representing contaminated samples. However, a more meaningful interpretation might be provided by examining what these complex data may indicate if focused on two basic, and related, problems. What do these data mean in relation to (1) an early indigenous process for domestication in this area, and (2) the overall human adaptation to changing ecological relationships brought about by domestication of plants and animals? A single date for deposits containing what were interpreted as domesticated sheep/goat that spans much of the eleventh and tenth millennia bc is not in itself sufficient evidence for postulating an indigenous process of domestication, but it is earlier than the numerous dates for early domestication from sites located further west (post-11,000 bp/9000 bc)23 in sites. However, this single date potentially assumes new significance when considered with the identification of wild predecessors of domesticates in the Upper Palaeolithic levels at the same sites, which have been dated to 16,615 ± 215 bp/18,630–17,590 cal. bc (Table 3.2). The Upper Palaeolithic finds suggest that the necessary animals were in the area and being exploited by humans at an early date, and the fact that apparently fully domesticated fauna can be identified by 10,210 ± 235 bp/10,630–9285 cal. bc implies that this process had already occurred by that date. In the first edition of this book, it was suggested that, since biological characteristics denoting domestication were identified among these fauna, it was feasible that even before this date the processes responsible for such changes were already in operation: that is, this date may be interpreted as a terminal rather than an initial date demarking appearance of domesticated animals in this area.24 However, these animal bones have not yet been reassessed with more recent methods, and although it is tempting to speculate that transitional or initial domesticates might be discovered in Afghanistan that date even earlier (i.e. between 17,590 and 9285 bc), it must be acknowledged that identification of domesticable wild species and one very early date for apparently fully domesticated species forms are somewhat tentative bases upon which to hypothesise an indigenous domestication process. It is also arguable that the radiocarbon dates themselves are problematic. Before addressing the second problem of ecological relationships and the apparent long continuity of these Neolithic cultures, it is necessary to examine the presence of other associated artefacts.

the oxus civilisation north of the hindu kush

From the available information, there do not appear to be any architectural remains associated with either the ‘Non-Ceramic’ or ‘Ceramic’ Neolithic. Because of sampling factors, it is difficult to state that there was an absolute absence of architecture. A possible significance of this lack of architectural remains will be discussed below. The major artefactual difference which distinguishes these cultural complexes is the presence or absence of ceramics (Fig. 3.2). Stratigraphically and chronologically the ‘Non-Ceramic Neolithic’ is the earlier complex. It is the lithic implements which provide continuity between the ‘Non-Ceramic’ and ‘Ceramic Neolithic’, and also distinguish it from the Upper Palaeolithic. Besides lithic remains the only other artefacts encountered were manufactured from bone. These bone artefacts included such simple tools as polished points, plain and decorated needles, awls (punches) and spatulas. However, it is lithic implements which dominate the culture inventory and which must be discussed in detail. There is little distinction between the ‘Non-Ceramic’ and ‘Ceramic Neolithic’ lithics, but together they are very distinctive when contrasted with the underlying Kuprukian Upper Palaeolithic assemblages.25 This contrast is highlighted by the introduction of ‘sickle blades’ with the characteristic sheen on the edges resulting from the cutting of grasses. Other distinguishing artefacts were: one pressure-flaked unifacial leaf-shaped point; a bifacial point fragment; many notched flakes; hoes; ground-stone querns and pounders; celts; and steatite bowl fragments.26 Notched flakes occurred in the Kuprukian Upper Palaeolithic but were fewer in number. Other artefacts which demonstrated continuity with the Kuprukian are large blades, points and dihedral burins. The presence of sickle blades is extremely important, for it demonstrates that populations who were apparently exploiting domesticated animals were also exploiting some sort of grasses, and a similar speculation might be presented for hoes, querns and pounders. Until botanical remains from the relevant deposits have been identified (and dated), it is impossible to know whether or not the grasses being exploited were domestic. However, the total set of circumstances suggests that they might well have been so. Ghar-i Mar Non-Ceramic Neolithic A and B are basically similar, but there are some important differences. Dupree described Non-Ceramic Neolithic A as having a smaller percentage of sickle blades and more closely resembling the Kuprukian than Non-Ceramic Neolithic B.27 In addition, Non-Ceramic Neolithic A did not present the stone hoes, querns, pounders and steatite bowl fragments located with Non-Ceramic Neolithic B. Such items were also lacking in the Non-Ceramic Neolithic complex defined at Ghar-i Asp. The lithics of Non-Ceramic Neolithic B are described by Dupree as follows: Same as Ceramic Neolithic, but more sickle blades, plus cores, microblades, end and side scrapers, points, burins, occasional backed blades. One pressureflaked, unifacial point.28

107

108

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.2  Neolithic material from northern Afghanistan.

Almost exactly the same description is used for the ‘Ceramic Neolithic’ at both Ghar-i Mar and Ghar-i Asp. It is interesting to note that the ‘Ceramic Neolithic’ at Ghar-i Mar also lacked stone hoes, querns, pounders and steatite bowl fragments. The overall impression of these Neolithic assemblages is one of c­ ontinuity.

the oxus civilisation north of the hindu kush

Furthermore, there is nothing in the available information on the lithics to indicate that radical new tool inventories were introduced with the appearance of domesticates. There is also nothing to indicate that these Neolithic lithic assemblages did not develop out of the preceding Kuprukian. The ceramics from these sites were only briefly described in the preliminary report and, as noted above, are now recognised as being much later, so must have been intrusive into the Neolithic strata, though it is also possible that the strata themselves may also have been misattributed. Some of the ‘Goat Cult Neolithic’ pottery29 is certainly related to the steppe Bronze Age handmade pottery and can be compared with the Shortughaï steppe pottery,30 and the ‘goat burials’ themselves are known at Shortughaï31 and in the Oxus Civilisation (see below). Some metal artefacts, such as the circular mirror, and some pottery from Aq-Kupruk IV are also probably Bronze Age or from a later historical period.32 Early Mobile Pastoralists in Northern Afghanistan In attempting to interpret the Neolithic assemblages, Dupree alluded to the possibility that these archaeological assemblages are the material remains of specialised mobile pastoralists.33 However, it is not possible to base such an identification on any specific artefact or set of artefacts, and except for the presence of a posthole pattern suggestive of a tent-like temporary dwelling there is nothing in the artefact inventory which is distinctively ‘nomadic’. Furthermore, this interpretation precludes the use of a very specific subsistence strategy, so it might be more appropriate to describe the Neolithic occupation as being transient, and suggest that it was perhaps related to mobile pastoralist activity. An argument might be made that the absence of architectural phenomena in itself is suggestive of transience, and the likelihood that these populations were interacting with settled populations is suggested by the presence of tools usually associated with agriculture (sickle blades, milling stones, hoes, etc.). Some of these tools would be needed by pastoral groups to process cereals obtained by interaction with agriculturalists, or they might be needed as part of their tool-kit in compliance with providing a source of labour for agriculturists. Such tools might also have been necessary to enable the pastoral group to take advantage of wild crops or areas that could be dry-farmed, or they might have been utilised in any combination of the above circumstances. Perhaps the one physical factor which does argue for people engaging in pastoral activities is physical location in caves. Traditionally, caves have been a favourite location for temporary encampments of pastoralists. Given the limitations of the available data it appears that Dupree’s interpretation of these assemblages as representing mobile pastoral groups seems plausible. The existence of such groups contemporaneous with earliest sedentary village agriculturists would significantly alter various interpretative models about the socio-cultural and ecological contingencies surrounding the development of domesticates and the subsequent development of urban centres or civilisation.34 Given the difficulties of i­ dentifying

109

110

THE ARCHAEOLOGY OF AFGHANISTAN

nomads in this context noted above, and the likelihood that the populations using these caves were interacting with settled populations, to suggest that they were nomadic pastoralists almost certainly over-interprets the evidence, and the more straightforward suggestion that they were mobile pastoralists is preferable. The process of Neolithisation in Central Asia and the relationship with northern Afghanistan have more recently been considered by Brunet.35 In light of the proposed pastoralist interpretation for these early assemblages the presence of steatite bowl fragments is extremely interesting. Steatite bowls have been singled out as an important trade item throughout the Iranian plateau in later prehistory.36 The available dates indicate that trade of such items may have had a greater antiquity than previously expected, though, as with much of this discussion, this suggestion is speculative. Taken at face value, their identification in such a context might be interpreted as indicating that such objects (and by analogy other objects such as lapis lazuli, shell, obsidian) were focal points of an exchange system which facilitated symbiotic interactions between pastoral and agricultural groups. One final factor which should be taken into consideration is the long chronological persistence of these assemblages. One must always be cognisant of an extremely limited sample, but nonetheless some suggestion of the continuity of these assemblages can be proposed. If they actually represent early manifestations of a successful specialisation in a pastoral economy then one might expect a rather high degree of consistency in the basic aspects of the associated material culture. This would especially be the case given the rather generalised tool inventory that pastoralists must have had prior to the introduction of metal tools. A pastoralist explanation is also applicable to the contingencies surrounding the (misnamed) ‘Goat Cult Neolithic’ which is discussed below in the consideration of the Bronze Age.37 Early Sedentary Agriculturists in Northern Afghanistan Up to this point the discussion here has centred upon the description of Neolithic archaeological assemblages which appear to represent groups engaging in pastoralism. Ethnographically it is known, however, that pastoralists maintain multiple systemic interactions with specialised sedentary agriculturists, resulting in development of symbiotic relationships between the two groups.38 Therefore, it is postulated here that the domestication of plants and animals on the eastern parts of the Iranian Plateau resulted in an initial stage of mixed farming which in turn, due to ecological and demographic circumstances, resulted in the contemporary development of specialised subsistence behaviour based upon sedentary agricultural and pastoralism,39 although the degree to which these groups interacted or were connected remains unclear. At this point it then would be logical to present the evidence for the existence of such agriculturists. Archaeologically, such sedentary farmers would be differentiated from pastoralists by the contem-

the oxus civilisation north of the hindu kush

porary existence of sedentary villages with substantial architecture and associated artefacts: but to date in Afghanistan not one sedentary farming village has been located which is contemporary with the Neolithic archaeological assemblages from the various cave sites. This lack of evidence is not unexpected considering the current state of prehistoric research in Afghanistan and the presence of sites in neighbouring countries, which makes it likely that such sites are yet to be discovered. In this respect a series of low mounds located north of the Hindu Kush in the Tashkurghan region which are covered with microliths and other artefacts might be suggestive of the existence of such early farmers,40 as well as the sites surveyed along the Oxus north of Dashli by the Soviet Mission, but it is not possible to be sure simply from surface collections.41 No doubt future research will reveal such mounds in other areas, and testing the lowest levels of some of the thousands of mounds dotting the Afghan landscape north and south of the Hindu Kush will almost certainly identify such early sedentary agriculturists. However, a look beyond the boundaries of Afghanistan for examples of early farmers provides an analogous picture of such groups, and gives indications of Afghanistan’s likely place in broad-scale interactive processes. Developments in Neighbouring Regions and Connections to Northern Afghanistan Comparable cultural manifestations of the Neolithic sedentary societies likely to have existed in Afghanistan come from several neighbouring regions. Perhaps the most analogous comes from Central Asia and the cultural complexes referred to as Jeitun (Djeitun),42 Kelteminar43 and the Hissar Neolithic in South Tajikistan.44 Cognate sites in other regions include Sang-i Chakmak in north-east Iran, as well as Mehrgarh and a number of other early village sites in the borderland regions of Baluchistan and Khyber Pakhtunkhwa (formerly NWFP) in Pakistan, and also Tepe Yahya and other sites in south-eastern Iran (see below). Central Asia The Jeitun cultural assemblage is geographically proximate to Afghanistan, and has a well-recognised similarity in cultural material and development to that seen in Afghanistan in later prehistoric periods.45 Given the assumption here that such later similarities are a result of indigenous cultural processes taking place across this area and parts of northern Afghanistan, it is equally valid to assume that a certain degree of similarity existed between these regions in earlier periods. The following brief description of the Jeitun material is presented to help fill an information gap created by lack of research on contemporary evidence in Afghanistan. Jeitun-style material46 has been identified at several sites in southern Turkmenistan. Several phases of development have been defined, the latest dating to 6210–5770 cal. bc47 and therefore roughly contemporary with the Afghan

111

112

THE ARCHAEOLOGY OF AFGHANISTAN

Neolithic from Ghar-i Mar, Ghar-i Asp and Darra-i Kur. Grains of domesticated barley and wheat as well as domesticated sheep and goat remains have been identified at Jeitun sites, including the site of Jeitun itself.48 However, in the early phase of development hunting continued to have an important role, as indicated by the identification of several species of wild animals (gazelle, onager, wild pig and sheep, fox and wolf). Jeitun houses were rather standardised in proportions, dimensions, construction and layout. Construction was of mud-brick and the overall shape of the structures was basically rectangular or square. The average size of these habitations was 20–30 square metres. Every house had a courtyard containing outhouse structures, and these courtyards varied in size and were occasionally shared by two-dwellings. Besides habitation structures and courtyards possible grain storage platforms were excavated.49 A basic feature of the lithic tool-kit was microliths, both blades and various geometrics (trapezes, lunates and triangles). Important tools manufactured from such microliths were side scrapers, spoke-shaves, gravers and sickle blades. In later phases, there is a notable decrease of such microliths in favour of larger implements, particularly denticulate-edged sickle blades, which are also known from Mehrgarh (see below). From analysis of wear-traces it appears that a large number of scrapers were utilised in animal skin processing. Bone tools usually consisted of awls, gouges, needles, points, spatulas and an occasional shoulder-blade scraper. Other non-ceramic artefacts included various beads (bone, shell and semi-precious stones such as turquoise) and a small number of stone animal figurines. Jeitun ceramics demonstrate a rather well-developed industry that is broadly related to the ‘soft-ware’ tradition seen across the Iranian Plateau and many of its adjacent piedmont zones.50 All the ceramics are either handmade or manufactured on a very slow wheel and made from a chaff-tempered paste.51 Vessel shapes included globular jars, bowls and even small beakers. The decoration consisted mainly of red geometric motifs executed on yellow-buff backgrounds, which appeared on polished surfaces. Initially, most decoration consisted of simple parallel rows of wavy lines and vertical bracket-like lines, with triangular patterns being rare. In the later phases these motifs decline in favour of increased usage of rectangular designs, dotted patterns and triangles. Finally, in the third period … designs were often painted also on the inner surface of the pots. The decoration is in the form of undulating lines, vertical zigzags, and tree-like patterns.52 Figurines were both anthropomorphic and zoomorphic and were similar to statuettes found in the later Chalcolithic/Eneolithic of Turkmenistan at Altyn Depe and Ilgynly Depe, and as far south as Baluchistan and Khyber Pakhtunkhwa (see below).53 Zoomorphic forms included sheep/goat and bovids. Human figurines consisted of human faces modelled in the flat and seated figurines (female?). These

the oxus civilisation north of the hindu kush

figurines will be an important point of discussion in respect of the later prehistoric periods. The overall impression of the Jeitun cultural assemblage was related to prosperous agricultural development that was adapted to local ecological contingencies. The presence of semi-precious stones such as turquoise indicates the possibility of interaction with outside areas. Although an identical cultural assemblage may not be identified in Afghanistan, it is likely that something similar will be eventually identified, most likely in the northern areas. There are a few other Central Asian cultural assemblages which might be utilised for comparison and analogy like the Jeitun. The Kelteminar cultural assemblage (roughly 6300–3500 bc) is associated with communities primarily engaging in hunting and fishing in the lower Zerafshan and Amu Darya Rivers, but also cattle and potentially camel breeding, which suggests that the nature of Neolithic development in this region was different from that seen at Jeitun and the regions further to the west and south.54 Kelteminar pottery decoration consisted of incised wavy, zigzag and triangular patterns. Settlements consisted of pisé huts of very large size that were potentially communal. Eastern Iran The site of Sang-i Chakhmaq on the northern part of the Iranian Plateau is the north-easternmost example of an aceramic Neolithic settlement (c.7140–6825 cal bc) that shows material similarities to sites seen further to the west in the Zagros and beyond.55 Domesticated wheat and barley are present from the earliest levels, and small ruminants and caprines dominate the animal bone assemblage; occupation at the site continued into the ceramic Neolithic (c. 6230–5475 cal. bc), and it is not clear whether there was a break in the sequence. 56 In south-east Iran, Period VI at Tepe Yahya is another geographically proximate manifestation of such an early agricultural population (c.5875–3900 bc).57 Most of the ceramic and lithic parallels from Tepe Yahya lie with other sites in southern Iran and Makran in Pakistani Baluchistan, but this area of south-eastern Iran also demonstrates the close cultural affinities in the later prehistoric periods with developments in Afghanistan, as do the prehistoric cultures of Central Asia. There is also even earlier occupation in the Shah Maran–Daulatabad region to its east, most notably the site of R37 (Tepe Gaz Tavila), which is a large settlement with traces of abundant mud-brick architecture visible on the surface.58 There is also an aceramic Neolithic site close to Bam, Tell-e Atashi, which has material similar to that seen in the early phases at Mehrgarh, though occupation only dates to the fifth millennium bc (see below).59 Analogies between Tepe Yahya Period VI and possible developments in Afghanistan as a whole might be misleading, but we know nothing about Neolithic occupation in south-west Afghanistan, so it is the closest comparison that can be made from that quarter. There are, however, other cultural complexes

113

114

THE ARCHAEOLOGY OF AFGHANISTAN

that are ­possibly analogous to developments in Afghanistan, particularly those in the western borderlands of Pakistan, such as Mehrgarh, and other sites in north Baluchistan and Khyber Pakhtunkhwa.60 Pakistan We have only a relatively limited understanding of the pre-sedentary early Holocene occupation of the intermontane valleys and piedmonts of the parts of the Indo-Iranian borderlands that now lie within the bounds of modern Pakistan. Unlike in Afghanistan there have been no major cave sites excavated in this region, and potentially contemporaneous sites have only been identified through the discovery of microliths on the surface in various locations in Baluchistan and Khyber Pakhtunkhwa.61 The earliest evidence for sedentary occupation, agriculture and pastoralism thus far discovered in South Asia has come from Mehrgarh, which is situated at the bottom of the Bolan Pass at the northern end of the Kacchi plain. Mehrgarh was the focus of excavations by the French Archaeological Mission to Pakistan (1974– 85, 1997–2000),62 and is of importance to our understanding of early village life immediately to the east of Afghanistan. Mehrgarh is an archaeological complex comprising several separate areas of occupation and has a sequence that has been divided into eight major phases (I–VIII), which definitely span the period from c. 6000–2000 bc. It has been suggested that the site was established as early as c. 7000 bc,63 but this dating is not supported by well-stratified radiocarbon dates.64 The site was established in the Neolithic, and the earliest occupation is divided into three phases: Periods I (aceramic), IIA and IIB (ceramic).65 This is followed by several Chalcolithic phases labelled Periods III, IV and V, and there are also later phases that are effectively Bronze Age in date (VI–VIII).66 Grain impressions and charred remains preserved in mud-bricks suggest that the earliest Period I levels at Mehrgarh are characterised by a barley-dominated agricultural economy.67 Domesticated goats were present from the earliest levels, but wild species, including gazelle, wild goat, wild sheep, deer, wild buffalo and wild cattle dominate the faunal assemblage for Period I.68 The earliest inhabitants of Mehrgarh thus appear to have been hunting alongside the cultivation of domesticated crops and use of some domesticated animals. Mehrgarh Period I appears to be at least partly contemporaneous with the earliest aceramic levels at Kili Gul Muhammad (see below).69 Mehrgarh Period IIa sees the appearance of the first fired ceramic vessels, substantial mud-brick buildings, clear evidence for specific craft activities, and a marked development of the agricultural subsistence economy.70 The small numbers of fragments of handmade bowls and jars are similar to the Burj Basket Marked pottery seen at Kili Gul Mohammad,71 and use a technology that is stylistically related to the soft-ware potting tradition attested across the Iranian plateau during the sixth millennium bc and also seen at Jeitun.72

the oxus civilisation north of the hindu kush

During Period IIb ceramic vessels begin to be decorated with simple and then more complex motifs executed on a red slip.73 White fired steatite beads and a small number of copper objects are also attested suggesting development of various pyro-technologies.74 Although it has been suggested that Periods IIa and IIb cover the sixth millennium bc,75 the available radiocarbon dates suggest that Mehrgarh Period IIb occupation dates between c. 4700 and 4000 cal bc,76 and corresponds with occupation phases at other sites in the frontier regions, including Kili Gul Mohammad II, Sur Jangal I, Surab I, Rana Ghundai I and possibly early Periano Ghundai (see below). In Period III, the ceramic decoration became increasingly sophisticated, seeing the appearance of friezes incorporating complex geometric and zoomorphic motifs, which also appear in Kili Gul Mohammad Periods II and III.77 During this period evidence of lapidary and shell working was exposed, and the first fragments of copper and crucible fragments appear, which indicate local copper production78 and have led to this period being designated as Chalcolithic. Mehrgarh III is dated between c. 4000 and 3500 bc,79 and appears to correspond to Kili Gul Muhammad III (see below). Kili Gul Mohammad (hereafter KGM) is situated at the top of the Bolan Pass in the Quetta Valley of Pakistani Baluchistan, which is just across the border from the prehistorically important Kandahar region. Like Central Asia, the Quetta valley demonstrates a close cultural affiliation with contemporary developments in Afghanistan during later prehistoric periods, and there is a pronounced similarity between Quetta valley material and that of Central Asia itself. Unfortunately, more than a little controversy has centred upon the interpretation of the material excavated from the site, both with regard to method of excavation and the fact that only an extremely small sample of material is available, but it broadly parallels the sequence at Mehrgarh. Fairservis separated the KGM sequence into four phases.80 The levels dating to the earliest phase, KGM I, were only exposed in a small (1.75 × 1.75 m) area at the bottom of the sounding, but lack ceramics,81 suggesting that they are contemporaneous with Mehrgarh I.82 Several hand-made ceramic wares appear in KGM II, with parallels at Mehrgarh II, though some wheel-turned wares were also present.83 Distinctively decorated ceramic wares begin appearing in KGM III,84 which indicate that this phase is comparable to Chalcolithic period Mehrgarh III.85 In addition to the ceramic material, a limited number of bone, flaked stone and ground stone objects were recovered from KGM I–III, and bone and ivory beads and a fragmentary copper object from KGM III.86 A site which shows closer connections to the regions of Afghanistan north of the Hindu Kush is Sheri Khan Tarakai (henceforth SKT) in the Jani Khel/Wali Noor area of the Bannu basin.87 It is effectively a Neolithic site, but absolute dates and relative comparanda for cultural material indicate that it was occupied from the late fifth to the early third millennium bc, which make it contemporaneous with Chalcolithic Mehrgarh.88 The main period of occupation, which comprises up to seven separate stratigraphic phases, appears to have been between 3800–2900 cal.

115

116

THE ARCHAEOLOGY OF AFGHANISTAN

bc. Cultural material was scattered across an area of up to 21 ha, but it is unlikely that the whole area was occupied at one time. The ceramic fabrics, forms and decorative and surface treatment styles attested at SKT are most akin to those used during the second major phase of ceramic technological development in northern Baluchistan, which is referred to as the Togau Phase, and is evidenced in Mehrgarh Period III, Kili Gul Mohammed III, but the closest parallels to the ceramics from SKT are with material from other sites in the Bannu basin. The bioarchaeological evidence from SKT implies that the settlement was a sedentary agro-pastoral community. The inhabitants engaged in a broad spectrum of subsistence practices, including the cultivation of barley and wheat, and the management of domestic sheep, goat and cattle. They also engaged in the collection of a range of wild plant and wood species, and the hunting of wild animals. The range of wood species found at the site implies that some of the inhabitants were engaging in some form of transhumant pastoralism, into the uplands close to the modern border with Afghanistan.89 There was a range of evidence discovered at SKT that indicates that its inhabitants were engaging with far-reaching networks of interaction. A range of wood charcoal from high-altitude species was recovered from the site,90 and lapis lazuli and marine chank shell were also recovered, coming from the neighbouring uplands, northern Afghanistan and the Arabian Sea coastline respectively.91 The evidence for wood charcoal and exotic raw materials is complemented by the anthropomorphic terracotta figurines, terracotta cones and handled weights. There are no close parallels to these objects found at other sites in South Asia, but parallels have been identified at various sites on the Iranian plateau and in adjacent areas of Central Asia, particularly at Ilgynly-Depe in Turkmenistan,92 suggesting that there were degrees of direct or indirect contact between these regions. The most direct route from SKT to these areas was via the intermontane valleys of northern Afghanistan and the mountain passes of the Hindu Kush, so there is a reasonable likelihood that contemporaneous settlements will eventually be found in those areas. This summary provides a possible picture of what early sedentary agriculturists might have been like in south-east and north-east Afghanistan, although when contemporary cultural complexes are defined in these regions, they will doubtless present differences from as well as similarities with those groups. The transition to a subsistence dependent upon utilisation of domesticated plants and animals stimulated processes of cultural, economic and ecological change which resulted in even more complex ecological and cultural adaptations, which in turn resulted in another equally important transition to a socially stratified way of life. Development of Stratified Societies Archaeological research in Afghanistan concerned with stratified societies has been recognised both north and south of the Hindu Kush (Fig. 3.1), and, as noted

the oxus civilisation north of the hindu kush

above, the discussion of this process has been divided into two parts, with consideration of the areas to the north presented here. In northern Afghanistan there are five sites, or groups of sites, relevant to this discussion: the sites in Eastern Bactria; the pioneering excavations at Shortughaï; the Chalcolithic or Bronze Age levels at Ghar-i Mar; the so-called ‘Goat Cult’ of Darra-i Kur; and the Dashli series of sites. Although not fitting into any neat chronological sequence, they are discussed in this order below for convenience. In southern Afghanistan research has centred on the Kandahar region, with excavations mainly at the sites of Mundigak and Said Qala Tepe, and more limited evidence from excavations at Deh Morasi Ghundai, Old Kandahar and Nad-i Ali, together with surveyed sites in the Helmand-Sistan region. These are discussed in Chapter 4. The Eastern Bactria Survey93 At the time when the DAFA was excavating Aï Khanoum, Jean-Claude Gardin organised a large-scale survey on the plain where the site is located. The initial aim was to understand the economic background behind such a large Greek colony, recording systematically all visible mounds (tepes) and irrigation canals, many of which were abandoned at that time (Fig. 3.3). A huge amount of pottery was collected to support systematic relative dating. The unexpectedly rich discoveries made during three seasons (1974–6) led to further developments (in 1977–8) aiming at a full comprehension of the history of the settlement on a much wider area, from the Qunduz River to Badakhshan, and the Amu Darya/Darya-i Panj to the Hindu Kush, that is, in the region to the east of ancient Bactria. This second phase of the survey was carried out in a much more extensive manner, but it profited from all lessons from the first phase. The canals were dated by association with the dates given to the sherds collected on the adjacent tepes (Fig. 3.4). Altogether 823 sites were visited and a small number were excavated to confirm the periodisations suggested by the sherds collected. Three volumes have been published concerning this survey: the first is a study of the landscape and the irrigation canals,94 the second is the study of the pottery and a reconstruction of the settlement pattern from the Chalcolithic to the Arab conquest,95 and the third presents a description of the sites together with a critical synthesis.96 Gardin was especially keen on methodology and he argued in detail each of the steps used for the survey: from the search of sites and canals (on very precise Soviet maps), through the collection of sherds and their typological study, to the different inferences that could be drawn from all the data and, finally, critical considerations about some historical interpretations. Nevertheless, it is clear that the survey was exclusively oriented towards the plains and valleys where irrigation was feasible, leaving aside possible caves or cemeteries in the hills.

117

118

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.3  The Eastern Bactria survey.

the oxus civilisation north of the hindu kush

Figure 3.4  Abandoned canal, east Bactria.

The survey results have provided very important and completely new data on all periods. Not only did the survey reveal aspects of settlement history that had hitherto been completely ignored, but it also gave, for the first time, an uninterrupted overview of the history of a large area, instead of the more isolated view that long-term excavations at just one site give. At the same time, the Soviet Mission headed by Irina Kruglikova and Viktor Sarianidi was also carrying out an intensive survey north of Balkh. Their aim was more to make an inventory of the sites in order to excavate the most interesting ones, and no full publication of the survey per se has appeared. Indeed, it was during the course of this work that Sarianidi discovered and excavated the first remains of the Bronze Age ‘Bactro-Margiana Cultural Archaeological Complex’ (BMAC – see below). The following discussion will describe the proto-historical periods highlighted by the East Bactria survey, from the Late Chalcolithic to the end of the Bronze Age. It is based upon Bertille Lyonnet’s (1997) primary publication, but additional information has been added where appropriate. Information about the Iron Age through to the period of Macedonian conquest is included in Chapter 5. The Late Chalcolithic/Early Bronze Age (c. 3500 to 2500 bc) The oldest sedentary occupation recognised in northern Afghanistan is dated to the end of the Late Chalcolithic and beginning of the Early Bronze Age, and stands at the heart of the Taluqan valley (Fig. 3.3), mostly on natural hills along what

119

120

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.5  Example of a river diversion.

has been considered an artificial diversion of the river (Fig. 3.5).97 The seventy sherds on which this occupation is mainly based come from eight different sites. The sherds have no preserved decoration, but their specific technological features (shapes and one potter’s mark) are without any doubt comparable with pottery from Mundigak III, and with the ‘Baluchistan-like’ material found at Sarazm in the Zeravshan Valley (Tajikistan)98 (Fig. 3.6; see Chapter 4). Comparisons made with this material and with that from sites south of the Hindu Kush and in neighbouring regions (such as Period I at Shahr-i Sokhta in Iranian Sistan) suggest relative dates of between 3500 and 2500 bc, based on a new series of radiocarbon dates.99 A fragment of an artefact made of red-brown marble100 found on another site of the same valley (T. 250), and a roughly made stone-handled weight discovered on the surface of a modern tomb in the valley of the Rud-i Shahrawan (T. 270) close to Taluqan, can also both be dated to this broad period.101 Unfortunately, none of these small settlements was excavated, due to the belated recognition of their date and the unstable political situation after 1978. The Taluqan discoveries bridge the chronological gap between the rare Neolithic discoveries made in Bactria that are related either to the Hissar Neolithic102 or to the post-Aq Kupruk Neolithic103 (Ghar-i Mar, Ghar-i Asp and Darra-i Kur, which are considered elsewhere to be linked to the Kelteminar culture)104 – or to yet other Palaeo-, Meso- and Neolithic ‘cultures’ along the Amu Darya,105 and the now better-known Bronze Age periods in these regions

the oxus civilisation north of the hindu kush

Figure 3.6  Chalcolithic sherds from Taluqan.

121

122

THE ARCHAEOLOGY OF AFGHANISTAN

(see below). Up until now, however, they are unique for this period in the region north of the Hindu Kush, except for the exceptional settlement of Sarazm, which lies even further to the north, well outside Afghanistan. We cannot preclude the possibility that the Taluqan sites were only preserved because of their situation upon a natural hill, and that many others existed but have been washed out or buried under alluvium or later occupation. As mentioned above, the diversion of the river along which they are positioned is not natural and represents a first step towards later complex irrigation systems, but we have no other data to support this proposition. The position of these sites on the most direct road to Badakhshan and Sar-i Sang (Fig. 1.43), where the best lapis lazuli raw material comes from,106 is certainly significant. As is well-known, this semi-precious stone was highly prized in the ancient world,107 at least since the end of the fifth millennium bc, and increasingly so into the third millennium bc, with lapis objects being found in Iran, Mesopotamia, Egypt and South Asia (Fig. 1.42). At Sarazm, where several cultural assemblages from very distant geographical horizons are attested, a kurgan-like tomb that is related to the Afanasievo steppe culture by its construction, and should be contemporaneous with the Taluqan sites, contained a range of exotic goods, including hundreds of lapis lazuli beads.108 It was with the end of the Uruk period at the close of the fourth millennium bc (i.e. at the time of our Taluqan sites) that tremendous changes were occurring in the ancient Near East, with Proto-Elamite material culture penetrating deep into highland Iran109 and further east, as highlighted by the discovery of a single proto-Elamite tablet and several cylinder sealings related to the Djemdet Nasr koinè at Shahr-i Sokhta,110 and distinctive vessel types such as bevelled rim bowls at Mahtoutabad in the Halil Rud region of southeast Iran,111 and ShahiTump in Pakistani Makran.112 The nature of long-range contact was further transformed in the third millennium bc, leading to what has been referred to as the ‘Middle Asian Interaction Sphere’.113 The late fourth and third millennium bc were thus periods of unprecedented long-range contact, and exchange and trade relations. The End of the Early Bronze Age and Middle Bronze Age (c. 2500 to 1500 bc) The second period recognised is dated to the end of the Early Bronze Age and to the Middle Bronze Age, but we are not able to say if it immediately follows the first period. This period was first identified by a totally unexpected discovery of a group of settlements related to the Indus Civilisation.114 These sites were situated in the plain of Aï Khanoum, beyond the reach of contemporary canals, in an area that was deserted at the time of the survey. Our attribution was confirmed when one of them, Tepe 209, which was subsequently given the name of Shortughaï, had quantities of Indus-like pottery on the surface and was subsequently excavated by Henri-Paul Francfort (reviewed below). The surface sherds were not exclusively

the oxus civilisation north of the hindu kush

Indus-like, and it was possible to distinguish several phases of occupation on this group of settlements, and this likelihood was further supported by the extension of the survey. On the basis of the excavations at Shortughaï and other comparisons, three major phases have been identified within this period. The settlement pattern for the whole period clearly shows a striking development with new irrigation canals (Fig. 3.3). Phase I corresponds to the first two periods of Shortughaï (see below). The material culture from the site is clearly that known in the Indus region, whereas only faint connections can be drawn with the Sapalli Culture in Surkhan Darya. On the plain of Aï Khanoum, the potsherds for this phase come essentially from Shortughaï, which is the only site where painted ceramics have been found (Fig. 3.15), but another group of sites produced similar shapes and is situated on one of the old terraces along the Qunduz plain that we named Asqalon. In the Taluqan valley, two sites also produced some similar sherds, with one of them also being thought to have been occupied during the previous period (T. 299). Two other tepes, one along the Kokcha River and the other being the old site of Rustaq, also produced a few sherds of this phase. Phase II corresponds to Period III of Shortughaï. The material culture is still related to that of the Indus region at the end of its existence, but connections with the BMAC (Dzharkutan and Kuzali phases) appear now clearly for the first time. The total number of sites in the region increased significantly, although these settlements were situated in the same areas, that is, the area around Shortughaï, the Asqalon terrace, the Taluqan valley and two isolated sites, one in the Farkhar valley and the other at Kalafgan. Phase III corresponds to Period IV of Shortughaï and may have followed after a short hiatus (see below). Two sub-phases have been identified. The first sub-phase displays continuity with Phase II. In the plain of Aï Khanoum, the previous group of sites around Shortughaï was still occupied, but those at Asqalon disappear, while new ones are occupied in the plain of Archi and others were scattered in different small areas. The second sub-phase attests to the intrusion of new material and the location of sites mostly along natural rivers. The excavations of Shortughaï clearly show relations with the last phase (Molali) of the BMAC, with the cultural assemblages of Bishkent and Vakhsh, and with steppe/Andronovo populations known north of the Amu Darya.115 The material from the survey is not as varied. It is probably to this second phase that the (unpublished) material excavated on the Bala Hisar of Aï Khanoum by Litvinskij (chantier XX) should be dated. It includes a few complete vessels (probably from a grave) with Andronovolike(?) features, such as oblique incisions on the inside of the rim of a squat pot, together with other material linked more with the handmade pottery of the Yaz I/Burguljuk cultures, such as a circular lid with a central handle (Figs 3.7, 3.17). Altogether, the data gathered for Phase III testify to a decline with an absence of distant external relations, and deep transformations in the population although probably not a sudden collapse.116

123

124

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.7  Andronovo/Burguljuk/Jaz I related sherds from Aï Khanoum.

From a chronological point of view, this succession of three phases and their different cultural orientation is interesting so far as the formation of the BMAC is concerned, though the radiocarbon dates obtained from Shortughaï span a broad chronological range (at least 2500–1700 bc), making it difficult to provide absolute dates for each phase (see below). A few Indus seals and other items had been discovered in the late Namangan sequence of Turkmenistan, which suggested the existence of contact between the Indus region and the Kopet Dagh piedmont, and

the oxus civilisation north of the hindu kush

the discovery of Shortughaï suggests that these relations need not have been direct with the Indus region. The chance discovery of the Fullol Hoard, which was found close to the Shamash Plateau in the survey area and is clearly related to the BMAC phenomenon, must also be emphasised within this context,117 and Lyonnet118 has argued that ‘international’ trade was behind this extraordinary upsurge of the BMAC. Not only was the discovery of Indus settlements in the plain of Aï Khanoum astonishing per se, but the position of the sites clearly necessitated the development of a complex irrigation system since they were far from the river(s) and stood along visible remains of canals. Similarly, the sites on the Asqalon terrace suggest the existence of a canal. As discussed below, the excavations made at Shortughaï itself provide further confirmation of the existence of artificial irrigation.119 In contrast to the simple diversion irrigation that was likely in use during the previous period in the plain of Taluqan, the canals built on terraces such as Aï Khanoum and Asqalon during Phase II would have been a far more complex enterprise. Furthermore, their existence implies that there were many more s­ ettlements than those that have been identified, as sites potentially existed all along the canals and their branches. It can only be assumed that they have been obliterated by modern occupation and/or sedimentation. Curiously, no site seems to have reached any large size. It is puzzling that local material culture is absent in Phase I, which the excavations at Shortughaï suggest was wholly oriented towards the Indus region. Even though comparisons can be made with the Bactro-Margiana area in Phase II, strong links can still be drawn with the Indus. This, added to the Taluqan data during the previous period and its unique cultural orientation towards the south of the Hindu Kush, highlights a clear difference in the cultural orientation of the original population of the area east of the Qunduz River when compared to that of the rest of Central Asia. Finally, the end of this period is certainly marked by the arrival of new peoples related to the north, although this looks more like a slow infiltration than a sudden migration. It foreshadows the radical change that characterises the next period (discussed in Chapter 5). Shortughaï120 The protohistoric Indus and post-Indus (c.2500–1700 bc, see below) site of Shortughaï was discovered in 1975 in the plain of Dasht-i Qala/Aï Khanoum during a survey conducted by Jean-Claude Gardin in conjunction with Pierre Gentelle and Bertille Lyonnet (see above). The site was completely deflated, but a mound that appeared to have the remnants of a Hellenistic farm on top attracted the attention of the team. After identification of the protohistoric material on the surface (see above), a trial trench was sunk in 1976121 and larger-scale excavations carried out between 1977 and 1979. The excavated material was studied in Kabul in 1980 and given to the National Museum in 1981. The final publication appeared

125

Figure 3.8  Plan of Shortughaï.

the oxus civilisation north of the hindu kush

in 1989122 and various specific papers before that. The excavation of Shortughaï and its material provided the opportunity for a reappraisal of three questions: (1) the origin of the protohistoric complex societies of Bactria and Central Asia; (2) a re-evaluation of the cultural and chronological position of the Central Asian Bronze Age (BMAC or Oxus Civilisation); and, finally, (3) a fresh appraisal of the transition towards the Iron Age. The term ‘BMAC’ (Bactro-Margian Archaeological Complex) was coined by Viktor Sarianidi at the time of the first discoveries in the early 1970s, before the further researches that led to a better appreciation of the large spatial extension (see below), and demonstration of the early date, its broad chronological duration, and all that is now understood about its various external relations, from the Indus to Iran, Mesopotamia and the Levant and from the Inner Asian steppes to the Persian Gulf. This has led to its re-evaluation as a real Civilisation, and not simply an ‘archaeological complex’, encouraging use of the term ‘Oxus Civilisation’ rather than the restrictive BMAC. Stratigraphy and Architecture Surface topography and magnetometry survey The site consists of two main shallow mounds (Fig. 3.8), A and B. Mound A (c. 11,250 square metres, of which 4,600 were excavated) is ‘Indus’, and Mound B (c. 10,000 square metres, of which 4,000 were excavated) is ‘post-Indus’. Five soundings were also excavated. The two periods possess different cultural characteristics that appeared after the excavation.123 The possible extent of ancient occupation was indicated by the topographic and magnetometry surveys.124 On Mound A, excavation was made complicated by a great number of deep cylindrical pits, which were dug down to the natural soil in antiquity (before the Hellenistic occupation). Six stratigraphic levels were identified on Mound A (levels 1 to 6), and when the stratigraphy, architecture and material changes were synthesised across the site, the sequence was separated into four distinct Periods (I to IV).125 The material culture from Period I is typically Indus or Harappan, indicating the establishment of a ‘colony’ from the Indus basin. However, most of the materials and artefacts analysed were locally made, apart from those clearly imported from the Indus region (e.g. long cylindrical and etched carnelian beads; see below). The walls of the structures were thick and long, and constructed of bricks in Indus proportions (Figs 3.9–3.12).126 The Period II levels show some development of the material assemblage, but the walls were reconstructed on the same architectural layout. Floor levels are numerous, indicating continuous occupation, and the size of the mud-bricks of Indus standard remains the same. Storage installations127 and the work of artisans appear clearly: copper and gold processing (crucible fragments), as well as chunks and chips of lapis lazuli (see below). Period II has

127

128

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.9  Shortughaï Period I–II.

since been shown to have close parallels from a neighbouring site in Farkhor (Tajikistan).128 Post-Indus burials (Mollali and Bishkent types) appeared later (see below) and a farm was constructed on the Protohistoric mound during the Hellenistic period. Five stratigraphic levels were recognised on Mound B, and all were postIndus,129 except, perhaps, one possible Indus pit for firing pottery dug in the natural ground.130 The three first levels (Period III; Fig. 3.13) appear to be a local Eastern Bactrian Bronze Age variant having affinities with the Oxus Civilisation. Apart from pottery, there was a sheep/goat burial.131 The walls of this phase were made of pakhsa, and large clay storage pits were discovered, as well as pottery production objects (such as moulds).132 The last level on Mound B (Period IV) was clearly related to the Late Bronze Age of the Oxus Civilisation, known in Tajikistan and Uzbekistan as the ‘Bishkent’, ‘Mollali’ and ‘Bustan VI’ phase.133 Two late intrusive ‘Mollali’ and ‘Bishkent’ burials on Mound A (Fig. 3.14) belong to the same Period IV,134 as well as deposits of pots and clay artefacts on both mounds.135 Five more soundings gave more information on the sequence (soundings: F; I; O; BD).136 One trench revealed a Bronze Age irrigation canal, just at the foot of Mound A, along the edge of the natural terrace, but about 30 km away from the source in the Kokcha River (see above).137

the oxus civilisation north of the hindu kush

Figure 3.10  Shortughaï Period I–II.

129

130

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.11  Shortughaï Period I jar in situ.

Figure 3.12  Shortughaï wall Period II and intrusive Period Ib burial.

the oxus civilisation north of the hindu kush

Figure 3.13  Shortughaï Period III.

Figure 3.14  Shortughaï Period IV burial.

131

132

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.15 Shortughaï Period I Harappan sherd.

Figure 3.16 Shortughaï Period II pedestal dish.

Pottery Typology and Comparisons The study of about 20,000 potsherds (out of a total of 1 million sherds excavated) provided a typology, a chronology and relevant comparisons.138 A wheel-turned, homogeneous fabric which remained the same for the great majority of pottery for all periods was attested,139 and appeared alongside a raw fabric, which was used to produce handmade vessels using rotation.140 Some vessels appeared with engraved or printed ornaments,141 or black on red painted designs, mainly Indus in type, with depictions of ‘pipal leaves’, peacocks, intersecting circles, or painted bands (Fig. 3.15).142 Distinctive Indus shapes were exclusively attested in Period I,143 including dishes on stands (Fig. 3.16).144 Large jars, perforated vessels (for cheese making?), terracotta scoops, graffiti on edges of jars, and pottery rings for

Figure 3.17  Shortughaï ‘Andronovo’ sherds.

the oxus civilisation north of the hindu kush

manufacturing the lower parts of jars, were also recovered. Specifically Bactrian pots from burials of Period IV were also attested (Fig. 3.14).145 The presence of such material indicates at the least contact with Bishkent/Mollali groups in this area, though analyses of some sherds showed that their place of manufacture was most probably regional, except for one ‘Andronovo steppe’ sherd with a fabric that looks ‘foreign’ (see petrographic analyses below). The handmade, homogeneous or raw fabrics include a handful of ‘Andronovo steppe’ grey pottery vessel fragments (Fig. 3.17).146 The assemblage included various other terracotta artefacts,147 mainly Indus in style,148 apart from some Oxus square boxes149 and a camel figurine that could be either Indus or Oxus.150 Periods III and IV display a simpler assemblage, with a smaller number of different forms that lack painting, but with a great number of cylindrical goblets and basins. Statistical analyses151 correlating the find spots (strata and soundings) with the most distinctive and representative types showed an evolution of the pottery.152 Continuity and correlation between the levels and the locations is clear,153 but possible (short) breaks in the sequence may have occurred between Periods II and III. On the basis of the analysis of the pottery assemblages it was possible to delineate the evolution of forms and to distinguish two broad phases: Phase A (Harappan or Indus, grouping periods I + II) and Phase B (post-Harappan or post-Indus, grouping periods III + IV).154 Petrographic Analyses of Pottery155 Thirty-seven ceramic samples were studied petrographically, and four groups were isolated. The analyses confirmed the local manufacture for the pottery, with nothing indicating imports from the Indus basin. One ‘Bishkent’ sherd appears to be not local, but it was probably still from Bactria. More recent analyses performed using XRF, ICP/MS and lead isotopes analysis156 show that one of the two Q wares (Q = ‘cooking’ pottery of ‘crushed calcite’ group, samples nos 10, 11) that was analysed comes from a completely different area, and was possibly related to the presence of some ‘Andronovo’ vases (and potentially peoples) originating from the north (Figs 3.7, 3.17). Small Finds157 Small finds from the excavations were divided into ten categories according to raw material: a. Stone158 includes grindstones, pounders, scrapers, axes, whetstones. One fragment is probably from an ‘Oxus’ long ‘stone staff’ or ‘stone sceptre’,159 which has many parallels in Bactria and Margiana. Stone scrapers, from Phase B,160 seem specific to eastern Bactria and are found in Kashmir, but also in China: their presence might relate to the introduction of millet at Shortughaï during Phase B (see below).161

133

134

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.18  Shortughaï Period I–II etched carnelian beads.

Figure 3.19  Shortughaï Period I long carnelian bead.

b. Lithics162 include cores, flakes, blades, micro blades and tools that all confirm the well-documented phenomenon of continuity of stone technology and of the use of flint and other stones during the Bronze Age in Central Asia, alongside the utilisation of metal. c. Beads163 include etched carnelian; long as well as discoid, ring, tubular thin, tubular, spherical, barrel-shaped and unique shapes, variously made of gold, carnelian, agate, amethyst, faience, lapis, steatite, terracotta and more common stones. Some of the most remarkable objects in the assemblage were the Indus long tubular and etched carnelian beads (Figs 3.18, 3.19);164 the gold winged discoid is similar to Near Eastern beads.165 d. Lead166 includes plaques, pins and rings, which testifies to the importance in Bronze Age Central Asia of this metal, utilised since the Chalcolithic (e.g. at Sarazm).167 e. Copper alloys168 include pins (Fig. 3.20), tools, hollow tools, bladelets, a ring, a hook, a razor, mirrors, an arrow-head and a spearhead; these have parallels in the Indus or Oxus regions.169 The copper technology changed from mainly tools in Phase A to include other objects in Phase B, but did not decline, while other technologies declined or collapsed (see below). f. Bone objects are mainly tools170 on both mounds in both phases and include handles, points and small points.

135

the oxus civilisation north of the hindu kush

Figure 3.20  Shortughaï Period III bronze pin.

Figure 3.21  Shortughaï Harappan seal.

g. Shell bangles:171 these 42 fragments seem made out of Xancus pyrum from the Indian Ocean. They are among the northernmost examples of this shell. h. Terracotta objects:172 tokens, ‘pin or drill head’ spindle whorls similar to many from Early and Middle Bronze Age contexts from Sarazm to Altyn Depe. i. Miscellaneous objects173 include seals and clay objects. An Indus square seal depicting a rhinoceros and two pictograms was found on Mound B, where it appears to have been reused or simply moved from its original context (Fig. 3.21).174 Another conical seal is similar to the shapes of the post-Harappan ‘Jhukar’ seals.175 j. Stone vases176 are quite rare. The general study of the artefacts177 shows a shift from ostensibly Indus-type material towards ‘Bactrian’ or ‘Oxus’ assemblages, from Phase A to Phase B, confirming what was observed with the pottery. Archaeobotany178 Irrigated agriculture and dry farming provided subsistence crops, and complemented the resources from the gallery-forest along the Panj River and from the hills. The biotope of Shortughaï is diversified and rich for foraging, husbandry and agriculture, especially with the presence of an irrigation canal. Among the usual cereals, wheat and barley, the presence Panicum miliaceum before 2000 bc is especially interesting.179 A reconstruction of the Bronze Age environmental capacities and exploitation of the Shortughaï plain has been proposed (Fig. 3.22),180 which incorporates different growing zones including: low terrace (wild olive, tamarisk, Lycium, Salix, Zygophyllum); irrigated high terrace (Trifolium, lentils, wheat, sesame, peas, legumes, fruit trees, lucerne, vine); dry farming and hills (wheat, barley, linen, Eruca, chickpea); steppe (Astragalus, Aegilops, Phlomis); and loess hills (pistachio, Ephedra, almond, juniper, hackberry, hawthorn).

136

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.22  Map showing a reconstruction of the Bronze Age environment of the Shortughaï Plain.

the oxus civilisation north of the hindu kush

Archaeozoology181 Domesticated species make up 98% of the bones recovered, and include sheep and goat, cattle and zebu (Bos indicus), but also water buffalo (Bubalus bubalis). Zebu was domesticated in South Asia, but is also present in south-east Iran from the Chalcolithic, so its presence is not unusual, but water buffalo is certainly a specific introduction from the Indus region. The wild species attested include wild sheep, wild goat, deer (Axis axis), gazelle and wild ass. But, surprisingly, no wild boar (while present in Margiana, for instance) and no camel bones appeared (in general bones are rare in settlements, but at Shortughaï one figurine of a camel has been recovered). Interestingly, techniques of husbandry indicate that male sheep were killed young (30% between 1.5 and 2 years) leaving only selected reproducers, and almost all female sheep were generally killed at around four years. Fish bones are present too. The introduction of water buffalo and the absence of boar indicate an Indus basis for the management of the fauna, in a space where most of the predatory species were apparently eliminated. Metals182 The presence of lead, iron slag, cinnaber and copper is attested (Fig. 3.23). Some copper alloys appear with tin and/or arsenic. Very important is the presence of ‘pure’ gold, probably coming from washing the auriferous sands of the Panj River, originating in the Samti vein. In fact the presence of the site of Farkhor on the

Figure 3.23  Shortughaï Period I–II crucible fragments.

137

138

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.24  Shortughaï: raw lapis lazuli.

Tajik side of the Panj, exactly opposite Shortughaï,183 and the analysis of osmium traces in the gold from the ‘Royal’ Tombs of Ur,184 suggest the interesting possibility of early exploitation of gold from this area, which may then have been moved extended distances. Physical Anthropology185 The ‘Mollali’ and ‘Bishkent’ burials (so named after the analogies with the tombs of the cemeteries of Mollali-Bostan-VI – Uzbekistan – and Bishkent – Tajikistan; Fig. 3.14) on Mound A, but dating to Phase IV, belong to females of the same populations. Interestingly, four infant burials were found, one inside the settlement of Mound B while it was still functioning, sealed in a pot inside a wall of Period III. Chronology Sixteen radiocarbon dates were obtained for Shortughaï (Table 3.2),186 and although the calibrated dates are roughly consistent, some are definitely problematic. The initial interpretation of the dates suggested that the site was occupied for an extended period of occupation spanning c. 2500–1700 bc, and attention was drawn to a clear later phase spanning c. 2200–1700 bc. The initial analysis argued that these ranges span from the late part of the Early Indus or early part of the Mature Indus (c. 2500–2200 bc) down to the late phase of the Oxus (c. 1700–1500 bc).187 The possibility of a gap between the Indus (Periods I and II) and the Oxus (Periods III and IV – this last after another gap) occupation has been examined and discussed in detail.188 There was no consensus about the Oxus Civilisation in the 1970s and 1980s, but new discoveries and absolute dates relating to the Oxus Civilisation (see below), and a reappraisal of the Indus material, suggest that there was a gap in the occupation.189

the oxus civilisation north of the hindu kush

Recalibration of the Shortughaï dates (Table 3.2) highlights that the established chronology does not account for the full chronological range of dates, suggesting that some of the dates may be problematic, already highlighted in the initial report. For instance, recalibration of the Period I dates produces broader ranges extending back further into the third millennium bc (e.g. Ny-425: 4040 ± 100 bp, 2880–2305 cal. bc; Ny-430: 4075 ± 95 bp, 2895–2350 cal. bc), and forward into the second millennium bc (e.g. MC-1727, 3570 ± 95, 2200–1685 cal. bc), though this latter date was initially thought to represent contamination from later deposits.190 The earliest ranges are seemingly too early in comparison with the relative dates for the Indus cultural assemblage at sites like Harappa.191 In general, the Period II dates are statistically similar to those of Period I, though several are chronologically earlier (e.g. Ny-429: 4375 ± 160 bp, 3510–2580 cal. bc, Ny-428: 4190 ± 125 bp, 3310–2460 cal. bc), which again does not suit the relative chronology neatly. The Period III and IV dates span from the Mature Indus period and stretch into the mid-second millennium bc (e.g. MC-1729: 3620 ± 105 bp, 2290–1693 cal. bc; Ny-421: 3432 ± 160 bp, 2201–1401 cal. bc; MC-1730: 3534 ± 92 bp, 2135–1640 cal. bc), which broadly matches the originally proposed chronology, but shows some overlap with the dates of Periods I and II. It is worth noting that the radiocarbon dates from Shortughaï have large uncertainties, and they would clearly benefit from some re-evaluation involving Bayesian analysis, which is beyond the scope of this discussion. Ecology The mineral, vegetal and animal resources of the surrounding area were investigated as part of the project.192 Environment and resources were mapped and annual cycles of activities were reconstructed on the basis of the traditional local economy, month by month.193 From these observations, it was possible to note that 8,000 hectares of irrigated land (if all of it was cultivated) would require 40,000 person-days for harvesting (i.e. 1,300 individuals for one month). This is larger than the expected population of the settlement, but does not take into account the possibility that other sites existed between the source of the water and the location of Shortughaï. Nonetheless, this land was probably not all in cultivation (the 20 km long canal being just an aqueduct during a part of its course), and people from outside the area might have been employed to help temporarily, potentially coming from the hills and mountains.194 The scale of the cultivable area and the potential for producing cultivated fodder tend to diminish the need for seasonal transhumance. Furthermore, the Panj and Kokcha Rivers cannot be forded by flocks except in winter, which potentially explains one of the environmental constraints impacting upon settled populations in this area, and highlights the possibility of seasonal migration towards Badakhshan, Darwaz and Pamir, which is important for understanding pastoral seasonality and the behaviour evidenced in Period IV.195

139

140

THE ARCHAEOLOGY OF AFGHANISTAN

Economy and Society at Shortughaï196 The excavations at Shortughaï prompted a number of questions about the economic and social context of the occupation related to ‘urban’ Phase A (Periods I and II) and post-urban Phase B (Periods III and IV), and whether this transformation represented a decline or not.197 The questions of urbanisation and class societies have been at the core of many interpretations of the Indus (urban undoubtedly) and of the Oxus (possibly urban) Civilisations since the 1980s, and are connected with discussions about ‘class societies’, ‘hierarchies’, ‘[Early] States’, ‘complex societies’, etc. This is why an attempt was made at Shortughaï to study in more detail, and in a new manner, because it apparently provided evidence of the operation of the economy over a period of cultural continuity. For this goal, all material collected – not just that used for the usual typological chronologies – was subjected to a detailed level of spatial and chronological analysis incorporating assessment of the ‘chaîne opératoire’ and statistical analysis by using special ec0nomic indicators.198 The detailed analysis showed that fewer objects were processed over time, typologies became less complex, processing became less sophisticated, and some crafts were abandoned, while others were, by contrast, transferred and developed, etc. An evaluation of this approach and an interpretation were also proposed.199 This analysis described as accurately as possible, but did not tackle the causes of, this transformation from ‘urban’ to ‘post-urban’, and recognised this change in almost all Central Asia at about the same time, or during the same period of time.200 It is notable that as specific data were not collected on ancient environment, aspects of environmental or climatic causality were not considered beyond P. Gentelle’s studies (1978; 1989). Ghar-i Mar Although Dupree referred to the limited archaeological assemblages at Ghar-i Mar for this period as Chalcolithic,201 the analysis of the associated metal artefacts by Caley clearly indicates that its composition was a low-tin bronze,202 suggesting a later date. There is thus some likelihood that two radiocarbon determinations from these levels (Hv-428: 7220 ± 100 bp, 6355–5895 cal bc; Hv-429; see above: 7030 ± 110 bp, 6095–5675 cal bc; Table 3.2) represent contamination from earlier deposits. The metal artefacts consisted of three fragments of sheet metal with an embossed motif, two fragments of a rectangular rod and one other sheet fragment.203 Caley maintained that the composition of the metal was soft enough for the embossed design to be produced by hammering on a soft substance (wood) when the metal was heated.204 The reconstructed original composition of these artefacts indicates a very high percentage of copper, with about 7% tin and traces of iron and nickel, a composition which, according to Caley, is characteristic of the early stages of bronze metallurgy. Lithic artefacts were summarily described by Dupree as consisting of flint cores, sickle blades, blades, possible burins, perforators and endscrapers on blades;

the oxus civilisation north of the hindu kush

no geometries but many micro-blades were also associated with this level.205 Bone artefacts included points, awls and needles. The fauna identified at the site included domesticated sheep, goat, cattle and possibly onager,206 but a detailed discussion has not been published. Dupree also noted the possibility that molluscs might have been of some dietary importance.207 Available ceramic descriptions are incomplete208 and have subsequently been identified as being of much later date, and should hence be presumed to be intrusive.209 The late sixth millennium bc dates for these deposits containing evidence of sophisticated bronze technology would make it one of the earliest dates for such technology recorded, and the fact that bronze does not appear at any other site in Afghanistan or its hinterland until one and a half millennia later, together with the clearly much later date of the ceramics, makes it likely that there were major problems with the stratigraphy. These factors make it impossible to create any reliable cultural perspective for the Ghar-i Mar material. Darra-i Kur The archaeological assemblages recovered at Darra-i Kur and identified by Dupree as the ‘Goat Cult Neolithic’ represent a classic example of an archaeological misnomer.210 The designation ‘Neolithic’ conveys an antiquity and cultural contemporaneity for this archaeological assemblage which is neither warranted by the material nor implied by the excavator. The two radiocarbon dates from these deposits (GX0910 [bone collagen]: 3780 ± 130 bp, 2580–1830 cal. bc; GX-0910 [carbon]: 3425 ± 125 bp, 2110–1440 cal. bc; Table 3.2),211 in conjunction with the presence of associated metal artefacts, clearly indicate that this archaeological assemblage is contemporary neither with the Non-Ceramic nor the Ceramic Neolithic periods. Indeed, its major feature, ritual goat burials, has its closest cultural affinity with the Dashli series in northern Afghanistan (see below).212 There are three categories of artefact associated with the ‘Goat Cult’ that make it very distinct from the earlier material. The first is metal objects: three fragments of a low-tin bronze artefact were associated with this assemblage. Two of these fragments appear to be part of a pin while the third is a ‘tapered rectangular rod broken off at one end and having a conical tip at the other’,213 and after making a component analysis Caley came to the conclusion that these items are probably of local (i.e. northern Afghanistan) manufacture.214 The second major distinctive category of artefacts is the presence of architectural remains in the form of postholes, which were described as follows: A series of at least 80 postholes 2–4 cm in diameter was noted just under the lip of the cave, and may indicate the use of culinary racks, windbreaks, other shelters, tethering posts, etc.215 The presence of such postholes potentially indicates the presence of tents. The final distinctive category of artefacts is the ‘goat burials’. Unfortunately, due to

141

142

THE ARCHAEOLOGY OF AFGHANISTAN

the preliminary nature of the available report, the description of these burials is very brief: Three intentional pit burials of domesticated goats were uncovered. Two skeletons had been decapitated; one had the skull articulated … Directly underneath and possibly in association with Burial 3, skull fragments and several long bones of one or two children were discovered.216 Whether the context of these burials warrants interpretation as a ‘cult’ is debatable, but the articulated burial of domesticated goats certainly indicates the important roles these animals held within the social group involved. It is now clear that sheep/goat burials are frequent in the Oxus Civilisation, notably at Dashli217 and Shortughaï in Afghanistan,218 and at Sapalli in Uzbekistan.219 Animal burials involving a number of species appear also in Margiana.220 They appear all during the Bronze Age of Central Asia. Some artefacts from Darra-i Kur could have been manufactured in more settled agricultural sites, such as Shortughaï or the Bronze Age sites of South Tajikistan, which are relatively close. The faunal remains from Darra-i Kur were not extensively studied or discussed, but domesticated goat, cattle, onager and horse were identified.221 Wild fauna remains included fox, marten, gazelle, birds, fish, rodents and tortoise. The majority of lithic implements were of two basic types:222 (1) excellent flint blades (some with fine alternate retouch), and possible sickle blades, and (2) a series of relatively large diabase points with thickened cross section which exhibited extensive use along both edges and the vertical ridge. Other lithic implements included celts, a slate knife and pendant, a broken jasper point, slate scrapers, limestone blade and bead (cylindrical), steatite spindle whorl, obsidian bracelet(?) fragment, basaltic hammerstones, and a series of quartzite pebble tools. Bone implements included awls, needles, gouges, spatulae, polishers, polished sheep astragali (gaming pieces?), and one perforated long bone which may have been either an ornament or an amulet. One shell or ‘limestone’ perforated disc bead was also recovered. This inventory seems to represent a rather generalised collection of tools not very different from those delineated for the Non-Ceramic and Ceramic Neolithic described earlier, but which is also known to have continued in use into the Bronze Age. The above brief description of the ‘Goat Cult’ of Darra-i Kur includes all currently available data. The existence of this late prehistoric possibly pastoralist assemblage demonstrates the plausibility of this ecological adaptation. Furthermore, the similarity in material culture (goat burials) and possible chronological contemporaneity with the more culturally complex and sedentary sites of Shortughaï and Dashli makes it likely that significant interaction occurred between pastoralists and sedentary agriculturists.

the oxus civilisation north of the hindu kush

Figure 3.25  Map of the Dashli and associated sites between 60 and 70 km north-west of Balkh.

Dashli Sites A series of sites were located near Aqcha in northern Afghanistan by the Soviet Archaeological Mission to Afghanistan. While the initial work was limited, many reports and studies have since been published including architectural, objects studies and metal analyses.223 The sites were designated by the name ‘Dashli’ and assigned numbers (1–10). All had extensive surface remains, but only two were excavated, and the following is a brief summary of available data (Fig. 3.25). Architecture Dashli 1 The major feature at this site was a large mud-brick rectangular fort (100 m on a side, and walls 3–4 m thick) and associated settlement (Fig. 3.26). At the corners and along the mid-walls of the fort were located circular and semi-circular towers. This is one of the earliest examples of this particular type of fort architecture for this part of the world. The fort interior was filled with a maze of small rooms apparently lacking any overall planning. A series of small rooms were located immediately against the fort walls, which might have been defensive in function judging from the large quantity of sling balls found in them. Habitation structures were plastered with a yellow clay, and were characterised by wall niches and ­interior

143

144

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.26  Dashli 1 fort.

hearths. Only two major building levels were identified, and both appeared to be basically contemporary. This association of a fort and settlement was repeated at other Dashli sites. Two ritual burials of rams were found at Dashli 1. These were complete articulated skeletons of adult rams placed in pits and accompanied by several ceramic vessels. Dashli 3 Excavations at Dashli 3 again revealed a basic site plan consisting of a fort and adjacent settlement area. A circular site was designated by the name of Dashli 3 ‘temple’ (Fig. 3.27), and a quadrangular one by Dashli 3 ‘palace’ (Fig. 3.28). The real functions of these ‘manors’ are not definitely established. However, in the centre of the settlement constructed area of the ‘temple’ there was a large circular (36 m in diameter) building, and connected to, but extending out from, the outer

the oxus civilisation north of the hindu kush

Figure 3.27  Dashli 3 ‘temple’.

circumferential wall were nine rectangular towers. Two metres inside the outer circular wall an inner circular wall was constructed. The space between the two walls was divided into compartments, some of which had access to the interior area, while they all had access to the towers. Only a single external entrance to the structure was definable.

145

146

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.28  Dashli 3 ‘palace’.

Abutting the interior wall were a series of small rectangular rooms, and in the centre of the enclosure is a large rectangular building. This building was divided into rooms, one of which had opposing decorative wall niches, while the other walls had large niches with traces of burning. It would appear that this circular building had some sort of special function as yet unknown, but quite possibly religious in nature. In the neighbourhood of the circular building of the ‘temple’ is another monumental building, the ‘palace’, with an inner square some 33 metres across, surrounded by a narrow circumambulatory corridor. The inner square shape is transformed

the oxus civilisation north of the hindu kush

Figure 3.29  Dashli 3 ‘sceptre’ in the National Museum.

147

148

THE ARCHAEOLOGY OF AFGHANISTAN

into a cruciform plan by the extension of T-shaped corridors on all four sides, further transformed into a segmented cruciform by eight L-shaped corridors at each corner of the square. Exterior walls are decorated with rows of projecting pilasters. The architectural device of projecting pilasters recalls a similar device used on the monumental buildings at Mundigak IV and at Gonur Depe, and the overall plan of the building has been compared to the design of contemporary bronze seals.224 The complex has been interpreted as a possible palace, though it has also been suggested that instead of a ‘palace’ it may have been a bazaar, a ‘temple’, or perhaps even a ‘caravanserai’.225 Interestingly, in the lower levels of the Dashli 3 ‘palace’, excavators discovered a broken long stone staff (‘sceptre’: Fig. 3.29),226 and fragments of an alabaster mosaic representing trefoils and the horns of a zebu, reminiscent of well-known Indus decorations (Fig. 3.30).227 Such alabaster wall mosaics are also known in monumental tombs at Gonur-Depe, Turkmenistan. At the Dashli 3 ‘palace’, a mould of an axe was also found, a testimony to metallurgy at the site.228 Terracotta figurines were discovered at Dashli 1 and 3, and stone (gypsum) animal statuettes at the Dashli 3 ‘palace’, as well as beads (or spindle whorls), some decorated with pointed circles; one kidney-shaped miniature chlorite vessel comes from the Dashli 3 ‘palace’; bone, antler and flint tools are quite common, as well as – notably – arrow-heads.229 Some seals were discovered at Dashli 1 in situ, but the great majority of seals came out of illegal excavations. Cemeteries were found both at the edge of the settled area and within the structures themselves. Burials on the edge of the site were simple pits covered with bricks. The bodies were flexed and oriented to the north and accompanied only by ceramic vessels. Burials located within the circular building were significantly different. The method of burial was the same, but these individuals were

Figure 3.30  Dashli 3 fragments of alabaster mosaic in the National Museum.

the oxus civilisation north of the hindu kush

a­ ccompanied not only by ceramic vessels but also by metal pins, mirrors, bracelets, rings and stone vessels. Also located inside the circular building were what are described as tombs which lacked bodies but had a high number of ceramic vessels. It is possible that the differences in burial practices reflect socio-economic differences in the interred populations. It must be emphasised that the majority, if not all, the burials discovered during the excavations of the Dashli 1 and the Dashli 3 monuments (‘temple’ and ‘palace’) by Sarianidi are posterior to the abandonment of the sites. The pottery of the tombs of Dashli is the only material to be published, because they are complete, and no potsherds were collected, which has created problems for dating the use of the buildings, which must have been earlier than the burials. A reappraisal of the Dashli material, and a typological classification of the potteries, has been undertaken,230 and a new proposal of chronological sequence for Bactria has been put forward, establishing that the Dashli 3 ‘palace’ is an early monument, for many reasons, as well as the pre-rampart layers of Dashli 1, where grey burnished Hissar 3 B-C pottery was found.231 Ceramics The excavators identified three major groupings within the ceramic assemblage – wheel-made, handmade and grey ware (Figs 3.31, 3.32). Wheel-made pottery was a light-coloured clay (buff?) and had vessel forms which included several

Figure 3.31  Dashli ceramics in the National Museum.

149

150

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.32  Dashli ceramics in the National Museum.

the oxus civilisation north of the hindu kush

sharply carinated pedestal forms. Handmade pottery seems to have consisted of the common types of general purpose utilitarian pottery encountered throughout the prehistoric period in Afghanistan. A small but important number of vessels were made from a grey paste, manufactured in the form of hemispherical cups and spouted bowls, and had incised decorations. During a visit to Dashli 3 ‘palace’ in the winter of 1974–5, it was possible to see on the spot sherds of big jars, including one with the cord marks typical of the Indus tradition within the pottery heaps on the surface of these sites, and it is notable that this material did not appear in the published funerary assemblages. It is likely that the actual pottery of the functioning occupation period of the Dashli monuments would show an aspect quite different from the complete pots found only in the tombs. Other Artefacts Lithic artefacts included various sizes and projectile points and blades. A wide variety of metal artefacts were found, including razors, bracelets, mirrors, pins, knives, daggers, serrated sickle blades, axes and a silver ring. Geometric compartmented seals were identified made of stone, ceramic and metal materials. Two stone seals were found which depicted zoomorphic motifs (a scorpion and winged lion). Ceramic and stone spindle whorls and bone punches were also found. Beads of semi-precious stones were found on the surface of Dashli 1 along with the only zoomorphic figurine fragment. Other Sites in Northern Afghanistan A slight comparative resemblance can be discerned between the Dashli material and the Bronze/Iron Age remains recorded from the vicinity of Tashkurghan (Fig. 3.33).232 However, until more is reported about the ceramic remains from these sites, it is impossible to make any meaningful correlations. Of particular importance to the Bronze Age of northern Afghanistan was the accidental discovery in 1966 of the Fullol Hoard (Figs 3.34–3.37) from northern Afghanistan.233 Although the cultural context is unknown, the hoard probably originated at the site of Khush Tepe in Baghlan Province, in the south-western border area of Badakhshan. The hoard consists of a series of gold and silver vessels, both plain and decorated, with motifs suggestive of extensive foreign contacts, including geometric motifs resembling those found on Quetta Ware, a bearded bull resembling Early Dynastic Mesopotamian motifs, and a serpent and vulture motif similar to those found at Tepe Sialk Period III in eastern Iran. Whatever the ultimate context may be for this hoard, it indicates the far-flung influence that the lapis lazuli trade might have brought to bear on Afghan prehistoric developments. The Fullol casual finds have been put in perspective with the toreutics of the Oxus Civilisation and its arts in general, which highlights that the material has a broad chronological range that was ignored at the time of the discoveries.234

151

152

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.33  Bronze Age sites north of Tashkurghan.

Figure 3.34  Fullol Hoard: gold vessels.

the oxus civilisation north of the hindu kush

Figure 3.35  Fullol Hoard: silver vessels.

Figure 3.36  Fullol Hoard: silver vessels.

During the survey of Eastern Bactria some ‘Oxus’ objects were collected, such as a fragment of a stone weight;235 at Aï Khanoum itself, in the Hellenistic settlement, Bronze Age objects were found, including one stone statuette in a temple and Vakhsh pottery on the citadel.236 The recent German survey of the Herat region identified an Oxus Civilisation site at Gulran, from where a Bronze Age chlorite bowl was recovered,237 and the Herat museum has some comparable

153

154

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.37  Fullol Hoard: motifs.

pieces. At Sang-i Mazar in Ghur in western Afghanistan, a bronze tripod with three ibex heads comparable to Bactrian and Luristan bronzes was found in the 1990s.238 Without any new formal excavations in Afghanistan, it is important to note that an enormous quantity of material of the Oxus Civilisation has appeared in various antiquities markets and museums, and subsequently has been published in many articles and books.239 The Oxus material of the Bronze Age is supposed to have been found at many places in Bactria, but unfortunately, with only a few exceptions, there is nothing to indicate that this material all comes from Afghanistan: it is greatly to be regretted that many of these artefacts and remarkable art objects have forever lost their archaeological, and therefore historical and cultural, contexts. Nonetheless, the massive numbers of objects looted from sites (mainly graves, estimated to number tens of thousands) in northern Afghanistan that are now mainly in private collections have huge importance for the Bronze Age in northern Afghanistan – even though they are even less well-provenanced than the Fullol Hoard. Such objects began to appear in the late 1970s displayed in Kabul antique shops and even on the sidewalks of Kabul streets, before being taken out of the country (Fig. 3.38). They have included objects in bronze, precious metals, ter-

the oxus civilisation north of the hindu kush

Figure 3.38  Bronze Age objects from illicit excavations in Afghanistan displayed on a Kabul street in 1975–6.

racotta, ceramic and stone, including seals, statuettes, pins, mirrors, weapons and vessels, many highly decorated, including large numbers of the distinctive limestone and chlorite statuettes, such as the so-called ‘Bactrian Princesses’. Many objects that passed through the Kabul bazaar were published by Sarianidi240 and recorded and catalogued by Marie-Helène Pottier in 1984,241 and many more have been published elsewhere (Figs 3.39–3.41).242 While these high-quality art objects tell us a huge amount about the Oxus Civilisation – not least about its richness, diversity, sophistication and also its international connections throughout Central Asia, the Indus region, Iran and Mesopotamia – they are also evidence of just how knowledge of this civilisation has been lost in recent years. Almost certainly from Afghanistan are some published objects of Oxus type,243 as suggested by the strong parallels with material recovered during the remarkable excavations in Margiana by Sarianidi and his team, particularly from the site of Gonur.244 False proveniences and forgeries are, however, quite common now among the ‘antiquities of Bactria’, but the regular excavations of the necropolis and the ‘royal cemetery’ at Gonur Depe give a better idea of the affluence and international relations of the elites of this civilisation.245

155

156

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 3.39  A Bactrian bronze seal in the Louvre.

Figure 3.40  A Bactrian decorated chlorite flagon in the Louvre.

Figure 3.41  A Bactrian bronze axe-head in the New York Metropolitan Museum.

the oxus civilisation north of the hindu kush

Conclusions about Bactria, Central Asia and the ‘Oxus Civilisation’246 The excavations at Shortughaï made it possible to carry out a fresh examination of all aspects of the Bronze Age of Central Asia in general and northern Afghanistan in particular.247 It was possible to investigate the eastern Bactria material in relation to that from western Bactria (architecture, ceramics, metal, stones, seals), and make a proposal for a new sequence in Bronze Age Bactria and Central Asia (Turkmenistan, Chorasmia, Tajikistan and Afghanistan, Margiana, Zeravshan, Ferghana). It was also possible to explore the nature of exchange relations by the southern road (via Mesopotamia/India, as evidenced by texts and archaeology) or via the northern road (‘lapis lazuli road’ from Turkmenistan, and across northeastern Iran, or via Kerman–southern Iran–Baluchistan, Elam, and then further west), and thence back to Bactria. It was thus possible to make observations on neighbouring or related economies and cultures. Some widely shared levels of craftsmanship or art forms were evidenced, but also differences in hierarchy of integration and levels of cultures. The collected data also made it possible to examine the transition from the Bronze to the Iron Age. This included also the interaction with Northern Steppe cultures of Andronovo type (Tazabagjab in Chorasmia, Kajrak Kum in the Syr Darya valley), and with other regions. The relations with other cultures, and the causes of the end of the Bronze Age, were able to be tackled: then the three Late Bronze and Iron known phases could be characterised: painted handmade ceramics (Yaz-1), pre-Achaemenid (Yaz-2), Achaemenid (Yaz-3). For this later period, written sources appear, which have been utilised by linguists, raising for archaeologists the difficult questions of the ethno-linguistic identifications of the Indo-Aryans and Indo-Iranians in Central Asia (see below). As a number of the results were directly or indirectly the consequences of the analyses of the Shortughaï excavations, mainly those related to Eastern Bactria, what follows is an integrated overview of the nature of local developments and connections with other regions (this is just a selection, and we do not include here the questions about the Oxus Civilisation in Margiana, studied over many years by Sarianidi and Dubova): a. It has been possible to establish a more correct chronological sequence for the Bronze Age of Central Asia, and critique the inadequate and obsolete sequence of Namazga-IV/V to Namazga VI, which is only valid locally in the Kopet Dash piedmont, and not suitable for building a generalised picture of all of Central Asia.248 Many radiocarbon dates have substantiated the early beginnings of the Oxus Civilisation (around 2500–2400 bc) and its duration until 1600–1400 bc. However, the exact phases are still to be defined more precisely, which is a difficult task without a good stratigraphy, such as at Shortughaï. The site of Girdai Tepa in Bactria, in which Sarianidi dug a trial trench, would have provided and still can offer such a stratigraphy. It is no longer valid to argue that a

157

158

THE ARCHAEOLOGY OF AFGHANISTAN

chronological slope from ‘west’ (Turkmenistan) to ‘east’ (Tajikistan) is a valid concept or an ‘explanation’ for a supposed migration or ‘colonisation’ of tribes of agriculturalists moving eastward (see below). b. Since the explorations and (relatively limited) excavations of Sarianidi in Bactria at two Dashli sites, and the studies he published on the material recovered in Bactria,249 considerable new evidence has emerged. It is now argued that there was an ‘Oxus Civilisation’, by analogy with the contemporary ‘Indus Civilisation’. This was more significant and had a larger extension (from the Zeravshan to Pamir and Khorasan, and with material being found in the borders of South Asia and in the Persian Gulf; see below) that is something more than just an assemblage of materials, as was initially suggested by the ‘BMAC’ term; the Oxus Civilisation has been tentatively identified with Marhashi by some,250 and as Shimaski by others.251 These are both regions mentioned in the cuneiform texts of late Akkadian and Ur III periods, and have relations with the Indus Civilisation on one side, and Elamite Iran to Mesopotamia on the other, which are attested by archaeological finds. c. It is now possible to speculate about the origin and evolution of the Oxus Civilisation, especially the identification of an ‘early’ or ‘formative phase’, including the sites of Mundigak, Taluqan and Sarazm, and early links between Eastern Bactria and Indus–Baluchistan areas from as early as the Chalcolithic– Eneolithic (see above).252 This ‘early phase’, which we may date before 2400 bc, is especially important because a number of constituent elements of the Oxus Civilisation appear quite early and are surprisingly widespread, such as typical monumental architecture or artefacts like stone ‘handled weights’ or grooved stones, as predecessors of the grooved miniature columns. This network of relationships anticipates the linkages related to the Oxus Civilisation, but proper understanding of this process requires still more research in order for it to be better characterised. For instance, Sarazm, which is located north of Bactria in the Zeravshan valley in Sogdiana, displays relations with the Kopet Dagh region (potteries), but also with the Aral Sea area (Kelteminar potteries), the steppes (Afanasievo-type burial architecture and perhaps pottery), the Caspian zone in Iran (grey pottery), the borderlands of Khyber Pakhtunkwa and Baluchistan (painted pottery and figurines), the Indian Ocean zone (shells), proto-Elamite Iran (cylinder seal), and possibly Mesopotamia (a gold rosette comparable with material from burials at Ur). This list can be completed by the finds of a pilaster façade at Mundigak, which is similar to architecture engraved on chlorite artefacts (see Chapter 4). There is much to learn from a more accurate and more documented study of this important period. We can hypothesise that Chalcolithic sites of this ‘formative’ phase should exist in the northwest Afghanistan piedmont of the Hindu Kush, between Aqcha, Andkhoy, Shiberghan, and up to Herat, in relation with what we know already in the Tedjen region, in the lower Helmand and Sistan region, in the North Kopet Dagh piedmont in Turkmenistan, and in Iran. The most plausible hypothesis

the oxus civilisation north of the hindu kush

for the presence of all these different cultures in the same environment almost at the same time (at least for some of them) is a search for minerals (abundant in the valley), and notably gold, which gave its name to the river (Zeravshan means ‘spreading gold’ in Persian). Tin, also present in quantities in the sands of the river, along with possible mining in the mountains (Musishtan, etc.), is also a possibility.253 d. It is now also possible to consider the Middle Bronze age in Eastern Bactria during the Shortughaï II period.254 The most important recent finds are those of the Farkhor cemetery in Tajikistan, on the right bank of the Panj River, near the confluence with the Kyzyl-Su. This is very close to Shortughaï. The form of Middle if not Early Bronze Age pots is similar to the very peculiar and rare assemblage of Shortughaï II (see above), but in some of the burials material appears that is similar to the Central Asian Namazga IV (rather than V?) of Turkmenistan, including pottery (quadrangular box), metal and stone artefacts (long staff, vases, beads). The more recent discovery of the Kangurt-Tut-II cemetery confirms this early presence of the Oxus Civilisation (its Middle/Early phase) in SE Tajikistan, exactly en face Shortughaï. As we can imagine, here again an explanation can be given by attractive minerals. Indus Civilisation and Oxus Civilisation peoples potentially met at a place where the Panj just becomes calmer after its torrential course from Wakhan, a place where the sands contain quantities of gold dust, available in winter when waters are very low. Is this a sufficient explanation? Perhaps, when we consider the long distances ancient people were going to obtain lapis lazuli or tin. The road to the Sar-i Sang lapis mines begins here too, but simply collecting lapis pebbles in the riverbed was also a possibility, as shown at Shortughaï. All of this is just a part of the story and more research is needed. e. There has been further discussion of the position of the Oxus Civilisation regarding the problem of the Indo-Aryans in relation to the Indus Civilisation, and its problems regarding the connections between the Indian Subcontinent and Central Asia, especially Bactria.255 The literature is abundant, but the archaeological material does not support the theory of the crossing of the Hindu Kush by Indo-Aryan or Aryan tribes, whether they are identified as the ‘Oxus people’, or with the ‘Andronovo tribes’. Andronovo-type pottery has been found at Shortughaï and in the Dashli sites in Afghanistan, as well as almost everywhere in Central Asia north of the Hindu Kush, after around 1800 bc, but there is no reported evidence for this ware to the south of the Hindu Kush. On the other hand, there is evidence for Oxus Civilisation material to the south of the Hindu Kush (see below), but no conclusive proof that the distribution of this material represents a migration rather than other types of exchange or trade. Anthropology and the study of ancient genetics have not yet provided any conclusive evidence of possible migrations, though this topic continues to attract interest and active debate, the review of which is beyond the scope of this volume.256

159

160

THE ARCHAEOLOGY OF AFGHANISTAN

f. It is also now possible to have new perspectives on the Late Bronze Age (equivalent to Shortughaï IV) in Bactria and Central Asia.257 This late phase of the Oxus Civilisation, after 1750 bc, makes it possible to consider issues of economic and social decline, implied by the end of the mature phase of the great network of exchanges and prosperity of the elites. Again, here we face the question of migrations from the north by the Andronovo tribes, and of a possible change in climatic and environmental conditions towards a drier environment. The presence of Andronovo hand-made pottery seems more abundant in this period, and research has been carried out in Margiana on such topics, including botanical and zoological analyses. Currently, there is a need for more environmental information and analyses in order to help characterise the nature of environmental (possibly local) or climatic (possibly regional or general) changes, and their precise causes. However, it should be remembered that many other causes for this ‘decline’, be they social and/or political, are also feasible. It is now clear that the Oxus Civilisation played a major role in the socio-­ economics and politics of the late third and early second millennia bc, extending far and wide across Central Asia, and exchanging and/or having contact with populations living in a number of other regions. Oxus material has been found far from Bactria and Margiana, in regions as diverse as Iranian Khorasan at Tepe Hissar, Tepe Damghani, Tepe Chalow, Nishapur, etc.;258 in southern Afghanistan at Shamshir Ghar,259 Nad-i Ali and elsewhere in Sistan (see Chapter 4);260 in southeastern Iran at Tepe Yahya and Shahdad; and also in Baluchistan at Shahi Tump,261 Mehrgarh and Sibri,262 and in the Quetta hoard,263 and also at sites in the Persian Gulf, including Tell Abraq264 – and this list is not exhaustive.265 There have been a number of syntheses aimed at a better understanding of various aspects of the Oxus Civilisation and its place in the archaeology of Bactria and Afghanistan and the greater region,266 as well as further precision about the originality of the mythological pantheon of the Oxus Civilisation and its arts.267 There is much more to be learned about the late prehistory of northern Afghanistan and the surrounding regions of Central Asia, and the ancient societies of these regions have the potential to reveal much about distinctive approaches to social, economic and political organisation and interaction in the Chalcolithic and Bronze Ages.

CHAPTER 4

The Development of a ‘Helmand Civilisation’ South of the Hindu Kush 1

Cameron A. Petrie and Jim G. Shaffer

Introduction Our information about late prehistory in the parts of Afghanistan that lie to the south of the Hindu Kush is limited to the Chalcolithic period and later, and comes mainly from substantial excavations at Mundigak2 and Said Qala Tepe,3 as well as more limited investigations of these periods at Deh Morasi Ghundai,4 Kandahar and Nad-i Ali.5 To these can be added sites found during several surveys in the Helmand-Sistan area (Fig. 3.1).6 While limited to the southernmost parts of Afghanistan, the sample size is sufficient to indicate that southern Afghanistan was an important area for understanding the transition to stratified societies and for comprehending the cultural processes which affected neighbouring areas. Together, these Bronze Age developments south of the Hindu Kush have been dubbed a ‘Helmand Civilisation’ by analogy with other contemporary ‘riverine civilisations’ such as the Oxus or the Indus, particularly with reference to its two main excavated sites, Mundigak and Shahr-i Sokhta (just across the border in Iranian Sistan), although the term is perhaps not as widely accepted as the other two.7 In contrast to the approach used in the first edition of this book, here the earliest phases of sedentary occupation for the whole country have been discussed in Chapter 3, and the overview of the development of stratified societies has been divided into two parts: discussion of the evidence from north of the Hindu Kush has been presented in Chapter 3, and the evidence from south of the Hindu Kush is presented here. The Mundigak sequence, and the supplementary information from Deh Morasi Ghundai (hereafter Deh Morasi) and Said Qala Tepe (hereafter Said Qala), provide crucial information for understanding the cultural processes linking the areas of Baluchistan and the Indus to Central Asia and eastern Iran. Conversely, excavations at sites like Shahr-i Sokhta,8 Shahdad,9 Tepe Yahya10 and Konar Sandal,11 all in south-eastern Iran,12 as well as at sites in Makran,13 Baluchistan,14 Khyber Pakhtunkhwa15 and the greater Indus region,16 have since greatly added to our knowledge of this vast area,17 and helped to further contextualise the material from Afghanistan.

162

THE ARCHAEOLOGY OF AFGHANISTAN

Geographical Factors The site of Mundigak comprises a series of mounds situated in a mountainous region approximately 55 km north-north-west of Kandahar, in the upper drainage of the Kushk-i Nakhud stream, which roughly parallels the Arghandab River as it flows west past Kandahar, eventually joining the Arghandab (Fig. 4.1). The Arghandab joins the Helmand approximately 130 km south-west of Kandahar at Bust. Like most areas of Afghanistan this region is arid. The Helmand then flows south and west then north until it reaches the large marshy, but today extremely arid, area of Sistan on the Iranian border, which has in Afghanistan only had surface survey for prehistoric sites, and limited excavations.18 Before the

Figure 4.1  Map of the Kandahar area in the Bronze Age.

a ‘helmand civilisation’ south of the hindu kush

Arghandab–Helmand Rivers turn to the south-west, they form the northern boundary separating the cultivable lands to the north from the arid Registan desert. Once the Helmand begins its south-westward course it passes through some of the most arid regions in the world, although satellite imagery has revealed a huge quantity of pre-modern remains in the sands hitherto unsuspected,19 and surface survey has revealed numerous settlements along long-abandoned river channels.20 The Helmand is actually the only major perennial river located between Mesopotamia and the Indus River, and its importance in prehistoric cultural developments throughout this vast area between the Euphrates and the Indus cannot be overemphasised.21 The location of Mundigak within the drainage of one of the main tributaries of this system is a major factor in understanding the cultural processes and phenomena which are reflected at this site. Similar conditions may be defined for the related sites of Deh Morasi and Said Qala, which are both located within 30 km of Kandahar and about 100 km south-east of Mundigak, on the flood plain separating the Arghandab and Tarnak Rivers. The Tarnak runs parallel to the Arghandab and flows into the Dori south of the sites. The Dori in turn flows west to join the Arghandab about 70 km southwest of Kandahar. The proximity and location within the same river system are important to the close relationships definable between these southern Afghan prehistoric sites. It is worth noting that research in south-east Iran22 and the piedmonts of the Suleiman Ranges in Pakistan23 has shown that many Neolithic and later settlements like Tepe Gaz Tavila (R37), Tepe Yahya, Mundigak and Sheri Khan Tarakai were situated on alluvial fans or terraces associated with perennial and ephemeral water courses. Given the similar geographical constraints that exist in southern Afghanistan, it is likely that a number of settlements will have been situated in similar contexts in order to exploit available water sources. For instance, Mundigak appears to have been situated on a relic alluvial terrace adjacent to a river, and at the foot of a number of large alluvial fans.24 Mundigak During the middle and late 1950s the DAFA conducted continuous and extensive excavations at Mundigak under the direction of Jean-Marie Casal. Over seventy years after they were completed (Figs 4.2, 4.3, 4.4), the results of these excavations still represent the major research effort concerned with the later periods of prehistory in southern Afghanistan, despite the inevitable limitations that result from the excavation, retrieval and analytical methods that were then in common use.25 Casal defined seven major occupation periods at Mundigak, and the first five of these are of concern here. Chronologically these periods represent a time span of approximately 3,000 years from the beginning of the fourth to sometime in the second millennium bc. During this span Mundigak developed from a small agricultural village (Periods I–III) to a major centre with signs of

163

164

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.2  Mundigak, general site plan of excavation areas.

incipient u ­ rbanism (Period IV–V) and then was abandoned during the Iron Age (when Kandahar probably emerged as the regional centre). The complete occupation sequence was defined only on Mound A, but during Period IV several adjacent mounds were occupied by habitations and ‘public’ structures. Although much of what follows is relatively detailed, and reiterates information from Casal’s ­original

a ‘helmand civilisation’ south of the hindu kush

Figure 4.3  Mundigak, view of mound A in 1956.

Figure 4.4  Mundigak, view of mound A in 1966.

165

166

THE ARCHAEOLOGY OF AFGHANISTAN

1961 p ­ ublication, it remains the most comprehensive and detailed discussion of the Mundigak sequence and associated architecture and material in English, and has been largely maintained in the original format prepared by Shaffer for the 1978 edition. Stratigraphy and Architecture Period I The initial occupation at Mundigak, Period I, was subdivided into five SubPeriods (I1–5), with the divisions being based upon structural and cultural correlations. Sub-Periods I1–3 are, however, very incompletely known since the total sample came from a sondage of 10 × 6 m, and all three sub-periods accounted for less than 1.5 m depth of deposits. Sub-Periods I4–5 are known from a much larger sample (maximum dimensions of 29 × 18 m with 2.5+ m deposits) and, therefore, are more completely known. Sub-Period I1 and most of Sub-Period I2 are located directly above virgin soil, while Sub-Periods I3–5 are superimposed directly atop each other. Sub-Periods I1–2 are devoid of any substantial architectural remains, and their delineation is based upon depositional soil differences and artefact content (see below). Casal referred to the possible existence of tent-like structures during these sub-periods, but the evidence is limited. The first substantial, or permanent, structure was encountered in Sub-Period I3, and consisted of two pakhsa walls. The first mud-brick structures were encountered in Sub-Periods I4–5. The SubPeriod I4 structures are all rectangular single room units of varying sizes. The two completely excavated units are small (3 × 2 m) when compared with the obviously larger but incompletely excavated units of that sub-period. Walls were constructed with single or double coursed bricks, and the two complete structures have doorways. Most structures have their walls reinforced with interior and/or exterior buttresses. Rectangular interior ovens or hearths are located in three of the units. These features were constructed through the use of pakhsa and protrude a few centimetres above the floor. Our knowledge of Sub-Period I4 structures is incomplete because they were disturbed by the constructions of Sub-Period I5. A significant alteration in construction technique occurred with the Sub-Period I5 structures (Fig. 4.5). These structures are characterised by the laying of a foundation of pakhsa sometimes mixed with stones below the immediate living surface (a wall trench), which should have strengthened the walls considerably. Houses continued to be rectangular in plan, and three were characterised by interior partition walls with doorways to allow access from one room to another. Access to the interior of the structures was via lateral doorways. However, some structures had no evidence of such doorways, indicating that access may have been through

a ‘helmand civilisation’ south of the hindu kush

Figure 4.5  Mundigak A: Period 1, 5.

an elevated doorway or via the roof. Two examples of rectangular pakhsa inner ovens are recorded. Of particular interest is the presence of two large mud-brick and pakhsa oval or U-shaped ovens in an exterior open area. Analysis of these ovens indicates that extremely high temperatures (600–1100° C) were produced in them. Their location in a large open space, exterior but adjacent to habitations, might indicate the beginning of functionally specific areas within the site. From the limited sample, it would appear that there is a significant increase of basic structure size, with those of Sub-Period I5 being almost three times larger (9 × 6 m). Sub-Period I5, and therefore the whole of Period I, was stratigraphically sealed by a deposit of varying types of debris, suggesting that this particular area of the site was not subsequently occupied for some time. The type of habitation structures described for Sub-Periods I4–5 seems to establish a pattern which, as will be shown below, is very consistent until Period IV. It should be noted here that Casal has been criticised for not making a sharper distinction, if not periodisation, between Sub-Periods I1–3 and I4–526 because of differences evident in the architecture and ceramics. The ceramics are discussed below, but on the architectural evidence, Casal’s designation of a single period seems correct. Given the extremely small sample and exposure of Sub-Periods I1–3 as compared with Sub-Periods I4–5, or other periods, it would have been premature to make a major stratigraphic division merely on the basis of the presence or absence of substantial architecture.

167

168

THE ARCHAEOLOGY OF AFGHANISTAN

Period II Period II is divided into three Sub-Periods (II1–3), the last of which is subdivided into units a and b (Figs 4.6, 4.7). Walls were of mud-brick, founded in deep trenches filled with pakhsa, or reusing wall remnants from previous structures (Sub-Period I5). Only rectangular structures were identified, but unlike in Period I they were often divided into two rooms, one smaller than the other (2 × 4·4 m v. 4/5 × 6 m),

Figure 4.6  Mundigak A: Period II, 2.

Figure 4.7  Mundigak A: Period II, 3b.

a ‘helmand civilisation’ south of the hindu kush

connected by doorways. During Sub-Periods II1–2, most rooms had an exterior entrance whereas in Sub-Period II3 only one such entrance was found. Interior and exterior wall buttresses were frequently encountered along with a few examples of small windows and wall niches. A very marked characteristic of Period II was a much greater density in the disposition of structures. Five structures in Sub-Period II1 continued to have a centrally located interior rectangular hearth set into the floor constructed of pakhsa, and surrounded by a clay border. However, these hearths now have a centrally located fire-pit, which continues to characterise such features throughout the remaining occupations at the site. A single structure was found with an interior rectangular pit constructed with pakhsa and filled with ash and debris, which occupied most of the floor area. A large open area was interpreted as a possible cattle pen. Finally, another room had a small semi-circular pakhsa structure attached to one wall, which was described as a seed trough. The total number of structures increased during Sub-Period II2 although the large open area remained free of structures. However, it is doubtful that this area was used as an animal pen since it now had a dividing wall and a substantial rectangular oven with two U-shaped chambers. A northern cluster of structures was distinguished by the construction of a well (1 m in diameter by 8 m deep) excavated into virgin soil (Fig. 4.6). The upper part of the well was enclosed by an octagonal mud-brick structure c. 1 m high while the well interior was pakhsalined. An area which separated the well from nearby structures was lined/paved with stones. Immediately south of the well was a small open space with a centrally located rectangular oven, suggesting a possible functional correlation between the two features. Another interesting correlation with these features was that every surrounding structure had an interior hearth, which again suggests that this area was functionally distinct. Sub-Period II3a underwent a decrease in the number of structures and an increase of more than 50% in open areas. Only one definite and one possible external entrance were located in Sub-Period II3a and none in Sub-Period II3b. Absence of such entrances led Casal to speculate that these walls functioned as supports for a wooden superstructure or an upper living structure, but the interior hearths and mud floors might indicate that entrance was via the roof. It is possible that rooms without an external entrance had a non-habitation specialisation, such as storage, for which an external entrance was not necessary and/or desirable. Another large rectangular mud-brick oven with two U-shaped chambers was constructed in an open area during SubPeriod II3a. This oven continued in use through Sub-Period II3b with the addition of a third U-shaped chamber. An additional asymmetrical mud-brick three-chambered oven was constructed in a different open area during Sub-Period II3b. Finally, a unique circular pakhsa basin was constructed inside one 3b structure. There is no reason to question Casal’s stratigraphic designations for Period II. A marked continuity seems to persist for the architectural developments represented in Periods I–II, which corresponds to a similar continuity in most categories of

169

170

THE ARCHAEOLOGY OF AFGHANISTAN

material culture. The overall picture is one of continuous rebuilding, reflecting internal population growth and shifts within a village settlement pattern. A significant development for Period II, however, is the possible existence of functionally distinct areas and structures within the settlement. Period III The remains of Period III in general are usually contrasted with those of Period II, and depicted as representing a period of ‘vitality and expansion’ as opposed to the ‘stagnation’ of Period II. Although Period III witnessed some significant changes in material culture, it also demonstrated a stratigraphic and architectural continuity with Period II. Casal divided Period III into six Sub-Periods (III1–6), with the last Sub-Period (III6) subdivided into units a, b and c. The Period III mud-brick structures are essentially similar to those found in Periods I–II. However, the previously typical structure with one large and one small room becomes extremely rare (one example in each of Sub-Periods III1 and III6c). During Period III an increased density in the number of structures per excavated area was discernible. This increased density is often interpreted as

Figure 4.8  Mundigak A: Period III, 2.

a ‘helmand civilisation’ south of the hindu kush

Figure 4.9  Mundigak A: Period III, 6a.

representing population growth, but this fails to take into consideration the possibility that not all structures were habitations. Sub-Period III1–5 structures demonstrate various types of rebuilding such as additional walls, interior features and repairs (Fig. 4.8). Walls are not, however, as firmly founded in Sub-Period III1–5 as in previous occupations. Throughout Period III several structures are found with common, or closely abutted, walls, which might indicate that such structures represent a socio-cultural unit. Open spaces between these structural clusters are very irregular, giving the impression that if such socio-cultural units are represented they took shape rather haphazardly or according to some agglutinative cultural process. Entrance to structures was either by a small lateral doorway or via the roof. It is interesting to note that almost every structure with an interior hearth was provided with a lateral doorway. Structures without such hearths were far more variable in this respect. The large rectangular multi-chambered mud-brick ovens are identified for the last time in Sub-Period III4. If these large, possibly domed ovens are potter’s kilns, as Casal suggests, then their final appearance in this period might be related to two important factors. First, the increased variety and sophistication of the ceramic industry in Period IV combined with the disappearance of ‘kilns’ in habitation areas might represent an intensified industrialisation in specialised areas of this category of material. Second, their disappearance in Period IV might be related to the special functional structures recorded for that period. Small windows were recorded throughout Period III, and in one instance a mud-brick ‘shutter’ was found in place. Three wells, or three building phases of a single well, were found in Sub-Period III2 (Fig. 4.9). Unlike the well of Period II these wells were simple circular pits

171

172

THE ARCHAEOLOGY OF AFGHANISTAN

excavated into virgin soil. However, like the earlier well, they were located in a courtyard surrounded by structures. A multi-chambered mud-brick oven was also located in the courtyard. Other interesting structural features found in Period III were: a central pillar of mud-brick within a room of Sub-Period III3, and a bench extending around three walls of a room in Sub-Period III4. A very important feature of Sub-Period III4 was the construction of a retaining terrace wall to the south-west. The wall was constructed from large blocks of clay and the area behind the wall was then filled. Casal maintained that this was done with the purpose of increasing the areal expanse available for the construction of structural units. The exact function of this wall remains to be determined although its presence must have had a significant bearing on subsequent developments in Period IV. Sub-Period III6 was stratigraphically distinguished by the intentional levelling of the surface left by structures of Sub-Period III5. Similarly, the end of III6 was intentionally levelled to facilitate the construction of Period IV structures. The initial occupation, III6a, is only slightly different from the earlier sub-periods of Period III (Fig. 4.10). Many structures had an interior floor 25 cm lower than the level of surrounding living surfaces. Examples of complete ‘entrances’ were found, all of which were very small: less than 1.0 m in height. Another structure in Sub-Period III6a had a large pit excavated into the interior surface. Located to the west of this structure, and constructed with mud-bricks, there was a small tomb containing a single individual. Two walls of the tomb abutted this other structure and an associated entrance had been sealed. The entrance to the tomb itself had been sealed and then another wall constructed in front of that entrance. Only a single ceramic container was found in association with the burial, but outside the tomb were the only examples from Mundigak of bronze axes and a single adze. This tomb is the only one located within a habitation area at Mundigak, and it provides a stratigraphic correlation for burials located on another mound. Two successive groups of burials were defined on Mound C located below structures of Period IV. The earliest (perhaps earlier than Sub-Period III6) consisted of single flexed burials placed in irregularly shaped pits excavated into virgin soil. No diagnostic material was found with these early burials, with grave goods being limited to flint tools, a single necklace of cow’s teeth and a single bead bracelet. The later(?) group of burials were placed in tombs similar to that identified in SubPeriod III6. These ossuaries were rectangular and constructed from mud-bricks. Puddled clay floors were constructed first and then brick walls were set onto them. The predominant form of interment involved multiple burials, although a few single burials were found. Skulls were carefully arranged in a row along one of the walls (e.g. Ossuary C. 27) while the remaining skeletal parts were scattered about the interior of the tomb. Only rarely were articulated limbs located. The skulls appear to have been placed in situ, while the rest of the body was either mutilated or de-fleshed in some manner and placed in the tomb at a later date. A single instance of a lamb’s remains was found deposited in a like manner in one

a ‘helmand civilisation’ south of the hindu kush

tomb. Besides a few vessels of decorated ware, isolated pendants and a few blue and white beads, grave goods were very rare. There was nothing to indicate any differential social status except perhaps the fact of the burial itself. Unfortunately, the size of excavations at Mound C was very limited. It is interesting to note, however, that this is the only mound at Mundigak besides Mound A to have materials datable before Period IV. This concentrated location away from Mound A might indicate that the concept of a specially designated area for burying the dead was well-developed by this time. Sub-Period III6b had only a few structures, all of which seemed to centre on a single row of rooms. However, several free-standing walls were found leading off to the north that might be rooms open at their northern end. Some of these structures had been filled in with mud-brick by the building activities of Period IV. Several structures were definable for the last occupation of Sub-Period III6c, among which was a single example of a rectangular structure divided into a large and small room. This final occupation underwent several modifications and rebuildings which centred on the gradual addition or alteration of open spaces surrounding various blocks of structures. After Period III6c there is a pronounced change in the function of the structures built on Mound A. From Period I through III6c the general impression has been one of structures and debris associated with multi-purpose activities necessitated by a sedentary agricultural way of life. After Period III, however, a very different picture emerges. Period IV On Mound A, the Sub-Period III6 structures were levelled to provide a surface for the construction of a building which marks a significant variation from previous architectural traditions. Its style could indicate that it is associated with a particular social segment or function. Equally important during Period IV was the occupation of new areas at the site, and the construction of special function structures. These newly occupied areas included Mounds B, D, E, F, G, H, and I (Fig. 4.10). Unfortunately, it is precisely these structures of Period IV which were heavily devastated by later erosion at the site, but fortunately many were preserved due to a fire that hardened the mud-bricks (giving the appearance of fired brick), or else they would likely have completely disappeared. There were no connecting stratigraphic trenches between major structures to confirm associations, but Casal established contemporaneity between structures on the basis of associated ceramics, and divided Period IV into three major Sub-Periods (IV1–3), the first of which is by far the most impressive and without parallel in prehistoric Afghanistan. Sub-Period IV1, Casal’s ‘Epoch of the Palace’, was characterised by the construction of large monumental structures and enclosing walls (Figs. 4.10–4.22). The ‘Palace’ was a large monumental building located on Mound A, which underwent several rebuilding phases. There is little evidence to definitely indicate that this structure represents a ‘palace’, but there can be no doubt that it was monumental,

173

Figure 4.10  Mundigak, general plan of the Period IV structures.

a ‘helmand civilisation’ south of the hindu kush

Figure 4.11  Mundigak A: plan and elevation of ‘colonnade’.

significantly different from previous and contemporary structures, and culturally important. However, to designate it as a ‘palace’ implies a degree and level of political organisation, which cannot be presently confirmed. The façade was embellished with a line of engaged semi-columns27 (henceforth ‘colonnade’). This distinctive architectural device is seen elsewhere in the Bronze Age such as at the ‘temple’ on Mound G at Mundigak (see below, where the engaged ‘semi-columns’ are projecting triangles) and at the ‘palace’ at Dashli 3, with its rows of external repeated projecting buttresses. It is possible that the device originated in fourth millennium bc at Uruk/Warka in Mesopotamia, in the cone-decorated engaged semi-columns at the ‘White Temple’,28 although such features might have originated locally in Afghanistan and subsequently have a long later history in Central Asia.29 This important structure was located at the highest point of the site – Mound A. When this building was constructed, Mound A must have been an imposing edifice with its 11 metres of elevation resulting from previous occupations and its surrounding(?) terrace wall constructed in Period III (Figs 4.3, 4.12, 4.13).30 From the top of the mound it was possible to see not only the other structures at the site, but also the surrounding countryside. The initial structure plan underwent at least three rebuilding phases of which only the last was a significant alteration. Until the last rebuilding, the area north of the colonnaded wall was kept free of other structures so that this building was equally visible in areas away from the mound. Although only the northern wall with its east–west colonnades remained it is possible that such colonnades existed on all building faces. Similarly, the only

175

176

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.12  Mundigak A: façade of the Period IV monumental building.

Figure 4.13  Mundigak A: façade of the Period IV monumental building.

a ‘helmand civilisation’ south of the hindu kush

Figure 4.14  Mundigak A: ‘palace’ first reconstruction.

remaining entrance was through the north wall, but it is impossible to rule out the existence of other entrances. The structure’s exterior walls appear to be aligned with the cardinal points of the compass, but such is not the case for the interior walls. A similar alignment can be defined for other monumental structures of Period IV except for one portion of a large enclosing wall which goes off at a slight north-east angle. This building orientation results in Casal’s ‘Palace’ and ‘Temple’ structures being located in the same east–west line, which is parallel to one formed by the enclosing walls. It is doubtful that such alignment was the result of chance. The primary focus of construction appears to have been the large exterior walls faced with partial colonnades of which only the north wall remains. The colonnades, like the wall, were constructed with bricks that were fired in a conflagration. Nonetheless, the colonnades’ exterior was plastered, painted white and topped with a brick frieze of opposing stepped merlons (Fig. 4.11). The wall remnant was 2.3 m high and originally taller. North of the wall was a broad brick walkway. South of the wall, and sometimes attached to the wall, were several small rectangular habitation(?) structures similar to those described for earlier periods. Access to these structures was either through a stepped entrance that opened onto a large ‘courtyard’ or through another entrance that led directly into one of the small interior rooms. These interior rooms were not significantly d ­ ifferent from those

177

178

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.15  Mundigak A: ‘palace’ final reconstruction.

found in Sub-Period III6, and except for the organisation imposed on them by the colonnaded wall and ‘courtyard’ they were haphazardly arranged. The first two rebuildings (in Sub-Periods IVla–b) are almost exclusively concerned with these interior structures. Successive rebuildings resulted primarily in a proliferation in the number of rooms and a thickening of walls (Fig. 4.15). The north end of the large courtyard became divided into a series of rooms distinguished by the construction of two substantial walls perpendicular to the colonnaded wall. These new structures continue to reflect habitation activities indicated by the presence of interior ovens, drains, wall-lamp niches and at least one kitchen area. In con-

a ‘helmand civilisation’ south of the hindu kush

trast, the third rebuilding, Sub-Period IVlc, represents significant architectural and, possibly, functional changes for this complex of structures. During Sub-Period IVlc major new structures are constructed north of the colonnaded wall. These elongated rectangular structures would have completely obscured the colonnaded wall from view. Moreover, structures on both sides of the colonnaded wall no longer appear concerned with habitation activities. These new structures are generally lacking the interior features which characterised the previous ones, and in one instance a room contained an unusual quantity of alabaster bowls and bronze projectile points. Stairways located in some structures indicate the presence of an upper storey, or at least reflect the importance of having access to or from the roof. The previous colonnaded wall continued to be used, but many of the structures south of it were filled with brick in the construction of a large platform which was made accessible by a series of stairways added on to the old entrance. Five metres behind and parallel with the old wall a new colonnaded wall was built. The construction of a new and taller colonnaded wall must have added significantly to the overall terraced effect of Mound A.31 The platform had received several coatings of white and/or red plaster. A wall trench had been excavated for the foundation of this new wall (it cut through Sub-Period IV1a–b) and then filled with large stones after construction of the wall. Structures located south of this new wall were no longer haphazardly arranged, but were organised along a grid pattern with varying sized rooms (Fig. 4.15). These rooms were extremely small and void of any internal features. Wall remnants indicate that the walls were never very high and it is doubtful that they were used as habitations. The floors of these units had been very carefully filled and levelled. Unfortunately, this last rebuilding was heavily eroded, making it difficult to assemble an overall plan. West of Mound A, Mounds B and D produced remains of an enclosing wall complete with square ‘bastions’ (Figs 4.16, 4.17, 4.21). These structures were erected directly on virgin soil and consisted of two thick parallel walls of mud-brick resting on foundations of stone and clay. Regularly spaced rectangular projecting buttresses characterised the exterior wall, while the interior separating the two walls was divided into small rooms. The floors of these rooms had been raised significantly above the level of the exterior living surfaces. The frequency of stairways associated with these rooms indicates that access to either the roof or an upper storey was of some importance. This arrangement is seen in Oxus Civilisation architecture, and is also similar to the later Achaemenid ramparts at Kandahar, where the ‘rooms’ were interpreted as casemates.32 Two completely separate examples of such enclosing walls were defined west of Mound A (Fig. 4.21). The nearest and most extensively excavated example was about 100 m away, while the second example was 150 m away, parallel to the first (Fig. 4.17). Both examples were oriented in the same cardinal directions as the walls of the ‘palace’ and the terrace walls of Mound A. The wall nearest Mound A was located on a natural elevation significantly higher than the second wall, which would

179

180

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.16  Mundigak B: angle of rampart and contemporary structures IV, 1.

have contributed greatly to the overall terraced, or stepped, profile that the entire settlement was now assuming. The easternmost extent of these walls is unknown, but it is reasonably certain that they incorporated at least Mound A and possibly Mound G, with its ‘temple’ structure (Figs 4.18, 4.19).33 At both corners bastion-like structures were found, constructed in the same architectural style as the walls. The south bastion was a simple rectangular room with exterior buttresses (Fig. 4.17) whereas the northern bastion was divided into four inner rooms, and was possibly two storeys high (Fig. 4.16).34 Similar one-room bastion-like structures were located two metres to the east along the exterior face of both west–east walls, and a similar but completely

a ‘helmand civilisation’ south of the hindu kush

Figure 4.17  Mundigak D: bastion IV, 1.

separate bastion-like structure or ‘barbican’ was found north of the north-west corner bastion. Abutting this bastion was another perpendicular enclosing wall which presumably extended westward until it joined the most western north– south wall. Excavations in areas adjacent to these walls and bastions disclosed many more typical house-like structures. There was no apparent organisation to these structures other than that dictated by the proximity of the large walls. These structures were characterised by numerous habitation-type features such as interior hearths. In the area of the north-west corner bastions, several different styles of ovens (oval, circular, rectangular, single and multi-chambered) were found, suggesting that this might have been an industrial area. In the same area were also several examples of drains. The limited excavations on Mound C, between Mounds A and B, suggest that the area enclosed by these large walls was densely occupied. On Mound G excavations revealed a large monumental structure referred to by Casal as a ‘temple’ (Figs 4.18, 4.19, 4.20).35 This structure had an orientation parallel to that of the ‘palace’ and was in a direct west–east line with it. The structure was built mostly on virgin soil and in a technique similar to that of the enclosing walls. Unfortunately, the southern part was heavily eroded, but there is no reason to believe that it differed significantly from the excavated portion. Two massive parallel external walls formed the structure’s perimeter. The internal area between these walls was divided into small rooms which were not used for habitation and probably served a purely structural function. Massive projecting triangular buttresses were constructed from mud-brick (‘briques crue’) along the external face of

181

182

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.18  Mundigak G: plan and section of ‘temple’.

the perimeter walls (Fig. 4.20). Only one construction phase could be determined, and it, like the last rebuilding phase of the ‘palace’, demonstrates a high degree of organisation. No entrance was located, but it might have been in the eroded south wall. A large rectangular structure with its eastern two-thirds divided into small rooms dominates the centre of the ‘temple’. The western part of this building

a ‘helmand civilisation’ south of the hindu kush

Figure 4.19  Mundigak G: view of the ‘temple’ from Mound A in 1966.

consists of a large open area or courtyard. Centrally located at the north end of this courtyard was a large basin that was considerably elevated above the surrounding living surface. The immediate area was ash-covered and located directly behind the basin was a ceramic drain which extended east–west between the main wall and smaller L-shaped wall associated with the interior building. This smaller wall formed the western boundary of a little chamber interpreted as representing a shrine complex. In the south-east corner was a large square rectangular masonry structure with white plastered benches. A similar bench was found along the east wall. In the centre of the chamber was a large rectangular hearth painted red with a small step on the west side. The rooms to the east were of various sizes and a few of them had interior hearths and other small features. Although there is nothing to indicate that this was a religious structure, it was certainly not a habitation either. Whatever the function, it presents an interesting contrast with the rest of the site. Sub-Period IV1 therefore demonstrates both continuity with and change from the preceding Periods at Mundigak. Perhaps the most distinctive change was in the appearance of special function architecture. While the specific function of the structures of Sub-Period IV1 is speculative, they clearly represent a monumental remodelling of the settlement, and this would presumably only have been possible as a result of specific transformations in the socio-economy of the inhabitants, which may indicate an increase in inequality and complexity. This suggestion is reinforced by the enclosing of the settlement with elaborate ramparts which with

183

184

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.20  Mundigak G: view of the ‘temple’ showing triangular ‘buttresses’.

its projecting buttresses – more closely spaced than required by purely structural purposes – likely has a decorative and monumental rather than a purely defensive purpose, recalling the buttressing/colonnading of the monumental buildings themselves. As reconstructed by Dumarçay36 and superimposed on Casal’s map, these ramparts comprise inner and outer enclosures with the ‘palace’ in the centre of the inner enclosure (Fig. 4.21), occupying the highest and most prominent position on the settlement. This arrangement anticipates the classic arrangement of the standard later Central Asian town of citadel and lower town, or arg and shahristan. Interestingly, the ‘temple’ is outside either of Dumarçay’s projected enclosures which – if it is a religious building – implies a separate ‘sacred enclave’. We must therefore consider the ‘ramparts’ as monumental structures in much the same way as the ‘palace’ and ‘temple’ are, part of an overall monumentalisation of Mundigak that marks Period IV. The command of resources to build these structures, plus the need to make a major architectural statement, implies a renewed status for Mundigak, of more than just a major settlement. Just what

a ‘helmand civilisation’ south of the hindu kush

Figure 4.21  Mundigak, conjectural reconstruction of the Period IV ramparts (compare with Fig. 4.10).

this status might be must remain a matter of speculation, but it does lend support to Whitehouse’s initial suggestion that Mundigak anticipated Kandahar as the regional centre.37 In this connection it is worth observing that a village and site just 3 km to the south of Mundigak (albeit with no material earlier than Parthian) preserves the name ‘Arukh’, derived from the Achaemenid Harahuvatiš/Greek Arachosia/Early Islamic ar-Rukhaj, the ancient name for the region.38 Casal maintained that the Sub-Period IV1 occupation met with a violent end resulting in partial abandonment of the site. The scattered examples of burned buildings reinforce the likelihood that the ‘palace’ and ‘temple’ appear to have fallen into disuse and did not regain prominence again during the later phases of Period IV. Contrastingly, several of the structures associated with the enclosing walls appear to have been continuously inhabited throughout Sub-Period IV1. In the first edition of this book, Shaffer suggested that given the problems created by large amounts of erosion and the lack of adequate stratigraphic data, the three-subperiod division of Period IV might be challenged,39 but new excavations that might clarify things are unlikely in the immediate future. Wherever Sub-Period IV2 was found (mainly Mound B) it was of shallow depth and the structures demonstrated continuous occupation from Sub-Period IV1. The few new structures which were located continued to utilise old walls, including the enclosing walls. Foundations when present were slight and the basic preparation was a simple levelling of the soil. Small sondages away from the enclosing walls indicated that there was some expansion into previously uninhabited areas,

185

186

THE ARCHAEOLOGY OF AFGHANISTAN

confirming the suggestion above of an ‘outer town’. The ceramics show a continuation of previous motifs and vessel forms with some changes and alterations, and metal artefacts demonstrate a marked frequency increase (see below). This sub-period seems to be distinguishable on the basis more of ceramics than of stratigraphic data. Casal proposed that the end of Sub-Period IV2 was caused by a possible earthquake, with the site being abandoned for a short period. Such an earthquake could also account for the evidence of destruction (and conflagration) noted at the end of Sub-Period IV1. Sub-Period IV3 is also known mainly from Mound B. Structures are concentrated around earlier walls, but a new enclosing wall was constructed parallel and a little to the north of the older ones. The new wall was effectively three major parallel walls, and the two exterior walls had a small rubble-filled core, thereby forming a single massive wall. As in the previous sub-periods, the remaining interior space was divided into small rooms used for habitation. Interior stairways again indicate that access to an upper storey or the roof was an important aspect of these dwellings. North of the wall was a large open space of unknown function. Most structures were described as being much more carelessly constructed than those of the preceding phases, but are otherwise very similar. Again, the major distinction for this sub-period is to be found in the ceramic artefacts and the disappearance of previous forms and motifs (most notably animal motifs; see below). Structures from this sub-period have also been located on other mounds, but these preserved parts are usually shallow and highly eroded. The single piece of stone sculpture from the site was found in the upper 15 cm of deposits attributable to Sub-Period IV3. This final sub-period of Period IV did not come to any violent end, and the difference between it and Period V again seems based mainly upon ceramic traits and the establishment of another monumental structure on Mound A. Up to this point the Mundigak sequence has basically demonstrated a record of continuous cultural development, albeit one beset with sampling and stratigraphic problems. However, subsequent to Period IV the site appears to have been abandoned, and Period V belongs to the late Bronze or early Iron Age. Almost all charts and discussions of prehistoric cultural development in Afghanistan terminate with the end of Period IV3,40 although the sequence continues for three more major periods. The interpretation that a major abandonment occurred after Period IV3 has been reinforced by the excavations at Said Qala, Deh Morasi, Shahr-i Sokhta and the Quetta valley, all of which failed to define a sequel to Mundigak IV-type material. Moreover, this hiatus in cultural development seems to be contemporary with the development of the Mature Harappan culture in the Indus valley, which is when there is evidence for contact between the Indus region, northern Afghanistan (at Shortugaï; see Chapter 3), eastern Iran41 and Central Asia.42 Throughout Mundigak III–IV there was evidence of some cultural interaction with the Indus region, and to further complicate the situation there is a pronounced dissimilarity between the material culture of Mundigak IV and V or, for that matter, between V and any other prehistoric culture yet defined

a ‘helmand civilisation’ south of the hindu kush

in the area. Therefore, the culture sequence at Mundigak following Period IV is extremely problematic. Period V No doubt many of the interpretative problems can be attributed to the heavy erosion witnessed by the upper levels of the site in conjunction with what appears to be disturbed stratigraphy. Casal referred to the possibility of earthquakes, a not uncommon phenomenon in this area. Mound A was reoccupied and utilised for the construction of a large monumental building directly atop remains of Sub-Period IV3 (Fig. 4.22). Construction involved the use of mud-bricks placed atop a stone foundation. Previous structures were filled in with debris or covered with brick to form a level surface. The old colonnaded structures were completely covered

Figure 4.22  Mundigak A: plan of the ‘massive monument’.

187

188

THE ARCHAEOLOGY OF AFGHANISTAN

with bricks in the construction of a massive platform on the northern part of the mound. On the top of this platform were two small rooms occupying the highest elevation at the site. Directly south of these rooms, and at a lower elevation, were two long and narrow rectangular rooms encompassing almost the entire east–west distance of the mound’s summit. South of these rooms was another series of small rectangular rooms which could be divided into two groups: small rooms on the west and slightly larger rooms on the east. Indeed, the plan for this southern sector is strikingly similar to that of the last rebuilding in Sub-Period IV3. The immediate summit area of the mound was terraced, and it appears that the perimeter of this whole building complex was defined by these terrace walls. Unfortunately, these upper levels were very heavily eroded, preventing a full structural plan from being delineated. There can be no doubt that the elevated position of the whole structure, the terracing, the massive platform with elevated rooms atop it and the fact that it was painted red and/or white made for a very impressive building. The function of this massive Period V monument is unknown, but it is reasonably certain that it was not a habitation, as it was completely devoid of the usual habitation features such as interior hearths. Furthermore, there is no indication that these rooms possessed any sort of roof structure. The only indication as to the possible purpose of this structure was the location of a ‘human sacrifice’ just outside the foundation of the surrounding terrace, where several human bones including an infant’s jaw bone were found in association with the terrace wall foundation. The human ‘sacrifice’, combined with the stepped-pyramidal shape created by the terrace walls and the massiveness of the structure itself, is certainly suggestive, as Casal noted, of the Mesopotamian ‘ziggurats’, but other monumental structures are now known elsewhere, including Nad-i Ali in Sistan and Konar Sandal in south-east Iran (discussed below). Nonetheless, the exact function of this interesting building remains to be determined. Although the monumental structure was the only extensive area from this period excavated, it was not the only structure associated with Period V. Casal maintained that a whole series of habitations were to be found to the east side of Mound A and extending eastwards to the river, and there appear to be dwellings in areas immediately outside the terrace walls. It should also be noted that the main structure itself underwent at least one sub-period of rebuilding. Architecturally and stratigraphically, there is no greater difference between Periods V and IV than there was between Periods IV and III. The major distinction between Periods IV and V is to be found in the ceramics. The only comparable ceramic industry is part of the Yaz I group in Bactria, which is geographically distant and chronologically later (early Iron Age). It appears that it was upon these ceramic comparisons that Casal based his contentions that a period of abandonment separated Periods IV and V. However, Period V is now generally regarded as much later than Casal originally claimed. Although there are some continuities between Periods V and VI, the latter is not included in this discussion of prehistoric Afghanistan. The presence of an

a ‘helmand civilisation’ south of the hindu kush

iron technology and the associated cultural–ceramic affiliations indicate a late chronology for this period. Chronology There are only a limited number of radiocarbon dates that provide insight into the absolute chronology of the Mundigak sequence. At the time of excavation Casal obtained a series of radiocarbon determinations for Periods I and III,43 which have been recalibrated: Period I5 Period II1 Period III1 Period III5

GSY-50: 3945 ± 150 bp, 2880–2040 cal. bc GSY-52: 3480 ± 115 bp, 2130–1520 cal. bc GSY-51: 2995 ± 110 bp, 1494–927 cal. bc GSY-53; 4185 ± 150 bp, 3330–2345 cal. bc

When these dates were published, Coursaget and Le Run noted that ‘there are numerous difficulties in this sequence of dates’,44 and this can be clarified by noting that they each have large uncertainty ranges, that the calibrated date ranges do not conform to the stratigraphic phasing, and that they also bear little resemblance to the expected ages derived from the relative comparandae (see below). The dates are, however, not modern, so there appears to have been some contamination, perhaps from later pits that were not adequately isolated in the initial excavation. Dales45 independently processed some wood charcoal collected from Period I strata in order to establish more accurate chronological parameters for the initial occupation, and these have also been recalibrated as follows (Tables 3.1, 4.1, 4.2): Period I2–3 (TF-1129: 4947 ± 107 bp) 3975–3520 cal. bc; Period I5 (TF1131: 4568 ± 102 bp) 3630–2940 cal. bc; and Period II, or I5 (TF-1132: 4805 ± 102 bp), 3795–3360 cal. bc.46 The Dales dates suggest that the initial occupation of Mundigak (Period I1–5) occurred during the first half of the fourth millennium bc (c. 4000–3500 bc). Casal’s Period II date is problematic, and the Dales Period II date is from an uncertain stratigraphic context and is statistically similar to his Period I2–3 date. Some appreciation of the chronology of Period II can be obtained by examining the recalibrated range for the Casal Sub-Period III5 date – 3330–2345 cal. bc, which should provide a terminus ante quem for Period II and much of Period III.47 Matters have not been helped by the two published dates for Mundigak Period IV-type material from Deh Morasi (Period IIb), which are dramatically different from each other (P-1493: 4414 ± 53 bp, 3330–2910 cal. bc; P-2292: 5680 ± 300, 5295–3960 cal. bc; see below), with one having a range that is broadly similar to the Casal Sub-Period III5 date and the other being earlier than the earliest date from either site.48 On the basis of ceramic comparisons, Said Qala is similar to Mundigak III and perhaps early Sub-Period IV1, but the radiocarbon dates from Said Qala are also confusing (see below). The three radiocarbon determinations from different strata have been

189

190

THE ARCHAEOLOGY OF AFGHANISTAN

Table 4.1  Radiocarbon dates from sites mentioned in the text Site

Period/phase

Lab sample #

Date BP

Uncertainty

Calibrated range BC

%

Mundigak

I2-3

TF-1129

4947

107

3975–3520

95.4

Mundigak

I5

GSY-50

3945

150

2880–2040

95.4

Mundigak

I5

TF-1131

4568

102

3630–2940

95.3

Mundigak

I5 or II1

TF-1132

4805

102

3795–3360

95.4

Mundigak

II1

GSY-52

3480

115

2130–1520

95.4

Mundigak

III1

GSY-51

2995

110

1495–930

95.4

Mundigak

III5

GSY-53

4185

150

3330–2345

95.3

Deh Morasi Ghundai

IIb

P-1493

4414

 51

3330–2910

95.4

Deh Morasi Ghundai

IIb

P-2292

5680

291

5295–3960

95.4

Deh Morasi Ghundai

IIc

P-2288

3780

233

2880–1660

95.4

Deh Morasi Ghundai

IIc

P-2289

4440

252

3760–2470

95.4

Deh Morasi Ghundai

IIc

P-2290

4090

214

3335–2040

95.5

Deh Morasi Ghundai

IIc

P-2291

4500

 68

3370–2935

95.4

Said Qala Tepe

I

DIC-22

3620

220

2620–1450

95.4

Said Qala Tepe

II

DIC-20

3710

 90

2455–1885

95.4

Said Qala Tepe

III

DIC-18

3800

220

2880–1695

95.4

recalibrated as follows: DIC-22: 3620 ± 220 bp, 2620–1450 cal. bc; DIC-20: 3710 ± 90 bp, 2455–1885 cal. bc; DIC-18: 3800 ± 220 bp, 2880–1695 cal. bc);49 and these ranges are all considerably later than expected. This brief comparison highlights the problems with the absolute dates from late prehistoric phases in southern Afghanistan, and indicates how tentative any assessment of the absolute dating for the Mundigak sequence is. There are also problems with the absolute dates from the earliest phases of Mehrgarh in Pakistan Baluchistan (see Chapter 3).50 The following chronology is based on the radiocarbon dates and comparative material from outside Afghanistan, which will be presented below. As noted above, Mundigak Period I likely spanned the early fourth millennium bc (c. 4000–3500 bc). Period II probably represents a rather short occupation at the site, which occurred about the middle of the fourth millennium bc (c. 3500 bc) and perhaps lasted for one or two hundred years at most. Mundigak III would date no earlier than the mid-fourth millennium bc and potentially continued into the first century of the third millennium bc (c. 3500/3400–3000/2900 bc). Sub-Period III6 and Period IV likely spanned the early to mid-third millennium bc (c. 3000/2900– 2500/2400 bc).51 The degree to which there was contemporaneity between the developments in Helmand taking place at sites like Shahr-i Sokhta (II–III) and Mundigak (III–IV) and those in the Indus river basin are much debated, and current views suggest that Mundigak Sub-Period IV1 dates to the early third millennium bc, pre-dating the urban (Mature) phase of the Indus Civilisation (see below).52

191

a ‘helmand civilisation’ south of the hindu kush

Page 1 of 1 Table 4.2  Calibrated radiocarbon dates for the Neolithic and Bronze Age in southern Afghanistan (MGK: Mundigak; DMG: Deh Morasi Ghundai; SQT: Said Qala Tepe) OxCal v4.3.2 Bronk Ramsey (2017); r:5 IntCal13 atmospheric curve (Reimer et al 2013)

Phase Mundigak R_Date MGK_TF-1129 R_Date MGK_GSY-50 R_Date MGK_TF-1131 R_Date MGK_TF-1132 R_Date MGK_GSY-52 R_Date MGK_GSY-51 R_Date MGK_GSY-53 Phase DehMorasiGhundai R_Date DMG_P-1493 R_Date DMG_P-2292 R_Date DMG_P-2288 R_Date DMG_P-2289 R_Date DMG_P-2290 R_Date DMG_P-2291 Phase SaidQalaTepe R_Date SQT_DIC-22 R_Date SQT_DIC-20 R_Date SQT_DIC-18 7000

6000

5000

4000

3000

Calibrated date (calBC/calAD)

2000

1000

1calBC/1calA

192

THE ARCHAEOLOGY OF AFGHANISTAN

Ceramics Perhaps no other aspect of the material found at Mundigak has received as much attention as the ceramics, and this material continues to act a primary reference point for understanding the relative chronology for southern Afghanistan and the surrounding regions. One of the basic problems centres on the relationship between Sub-Periods I1–3 and Sub-Periods I4–5, and whether the latter should be a more clearly distinguished stratigraphic-cultural unit, which derives from the fact that no substantial architecture was revealed in the deposits until Sub-Periods I4–5. The Period I pottery (Fig. 4.23) is a buff-red ware, predominantly wheel-made (90%) with small quantities of handmade (10%) varieties; however, by Sub-Period I4 pottery that had been shaped through slow rotation (perhaps on a tournette; i.e. ‘turned’ rather than ‘thrown’) had decreased significantly in frequency to 70%, though it increased again in Sub-Period I5 to 80%. Vessel forms are limited to straight-angular or curved wall bowls in Sub-Period I2–3, and only straight-sided vessels in Sub-Period I4–5. In Sub-Periods I4–5 collared globular jars appear for the first time. Little attention was given to the undecorated pottery in Casal’s original report, so as for most sites excavated in this and earlier periods discussion must centre on decorated pottery. The basic decorative pattern for Period I involved with a few rare exceptions the execution of black geometric motifs on a red background. Important geometric motifs include carelessly drawn vertical or overlapping diagonal lines, and more carefully drawn: (1) triangles with interior cross-hatching with triangular or circular dotted tips; (2) vertical lines; (3) large festoons; and (4) a zigzag pattern with interior cross-hatching. From these major motifs only the triangular (1) and a modified festoon (3) persist into Sub-Periods I4–5 and provide continuity for the entire period (Sub-Periods I2–5). Fairservis argued53 that Sub-Periods I4–5 are ceramically different and therefore stratigraphically distinct from Sub-Periods I1–3. This distinction is based upon initial appearance of jar and cup(?) forms and a distinctively different design repertoire incorporating bichrome motifs in the later phases. Given the small exposure and associated sample of Sub-Period I1–3, however, it seems premature to rule out the existence of jar or cup forms in those phases. The argument for a significantly different design repertoire is simply not convincing on the basis of published data, as the predominant motifs in Sub-Periods I4–5 were all found in the earlier phases. Striking new motifs, including bichromes, appear, but are all listed as unique examples and total eleven sherds. Although the total ceramic sample size is unknown, if one takes into consideration the area excavated for Sub-Periods I4–5 and sherd counts from comparable deposits at Said Qala Tepe,54 the total sample size of excavated sherds must have been into the thousands. It is difficult to weigh the significance of eleven sherds as being indicative of a new design repertoire, especially in view of the continuity demonstrated by other motifs and vessel forms. Such sherds are certainly important as possible indicators of

a ‘helmand civilisation’ south of the hindu kush

Figure 4.23  Mundigak, pottery from Period I: (1–2) handmade pottery; (3–9) wheelmade decorated pottery; (10–11) bichrome sherds; (12, 17, 18) Kot Dijian style sherds; (13–14) sherds of ‘Togau’ style; (15–16) possible intrusive sherds.

communication with groups having different ceramic stylistic traditions, or of the development of new locally produced vessel types, potentially with a special function and/or limited market, but by themselves are not sufficient indicators of significant cultural changes. The ceramics of Mundigak Sub-Periods I1–5, then, seem to represent a single cultural development demonstrating a limited degree of internal change with a wide range of variation. Discussions of the Period II ceramics (Fig. 4.24) have centred upon the seem-

193

194

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.24  Mundigak, pottery from Period II: (1–4) handmade pottery; (5–7) decorated wheelmade vessels; (8) sherd showing ‘Rana Ghundai’ style.

ingly abrupt change in the relative percentages of handmade versus turned pottery. Handmade pottery in Sub-Period II1 increases to 97% of the total, as against 3% for turned. These percentages are in marked contrast to those noted for Sub-Period I5 above. Casal interpreted this change as representing a phase of cultural stagnation and this was initially accepted, but there are many problems with this interpretation of the data. First, the sample size that the 97% relates to is not given, and it is not clear whether the sample sizes from Sub-Periods I1– 3 with their greatly reduced excavation area, and presumably reduced sample, are directly comparable to the larger sampling of Period II. Data from Sub-Periods I4–5 indicate that the ratio of handmade to turned was variable between those two sub-phases (see above). Second, Casal notes that decorated turned pottery was replaced almost entirely by crude handmade undecorated pottery. It is not clear, however, whether these frequency variations reflect a shift in ceramic technology or a change in the relative frequency of undecorated versus decorated pottery, or a combination

a ‘helmand civilisation’ south of the hindu kush

Figure 4.25  Mundigak, pottery from Period III: (1–2) handmade pots; (3–12) wheelmade, decorated pottery.

of both. A shift to undecorated pottery might relate to shifts in areal functional activities, as is partially suggested by the architecture. Casal’s description and Shaffer’s examination of these handmade ceramics has indicated that they are very similar to Said Qala Coarse55 and Quetta Slate Temper56 pottery types, which are known to persist into Mundigak IV. At Said Qala Tepe57 it was demonstrated that

195

196

THE ARCHAEOLOGY OF AFGHANISTAN

these types were very consistent in vessel form (similar to ones at Mundigak), but that their relative frequency fluctuated significantly within any given occupation horizon.58 Similar fluctuations can be found in Period II, where by Sub-Period II3 handmade pottery has a frequency of only 60%. Moreover, continuity in architecture and other artefact types, including an increased frequency in metal and luxury objects, can hardly be said to reflect cultural ‘stagnation’. Therefore, the changed frequency of handmade versus turned pottery may reflect important cultural factors, but it does not indicate a well-defined cultural change from Period I and certainly not a period of cultural ‘stagnation’ or ‘devolution’. All pottery in Period II was manufactured from a buff-red paste, but the handmade pottery had coarse tempering materials (chaff and/or crushed rock) added. Among the handmade pottery the predominant vessel forms are the simple angular-walled bowls, and a large, wide-mouthed straight-sided jar with a pinched rim (Fig. 4.24 nos 1–2), which appear for the first time in Sub-Period II1 and continue through to Period IV. A shallow flat-bottomed bowl with vertical walls (Fig. 4.24 no. 3) appears in Sub-Period II2, and another bowl with sharply incurving walls (Fig. 4.24 no. 4) appears in Sub-Period II4. Both these handmade vessel forms continue through to Period IV. A single example of a handmade pedestal vessel was found in Period II. Casal mentions that a few handmade examples were decorated, but notes that the overwhelming majority were undecorated. Both decorated and undecorated examples of turned pottery were found, however, and by Sub-Period II3 the frequency of turned pottery had reached 40%. The most common vessel forms were the angular-walled bowls and the globular collared jars recorded for Period I. Ring bases were common on bowls and a single pedestalled vessel was found. Decoration was found mainly on bowls whose surface had been treated with a thin white or buff-coloured wash. Motifs were confined to black geometries, the most common of which included: festoons below the rim (found in Sub-Periods I4–5); undulating horizontal lines interspaced with horizontal lines (Fig. 4.24, no. 6) (frequent in Sub-Periods III1– 3); and an undulating line bordered by a lower horizontal line and filled with hatches (Fig. 4.24, no. 5) (also frequent in later periods). A unique sherd with a motif having a straight line with opposing perpendicular lines at opposite ends (Fig. 4.24, no. 7) was found in Sub-Period II4. This particular motif became increasingly popular in later periods. A single buff paste sherd with a ‘grey slipped’ exterior (perhaps overfired?) and light red interior was found in Sub-Period II1. It was decorated with a brown motif of parallel horizontal lines bordering a panel characterised by a central undulating line offset on both sides by opposing hatch marks (Fig. 4.24, no. 8). This sherd, like the eleven unique sherds from Period I, was significantly different from the rest of the ceramics. Casal noted its strong resemblance to pottery from Period II at Rana-Ghundai in northern Baluchistan.59 However, the sherd is even more similar60 to de Cardi’s ‘loop-and-tassel’ design61 seen on Togau ware found at Siah II. Togau ware of this period at Siah is also associated with later Kot Dijian and Amri pottery that is similar to the unique sherds found in Mundigak Sub-Periods

a ‘helmand civilisation’ south of the hindu kush

I4–5 (see discussion below on external relationships). To resolve the problem of different but contemporary intrusive types occurring in two stratigraphic levels at Mundigak, Mughal stated that ‘the single specimen from Mundigak II1 level may also be placed with the assemblages of level I’.62 However, Casal gives an exact provenance for this sherd and it appears to be definitely associated with the fill of a structure from Period II1. The specific question which this sherd, and the entire Periods I and II ceramic assemblage, focuses on is: ‘How culturally distinct are Mundigak I and II?’ Are these separate and distinct cultural periods or do they represent a cultural continuum which happened to be separated by a midden deposit? It is argued here that Mundigak I1– 3 are known only from a very limited sample and that I4–5 is not much different from II1–3 if the question of handmade versus turned pottery is placed in proper perspective. It is proposed here that Mundigak Period I (and especially Sub-Periods I4–5) and Period II represent the same basic period of cultural development, demonstrating a continuum of change. A similar cultural continuum can be determined for the ceramics of Mundigak Periods II–III. Although Period III represents a continuous development from the previous periods, it contained some strikingly new ceramic styles (Figs 4.25–4.29). Turned pottery gradually became dominant, increasing from 45% in Sub-Period III1 to 85% in Sub-Period III6. It is interesting to note that handmade pottery never disappeared from the Mundigak sequence and assumes a stable relative frequency similar to that found in Sub-Period III6 (15%). More importantly, it is during this period that the Quetta style of pottery (Quetta Ware) is found, which is similar to Geoksyur and Shahr-i Sokhta pottery,63 and thus indicates connections with southern Turkmenistan, eastern Iran and northern Baluchistan.64 However, the ceramic diversity found in this period makes a descriptive summary of Period III ceramics difficult. To facilitate description and comparisons the following categories have been imposed on the data: Handmade; Turned, Undecorated and Decorated; and finally, Intrusive or Special Function. Period III handmade pottery (Fig. 4.25, nos 1–2) The paste and vessel forms for handmade pottery remain unchanged from previous periods. The major vessel forms are the large mouth, straight-sided globular jars, and straight-angular walled bowls. The shallow bowl forms become more infrequent. Exterior surfaces are almost always heavily stained with carbon, indicating use near an open fire. Vessels made with the rock tempered paste quite often have a thin buff wash applied to the exterior. Period III turned pottery, undecorated Undecorated wheel-made pottery was manufactured from a buff-red paste with either a self- or a sand-temper.65 Two variants were found, one whose surface colour is identical to the paste, and another where the surface had a thin buff

197

198

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.26  Mundigak, decorated pottery from Period III.

wash. Globular jars with simple everted rims were found in the earliest phases and collared jars appear initially in Sub-Period III5–6. The straight-angular walled bowl predominates between Sub-Periods III1–4, but begins to co-vary with an S-profile wall bowl in Sub-Periods III5–6. It seems that a small relatively straight-sided beaker was also manufactured. Period III turned pottery, decorated (Fig. 4.25, nos 3–12) The vast majority of decorated turned pottery is of a single basic type with two variants. Both variants have a buff-red paste with self- or sand-tempering. The

a ‘helmand civilisation’ south of the hindu kush

major difference is that one has its surface coated with a very thin white translucent wash or no wash at all, while the other’s surfaces are coated with a cream-buff opaque wash.66 At present it is impossible to determine if these variants were the result of conscious production or merely firing, paste, or other variables, which were allowed to fluctuate because of their cultural insignificance. Identical vessel forms and vessel form changes as found among undecorated pottery are delineated among the decorated pottery, but there is a higher frequency of ring and pedestal based bowls among decorated pottery. Black or red motifs are almost exclusively geometric until Sub-Period III6, when what appear to be plant motifs are identified (palm fronds/pipal leaves or a stylised leaf). Particularly popular motifs include: festoons; festoons and horizontal, vertical or zigzag lines; horizontal and vertical zigzag lines; open panels formed by multiple lines; and, more rarely, stepped triangles and unaligned stars or undulating lines. The open spaces within these motifs were commonly filled with hachures, undulating lines and, rarely, cross-hachures. Motifs often divided the vessel into thirds or quarters and, in the case of bowls, incorporated large open areas into the overall design. Often, however, decoration was confined to a few horizontal bands around the vessel rim. An important, but infrequent (less than 10%), type of decorated pottery which appears first in Sub-Period III1 and occurs with increasing frequency throughout Period III is Quetta Ware (Figs 4.27, 4.28).67 Its paste colour is generally buff (although a light red is not uncommon) and it has a self- or sand-temper. The surfaces are coated with a thick to thin buff slip68 which ranges in colour from white to cream and occasionally has a grey, or even greenish, colour resulting from either primary or secondary firing. The primary vessel forms seem to be beakers and to a much lesser extent bowls and jars. However, the most distinctive trait of this pottery is its motifs. The black, and more rarely red, motifs are entirely geometric, and Casal and others distinguished two variants based upon differing emphasis on either solid or linear motifs. The solid variant (Fig. 4.27) emphasises solid geometric motifs to the extent of sometimes producing a negative effect. Common motifs of this variant are: stepped triangles; triangles; opposing triangles; diamonds with an internal square and black dot; chequerboard diamonds; and light coloured ellipses and crosses with, or surrounded by, stepped triangles. The linear variety (Fig. 4.28) is characterised by closely set thick lines forming opposing stepped triangles, parallel zigzag lines, and parallel zigzag lines infilled with crosshatching. In both variants, the motifs are executed in a single horizontal band near the rim, which is sometimes divided into panels. As noted above, Quetta Ware demonstrates a generic similarity to other Mundigak ceramics, but is also related to Geoksyur and Shahr-i Sokhta pottery.69 Period III turned pottery, intrusive or special function Included in this category are those ceramics which, because of their rarity or aberrant motifs and/or vessel form, indicate that they are intended for limited/

199

200

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.27  Mundigak, decorated pottery from Period III, showing Quetta ‘solid’ style.

special functions or represent foreign imports (Fig. 4.29). Amri (in Sind) style bichromes, identified for the first time in Period I, are found in small quantities throughout Period III (Fig. 4.29, nos 1–5). Of particular importance was the location of a double-rim white slipped Amri vessel in Sub-Period III4 (Fig. 4.29, no. 12). Also found in Sub-Period III4 were sherds decorated in the style of Amri or Nal (central Baluchistan) bi- or polychromes (Fig. 4.29, nos 6–8). These sherds had large black geometric motifs infilled with different colours (yellow, red-orange and white). Another important sherd from Sub-Period III4 has a very dense buff paste with an excellently executed pipal leaf motif (Fig. 4.29, no. 10) in the style of Faiz Mohammad-type pottery from the Quetta valley.70 Finally, from Sub-Period III5 are several sherds from a vessel with a large horizontal brown band near the rim resembling closely the Kot Dijian pottery from the Indus region (Fig. 4.29, no. 9).

a ‘helmand civilisation’ south of the hindu kush

Figure 4.28  Mundigak, decorated pottery from Period III, showing Quetta ‘linear’ style.

The ceramics from Period IV (Figs 4.30–4.40) are marked by several changes in vessel forms, motifs, the appearance of a new red paste and the disappearance of Quetta Ware. Among the turned pottery a true red paste increases in frequency at the expense of the previous buff-red paste. This transition in ceramic paste colours is potentially related to the new types of large ovens (‘kilns’) also associated with Period IV, though the red paste pottery is also characterised by an increasing utilisation of a red slip decorated with distinctive, but related, black motifs. There is also an increased frequency and variety of limited or special function pottery.

201

202

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.29  Mundigak, pottery from Period III: (1–5) Amri style bichrome ware; (6–8) sherds of Amri or Nal style bichrome or polychrome decoration; (9) Kot Dijian style vessel; (10) sherds of Faiz Mohammad style; (11) decorated vessel of canister form; (12) double rim of Amri style.

a ‘helmand civilisation’ south of the hindu kush

Figure 4.30  Mundigak, wheelmade and decorated pottery from Period IV1.

Period IV handmade pottery Casal commented that handmade pottery becomes rare during Period IV. However, examination of the Mundigak pottery at the National Museum in Kabul in the 1970s indicated that it was still present in some frequency. Moreover, Casal’s frequency chart71 indicates that the major jar vessel form which characterised the

203

204

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.31  Mundigak, wheelmade and decorated pottery from Period IV1.

handmade pottery in earlier periods maintained the same frequency level in Period IV as it had at the end of Period III.72 These factors suggest that handmade pottery continued to be manufactured, albeit at a reduced frequency (less than 10%?), and utilised as utility vessels. Period IV turned pottery, undecorated Besides the introduction of red paste, undecorated wheel-made pottery underwent significant vessel form changes. It is important to note that these vessel form changes are equally as applicable to wheel-made decorated pottery. A major

a ‘helmand civilisation’ south of the hindu kush

change occurs among bowl forms in that the previous angular-wall form is essentially replaced by the S-shaped wall form encountered first in Sub-Periods III5–6. Recorded for the first time in Period IV is a simple hemispherical form which increases in frequency over time, and more rarely a sharply carinated form. Except for a large relatively straight-sided variety, jar forms are almost exclusively globular with collars having straight or everted lips. Beakers are now predominantly manufactured with an S-shaped wall profile. Some of the smaller beakers may have pedestal bases with flaring or carinated walls.

Figure 4.32  Mundigak, wheelmade and decorated pottery from Period IV1.

205

206

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.33  Mundigak, wheelmade and decorated pottery from Period IV1.

Period IV turned pottery, decorated The most striking feature of the decorated wheel-made pottery was the significant frequency increase of stemmed goblets (Fig. 4.32) with incurving sides (first identified in III5) in Sub-Period IV1 and its equally significant decrease in Sub-Periods IV2–3. These vessels were decorated with an extremely high frequency of zoomorphic motifs, which are themselves a significant addition to the Mundigak motif

a ‘helmand civilisation’ south of the hindu kush

Figure 4.34  Mundigak, decorated pottery of Quetta style, Period IV1: (1–7) ‘linear’ style; (8–10) ‘solid’ style; (11–12) ‘Bucranium’ style.

repertoire. Horned caprids (ibex?), felines, birds and fish were depicted in a solid or hatched style in a wide horizontal band. Equally frequent were solid or hatched floral motifs, especially the pipal leaf. Geometric motif combinations of parallel zigzag and tight undulating lines executed diagonally were also recorded on these vessels. A related vessel form, which had the same frequency pattern as the above form, is a short-stemmed goblet with an S-shaped wall, sometimes marked by a sharp carination. This form is usually decorated with geometric motifs, particularly a band of parallel zigzag lines. Another important change was the decreased frequency of Quetta Ware in SubPeriod IV1 (Fig. 4.34) and its total absence from Sub-Period IV2–3. Those examples

207

208

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.35  Mundigak, special function and/or intrusive pottery, Period IV2: (1–5) Faiz Mohammad style; (6–9) Amri and Kot Diji style; (10–13) Bichrome style.

which are found lack the characteristic solid geometric motifs so common previously. Even the linear style of Quetta Ware lacks the usual geometric elements such as stepped triangles, crosses, etc. Motifs on the few examples found are mostly curvilinear or quasi-floral (termed bucranium by some). Interestingly, a certain degree of stylistic similarity can be distilled between the previous Quetta style and the motif combinations found on the remaining decorated pottery of Period IV. The remaining decorated pottery of Period IV, especially that from Sub-Period IV1, demonstrates a generic continuity with that in previous periods (Figs 4.30, 4.31, 4.33). There is a definite tendency to utilise a more opaque white or creamcoloured wash and the motifs are predominantly black. In general, there is a much greater use of hatching to fill in large open areas, eliminating some of the ‘openness’ of the preceding period. There is also a much greater usage of

a ‘helmand civilisation’ south of the hindu kush

parallel zigzag lines both horizontally and vertically. Tightly undulating lines are used as free motifs in place of simple solid lines, and also as a sort of hatching. New geometric motifs are limited and include circles within circles and, more rarely, intersecting circles. The previously mentioned animal and plant motifs are significant additions, but the animal motifs are infrequent in Sub-Period IV2 and disappear by Sub-Period IV3. Plant motifs, especially the pipal leaf, continue throughout the period in decreasing frequency. Solid geometric motifs are rare and usually confined to beakers. The later Sub-Periods IV2–3, are distinguished by an increasing quantity of redslipped red ware decorated with black motifs. These motifs are not radically different from those already recorded and would if anything appear to be generic with the previous ones. Perhaps one of the more distinguishing aspects of this decorated pottery is its correlation between shallow hemispherical bowls with everted rims and shallow sharply-angular walled bowls (Fig. 4.38, nos 2, 3). The co-variance between this black-red slipped red ware and Quetta Ware has not yet received adequate attention, but it is interesting to note that in the same sub-periods of Period IV, several interesting possible intrusive pottery types are recorded from regions to the south-east. Period IV turned pottery, intrusive or special function This category of pottery continued to indicate the possibility of interaction between Mundigak and regions to the south-east, particularly northern Baluchistan and the Indus valley. Throughout Period IV sherds could be isolated which resembled material produced in these regions (Figs 4.35, 4.40). Sub-Period IV1 had several examples of a large collarless jar with everted rim similar to pottery found at Kot Diji.73 Two rim sherds from what appear to be S-shaped beakers with a sharply carinated double-rim and fish-scale motif74 are very different from anything else found at Mundigak, but similar in style and form to Amri and Kot Diji material. Several examples of the Faiz Mohammed Painted wares were also found in Sub-Periods IV1–2. Also located in Sub-Periods IV1–2 was a red and black bichrome pottery which was similar in manufacture and style to the pottery definitely indigenous to Mundigak. Faiz Mohammad Painted pottery was also found in Sub-Period IV3, as was a complete vessel (Fig. 4.40, no. 3) of Quetta Wet-Ware-type pottery, which is typically found in Baluchistan and the Indus region in pre-Harappan contexts.75 Also found in Sub-Period IV376 was an isolated example of a pipal leaf motif executed in black with red infilling similar to the Sothi Ware designs seen at the eastern Indus site of Kalibangan, in northern Rajasthan.77 There are also examples of an intersecting circle motif infilled with hatching,78 which is very similar to late Amri and Kot Dijian examples. It is clear from these specific examples, and others, that during Sub-Periods IV1–3 there was some form of communication between Mundigak and Baluchistan and the Indus region.

209

210

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.36  Mundigak, wheelmade and decorated pottery from Period IV2.

a ‘helmand civilisation’ south of the hindu kush

Figure 4.37  Mundigak, polychrome vessel from Period IV2.

Unlike previous fluctuations in handmade pottery (Periods I–III), which were confined to the coarse tempered utilitarian vessels (also persisting into Period V), handmade pottery in Period V had new vessel forms and black-on-red slipped decoration, and was made from a finer paste (Fig. 4.41). It is difficult to evaluate the significance of this change since no relative frequency data are available for this period. Most of the handmade ceramics appear to be bowls, but not all bowls were handmade. Overall, regardless of manufacturing method, these new forms were deeper and more curvilinear. Among the various varieties, three basic bowl forms can be defined: (1) simple hemispherical; (2) a form with the lower three-quarters being a straight-angular wall with a slight convex curve ending in an S-shaped, straight or incurving lip; and (3) a more globular bowl with an S-shaped profile. Simple circular handles are a common feature on many vessels, particularly the last vessel form (3). In contrast, jar vessel forms, both decorated and undecorated, were predominantly wheel-made. Both collared and uncollared globular jars were found with either everted or simple rims. Judging from the illustrations, jars appear to be mainly large in size and sometimes manufactured with spouts. The almost completely geometric black or violet motifs are executed on a red slip which on some examples is extremely thin, almost a wash. A few highly stylised ‘stick’ figure zoomorphic motifs were found, and, more rarely, a horizontal frieze of horns. The basic decorative style is highly conventionalised, consisting of

211

212

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.38  Mundigak, wheelmade and decorated pottery from Period IV3. Nos 6 and 9 are greywares.

a ‘helmand civilisation’ south of the hindu kush

Figure 4.39  Mundigak, wheelmade and decorated pottery from Period IV3.

213

214

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.40  Mundigak, special function and/or intrusive style pottery from Period IV3: (1–2) Quetta style; (3) Quetta ‘Wet’ ware style; (4) Faiz Mohammad style; (5–6) Late Amri style; (7) Bichrome style.

a single horizontal band of cross-hatching near the vessel rim. Sometimes multiple bands were found separated by open spaces which could be filled with additional simple geometric motifs. Extending downward from these bands, and dividing the vessel surface into panels, are a series of solid or multiple-lined isosceles triangles or, more rarely, diamond motifs. Sometimes these elements are joined at the top by a festooned band. These motif combinations are found on both bowls and jars.

a ‘helmand civilisation’ south of the hindu kush

Figure 4.41  Mundigak, decorated pottery from Period V.

215

216

THE ARCHAEOLOGY OF AFGHANISTAN

However, on jar forms a more simple pattern of cross-hatched triangles can be found as well as cordons emphasising the motifs. Lithic Artefacts The lithic artefacts that are discussed here are those which have a utilitarian function. Those artefacts, which because of the nature of the object produced (beads, amulets, small vessels, etc.), stone of manufacture (semi-precious or precious) and comparatively small quantity indicate a non-utilitarian function, are discussed under Miscellaneous Artefacts.

Figure 4.42  Mundigak, miscellaneous objects: (1–2) retouched flakes; (3–4) chert blades; (5) grooved stone; (6–7) engraved stone objects; (8–10) chert points, Periods I–VI; (11–13) chert points, Periods III6VI; (14) bone point; (15–17) conical ceramic spindles; (18–19) disc-shaped ceramic and stone spindles.

a ‘helmand civilisation’ south of the hindu kush

Rectangular milling stones and associated rectangular to oval hand stones were recorded in all occupational periods at Mundigak. Milling stones were manufactured from medium to coarse grain basaltic boulders. They had a single level working surface (or concave depending upon degree of use), sometimes with one raised end. Many of these stones had their unworked surface smoothed by water action. Hand stones were manufactured from a medium to fine grain basaltic cobble and had grinding scars on one or both surfaces. Besides these milling and hand stones some crude mortars and pestles were identified for Period I. Another type of ground stone artefact found in Sub-Periods I5–III6 was a perforated, large white limestone ‘weight’. Stones with a long axis groove appear in Sub-Period III4 and increase significantly in remaining occupations (‘counter-weights’?) (Figs 4.42, no. 5). Large trapezoidal hoes retouched for hafting and having a polish resulting from use were located from the end of Period I until the final occupation. Beginning in Sub-Period I4 and persisting throughout the rest of the sequence at Mundigak is a type of knife-scraper manufactured from large cortex flakes (Fig. 4.42, nos 1–2). These flakes were usually retouched along one edge and were occasionally polished through usage. Flint artefacts appear for the first time in Period II, in the form of blades and points. Initially these blades are large and have triangular (Period II) (Fig. 4.42, no. 3) and trapezoidal (Period III) (Fig. 4.42, nos 3–4) cross-sections, but are very rare. However, in Sub-Periods III5–6 microlithic (2–5 cm) blades are introduced in quantity and are located throughout the remaining periods of occupation. Bifacially flaked lanceolate flint points were found first in Period II2 and continue throughout the sequence (Fig. 4.42, nos 8–10). Two types are distinguishable on the basis of size, large and small, with several examples of extremely large points located in Sub-Period IIIe. In Period IV a new type of point is introduced which is triangular with a rounded base and persists through Period VI (Fig. 4.42, nos 11–13). It is important to note that points of all types are found in some quantity. Period IV is also distinguished by the appearance of several examples of small concave-sided cylindrical objects with central perforation, and Quetta Ware-style geometric motifs engraved on their sides (Fig. 4.42, nos 6–7). These objects are manufactured from alabaster and other hard stones. A few pieces of unworked galena are also found in Period IV, but their purpose is unknown. Bone Artefacts Bone artefacts were limited in variety and were found with frequency only in Periods I–III. Casal maintained that the distribution of bone artefacts, particularly pointed objects, co-varies with increased utilisation of bronze points.79 The most numerous bone artefacts were hundreds of pointed awl or punch fragments (Fig. 4.42, no. 14). These awls, or punches, were manufactured from the long bones of various animals (sheep/goat?) and were polished from use. The first example is recorded in Sub-Period I3, reaches its greatest frequency of complete examples in Sub-Period II3 (88), progressively declines throughout Period III and becomes rare

217

218

THE ARCHAEOLOGY OF AFGHANISTAN

by Period IV. Another bone artefact found only in Period I (three examples) was a small polished rectangular piece of bone with several perforations whose function is unknown, though it may have been used in bead drilling. Bone spatulas are frequently found in Period III and, unlike the awls or punches, are encountered throughout Period IV. Rare examples of bone tubes were also found in Periods III–IV. Metal Artefacts Metal artefacts were first found in Sub-Period I2 and increased in frequency and variety throughout the rest of the sequence. The earliest example was a flat bladelike instrument which might have had a hafting tang (Fig. 4.43, no. 10). However, the most frequent metal artefact of the entire Mundigak sequence was a simple type of bronze point or punch with a circular cross-section (Fig. 4.43, no. 11). It was first identified in Sub-Period I4 and is so easily available that it replaces the bone awl/punch in Period IV. A single example of this point/punch was found in SubPeriod IV3 still hafted into a bone handle, confirming its functional designation as a punch. The first example of a true projectile point was identified in Sub-Period II3 and was lanceolate in shape with an elliptical cross-section: the type became increasingly frequent in later periods (Fig. 4.43, nos 12–13). In Period III a tanged lozenge-shaped point was introduced, and in Period IV a tanged oval-shaped point. Other possible weapons found in Period IV were a large lance head (Fig. 4.43, no. 16) and knife (sword) (Fig. 4.44, no. 14), but these artefact types were very rare. The first examples of ‘luxury’ metal artefacts were found in Sub-Period II3. These two artefacts were pins, one with a double-volute end while the other had a flattened and perforated end (Fig. 4.44, nos 6–7). Similar pins with flat ends and twisted shafts were also found in Period III (Fig. 4.44, no. 8). However, the greatest number and variety of ‘luxury’-type objects were found in Sub-Period IV1. Among such objects identified were: concave discs (mirrors) (Fig. 4.44, no. 13); doublevolute, lozenge and broad-flat headed pins (Fig. 4.44, nos 18–20); handles for discs (mirrors); and a buckle. It is important to note that in at least two instances smelted(?) iron decorative buttons were found on objects in Period IV. Frequently encountered utilitarian objects in Period III–IV were small curved knives or sickles (Fig. 4.44, nos 1–2) and chisels, and, only in Period IV, a few barbed hooks. Three important metal artefacts located in Sub-Period III6 were the only examples of socket-hole axes (two) and an adze (Figs 4.44, nos 3–5; 4.38a). Most of the more common utilitarian and ‘luxury’ metal artefacts were also identified in Period V. Elemental analysis of some of these artefacts indicated some interesting aspects of the metallurgical sophistication that they represent. Analysis of one artefact from Sub-Period 15 demonstrated that it was a very low-tin bronze, with tin accounting for only about 1%, iron 0.15% and the remaining material being copper. Such a composition is a striking contrast to the high tin content of the apparently chronologically earlier material from Ghar-i Mar. Of the artefacts

a ‘helmand civilisation’ south of the hindu kush

Figure 4.43  Mundigak: (1–2) stone bowls of common form, Periods I–VI; (3–9) additional forms of stone vessel, Period IV; (10–11) copper or bronze objects, Period I; (12) bronze point, Period III; (13) bronze point, Period IV; (14–15) bronze points, Period V; (16) bronze lance head, Period IV.

219

220

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.44  Mundigak, metal objects: (1–2) bronze sickle blades; (3–5) bronze axes and adze, Period III; (6–7) bronze pins, Period II; (8) bronze pin, Period III; (9–12) bronze pins, Period IV; (13) bronze short sword (?), Period IV; (14) bronze mirror, Period IV.

a ‘helmand civilisation’ south of the hindu kush

subjected to analysis from Period III, only the axes and adze had a composition approaching that usually associated with bronze, with a tin content of almost 5%, making it the highest recorded at Mundigak. This contrasts with the composition of other artefacts from this period, which is similar to that of Period I. Axes and adzes rely on weight, force and hardness for their efficiency as tools, and these characteristics are maximised by a high tin content. Other items such as points, pins or knives have different usages requiring different characteristics, and this is reflected in their lower tin content. Therefore, it appears that by Period III the differential characteristics of various alloying compositions had been recognised and metallurgy was entering, or had entered, a new phase of sophistication and presumably specialisation. During Period IV the tin content of such items as pins increases slightly (1.5%) and traces of lead (1.6%) and other elements are found. Miscellaneous Artefacts Spindle whorls During Sub-Periods I4–II two types of ceramic spindle whorls were found. The most frequent type was a cone shape (Fig. 4.42, no. 15) and the other was a truncated cone shape (Fig. 4.42, nos 16–17). Both types had a central perforation. In Period III both of these conical types decrease in frequency, and from Sub-Periods III5–V they are replaced by ceramic and stone (steatite) disc-shaped whorls (Fig. 4.45, nos 18–19). Stone vessels The predominant type of stone vessel is the simple bowl with straight-angular walls manufactured from alabaster (Figs. 4.43, nos 1–2; 4.45a; Fig. 4.45). It was first recorded in Sub-Period I2, increased in frequency in Sub-Period III6, and was located

Figure 4.45  Mundigak, stone vessels in the National Museum.

221

222

THE ARCHAEOLOGY OF AFGHANISTAN

in even greater quantity during Period IV. There was also a significant increase in the variety of stone vessel forms during Period IV (Fig. 4.43, nos 3–9), including: small beakers with straight and carinated sides; bowls with curving sides; pedestal bowls/goblets; bowls with short vertical sides; pentagonal canisters; and stemmed goblets. Many examples were found with geometric motifs similar to that found on the pottery engraved on the sides. White veined marble was also used in the manufacture of these later varieties. Beads and pendants Only a limited number of beads were found in Sub-Periods I3–II. Most of the shapes identified for these early periods will persist throughout the sequence, and the shapes included: simple-disc; tubular; trapezoidal; cylindrical; biconical; and lozenge. The material of manufacture included the following types of stones: various types of siliceous white limestone; shell; lapis lazuli (3); jade(?); blue stone (turquoise?); black or grey stone (steatite?); carnelian (2); quartzite; and pottery. Period III witnessed a significant quantitative increase in such beads, half of which were in association with the burials of Sub-Period III6, and most of the remaining examples were also located in a Sub-Period III6 provenance. Predominantly, the beads of this period were manufactured from steatite (11), quartzite (13) and siliceous limestone, with a few of lapis lazuli (4). The most prolific period for beads was Period IV, and particularly Sub-Period IV1. Besides already existing shapes, the following new ones were identified: small biconical; multifaceted biconical; eight-point star; truncated cone; rectangular with elliptical cross-section; oval with star cross-section; and irregular. Among materials of manufacture significant quantitative increases could be determined for lapis lazuli, carnelian and shell. Inlaid beads occur for the first time, represented by a single rectangular carnelian bead with a figure eight in white. Bronze beads are also encountered for the first time in the form of a simple tube and one small bell with an iron ball in the centre. Similar beads (except lapis lazuli) can be identified for Period V, but in lesser quantities. Seals (Fig. 4.46) A single example of a steatite compartmented seal with geometric motifs was found in Sub-Period II2. Such seals underwent a dramatic frequency increase in Sub-Period III6 (30), although they were infrequently found in the earlier phases of that period (Sub-Period III4 = 2, Sub-Period III5 = 3). The majority of these seals were rectangular with two central perforations. Identical seals made from bone were also found in Sub-Period III5 (1) and Sub-Period III6 (3). Yet another significant increase in seals occurred in Period IV, which is also distinguished by the introduction of several new geometric shapes of seals. The most noticeable addition was of notched edges encompassing the seal circumference. Metal seals were found in

a ‘helmand civilisation’ south of the hindu kush

Figure 4.46  Mundigak, seals, spindle whorls and Period IV ceramics in the National Museum.

the last phases of Period IV but were rare, and one had a zoomorphic motif. Only a single compartmented seal was identified as belonging to Period V. Ceramic figurines Only four figurines, of humped bulls, were found in Sub-Periods I3–5. Casal stated that such figurines increased in frequency during Period II, but no quantification was given. A single example of an anthropomorphic figurine was found in Period II. This is a crudely modelled human torso of indeterminate sex. Both zoomorphic and anthropomorphic figurines are numerous in Period III. Bull figurines are particularly numerous in the early phases of this period (Sub-Period III1 = 71), but then progressively decline in the later phases (Sub-Period III6 = 30). Two figurines resembling goats were also found in Sub-Period III1. A similar increase can be defined for the anthropomorphic figurines. Unfortunately, Casal gives no actual counts but refers to them as being abundant (Sub-Period III1 = 15). These figurines were predominantly females with prominent breasts. All figurines are highly stylised in a standing position. Arms are represented by mere wing-like projections while the lower portions are distinguished by broad flat hips. The body below the hips was not modelled at all. A single figurine had some punctated motifs in the neck region, which presumably represented a necklace. Period IV was the final occurrence of figurines in any significant quantity, and included a white calcite human head in Sub-Period IV3 deposits. Two important points need to be made about the Period IV figurines. First, considering the horizontal area of excavations in contrast to that of previous periods, the quantity of

223

224

THE ARCHAEOLOGY OF AFGHANISTAN

figurines seems very small, though it still appears that such figurines were actually less frequently made than in earlier periods. Second, according to available information, these figurines were conspicuously absent from both the ‘palace’ and ‘temple’ structures. In Period IV, figurines were confined to the habitation areas surrounding the enclosure walls, or in the same contexts as they were located in the previous periods. Most zoomorphic figurines were highly stylised representations of the humped bull/cow. Several were found with painted decorations, including polychrome, or an appliqué collar. A few figurines represent caprids such as sheep/goat or ibex and there is a single figurine of a pig. Anthropomorphic figurines were predominantly female. A notable exception is a sculptured in the round male figurine decorated with polychrome paint. Most female figurines were highly stylised with pinched faces, prominent breasts, appliqué eyes, winged arms, broad hips, and otherwise a rather flat profile. Appliqué and painted necklace and coiffures are recorded on many examples. All figurines are standing, with one exception, which has its legs crossed and extended out in front, apparently to act as a support. Two very fine examples of female figurines modelled in the round in a style usually referred to as ‘Zhob’ figurines were located in Period IV. Only the upper torso was found, but such figurines are usually depicted as seated when found at other sites. In contrast to Period IV, only a single female figurine was found in Period V. Faunal and Floral Remains Domesticated animals were identified initially in Period I and included sheep, goat, cattle, ass, horse and dog. Wild animals were represented by gazelle, ibex and lynx. Period II had the same complex of domesticated and wild animals, with the addition of a wild bird of prey. More important, the first cereal remains identified at Mundigak come from this period and included domesticated wheat (Triticum compactium). The faunal remains in later periods include the same animals, but no plant remains were identified. It is thus clear that the inhabitants of Mundigak were exploiting domesticated plants and animals from the initial occupation of the site, which is not surprising considering the developments that had taken place in the surrounding regions in the preceding millennia (see Chapter 3). Said Qala Tepe Said Qala Tepe is located approximately 96 km south-east of Mundigak near Kandahar. This site is not comparable to Mundigak in terms of its absolute size (8 m of elevation and 200 m in diameter), areal extent of excavations (two 10 m squares and a 6 × 2 m sondage) or extensiveness of the cultural sequence (all occupations correspond to Mundigak III5–IV1). The site was first tested by Fairservis;80 however, the major excavations were conducted by Shaffer in the 1970s, almost twenty years later.81 Shaffer’s excavations made use of somewhat improved

a ‘helmand civilisation’ south of the hindu kush

methods of artefact retrieval and quantitative analysis in comparison to what was attempted at Mundigak, but these were still very much of their time. The discussion presented here differs from previous reports on these excavations, principally in that previous reports retained the field designations for major occupations which were numbered from top to bottom, whereas here major occupations are numbered from the bottom to reflect the site sequence. Three radiocarbon dates and hindsight have also significantly altered the original interpretation. Stratigraphy and Architecture The initial occupation at Said Qala Tepe, Period I, is known only from the lowest 3 m of deposits in the sondage (Fig. 4.47). These deposits are characterised by several layers and lenses of differential soil deposition which can be distinguished on the basis of different colours, textures and cultural content. Although different, these depositions indicate continuous occupation throughout this period, and analysis of the associated artefacts adds validity to grouping these deposits into a single occupation period. The limited areal extent of the sondage precludes any definitive statements being made about architectural characteristics of this period, but it does appear that simple rectangular structures of mud-brick were constructed. Period II is likewise primarily known only from the sondage, although a final occupational phase was delineated in the other excavated areas. The outstanding feature of this period was the construction of a large mud-brick wall of unknown function. Previous deposits of Period I were levelled in preparing the area for construction of the wall. The wall was constructed entirely of mud-brick and clay fill. After levelling, a very wide wall (3.3+ m and estimated up to 6 m) was constructed of solid mud-brick up to a height of 80 cm. At this height two separate wall faces were constructed of mud-brick (1.5 m thick) up to a height of 3+ m. The area between these two walls was filled with undifferentiated clay free of any cultural debris. Only the south face of this wall was exposed in the excavations and there is no indication as to its function or overall shape. Several layers and lenses of differentiated clay soils abut the south face, and the remains of at least one mud-brick structure can be determined. These deposits are then sealed by a significant amount of wall fall from this wall, providing a stratigraphic boundary between Periods II and III. Additional structures from Period II were located in the other excavated area. Two large trapezoidal mud-brick structures with lateral entrances were separated by a large open space from a series of small rectangular room structures (Fig. 4.48). These small rooms had several common walls and a rather haphazard arrangement. The most distinctive feature found in these rooms was a large rectangular interior oven constructed from mud-bricks and abutted against the wall. This feature had an interior clay-lined fire-pit with a long narrow trench providing access to the interior fire-pit from outside the oven. Such a trench might have functioned as a bellows providing oxygen to the fire. Judging from the

225

Figure 4.47  Said Qala, sondage profiles.

Figure 4.48  Said Qala, example of Period II architecture.

228

THE ARCHAEOLOGY OF AFGHANISTAN

stratigraphic position of these structures they must represent a final occupational phase of Period II. The final major prehistoric occupation at the site, Period III, was located in both excavation units. In the sondage it consisted of a series of differentiated deposits sandwiched between the wall fall and a depositional layer which contained ‘Kushano-Sasanian’ pottery and correlated with the large cemetery of that period located in the other excavated areas.82 Excavations outside of the sondage indicated that at least four phases of occupation could be determined for this period (IIIA–D). The first three phases represented continuous building and rebuilding of small rectangular mud-brick structures similar to those found at Mundigak III (Fig. 4.49). Walls were constructed with both single- and double-coursed mud-brick and only rarely were entrances definable. Several structures had interior ovens similar to those found at Mundigak, while others only had a simple clay-lined fire-pit. A single structure in Phase IIIA had two milling stone fragments set into the floor with a centrally located groove on the upper edge. Between these two stones were extensive ash and bone deposits, suggesting that it represents a spit-roast. In Phase IIIC some of these structures had their walls founded in a trench with the first few wall courses being headers with the remaining wall constructed of stretchers. The final occupation phase, IIID, was represented by a series of small mud-brick and pakhsa structures which appeared to function as storage bins or work areas but were not related to any habitation structures. Also located in this final phase were several extremely large rubbish pits. The areas of the site tested were not occupied after this period until utilisation as a cemetery by a much later ‘KushanoSasanian’ group. However, it is possible that other areas of the site, or around the site, continued to be occupied as the cemetery cut through a deposition layer that contains extensive prehistoric ceramics but no occupational features. The ceramics from this level have been analysed under the designation of Period IV while the cemetery constitutes Period V and the modern surface of the mound Period VI. Chronology Three radiocarbon dates have been processed from the site, one from each different period. The dates from each period have been recalibrated: Period I (DIC-22: 3620 ± 220 bp) 2620–1450 cal bc, Period II (DIC-20: 3710 ± 90 bp) 2455–1885 cal. bc and Period III (DIC-18: 3800 ± 220 bp) 2880–1695 cal. bc (Tables 3.1, 4.1, 4.2). As can be seen from the calibrated ranges, the dates are in reverse chronological order, and a possibility of ground-water contamination was noted by the processing laboratory, making the absolute dates somewhat suspect. Although all three Said Qala dates are basically contemporaneous, they do not accord with the dates from Mundigak, despite there being material correlations that suggest that all occupation phases at Said Qala may essentially be equated with those of Mundigak Sub-Periods III5–6–IV1. This further highlights that there are problems with the available late prehistoric radiocarbon dates from southern Afghanistan.

Figure 4.49  Said Qala: (a) example of architecture, Period III; (b) oven from above structure (note ‘linear style’ Quetta black-on-buff sherd).

230

THE ARCHAEOLOGY OF AFGHANISTAN

Ceramics Said Qala Period I ceramics are known only from the sondage. Here, as at Mundigak, the useful dichotomy of handmade and turned ceramics is made to facilitate description and analysis. The paste for all the ceramics at Said Qala is essentially the same as was encountered at Mundigak, being a buff-red in colour and self- or sand-tempered, with a few noted exceptions. Two varieties of handmade pottery could be distinguished on the basis of surface treatment, although both have chaff tempering and the same vessel forms as those described for Mundigak handmade pottery. The most frequent type had a basket-impressed surface while the other was simply smoothed. Together these handmade potteries accounted for less than 2% of Period I pottery. Turned pottery was identical, in terms of varieties and vessel forms, with that described for Mundigak Period III. The decorated varieties had a relative frequency of about 3% with the similar, but more limited, variety of motifs. Only rarely were a few sherds of Quetta Ware found. A few sherds of a black-on-red slipped pottery, called Kili Gul Mohammad after a type in the Quetta valley,83 and Faiz Mohammad Grey ware (Fig. 4.52, no. 1) from the same area were also found. Period II is known from both excavation areas, as are all subsequent periods. Major ceramic changes were confined almost entirely to the handmade ceramics. There was a major increase in the relative frequency of handmade vessels, to almost 40% in the sondage and about 10% in the habitation area, which is probably more representative. Moreover, the variant with the smoothed surface is the predominant type, with only rare sherds of the basket-impressed variety being found. Equally rare were sherds of a handmade pottery with crushed rock tempering (similar to ‘Quetta Slate Temper’).84 There were no changes in associated vessel forms as these remain constant throughout the sequence for handmade pottery. Among the turned ceramics there was little noticeable change. The decorated varieties now represented about 10% of the total pottery assemblage, a frequency which remained relatively constant throughout the rest of the sequence. The larger sample from the habitation area indicates that most of the motifs described for Mundigak III pottery are present at Said Qala Tepe. Quetta Ware reaches a frequency of about 1%, which then also remains constant throughout the rest of the sequence. Several new types of special-function or intrusive pottery can be added to the two already listed in Period I. These new types (Figs 4.50–4.52) come primarily from the habitation area and include: Quetta Wet variants; Kechi Beg white-on-dark slip;85 Kechi Beg Polychrome/Bichrome;86 Amri Polychrome;87 Nal black-on-buff;88 and a few sherds resembling the polychromes from Gumla III and Mehrgarh IV–V.89 Each variety was represented only by a very few sherds. The last prehistoric occupation periods, Periods III–IV, will be discussed together since the detectable changes in the ceramics represent a continuum between the two. Handmade pottery decreases as regards the frequency of the crushed rock tempered variety. It is interesting to note that this latter variety was never

a ‘helmand civilisation’ south of the hindu kush

Figure 4.50  Said Qala, intrusive or special function sherds: (1–2) Nal style, black-on-buff pottery; (3) Nal style polychrome sherd.

identified in the sondage. Among wheel-made pottery vessel forms, changes occur which parallel those of Mundigak Sub-Periods III5–IV1. Bowls change from an angular wall to an S-shaped wall form. Jars are predominantly globular with collars. Beakers change from a parallel wall to an S-shaped wall profile. A few stemmed

231

232

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.51  Said Qala, intrusive or special function sherds: (1) Amri style polychrome; (2) Gumla style polychrome; (3) Amri (?) style polychrome pot lid.

vessel forms are recorded, but never the small pedestal-based beaker with flaring or carinated walls that was found in Mundigak IV1. Similar geometric motifs are identifiable on Said Qala Period III–IV and Mundigak Sub-Periods III5–IV1 pottery. Although floral motifs are found on Said Qala pottery, zoomorphic motifs were completely absent. Likewise, the Quetta Ware motifs are more similar to those identified in a Mundigak Sub-Periods III5– 6 context than in Sub-Period IV1. These latter factors argue for a closer correlation between the ceramics of Said Qala Period III–IV and Mundigak Sub-Periods III5–6 than between the Said Qala Period III–IV ceramics and those of Mundigak Sub-Period IV1, but these boundaries are very elusive. The only significant addition to the special function- or intrusivetype category is a single sherd of Nal Polychrome (Fig. 4.50, no.3). Overall, the ceramics at Said Qala support the ascription of all prehistoric occupations at the site as being contemporary with those of Mundigak Sub-Periods III5–6–IV1.

a ‘helmand civilisation’ south of the hindu kush

Figure 4.52  Said Qala, intrusive or special function pottery: (1) Faiz Mohammad red or black grey ware; (2) Kechi Beg or Gumla (?) style bichrome ware.

233

234

THE ARCHAEOLOGY OF AFGHANISTAN

Lithic Artefacts The lithics at Said Qala are not very different from those identified at Mundigak except perhaps in the more limited nature of the types of artefact found. Milling and hand stones similar in all respects to those at Mundigak were found throughout the sequence. Likewise, large cortex flakes with one or more retouched edges and demonstrating extensive use as knives or scrapers were found in all occupation periods. It is interesting to note, however, that these lithic artefacts begin to co-vary with similarly retouched and utilised pottery sherds in the last occupations (Periods IIID–IV). Actual flint artefacts were noticeable by their absence. Only a few waste flakes were found in the lower sondage levels and seven blade fragments from the habitation areas (Periods II–IV). Except for one trapezoidal example they were all triangular in cross-section. All specimens were fragments of whole blades, so it is impossible to determine the presence of microblades. The complete absence of any flint points represents an interesting contrast to the situation recorded at Mundigak. Other interesting lithic artefacts found throughout the Said Qala sequence included large trapezoidal-triangular hoes (Fig. 4.53), which were manufactured from extremely large cortex flakes of basaltic material. Extensive retouching could be observed on all edges, but mostly along the base (butt area), and the upper sides

Figure 4.53  Said Qala, examples of stone hoes.

a ‘helmand civilisation’ south of the hindu kush

were concave, to facilitate hafting. Often the basal edge was well-polished through use. Although infrequent, these artefacts were recorded from all prehistoric periods at the site. Another interesting artefact located only in the habitation areas was a large ‘cleaver’-type object made from basaltic rocks. These infrequent artefacts were bifacially retouched to a blunt cutting–crushing edge and were not hafted. Finally, there occurred at Said Qala in the habitation areas several examples (29) of small hammer stones (never greater than 10 cm in any dimension) made from lumps of natural iron ore. Bone Artefacts Bone artefacts were limited in variety and located in a rather constant frequency throughout Periods I–IV. As at Mundigak, the most numerous types of bone artefacts were simple point and awl fragments, with complete examples being rather rare. A few other types of bone artefacts were noted under various small miscellaneous artefact categories. Metal Artefacts Bronze artefacts are confined to the latest prehistoric occupations (end of Period II– IV) and were located predominantly in the habitation areas. Identifiable functional artefacts included: sickles (Fig. 4.54, nos 1–2), blade fragments(?), a lanceolate or lozenge tanged point and a point or punch with circular cross-section. ‘Luxury’ items in the form of pins were found in Periods II–IV. Although most examples were fragmentary the following styles could be identified: double volute head (Fig. 4.54, no. 5); flattened with forked end (Fig. 4.54, no. 6); and a simple forked end with twisted haft (Fig. 4.54, no. 4). A single example of a bronze handle(?) fragment with a rounded and perforated distal end was found (Fig. 4.54, no. 3). In general, the artefacts are very similar to those found in Mundigak Period III. Small Miscellaneous Artefacts Spindle whorls Truncated cone-shaped ceramic spindle-whorls with central perforation were found throughout the prehistoric periods. Rarely, these had an incised or painted motif. Perhaps somewhat more frequent were simple disc-shaped whorls made from pottery or stone (steatite, and more rarely alabaster). Stone vessels Several fragments of straight-angular walled bowls made from alabaster were found in the final occupation of Period II through to Period IV. Only two variants

235

236

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.54  Said Qala, metal objects: (1–2) bronze sickle blades; (3) bronze handle; (4–7) bronze pins.

were found: one was a small shallow oval bowl made from white limestone, and the other a rectangular vessel leg(?) with square cross-section also made from limestone. Beads and pendants Only a limited number of beads were found, and these were recovered in the habitation areas (Periods II–IV). The only shapes identified were: simple-disc; cylindrical; biconical; and circular. Materials of manufacture included: steatite, alabaster, marble, amber(?), carnelian, rock crystal, lapis lazuli (two examples) and ceramic. Pendants were very rare, though two rectangular pendants were found (alabaster and bone), and one oval bead with serrated edges made from turquoise.

a ‘helmand civilisation’ south of the hindu kush

Seals Several examples of compartmented geometric seals were found, all within the habitation area. Only three examples were not made from steatite and these were manufactured from bone and a grey-brown siliceous stone. All had two central perforations and were predominantly rectangular or square. Other shapes identified were: circular, triangular, lozenge and oval. One example had deeply serrated edges. Several examples of worked steatite were found, which must have represented blanks for the manufacture of such seals. Ceramic figurines Bull or cow figurines were found throughout Periods I to IV, with the majority of examples coming from the habitation areas (Fig. 4.55, no. 4). These figurines are similar to the same types found at Mundigak. One fragmentary example is of particular note because of its size. Only the forward half was found, but this fragment was 13 cm long and stood 13 cm high and had faint red painted motifs. In its complete form this figurine was approximately four times larger than the usual bull type. A single example of a possible bird figurine was also found (Fig. 4.55, no. 3). Except for one questionable example from Period II in the sondage, all the anthropomorphic figurines were found in the habitation area. Moreover, out of the twelve examples, all except three (two: Period IV; one: Period III) were associated with the final occupation phase determined for Period II. All examples were fragmentary and confined to the lower torso. Even with such a limited sample it is obvious that the Said Qala figurines are markedly dissimilar to those found at Mundigak. The most common figurine type had the lower torso stylised as a trapezoidal cone which also functioned as base. One example had an incised pubic area. More rarely found were examples of seated figurines with bent legs extended out to the front (Fig. 4.55, no. 1). A single example of a standing figurine with outstretched arms was also found (Fig. 4.55, no. 2). With the possible exception of the standing example, all figurines represented females. These figurines are very similar to those found at Gumla II in northern Baluchistan and Namazga III in Turkmenistan. Faunal and Floral Remains The faunal and floral remains from Said Qala Tepe have never been analysed.90 Deh Morasi Ghundai Deh Morasi Ghundai is located 16 km south-west of Said Qala Tepe.91 It is only about half the size of Said Qala, and the excavations under Dupree in 1951 were

237

238

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.55  Said Qala, terracotta figurines: (1) Female figurines; (2) Standing figurine; (3) Bird figurine; (4) Bovines.

very limited. Besides a sondage (6 × 2 m) three other small test pits were opened, but were incompletely excavated.92 Dupree was able to define four major occupational periods, of which Periods I–III are of major concern here. As with Said Qala it is felt here that all the prehistoric occupations are essentially of a single major period. However, Deh Morasi is later than Said Qala and represents a Mundigak Sub-Period IV1-type occupation.

a ‘helmand civilisation’ south of the hindu kush

Stratigraphy and Architecture The limited nature of the excavations prevented the delineation of any structures, and most stratigraphic observations are based on artefact content and soil depositional levels. Period I was characterised by laminated silt-clay layers of soil which contained a few artefacts but was otherwise devoid of features. The major occupation at Deh Morasi was Period II, which is divided into three phases, IIa–c. The only architectural feature of significance found in the excavations was recorded in Period IIa. This was a small (45 × 28 cm) mud-brick structure (Fig. 4.56a), trapezoidal in shape with the following artefacts in direct association: ceramic female figurine, copper tube and seal, goat bone and horn, utilised magnetite nodule (Fig. 4.56b) and pottery. Dupree interpreted this structure as a ‘household shrine’, as well it might be. The entire feature was surrounded by a prepared clay floor. Deh Morasi Phase IIb was distinguished by a semi-circular mud-brick oven associated with a prepared floor and the only section of mud-brick wall defined at the site. Period III was a unit of unstratified mound fill, the major features of which were three intrusive burials. The final Period was a series of highly disturbed deposits containing Early Islamic glazed pottery. Chronology There are six radiocarbon dates from Deh Morasi Ghundai, two from Phase IIb and four from Phase IIc (Tables 3.1, 4.1, 4.2).93 These determinations have been recalibrated as follows: Period IIb (P-1493: 4414 ± 53 bp) 3330–2910 cal. bc and (P-2292: 5680 ± 300, 5295–3960 cal. bc); Period IIc (P-2288: 3780 ± 233 bp) 2880–1660 cal. bc (P-2289: 4440 ± 252 bp) 3760–2470 cal. bc (P-2290: 4090 ± 214 bp) 3335–2040 cal. bc and (P-2291: 4500 ± 68 bp) 3370–2935 cal. bp. As noted above, the Period IIa dates, which should correspond to the Mundigak Period IV, are not consistent with each other, and while one is statistically similar to Casal’s Sub-Period III5 date the other is earlier than any of the other dates from Mundigak. As with the dates from Said Qala Tepe, the Period IIb dates from Deh Morasi Ghundai highlight the problems with the extant radiocarbon chronology for the late prehistoric periods in southern Afghanistan. With the exception of P-2288, which appears to be somewhat later than the other dates, the Period IIc determinations are statistically similar, though three of the dates have an uncertainty in excess of 200 years, making them imprecise. Ceramics Period I contained only the crude handmade chaff tempered pottery referred to as Said Qala Coarse, and is found at both Mundigak and Said Qala throughout most occupations. Otherwise the pottery is comparable with that identified in Mundigak IV1. With the exception of the Said Qala Coarse ware, it is all

239

240

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.56  Deh Morasi Ghundai: (a) shrine complex, Period Ila; (b) used magnetite nodule, Period IIa; (c) female figurines, Period IIa.

­ heel-made and has a buff-red paste with sand- or self-temper. Decorated motifs w are almost entirely confined to geometries, which are very similar in their overall style to those of Mundigak Sub-Periods III5-6–IV1. The motifs are painted in black and executed either directly on the red surface or on top of a thin buff-white wash. However, a more definite association with Mundigak Sub-Period IV1 is based upon the predominance of bowl and beaker vessel forms. These vessel forms had their

a ‘helmand civilisation’ south of the hindu kush

highest relative frequency during Mundigak Sub-Period IV1. Equally convincing of a Mundigak Sub-Period IV1 association is the low relative frequency of Quetta Ware, which is confined to the ‘linear style’ of decoration and a few examples of the curvilinear or floral (bucranium) style. Comparable Quetta Ware was not found earlier than Sub-Period IV1 at Mundigak. In addition, a zoomorphic motif was found on Quetta Ware in Deh Morasi IIb. Such motifs do not appear before Mundigak Sub-Period IV1 and were completely absent at Said Qala Tepe. Other Artefacts Lithic artefacts were few in number and found only in Periods I–II. Milling stones and pestles were found, as were a few examples of stone hoes. Also recorded were examples of stone celts and retouched flakes. In addition, there was the single utilised magnetite nodule found in the ‘shrine complex’, which is comparable to the iron ore hammer stones found at Said Qala. Only a very few bone artefacts were found in Periods IIb–c. These were the simple punches or awls encountered in the other sequences, as well as a similar artefact which Dupree refers to as a scraper. All the metal artefacts were copper and confined to Periods IIa–c and IV.94 Associated with the ‘shrine complex’ in IIa were two fragments of a hollow tube and a handle fragment. Several fragments of simple pins were located in Phases IIb–c, and a single fragment of a compartmented seal was found in IV. Miscellaneous Artefacts Spindle whorls Only disc-shaped spindle whorls were found, manufactured from ceramics and steatite. Stone Vessels. Two fragments of simple angular-walled alabaster bowls were found in Period II. Beads–pendants. An elongated bone pendant and a disc bone bead were found in Period II. Seals. Three fragments of compartmented seals with geometric motifs were found. Two from Period II were made from steatite, and one from Period IV was made from copper. Ceramic figurines Zoomorphic figurines were confined to a leg fragment in Phase IIb and a bird figurine in Phase IIe. The female figurine found in association with the ‘shrine complex’ of Phase IIa was a classic example of the ‘Zhob’-style figurine found in a Period IV context at Mundigak. Another possible ‘Zhob’ figurine was found in Period IV and a fragment of a seated figurine was located in Phase IIc.

241

242

THE ARCHAEOLOGY OF AFGHANISTAN

Faunal and Floral Remains The faunal remains from Deh Morasi have never received adequate analysis, although Dupree reported sheep/goat and large bovid bones from Phase IIa, and there is no reason not to assume that they were not domesticated. More important are the plant remains.95 A brick from the ‘shrine complex’ of Phase IIa was found to contain seed remains of a fodder grass (Aegilops tauschii syn. A. squarrosa) related to wheat, and domesticated six-row barley (Hordeum vulgare var. afghana). It would thus seem that the occupants of Deh Morasi Ghundai were exploiting both domesticated plants and animals from throughout the occupation, which as with Mundigak (and probably Said Qala Tepe), is not surprising given the developments in the surrounding regions in the preceding periods (see above and Chapter 3). Kandahar The site of Shahr-i Kohna or Old Kandahar was the focus of several seasons of excavations in the 1970s,96 and the excavations of the first millennium bc and later occupation levels will be discussed in the next chapter. For the moment it is worth remarking that what appears to be a decorated Mundigak Period IV bowl was found as a residual object in a later (Achaemenid) context (Fig. 4.57).97 This find likely confirms the existence of a Bronze Age settlement at Kandahar, which

Figure 4.57  Mundigak IV bowls from Kandahar (top) and Mundigak.

a ‘helmand civilisation’ south of the hindu kush

was first suspected from the study made by Gardin and Lyonnet of the pottery collected by Casal from the site in the 1950s.98 Given the small quantities of material recovered, it is unlikely that this was a major settlement, and it may not have been as large as other Bronze Age settlements in the region, such as Deh Morasi Ghundai, though the remains may also have been completely buried by later occupation. A Bronze Age date for an occupation phase at Kandahar therefore does not alter the overall regional picture for the eastern part of Helmand, which is that the main centre was at Mundigak. However, given the subsequent decline of Mundigak and rise of Kandahar, the inference that the one replaced the other as the main regional centre seems plausible. A number of other Bronze Age sites around Kandahar have also been recorded (Fig. 4.1). In addition to Said Qala Tepe and Deh Morasi Ghundai, a large 40-metrehigh mound about 80 to 100 metres in diameter was recorded at Bagh-i Pul Ghundai on the left bank of the Arghandab, fourteen kilometres west of Kandahar, which produced solely Bronze Age material.99 Further down on the right bank is the large urban site of Spirwan surrounded by ramparts and with a citadel, where Casal collected Bronze Age pottery. However, Indo-Parthian and Timurid pottery were also recorded, so the ramparts and citadel may be later.100 Bronze Age (as well as later) material was also collected by Casal from three small mounds at Mundi Hisar on the right bank of the Tarnak River, sixteen kilometres south-east of Kandahar.101 Fourteen kilometres north-east of Kandahar, near the main road to Kabul, Bronze Age and later material was recorded at the large mound of Bad-i Sah Ghundai,102 and a similar corpus was also recorded at the major site of Shahr-i Safa by the road some one hundred kilometres further north.103 However, the pottery collected from these sites was only subject to preliminary analysis, so there is no more precise dating apart from a broad ‘Bronze Age’ designation. Nad-i Ali Excavations at the site of Nad-i Ali (Sorkh Dagh) by Roman Ghirshman in the 1930s104 revealed a massive construction in Period II which was probably built against the natural mound in order to make available a larger area (Fig. 4.58). The outer walls were of mud-bricks (35 × 35 × 9 cm), while the inner structure was made up of alternating unbaked and baked bricks (57 × 28 × 9 cm), resting on a basement consisting of ten layers of baked bricks. Altogether this structure covers an area of approximately 200 × 50 metres and rises nearly forty metres above the plain. Dales carried out additional excavations at Sorkh Dagh in the 1970s, further confirming the massive size of the platform.105 It was initially ascribed to the Iron Age by Ghirshman by comparison with the massive monument at Tepe Sialk in western Iran. However, even by the time Dales was writing in the 1970s the Bronze Age material of Central Asia was not generally familiar to Western scholarship. In a subsequent study by Besenval and Francfort,106 a jar published by Ghirshman that sealed the top of the platform was recognised as belonging to

243

244

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.58  Nad-i Ali, Sorkh Dagh mound.

a Bactrian Bronze Age type. This, as well as a reinterpretation of the brick sizes, indicated that this monument should be dated over a thousand years earlier, c. 2300–1700 bc.107 Besenval and Francfort’s re-examination of the pottery recorded on the site by both Fairservis and Dales further confirmed a substantial Bronze Age presence there, drawing comparisons with the pottery from Shahdad and Khurab in south-east Iran. The massive structure at Nad-i Ali, which must have been at least forty metres high and some fifty metres at its base, can be compared to similar massive Bronze Age platforms at Namazga V, Altyn Depe, Hisar III and, especially, Mundigak Period IV. However, Besenval and Francfort emphasised that it is substantially different from and larger than these platforms, having more in common with the ziggurats of Mesopotamia. If so, the platform at Nad-i Ali extends the spread of this building type by some 1,300 kilometres further east of Chogha Zanbil, which was previously the easternmost-known ziggurat. Since this interpretation was published, the later excavation and possible identification of the monumental platform/building at Konar Sandal in south-eastern Iran have supported the supposition,108 though it may be unhelpful to make such a direct typological connection to the ziggurats of Mesopotamia and Iran. Whatever this building might have been, the existence of such a major monument could not have been in isolation, and was likely built as a part of an important urban centre, which may well have been a critical node on the communication routes that linked south-eastern Iran and Sistan with Central Asia and the Indus region.

a ‘helmand civilisation’ south of the hindu kush

Helmand-Sistan Surveys Apart from Nad-i Ali, and the excavations in the Kandahar region, the prehistoric remains identified in the greater Sistan Basin in south-eastern Afghanistan are known primarily from surface collections. Due to the nature of the resultant information the discussion for this region must centre almost exclusively upon ceramic data. Therefore, the previous format for presenting the data will be dropped in favour of a more generalised discussion. British Boundary Commission reports in the late nineteenth and early twentieth centuries first brought attention to the huge archaeological potential of Sistan.109 Aurel Stein visited the region in 1916110 and was the first to locate prehistoric remains there (including Sharh-i Sokhta), albeit on the Iranian side of the border. Subsequent investigations on the Afghan side were carried out by Ghirshman and Hackin in 1936,111 and their work was followed in the 1950s, ’60s and ’70s by Fairservis,112 Hammond,113 Dales,114 Fischer (focusing mainly on Islamic remains)115 and Trousdale.116 Fairservis was able to identify several prehistoric sites located primarily in the ‘Southern Delta’ associated with the Rud-i Biyaban River. Most of these sites appear to represent small villages although some are of considerable size, with signs of copper smelting and pottery manufacturing spread over a considerable area around Gardan Reg.117 On the basis of surface collections only, Fairservis found a great material uniformity persisting between various sites,118 which led him to speculate that the prehistoric occupation of Sistan was rather short. Of the many ceramic types defined by Fairservis for this area, two are of particular importance here. The first, and the most characteristic pottery, is that of ‘Gardan Reg Decorated’.119 Basically this a buff-red ware with black to reddish-brown motifs applied directly to the surface, or more rarely to a buff slip. In vessel form and motif style this ware is similar to that found at Mundigak Sub-Periods III6–IVl, Said Qala II–IV and Deh Morasi Ghundai. The second type was that of ‘Emir Gray’.120 This finely made grey ware with black geometric designs is not found at the sites to the east. The geometric motifs are not much different from that of Gardan Reg Decorated and would appear to be a local development of this more broadly distributed buff-red ware. Also identified on these sites were examples of Faiz Mohammad Painted Wares and Quetta Wet Wares similar to those from Mundigak and northern Baluchistan. North of Sistan near the city of Farah a single site was found with a similar black-on-buff type of pottery at Tepe Barangtut.121 Besides the ceramics, large quantities of cuprous slag122 were found on the surface of prehistoric sites, which suggest that copper was locally available and exploited. Fairservis also located a series of graves at his Site 109, G.R.6 (G.R. for Gardan Reg), which by their ceramics and associated metal artefacts show affiliation with Mundigak and Shahr-i Sokhta. The last two seasons of Dales’ work in Sistan were spent surveying the southern areas of Afghan Sistan (covering many of the same sites that Fairservis visited

245

246

THE ARCHAEOLOGY OF AFGHANISTAN

earlier). From the south-western corner of Afghanistan, on the Shela Rud, Dales identified some decorated wares and basket-impressed pottery which almost certainly indicates the presence of prehistoric remains in this remote region. Of particular interest were many alabaster objects, including columns, discs and weights, similar to Tepe Hisår IIIc material in Iran, scattered around the site of Gudar-i Shah. These might have been brought there by nomads from the nearby site of Dam, where basketware, grey ware and painted sherds were collected.123 The survey by Hammond in north-eastern Sistan along the Helmand Rud at first located a single site with possible prehistoric occupation near Darweshan.124 Subsequently, new studies by Besenval and Francfort reinterpreted Hammond’s Type 10 ‘cord marked’ ware as being Harappan.125 This ware is described as pink fabric with small inclusions, decorated by ‘parallel impressions of a twisted string of cord, evenly spaced or paired’.126 Although only five sherds of this group were recovered, it has led to the identification of three more Bronze Age sites in his survey, Zindan, Qal’a-i Hindu and Qal’a-i Sultan, as well as the originally identified Akram Qal’a.127 Trousdale subsequently collected Bronze Age pottery from another of Hammond’s surveyed sites in the Darweshan area, Khwaja Hasan,128 and also recorded a number of new Bronze Age sites further down the Helmand in Sistan, in addition to re-examining those recorded by Fairservis and Dales.129 External Relationships with Southern Afghanistan130 Having described the prehistoric developments in southern Afghanistan it is now necessary to delineate what other cultures, and areas, were exerting an influence on or were being influenced by these developments. Because of the nature of archaeological data such relationships are difficult to establish with any high degree of certainty, and are made doubly difficult due to the lack of reliable absolute dates from the sites discussed above. Generally, such external relationships are determined by similarities in ceramic industries between geographically disparate sites, or the occasional intrusive sherd which indicates that some form of interaction was taking place. More rarely, such relationships can be established by the identification of objects manufactured from a material which had a limited source of origin such as lapis lazuli or marine sea shells. Identification of such objects at sites geographically distant from the source areas clearly indicates that some form of direct or indirect interaction between the two areas was taking place. Both types of relationships will be discussed here. The cultural and chronological boundaries for these external relationships will be based on the protracted Mundigak sequence.131 No discussion of the Sistan Basin is complete without some mention of the extremely important prehistoric site of Shahr-i Sokhta.132 This huge urban site has produced a material culture (Periods II–III) which is so remarkably similar to that found at Mundigak Sub-Periods III5–6–IV1, Said Qala I–IV, Deh Morasi I–IV, and in the Quetta Valley that this material has been identified as representing a ‘Helmand

a ‘helmand civilisation’ south of the hindu kush

Civilisation’.133 Ceramically, the vessel forms, motifs and style of execution are nearly identical among the above-mentioned sites. The same similarity can be found in metal artefacts, lithic artefacts, ceramic figurines, compartmented seals and architectural features. The similarity is so pronounced as to indicate that extensive interaction and communication was maintained between these sites. The most extensive remains at Shahr-i Sokhta come from Periods II–III. Besides the strikingly similar examples of Quetta Ware and the more common black-onbuff/red pottery, smaller quantities of a black-on-grey pottery similar to Emir Gray from Afghan Sistan were also recovered. During Period III the black-on-buff pottery motifs become so highly standardised as to be monotonous by their repetition. Also during Period III the black-on-grey pottery increases in quantity. Besides pottery, Shahr-i Sokhta II–III yielded many other non-perishable and perishable items, such as: objects of wood; basketry; textiles; and large quantities of alabaster, carnelian, chalcedony, lapis lazuli and turquoise objects. Not only were finished artefacts of semi-precious stones found (e.g. beads), but also waste flakes and tools of manufacture, suggesting that the site was a manufacturing centre for such objects, though the nature of the trade in these objects continues to be debated.134 Alabaster was available locally, as were probably chalcedony and steatite; carnelian has a variety of sources which might have been local, as could have been the jasper drill-heads for working such material.135 Turquoise is also known from several localities in Iran. The lapis lazuli came from Badakhshan – close to the site of Shortughaï (see Chapter 3) – and this was the major source for this blue stone throughout the prehistoric Middle East,136 with significant quantities being imported into Mesopotamia during the Jemdet Nasr period. The Jemdet Nasr context is especially interesting since clay cylinder sealings of this period were found in Period I at Shahr-i Sokhta. It is also notable that a Proto-Elamite tablet was recovered from Period I deposits,137 which suggests specific types of connectivity within greater Iran rather than Mesopotamia directly. It has previously been suggested that Sharh-i Sokhta acted as a major centre of manufacture and distribution for lapis lazuli to the western markets,138 though this suggestion has now been contested.139 Clear examples of a more local product exchange system can also be defined at Shahr-i Sokhta in Periods II–III. Excavations at Rud-i Biyaban, 20–30 km south, revealed what appeared to be a potters’ village with numerous Period II–III ceramics and 50 kilns. A similar specialised site explanation has been proposed for the concentrations of copper slag at sites in Afghan Sistan.140 There is now a substantial number of radiocarbon dates from Shahr-i Sokhta, which attest that the Period II and III deposits were occupied between c. 2750 and 2200 bc.141 These dates correspond to the two radiocarbon determinations from Rud-i Biyaban 2.142 These dates largely conform to the suggestion that there was a chronological correspondence between Shahr-i Sokhta Periods II and III and Mundigak Sub-Periods III6 and VI1–3 (see above).143 The final occupation phase at Shahr-i Sokhta, Period IV, was dominated by the construction of a large building of over 650 sq. m. During this period painted

247

248

THE ARCHAEOLOGY OF AFGHANISTAN

pottery almost completely disappeared. Vessel shapes are distinguished by sharp carinations, probably due to the use of a potter’s wheel. The paste is a buff-red colour and surfaces are often slipped or burnished – techniques absent in previous periods. Among the limited examples of decorated pottery some parallels may be seen with the pottery of late Mundigak Sub-Periods IV2–3. Mundigak: Periods I–II These two periods should be combined, because: (1) the limited data concerning external relationships are similar for both periods; (2) the chronology and culture for both periods represent a continuum; (3) although the distinction between Periods I and II was useful in understanding the developmental sequence at Mundigak, it is doubtful that a similarly useful distinction can be defined for external relationships; and (4) the external relationships definable for Periods I–II provide a significant contrast with those definable in subsequent periods. Combining the evidence from these two periods suggests that significant external relationships existed with the populations of the Quetta valley in Northern Baluchistan. Traditional comparative studies have correlated the Mundigak Period I–II material with that of Kili Gul Mohammad (hereafter KGM) periods II–IV (Fig. 4.59).144 Shaffer originally argued that the correlation should actually be made between Mundigak I–II and only KGM III–IV,145 with the elimination of KGM II being based primarily upon the identification of a metal dagger blade in KGM III, but also on the occurrences of alabaster bowl fragments and ceramic characteristics. KGM II is also comparable to Mehrgarh Period II, which is characterised by the predominant use of chaff-tempered soft-ware ceramics.146 KGM III–IV, which corresponds to periods III and IV at Mehrgarh,147 is distinguished by a black-on-red slipped pottery with a preponderance of the same triangular motif recorded in Mundigak Sub-Periods I3–5. This pottery made an initial appearance in KGM II, reached its maximum frequency in KGM III and persisted into KGM IV. KGM III–IV are also characterised by an overwhelming number of handmade ceramics, a situation somewhat analogous to Mundigak I–II, though Mehrgarh III–IV see the introduction of the tournette for surface smoothing.148 This Mundigak/KGM correlation is, however, complicated by Fairservis’s emphasis upon distinguishing aspects of the ceramics from KGM IV and their equation with the Damb Sadaat (hereafter DS) I material (Fig. 4.60).149 Fairservis based his argument for a KGM IV–DS I correlation on the persistence of similar pottery types at both sites, but there are also indications that DS I is distinct from KGM IV. It has been argued, for instance, that the frequency of similar types is so low in DS I as to argue that the DS I assemblage represents a distinct and perhaps significantly later cultural development,150 and there are significant additions of various new types during DS I which also reinforce its distinctiveness from KGM IV. Furthermore, KGM IV is distinguished by the presence of a polychrome, or bichrome, type of pottery. This fine-turned buff pottery with well-executed geometric motifs in black and red, called Kechi Beg

a ‘helmand civilisation’ south of the hindu kush

Figure 4.59  Characteristic artefact styles of the Quetta valley: (1) Kile Gul Mohammad I and II; (2) Kile Gul Mohammad III; (3) Damb Sadaat I.

Polychrome,151 is an extremely striking and important pottery type. Several of the Kechi Beg-type motifs are similar to the ones identified on the bichrome sherds located in Mundigak I5. Likewise, there is a basic similarity between this material and the bichrome pottery identified at Amri IA–C in the lower Indus valley.152 The

249

250

THE ARCHAEOLOGY OF AFGHANISTAN

­ ecorative style of Kechi Beg is similar to that of Amri, but is at the same time disd tinctive in its own right. The possibilities cannot be ruled out that the bichromes found in Mundigak I5 might actually be Kechi Beg rather than Amri types, or that Kechi Beg itself reflects interaction and/or communication with the much more prevalent bichrome ceramic styles found in the Indus valley (Amri and Sothi). There is an overall striking dissimilarity to be found among the majority of ceramics described for Mundigak Periods I–II and KGM III–IV when contrasted with that recorded for these sites in the Indus region. Although the exact significance of these parallels remains to be determined, they seem to indicate that some interaction was present between southern Afghanistan and the Indus region during the mid-fourth millennium bc.153 Ceramic evidence, then, indicates that cultural interaction/ communication was taking place over a distance of approximately 800+ km to the south-east of Mundigak, or approximately the same distance as is indicated by the presence of lapis lazuli found at Mundigak I which must have originated in northeastern Afghanistan (a single lapis lazuli bead is reported from Amri I). Existence of such interaction is further supported by additional ceramic parallels noted for other, less completely known, sites in Baluchistan.154 West of Afghanistan, in eastern Iran, the nearest site which is chronologically comparable to Mundigak Periods I–II is Tepe Yahya.155 However, the fourth millennium bc occupations at this site (Periods VI and V) seem to have little in common with the Afghan material. True, a basic similarity may be distilled among the simple geometric motifs found on the pottery; but it would not be a

Figure 4.60  Characteristic artefacts at Damb Sadaat II: (1–5) pottery; (6, 8–9) figurines; (7) stamp seal; (10) clay rattle seal; (11) clay house; (12) bronze knife; (13) bone spatula; (14) clay ladle; (15) alabaster vessel.

a ‘helmand civilisation’ south of the hindu kush

convincing correlation, and is probably the result of the basic stylistic limitations imposed by a geometric repertoire (i.e. how many ways a triangle may be drawn). The only connection with Afghanistan is the presence of lapis lazuli.156 There has been a general trend to equate the Turkmenistan cultural complexes of Namazga II with Mundigak I–II.157 However, the position here is that these fifth/early fourth millennium bc metal-using cultural groups of Turkmenistan present more of a contrast than a similarity with Mundigak. A basic similarity may be defined among several categories of artefacts (i.e. lithics, architecture, beads, etc.). Much of this similarity, however, might be attributed to analogous ecological-social-economic responses to the contingencies imposed upon a sedentary agricultural way of life under broadly similar environmental circumstances. On the other hand, the respective ceramic stylistic traditions are, if taken in their entirety, significantly different. Certainly, the ceramics do not reflect the same degree of similarity demonstrated by the more parallel styles found in Baluchistan. The importance of these differences cannot be appreciated without reference to the very pronounced ceramic similarities delineated for Mundigak III and Namazga III. Mundigak: Period III The external relationships of Mundigak III, and by analogy Said Qala I–IV, are characterised by an intensification of interaction/communication with Baluchistan and the Indus valley. Unlike with the preceding periods, external relationships can now be defined with some of the eastern Turkmenistan cultural developments and the development of intensive contacts with north-east Central Asia (e.g. Taluqan sites and Sarazm; see Chapter 3). Most of these external relationships are again based upon ceramic correlations and in particular the identification of Baluchistan and Indus valley pottery types at Mundigak and Said Qala rather than vice versa. In contrast, the relationship with Turkmenistan is based upon similarities in decorative style discernible on indigenously manufactured ceramics. At Mundigak and/or Said Qala the following distinctive Baluchistan pottery types were identified: Kili Gul Mohammad black-on-red slipped; Faiz Mohammad Painted wares; Kechi Beg Polychrome or Bichromes; Quetta Wet wares; Nal blackon-buff and Polychromes. Except for the Nal types all these various types are also present in the Quetta valley of northern Baluchistan. In addition, there is a very pronounced similarity among all the vessel forms and decorative motifs found on Quetta valley pottery designated as DS I by Fairservis. As at Mundigak, it is during DS I that Quetta Ware-type pottery begins to appear in frequency. Indeed, Mundigak III, Said Qala I–IV and DS I are so similar that they could be considered a single cultural complex. Parallels and correlations can also be made for other areas of Baluchistan,158 but none is as dramatically similar as the Quetta valley material given the available information. Two radiocarbon dates from DS I159 have been recalibrated as follows: 3935–2040 cal bc (L-180B: 4348+-350) and 3330–2710 cal bc (UW-59: 4330+-70 bp); the latter of these is statistically similar

251

252

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.61  Namazga pottery from Geoksyur.

to Casal’s Mundigak Sub-Period III5 date, which provides some reinforcement of the correlation. Artefacts indicating interaction with the Indus region proper appear to be somewhat more infrequent than those that attest to interaction with Baluchistan.

a ‘helmand civilisation’ south of the hindu kush

Found at Mundigak and DS I (but not Said Qala) were examples of a globular jar with short neck and everted rim decorated with simple horizontal bands that are similar to Kot Dijian-type pottery found in the early occupations at sites such as Kot Diji,160 Harappa,161 Amri,162 Gumla II–III,163 Jalilpur II164 and Sarai Khola II,165 as well as at sites in the Gomal plain and Bannu Basin, which lie along the piedmont of the Suleiman Ranges.166 Wet wares have also been found in the Indus region in the early levels at Mohenjo-daro,167 and Harappa,168 and also at Gumla II,169 Sheri Khan Tarakai, Jhandi Babar A and other sites in the Bannu and Gomal regions,170 and Jalilpur II.171 Although it has been argued that wet-ware is a marker of cultural connection,172 it must be noted that this type of surface treatment was extremely widespread and long-lasting, and is unlikely to represent an integrated cultural tradition.173 In addition, several of the bi- or polychrome wares identified at Mundigak and Said Qala are very similar to those of Amri I. However, many of the black-and-red on white polychromes identified at Said Qala are more directly comparable to the pottery of Gumla II–III than Amri I. There are, however, a range of polychrome wares used throughout the Indo-Iranian borderlands region, and the Gumla material is similar to that seen at Rehman Dheri and is now regarded as belonging to a localised polychrome tradition that characterises the late fourth and early third millennium bc occupation in the Gomal and Bannu regions, and has been referred to as Tochi-Gomal ware.174 The female figurines identified at Said Qala are strikingly more similar to those found in Gumla II than those of Mundigak III. Similar figurines were found also at Jalilpur II. At Jalilpur several examples of lapis lazuli beads were found, and a single example was identified at Sarai Khola II.175 Although the evidence remains fragmentary, it is reasonable to conclude that some degree of interaction/communication was maintained between southern Afghanistan and the Indus region during Mundigak Period III. West of Afghanistan, the only directly related material is to be found at Shahr-i Sokhta I in Iranian Sistan (see above).176 Unfortunately, this period has had only limited testing. Even so the ceramics are similar to the Quetta Ware recorded at Mundigak Sub-Periods III5–6, Said Qala I–IV and DS I. Moreover, the pottery demonstrates the pronounced similarity with Namazga III material in Turkmenistan that is also noted for Mundigak (Figs 4.61–4.63). Shahr-i Sokhta I has been dated

Figure 4.62  Comparison of decorated pottery of Namazga III (1–6) and Mundigak III (7–12).

253

254

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 4.63  Comparison of decorated pottery, seals and figurines of Damb Sadaat II (1–7 and 15–21) and Namazga III (8–14 and 22–28).

to the end of the fourth and the start of the third millennium bc and is therefore contemporary with DS I.177 Limited amounts of a painted grey ware comparable to that of Faiz Mohammad Painted in the Quetta valley were also found in Shahr-i Sokhta I.178 South of Shahr-i Sokhta, in Iranian Baluchistan, is the site of Bampur,179 and the small amount of material recovered from the earliest occupations (Bampur I–II) indicates some parallels with Mundigak Period III material. However, the sample is small and the parallels general. Further west at Tepe Yahya the ceramics continue to indicate a rather unrelated cultural tradition. The few limited parallels which are definable are in Periods IVC2–IVB6. A small quantity of pottery from this period had geometric designs in black on a red, buff or grey surface similar to Bampur II–III, and these Western Baluchistan Ceramics are only otherwise seen in Kech-Makran and at Shahr-i Sokhta I,180 making them contemporaneous with Mundigak III. Better parallels may be seen among a decorated grey ware which is similar to that of Faiz Mohammad Painted pottery in Baluchistan. This grey ware becomes increasingly more frequent at Tepe Yahya in Period IVB.181 Another interesting item found in IVC was a straight-sided bichrome pot resembling those found in central Baluchistan and belonging to the Nal stylistic tradition, and dating to Sohr Damb Period II (c. 3200/3100–2700 bc).182 It should also be mentioned that a sherd resembling Amri ID type of pottery was found on the surface of Tepe Yahya, and should broadly have a similar date to the Nal-ware vessel.183 Period IVC at Yahya can be dated to the end of the fourth millennium bc on the basis of associated artefacts (Proto-Elamite tablets, cylinder seals, Jemdat Nasr polychrome, and bevelled rim bowls).184 Although there is some indication of limited interaction between

a ‘helmand civilisation’ south of the hindu kush

Tepe Yahya and with the southern Afghanistan cultural assemblages, it does not appear to have been extensive. Tepe Yahya, while important, thus seems to have been beyond the western fringe of Mundigak-related cultures.185 In contrast to the previous periods, some very significant similar developments can be delineated between southern Turkmenistan cultures grouped under the heading of Namazga III and Mundigak III. Indeed, accounting for these similarities has long been a focal point for explanatory models encompassing this vast area.186 Previously, it was thought that Namazga III represented a distinct break with the cultural traditions found in Namazga I–II. However, it is now clear that there was continuity of development with the preceding periods alongside phases of innovation which may have been influenced by outside contact.187 On the basis of ceramics, a division is usually made into a western group, represented by the Kara-depe material, and an eastern group, the Geoksyurian sites. The western Turkmenistan sites are characterised by a brown-on-buff or red pottery and more rarely a red slipped or polychrome decorated pottery. The major aspect of these ceramics was utilisation of zoomorphic motifs which were absent from ceramics from the eastern sites. Zoomorphic motifs were, however, rare and could occur alone or in combination with geometric motifs. Actually, the geometric style of decoration is not dissimilar to that of Quetta Ware, but the presence of zoomorphic motifs has resulted in comparisons with the Hissar and Sialk sequences of eastern and western Iran. The red-slipped pottery and polychromes were limited to only geometric decorations. Geometric compartmented seals appear for the first time in the Turkmenistan sequence, and there is a significant increase in the use of alabaster for bowls, beads, figurines and other artefacts. Female figurines were rare but provide an important comparison with the Afghan ones. They are seated with the head having ‘pinched’ features including large eyes and noses. Some figurines were found without heads, arms or breasts: those with heads often had elaborate coiffures. Only rarely were male figurines found. Eastern Turkmenistan, or Geoksyur, sites are usually represented as distinct from western sites and more directly comparable to the material in Afghanistan. Their major characteristic is a buff pottery with red slip decorated with black and red polychrome geometric motifs (Figs 4.61–4.63). Zoomorphic motifs are not unknown but are extremely rare and highly stylised. An interesting contrast to Mundigak ceramics is that only during Namazga III are the first indications of turned pottery identified, a marked contrast to the high frequency of wheel-made pottery in Afghanistan. Female figurines are found in great frequency (300 at Geoksyur I) and are highly stylised (similar figurines occur in Namazga II). They are all seated with elongated heads, large noses and long necks. The figurines from both eastern and western sites are very similar to those found at Said Qala and Gumla II–III. Interestingly, this Geoksyur pottery is found in significant quantities at Sarazm (Tajikistan) far from its original location. At the same site, and together with it, were also found in increasing quantities (especially numerous in periods III and IV) ceramics undeniably correlated with southern Hindu Kush, comparable

255

256

THE ARCHAEOLOGY OF AFGHANISTAN

with Mundigak III5–6 to IV1–3 and with other Baluchistan pottery styles. In a lesser quantity, pots related with north-eastern Iran have also been found there.188 In north-eastern Afghanistan, in the Taluqan valley, sites with the same southern Hindu Kush pottery have been discovered during a survey (see Chapter 3). All this testifies of an intensification of long-distance relations where some sites, like Mundigak, played a key role. Many of these sites were situated within areas rich in semi-precious stone or metal ore (e.g. Sarazm, and Shortughaï and the Taluqan sites close to the lapis lazuli mountain), making the search for metal and this valued stone the possible targets of these networks. Mundigak: Period IV The continuity and overlap defined between late Period III5–6 and IV at Mundigak and Said Qala make it difficult to determine two distinct and separate sets of external relationships for these two periods. Therefore, many of the observations regarding external relationships for Period III can also be applied to Period IV, and by analogy with Deh Morasi I–IV. Here, as in previous periods, parallel developments appear to have been taking place in the Quetta valley and other parts of Baluchistan. Mundigak IV is best compared to Mehrgah VI and VII189 and DS II–III in the Quetta sequence (Figs 4.60, 4.64, 4.65), although DS I also displays many characteristics of this period. There are clear similarities in figurine types,190 but perhaps the most convincing correlation is provided by the introduction of zoomorphic motifs in DS II–III, which are similar to those at Mundigak. As at

Figure 4.64  Pottery of Damb Sadaat II.

Figure 4.65  ‘Zhob mother goddess’ figurines: (1) Damb Sadaat III; (2) Deh Morasi Ghundai Ha; (3) Sur Jangal III (?); (4) Mundigak IV1? Periano Ghundai; (6) Dabar Kot.

a ‘helmand civilisation’ south of the hindu kush

Mundigak, DS III,191 Quetta Ware demonstrated a marked decline in the solid geometric elements and an increased usage of simple horizontal lines and bucranium motifs. Overall, the ceramics of DS II–III are very similar to those of Mundigak Period IV, and especially Sub-Period IV1, in vessel forms, styles of decoration and variety of ceramics located. It should be noted that the zoomorphic motifs at both Mundigak and DS II–III are also similar to the so-called ‘Kulli’ pottery found in southern Baluchistan, particularly the sites of Kulli, Mehi and Nindowari.192 The most recent overview of the ceramics of Kech Makran, Baluchistan and the surrounding regions in this period has been published by Aurore Didier.193 Another important parallel with Mundigak Sub-Period IV1 is the presence of ‘monumental’ architecture. At Damb Sadaat, as at Mundigak, the previous areas of occupation were levelled for the construction of a large brick platform. Associated with the platform were rough limestone block stone drains and a bench against the southern wall. There is some evidence that spur walls (3 m thick) connected the platform with lower portions of the mound. The main wall of this platform was resting upon a small hollow constructed from stones and containing a human skull minus the lower jaw. In the immediate vicinity of this structure were located eight female figurines, among which some were in the ‘Zhob’ style. Found for the first time in DS III levels were cattle figurines, one of which had a ‘yoni’ motif painted on its forehead. Other artefacts included: model houses (also found in Mundigak IV1 and Said Qala III); clay rattles; small metal artefacts; alabaster bowls; compartmented seals; and beads (lapis lazuli, carnelian and turquoise). DS III appears to be the last major period of prehistoric occupation in the Quetta valley, but its chronological placement is not constrained well by the single radiocarbon date (UW-60: 4030+-160 bp, 3010–2090 cal. bc). The existence of interaction/communication between southern Afghanistan and the Indus region during the early–mid third millennium bc is, however, very limited.194 Indeed, most of the evidence comes from Mundigak itself and the identification of infrequent examples of Kot Dijian and Amrian-type ceramics. There thus appear to have been differing zones and dynamics of interaction between the two regions during the formative stages of Indus urbanism. Period II and III at Shahr-i Sokhta clearly represent a major urban centre for the Mundigak IV-type culture. However, Period IV at Shahr-i Sokhta cannot be as convincingly correlated with developments at Mundigak, although limited motif parallels can be found among later Mundigak Sub-Periods IV2–3 pottery.195 Likewise, the Bampur sequence196 of south-eastern Iran has numerous parallels with Mundigak Sub-Periods IV1–2 and Shahr-i Sokhta II–III, and these parallels have been summarised by de Cardi.197 Many of the ceramic characteristics of Shahr-i Sokhta IV can be also paralleled at Bampur V–VI. Of particular importance here is a black-on-grey canister jar found in Shahr-i Sokhta IV with horizontal friezes of stylised animals and geometric motifs, which is strikingly similar to vessels found in Bampur VI and sites on the Oman Peninsula in the Persian Gulf.198

257

258

THE ARCHAEOLOGY OF AFGHANISTAN

Shahr-i Sokhta and Bampur also provide some parallels with late Period IVB at Tepe Yahya. Although many of the ceramics found at Yahya IVB and Shahr-i Sokhta are mutually exclusive, an important link between the two sites is provided by the black-on-grey ware just mentioned. These black-on-grey wares are not generally found west of Yahya, but have, on the other hand, a very wide distribution to the east incorporating Baluchistan and south into the Persian Gulf.199 This pottery provides a basic, but tenuous, correlation between Yahya IVB, Shahr-i Sokhta III–IV and the southern Afghanistan sites. Also found in Yahya IVB is a great proliferation of incised steatite vessels in various stages of manufacture. Similar steatite vessels have been found in Mesopotamia during Early Dynastic II–IIIa times, and although Lamberg-Karlovsky initially suggested that Tepe Yahya was a centre for the production of these steatite vessels,200 it is now more likely that more substantial centres for the production of these vessels lay in the Halil Rud basin at sites like Konal Sandal.201 There is considerable scope for further work into understanding the various scales of interaction and exchange that were in operation during the third millennium bc, and the sites in southern Afghanistan will have a key role to play in the discussions. In Turkmenistan, there must have been considerable interaction/communication between the late Namazga III/Geoksyur Period cultural developments and those represented by Mundigak IV, and the scope of this interaction stretched as far north as Sarazm. At the end of Namazga III, the sites of the Tedzen Delta in eastern Turkmenistan were abandoned and, with this, the close cultural similarity between these areas ceases. Somewhat comparable material can be identified in Namazga IV sites in central Turkmenistan.202 Both large and small sites have been found during this period, several of which are resting directly on virgin soil, which is a factor that has caused speculation that they were settled by people from the Tedzen area. Ceramically this period is marked by a black-on-red slipped pottery decorated with geometric motifs. Motifs and some vessel forms are similar to those found at Mundigak IV2–3, but the degree of similarity is not nearly as pronounced as in Namazga III. A light coloured slip is also found at some sites, suggesting continuity with the previous ceramics. A partially burnished grey pottery with incised decorations and sharply carinated vessel forms also occurs in small quantities, particularly in the western sites. However, no painted grey wares have been found. Although the close similarity in material culture which had marked Namazga III and Mundigak III–IV1 disappears in the Namazga IV period, there are i­ ndications that at least limited interactions were maintained. The black-on-red slipped pottery bears some resemblance to that of Mundigak IV2–3, as do the figurines and compartmented seals. However, it is difficult to evaluate this similarity given the chronological problems and lack of data from northern Afghanistan (see Chapter 3).

a ‘helmand civilisation’ south of the hindu kush

Mundigak: Period V The period of transition between the Bronze and Iron ages in the greater region is particularly significant, but understanding is limited. The plains of Central Asia to the north of the Hindu Kush following the decline of the Oxus Civilisation are marked by a proliferation of regionally distinct painted pottery traditions, such as those seen at Yaz Depe and Tillya Tepe.203 The relationship between these developments and those happening to the south of the Hindu Kush is not entirely clear, and it is notable that there is nothing directly comparable to Mundigak V anywhere in Afghanistan or the immediately surrounding areas. This dynamic is complicated by the fact that besides Mundigak no other prehistoric site in southern Afghanistan which has been excavated appears to have been occupied after Period IV. Casal referred to some parallels with the Chust culture of the Fergana valley of Uzbekistan, which are now dated to the Yaz I period, from c. 1400 bc,204 which is notable as the pottery of Mundigak V appears to have more relationships with that of Mundigak VI, which is associated with the use of iron technology, than it does with Mundigak IV. Johanna Lhuillier has noted, however, that the differences are greater than the similarities, suggesting that the Mundigak IV material is likely to have been a regional manifestation of a broader phenomenon. There is only limited evidence for similar painted wares in the regions surrounding southern Afghanistan. For instance, excavations at Pirak between 1968 and 1974, again led by Casal, revealed evidence for this period of transition on the Kacchi plain in Baluchistan, and produced evidence for the proliferation of hand-made ceramic vessels decorated with monochrome and bichrome geometric motifs.205 This material is best paralleled with Tillya Tepe and other sites in Margiana and the Kopet Dagh piedmont.206 There are also clear similarities between the Yaz I ceramics and vessels recovered from the sites of Akra and Ter Kala Dheri, in Khyber Pakhunkwa, Pakistan.207 The significance that these linkages between populations living in various regions on the southern and northern side of the Hindu Kush have for discussions of connections, interaction and possibly even migration will need to be a topic for future research.

259

CHAPTER 5

The Iron Age, Achaemenid and Hellenistic Periods Warwick Ball, Simon Glenn, Bertille Lyonnet, David W. Mac Dowall and Maurizio Taddei Historical Background Although the Iron Age is not technically ‘historic’ (at least in Afghanistan), it is considered here together with the Achaemenid rather than with the later Prehistoric of the previous chapter. This is because the Iron Age generally marks a distinct break from the Bronze Age but a material continuity into the Achaemenid period. Indeed, in terms of settlement, there are few sites or remains that continue from the Bronze Age into the Iron Age and historic periods, but many of the great historic urban centres of Afghanistan and elsewhere in Central and South Asia have their roots in the Iron Age. The invasion of Alexander III of Macedon, however, does mark a distinct change in Central Asia, both historically and materially, so can be considered separately. The East Bactrian survey has shown that this is not as consistent a pattern as was once thought, particularly in view of the continuity of the canal systems recorded there, which had their origin in the Bronze Age and remained more or less in constant use into the Hellenistic. It is appropriate, therefore, that we begin our discussion of the archaeology with this, but first it is necessary to consider what state structures, if any, the Achaemenids inherited when they entered the region. The Question of Iron Age States Ever since the nineteenth century there has been speculation about the existence of a pre-Achaemenid kingdom of Bactria.1 This rests largely on the account in the Avesta of Vishtaspa, the supposed king of Bactria and patron of Zoroaster. Classical accounts also describe how Cyrus’s conquest of Bactria met with stiff resistance, implying the existence of an organised state, and Bactria – as well as perhaps Arachosia – seems to have been of high importance to the Achaemenids: Bactria alone supplied a cavalry force of 30,000 to the Achaemenid army, it paid 300 talents annually into the treasury, the highest amount of any single province in the empire, and it offered the stiffest resistance to Alexander’s invasion. However,

the iron age, achaemenid and hellenistic periods

Vishtaspa’s ‘kingdom’ and the homeland of Zoroaster may not have been Bactria, and interpretations have swung between various Central Asian localities and Sistan.2 The Avesta furthermore conveys an impression of a fairly archaic society that belies a developed state system. Dandamayev rightly expressed the opinion, therefore, that ‘the question of the existence of an ancient Bactrian kingdom remains open’.3 Archaeology, however, has lent weight to the suggestion. The evidence reviewed in the last chapter suggests a highly developed civilisation in Central Asia in the Bronze Age – the ‘Oxus Civilisation’ – that raises the possibility of states developing in the Iron Age. More particularly, the East Bactria surveys in Afghanistan (also discussed in Chapter 3 and further below) have recorded a highly sophisticated system of irrigation canals dating from the Iron Age and earlier that can only have been built by large-scale co-operation and centralised organisation, implying a state system. This system is characterised by marked continuity into the Achaemenid and Hellenistic periods. Excavations at Kandahar (also discussed below) have confirmed that the first construction of the massive urban fortifications and perhaps the citadel date from the pre-Achaemenid period, probably on a preconceived plan, earlier than any other fortified city in Early Historic South Asia, also suggesting more than just a city-state or trade emporium in Arachosia. A pre-Achaemenid fortified settlement was excavated at Tillya Tepe in Bactria, a site better-known for its first century ad tombs (discussed below). The early settlement of Tillya Tepe is a fortress with circular towers; the material collected in a stratified sequence is both of Yaz-I type (c. 1400–900 bc) and of Yaz-II (preAchaemenid) and Yaz-III (Achaemenid). It compares with the architecture and material of Kuchuk-Tepa (Uzbekistan) and Yaz Depe (Turkmenistan).4 Mundigak V might be related to this early period at Tillya Tepe. Whatever existed in either Bactria or Arachosia, it is clear that when the Achaemenid Persians conquered them the elements of urbanism and sophisticated state structures were already in place. The Achaemenid Empire5 Cyrus the Great (559–529 bc), the founder of the Achaemenid Persian empire,6 after defeating Croesus of Lydia and conquering Asia Minor, gained control of all the Afghan plateau in a series of campaigns to the north and east of Iran. We see the extent of the Achaemenid empire in the three lists of the satrapies of Darius I (521–486 bc): from the Bisitun inscription c. 516 bc, from Darius’ palace at Persepolis, and on his tomb nearby at Naqsh-i Rustam. The eastern territories of the Achaemenids included modern Afghanistan (Fig. 5.1) and were organised into the satrapies of: a. Haraiva (Greek Areia), modern Herat region, which was the important centre of eastern Iran. It had probably been, as Frye suggests,7 the main separation

261

262

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.1  Map of the eastern Achaemenid Empire showing satrapies.

point of the Aryans who migrated to India from those who moved to the west, and its name was subsequently applied to the southern group of eastern Iranian territories as far as the Indus. b. Baktrish (Greek Bactria), the fertile country of northern Afghanistan and southern Uzbekistan centred on the Amu Darya with its capital at Bactra, commanding the routes from Merv and Herat in the west to Sogdiana in the north, China in the east and across the passes of the Hindu Kush mountains to the Kabul valley and India in the south. c. Zranka (Greek Drangiana), the steppe and semi-desert country of the lower Helmand River and Hamun lake – modern Sistan (i.e. Sakastene), which derives its name from the Saka tribes who were settled there after their invasion in the second century bc. d. Harahuvatish (Greek Arachosia), to the east of Drangiana – the valley of the upper Helmand, modern Kandahar and the centre of the Achaemenids, ruling over the tribes as far as the Indus in the east and the sea to the south. e. Satagush (Greek Sattagydia or Paropamisus), the mountainous area of the southern Hindu Kush including the Kabul, Bamiyan and Panjshir valleys later known as Paropamisus, the land ‘above the eagle’.

the iron age, achaemenid and hellenistic periods

f. Gandhara, the area of modern Jalalabad and the Khyber Pakhtunkhwa Province of Pakistan. To the north of Afghanistan, the Achaemenid empire included Margush (Margiana) and Suguda (Sogdiana), both of which had close links to Bactria; later in his reign Darius I added the Panjab and Sind to constitute the satrapy of Indush. The whole of the empire was divided into provinces governed by satraps. Its economy was soundly based, trade was encouraged, and an effective road system was established. Aramaic was the language used for official business, and it is from the Aramaic script that the local Kharoshthi script was subsequently developed in Afghanistan and the Panjab.8 Bactria and Arachosia in particular were closely integrated into the Achaemenid Empire. Alexander III of Macedon and his Successors It was the defeat of the last Achaemenid king, Darius III, at the battle of Gaugamela on the plains of Mesopotamia in 331 bc that made Alexander III of Macedon the new ruler of the Achaemenid Empire, including its satrapies in Afghanistan. After capturing the Persian capitals in the west, seizing their treasures and burning Persepolis, Alexander marched into Afghanistan and crossed the Hindu Kush mountains to conquer the satrapies of Bactria and Sogdiana (330–327 bc).9 To secure his lines of communication he supposedly established permanent posts at Alexandria in Areia (Herat), Alexandria in Arachosia (Kandahar) and Alexandria sub Caucasum (Begram or a site nearby in the north of the Kohistan of Kabul), although the Alexandrian foundation of many of the so-called ‘Alexandrias’ is now doubted.10 Returning to Begram he marched east to conquer Swat and the Panjab (327–325 bc). On the banks of the Beas, his army refused to advance further. Alexander built a fleet, sailed down the Indus to the sea and returned to Persia in 324 after a dangerous land march through Baluchistan. He left behind him satraps and governors, but his administration did not long outlast his early death in 323. Although his campaigns and short rule have left no direct traces in Afghanistan,11 his conquests had far-reaching consequences and mark a watershed in the history of western Asia. In the north the old Achaemenid satrapies of Parthia and Bactria became provinces of the Hellenistic Seleucid kings; while the vacuum created by Alexander’s withdrawal from the Panjab enabled Chandragupta, the new Mauryan King of Pataliputra (Patna), to extend his kingdom to north-west India and eastern Afghanistan (Fig. 5.2).12 After Alexander’s death, Seleucus I Nicator emerged from the war of succession as King of Syria and most of western Asia; but when he tried to recover the territories of south-east Afghanistan and the Indus, he was forced to make peace, acknowledge the sovereignty of Chandragupta Maurya and cede Gandhara, Arachosia and Paropamisus in return for 500 elephants and a matrimonial alliance (c. 304 bc). These satrapies remained under Mauryan rule during the third century

263

264

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.2  Map of Hellenistic Afghanistan and adjacent regions, with the route of Alexander.

bc in the reigns of Bindusara (298–273) and Ashoka (273–232), until the Mauryan empire began to break up with disputes over the royal succession and its great provinces established their independence. The Graeco-Bactrians13 Bactria had been an important base for Alexander’s campaign against Sogdiana, and presumably retained an important military role under the Seleucids with a strong garrison. Perhaps around 250 bc (the exact date of Graeco-Bactrian independence is debated, with high and low chronologies) the northern satrapies of the Seleucid empire, Sogdiana, Parthia and Bactria, rebelled and became independent kingdoms – Parthia under Arsaces, Bactria under Diodotus and Sogdiana without any named ruler.14 The next certain date comes from the Greek historian Polybius, who tells us that Antiochus III, the Seleucid king, tried to take punitive action against Bactria in 208 bc and besieged King Euthydemus I in Bactria; but in the end he was obliged to withdraw and formally recognised Bactrian independence. Before returning to the west, Antiochus crossed the Hindu Kush, and renewed his ancestral friendship with the ruling king, Subhagasena – a reference to the alliance that

the iron age, achaemenid and hellenistic periods

Seleucus Nicator had made a century previously with Chandragupta Maurya. But the visit was hardly friendly. Antiochus revictualled his army at Subhagasena’s expense, robbed him of all his elephants and imposed an indemnity. After this episode Graeco-Bactrian history becomes much less clear, with very few mentions of historical events, or even the names of kings, in literary sources. Coins become the best evidence for the history of the period, but great caution must be exercised in attempting to reconstruct history solely on the basis of the numismatic evidence. Often much less is known than earlier historical works suggest. The coins in isolation can only provide a small part of the historical picture, while guesswork, however inspired, should not be allowed to fill in the rest. To take just one example, the reconstruction of civil wars between Graeco-Bactrian kings of different reconstructed dynasties, and a Seleucid mission of reconquest under Eucratides I, must be disregarded.15 What is clear from the coins and the fragmentary literary references is the appearance of Indo-Greek rulers south of the Hindu Kush, perhaps around 180 bc. Subsequently, the Graeco-Bactrian kingdom lost most of its western territories to the Parthians. We see several series of more local Greek rulers, until their kingdoms are progressively overthrown by the invasions of the Yuezhi and Sakas from the north and the Pahlavas from the west. The Parthians The other major group to take advantage of Seleucid weakness in the east was the Parthians, who broke away from the Seleucid kingdom at about the same time as the Graeco-Bactrians did. In about 238 bc they seized control of the trans-­Caspian territories of the Seleucid Empire under the founder of their dynasty, Arsaces. Ruling first from Nisa (near Ashkhabad), by the middle of the second century the Parthians invaded Seleucid Iran, culminating in Mithradates I being crowned king in Seleucia in 141 bc. The Parthians soon re-established most of the old Persian empire, expanding into western Afghanistan around Herat and into Sistan in the south. In their expansion into the east they came into occasional conflict with the Graeco-Bactrian kingdom, but both the chronology and the boundary between the two kingdoms is unclear. Although the Parthians never expanded into Bactria, the effect upon the Greeks of Bactria was enormous: it cut them off from the Hellenistic kingdoms of the west. Sistan became a major Parthian power base where Parthian princes eventually outlasted the collapse of the main dynasty. The Parthian princes of Sistan expanded into Arachosia and Gandhara, where one of the descendants, Gondophares, founded the Indo-Parthian kingdom in about ad 20 after the collapse of the Saka kingdom (see below). Yuezhi and Saka Invaders In the mid-second century bc the group of tribes known in Chinese sources as the Yuezhi had migrated westward from the borders of China into the Ili region

265

266

THE ARCHAEOLOGY OF AFGHANISTAN

in Central Asia, after being defeated by the Xiongnu, a neighbouring nomadic confederation who were probably forerunners of the Huns. Their arrival on the Ili seems to have dislodged several Saka groups, some of whom moved southwards into Bactria and others south-westward to Herat and Sistan, which bears their name (Sakastan/Sistan) and eventually to the southern Indus. The region hence became known in Hellenistic sources as ‘Indo-Scythia’ and in Indian sources as Saka-dvipa. Classical sources refer to these migrating groups coming into Bactria as the Tocharoi, Asii, Sacaraucae, Iaitoi and others. The Yuezhi have been identified with the Tocharoi (or Tocharians) and the Asii have been identified by some with the Sakas of the Azes dynasty of Gandhara. The Yuezhi themselves occupied northern Bactria between 145 and 138 bc, then moved south across the Oxus c. 120–100 bc, occupying the Graeco-Bactrian territory north of the Hindu Kush mountains. By about 70 bc they had entered the Kabul region, occupying much of Gandhara by c. 55 bc, having founded the Kushan kingdom (discussed in the next chapter).16 About 80 bc we find the first Saka king, Maues, ruling at Taxila and controlling the provinces of the Panjab. They are a separate group of Saka from those of the southern Indus, having probably migrated down through the Karakoram passes from Saka lands in Xinjiang leaving large numbers of petroglyphs along the route (although the Sakas of the petroglyphs might have arrived earlier during the Achaemenid period).17 The subsequent Scythian kingdom of the upper Indus and eastern Afghanistan is a shadowy one, known mainly from its coinage and the occasional mention in Indian sources. It did not extend much into Afghanistan, probably no further than the Jalalabad and Gardez regions. It was distinct from the near-contemporary Scythian kingdom of Arachosia and the lower Indus in Sind mentioned above, and the two are often confused with each other. Maues was succeeded by other Sakas who are Kings of Kings – Azes I, Azilises and Azes II, after whom the coinage stops: the kingdom seems to have only lasted into the early first century ad. The end of Hellenistic rule in the Paropamisus is much less clear. The last known king was Hermaeus, but his coinage is extensively copied, and it is not clear who the issuers were. In Arachosia, after the rule of the Scythian kingdom, we find a line of kings with Pahlava names, Spaliris, Spalagadama, Spalahores, the latter owing allegiance to Vonones as King of Kings, who is sometimes identified with Vonones I of Parthia (ad 10–12).18 Meanwhile, the Parthian dynasty of Arsaces had maintained its independence from the Seleucids, and under Mithradates I had become a major power. After the conquest of Media c. 155 bc Mithradates I campaigned in Arachosia and took some border provinces from Eucratides. At this time the Parthians probably ruled Herat and Sistan. The Parthians had serious problems with the Saka migrations, but Mithradates II (123–88 bc) was successful in settling them, receiving allegiance from them and establishing Parthian rule generally in the east. Isidore of Charax,

the iron age, achaemenid and hellenistic periods

who died c. ad 25, describes Arachosia and Kandahar as the easternmost part of the Parthian empire.19 The Indo-Parthians About ad 25 we see the emergence of a major new power in southern and eastern Afghanistan, the Indus valley and the Panjab established by the IndoParthian King Gondophares. This dynasty is quite distinct from the rulers of the Parthian empire, and there is much to commend Herzfeld’s hypothesis that the Parthian Suren in Sistan broke away to establish an independent kingdom. Gondophares (c. ad 32–60) controlled Sistan, Arachosia and the Paropamisus in Afghanistan, as well as extensive territories in the Indus valley that he captured from the Sakas and the Indo-Greek kingdom that he conquered from the successors of Strato II. His nephew Abdagases continued to rule most of this extensive empire, but their successors lost the Paropamisus, Arachosia, Gandhara and the Indus territories to the Kushans around ad 78. A much-attenuated Indo-Parthian Kingdom, represented by Orthagnes, Pacores, Gondophares II and Sanabares I and II, continued to rule in Sistan throughout the second century ad and seems at times to have controlled both Herat and Merv.20 Settlement, Material Culture, Architecture and Art The Iron Age and Achaemenid Periods The regions that are included within the boundaries of modern Afghanistan were involved in the political events of Mesopotamia even before Baktrish/Bactria was incorporated into the Achaemenid empire in the time of Cyrus the Great, along with Zranka/Drangiana (modern Sistan), Haraiva/Areia (the region of Herat) and Harahuvatish/Arachosia (the region of Kandahar), as well as the territories farther east, as far as Gandhara (Fig. 5.1).21 Formerly, archaeological documentation from Afghan territory concerning the Achaemenid period was so scarce that the most useful evidence for that period in Afghanistan was yielded by excavations and finds in neighbouring countries, principally Iran, but also Central Asia and Pakistan. More recent discoveries, however, have considerably enhanced our knowledge of the Achaemenid period for Afghanistan, particularly in Bactria, Arachosia and – most recently – Areia. These areas may be distinguished, each with its own characteristic features. Bactria First, the Bactrian area, which covers parts of Uzbekistan and Tajikistan22 and the northern part of Afghanistan between the Amu Darya and the Hindu Kush.23 We may begin our summary of Bactrian archaeology in Afghanistan for this period

267

268

THE ARCHAEOLOGY OF AFGHANISTAN

with the important evidence from the East Bactria survey directed by Jean-Claude Gardin between 1974 and 1978 (the earlier periods have already been discussed in Chapter 3).24 East Bactria  The period c. 1500–300 bc from the end of the Bronze Age to the Achaemenid (Fig. 3.3), corresponds to what is called Yaz I to III in the rest of the southern part of Central Asia, after the name of the type site Yaz Depe in Turkmenistan.25 The position of Yaz I within the late Bronze Age rather than within the Iron Age is a matter of debate and depends, on the one hand, on the extreme rarity of iron objects discovered up until now and, on the other, on the radical change in the culture in comparison to the previous one.26 The division into the three phases adopted by our Russian colleagues was followed, but it must be emphasised that the length and absolute dating of each of them is purely arbitrary. Overall, the number of sites and irrigation canals continue to increase (Fig. 3.3). This period is still silent as far as written documents are concerned, except for the very few recently discovered Aramaic documents dated to its very end.27 We must otherwise rely upon doubtful Classical accounts (Ctesias and Diodorus) which refer to the conquest of Central Asia at the time of the Assyrians. As for the Persian inscriptions, they are uncontested and mention its conquest under Cyrus and Darius I. But no specific items (coins, seals, metal- or earthenware) related to these invasions/conquests have been found in good stratigraphic contexts, so many uncertainties surround this period. Furthermore, the pottery does not present a wide spectrum of forms, and they are all of long duration. To this is added the absence of radiocarbon dates, so it is not possible to precisely date much of the first millennium bc (as is well-known). Phase I is not well-represented in the East Bactria survey. It certainly follows on immediately from the last phase of the Middle Bronze Age and begins around the middle of the second millennium. But, contrary to the case with the rest of Central Asia, there is no clear evidence here of the typical set of hand-made and painted pottery that characterises the Yaz I period, except for a few sherds with a band of red slip along the rim. Nor have similarities been noticed with the material from contemporaneous Mundigak V. However, as in the rest of Bactria and Margiana, some of the pottery is still ‘wheel-turned’ as if the dark previous episode were already over. Both the plain of Aï Khanoum and the Asqalon terrace seem to have been occupied then, as well as a few other small plains (Fig. 3.3). The most interesting innovations are the first traces of occupation along irrigation canals in the northern part of the Taluqan valley. This will eventually result, in the following phase, in the creation of the Rud-i Shahrawan through which will flow part of the Rud-i Taluqan, whether due to natural development or other forces (such as earthquakes) cannot be determined. Phase II certainly begins considerably before the Achaemenid conquest. Although the number of pottery types that we can safely relate to it is scant, occupation has

the iron age, achaemenid and hellenistic periods

Figure 5.3  Plan of Kohna Qal’a, the Iron Age ‘city’ adjacent to Aï Khanoum.

been identified in almost all the main plains of the survey and highlights the creation of new irrigation canals, except for the plain of Imam Sahib (but see below). This phase appears as one of intensive economic development with a high increase in population. It is at that time that some important circular and fortified cities are built, such as Kohna Qal’a along the Darya-i Panj in the plain of Aï Khanoum (Figs 5.3, 5.4), Qurgan Tepe along the Taluqan River and the Bala Hisar of Qunduz (Fig. 5.5). Phase III certainly concerns the Achaemenid period, although it should not be confined to it as many have assumed: as noted above, we have no precise dates associated with it. Furthermore, its pottery was certainly still in use at the time of the Macedonian conquest and until the arrival of new colonists under the Seleucids.28 As during Phase II, all the areas of the survey were occupied except, for the first time, the plain of Imam Sahib – the only one watered directly from the Darya-i Panj, which presents few traces of occupation. It should be mentioned, however, that evidence from recent research in the Bukhara area suggests the possibility of exceptional floods around the fourth century bc that would have changed the course of the Zeravshan River.29 This suggestion may find confirmation in the fact that, in the Samarkand area, ‘Achaemenid’ layers are deeply buried under several metres of alluvium.30 We cannot, therefore, exclude the possibility

269

270

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.4  The circular ramparts of Iron Age Kohna Qal’a.

that a similar phenomenon had happened on the Darya-i Panj or one of its tributaries, and that earlier occupation in the Imam Sahib plain is hidden under alluvium. We have no explanation for the apparent ‘absence’ of Yaz I typical material culture east of the Qunduz River. In Tajikistan its occurrence is also very rare compared to that known in Uzbekistan, Kyrgyzstan or Turkmenistan.31 The extraordinary development visible in Phase II has to be related to that also observed elsewhere in Bactria, as well as in Margiana and as far as Khwarazm. All over Central Asia, except for Sogdiana which was then under the control of semi-mobile Scythian groups,32 the material culture becomes unusually uniform. Some authors such as D’jakonov33 had considered the possibility of the existence of a palaeo-Bactrian entity, powerful and rich enough to be coveted by the NeoAssyrian Empire as first mentioned by Ctesias. No traces of the Assyrians have yet been found except at Ulug-Depe, in Turkmenistan, where the architecture of the citadel is similar to that of several Median Zagros forts34 and where sealings and bullae discovered in a good stratigraphic context are compared to seals of the last Neo-Assyrian king dated to 613 bc.35 A cylinder seal discovered by chance near Samarkand and another near Herat could also date to this period,36 and other finds said to come from plundered graves in northern Afghanistan have been compared with Neo-Assyrian glyptics.37 But these discoveries are altogether too scant to testify to the development of an administration in southern Central Asia before the Achaemenids, be it related to the Assyrians and Medes or not. The image given

the iron age, achaemenid and hellenistic periods

Figure 5.5  Satellite image of Qunduz Bala Hisar.

for this period is rather one of a rich and irrigated rural land under the control of circular and fortified cities. This picture, however, has to be moderated by the fact that the burial practices of that time, with exposure and no graves, prevent the discovery of funerary gifts and necessarily obliterate the eventual differences in the social status of the people. The same comments can be made for Phase III, although it certainly includes the period of Achaemenid control over this area. Apart from the Achaemenid royal inscriptions, the recent discoveries of written documents in Aramaic script mentioning the satrap of Bactria and its governor leave no doubt about Achaemenid domination. However, objects of original Persian manufacture remain extremely rare in Central Asia: the notable exception is the Oxus Treasure, although the exact provenance, composition and date of deposition are still a matter of debate. As has been long emphasised,38 no Persian influence is visible in the material culture. Only very rare pottery finds present an Iranian shape, but none has been discovered in a well-dated context. The arrival of Alexander of Macedon and his brutal conquest of Central Asia would lead to major changes with the foundation of colonies, the introduction of a monetary system, the replacement of the local material culture by that of Hellenistic models, and so forth. During the third century bc the irrigation

271

272

THE ARCHAEOLOGY OF AFGHANISTAN

systems were probably maintained as they were under the Achaemenids, but the picture changes when Aï Khanoum becomes the capital of Eucratides I in the second quarter of the second century bc. It is probably to this very last period of Greek rule that the construction of many new canals in order to irrigate even small areas all around the city should be dated, the maximum ever reached in the whole history of the region. Elsewhere in Bactria  Many remains of the Achaemenid period were located by the work of the Afghan–Soviet Archaeological Mission, mainly in the Dashli oasis north-west of Balkh.39 Among them is Dilbarjin, ‘a town of which the fortifications and a citadel are clearly recognizable’ (although most of the town belongs to the Kushan and Sasanian periods).40 Other important Achaemenid centres in this area located and excavated by the Mission include Altin 1, Altin 10 (discussed further below), the circular city of Altin Dilyar and a massive circular building, possibly a temple, at Kutlug Tepe.41 The researches of the Afghan–Soviet Mission were among the first to throw light on the Achaemenid period in Afghanistan, a period in which the contacts between Bactria and the empire of the Great King also involved a Greek component.42 These links between the Greek world and Bactria were later to be much strengthened, as we shall see. More interesting from the point of view of monumental architecture is the Bactrian site of Altin-10, excavated by the Afghan–Soviet Archaeological Mission; two buildings there appear to be of the greatest importance and show links with the architecture of Dahan-i Ghulaman (for which see below) – a rectangular ‘summer palace’ (No. I), 80 × 55 m, divided into two ‘palaces’ with a fourteen-pillar portico on either side (Fig. 5.6), and a square building (No. II), 36 × 36 m, with rooms

Figure 5.6  Plan of Altin 10, Palace 1.

the iron age, achaemenid and hellenistic periods

Figure 5.7  Plan of Altin 10, Palace 2.

arranged round a central courtyard from which they are separated by a corridor running on three sides (Fig. 5.7).43 The Afghan–Soviet surveys in the Balkh region were complemented by the French surveys. First were some limited surveys by Philippe Gouin north-east of Balkh, now supplemented by more recent investigations in the Balkh oasis which identified an Achaemenid period irrigation network flowing northwards from Balkh, that included a 4 m wide aqueduct stretching for several kilometres north of Altin Dilyar Tepe.44 Most significant were the detailed East Bactria surveys (described above), which recorded a high degree of settlement in the Achaemenid period and before in a fairly smooth sequence from the Bronze Age through to the Hellenistic. The most important feature recorded was a sophisticated canal-fed irrigation system comprising eighteen ‘irrigation zones’ dating from the Bronze Age but maintained to full operational capacity through the Achaemenid period, suggesting a high level of administrative control. This is amplified by the high number of fortified sites recorded. Many of these were substantial settlements with impressive ramparts that can be described as urban, or at least proto-urban. Notable are Kohna Qal’a, a huge circular settlement with inner and outer ramparts 900 m across on (and cut by) the Amu Darya just upstream from Aï Khanoum (which itself was the site of an Achaemenid garrison, although this might date from the time of Alexander)45 and probably forming its urban precursor (Figs 5.3 and 5.4), and Yangi Qal’a further upstream near Khwaja Hafiz, consisting of a lower and an upper city on a promontory overlooking the Rud-i Jilga where it joins the Amu Darya. ‘Achaemenid’ remains (in fact Yaz III period) were also excavated by the Aï Khanoum team at the settlement of Zulm nearby.46 Clearly, Achaemenid (and probably pre-Achaemenid) Bactria was urban, prosperous and highly organised.47 A particular urban feature of Achaemenid and pre-Achaemenid Bactria is that many settlements were planned circular towns. These included Altin Dilyar (Fig. 5.8), the Balkh Bala Hisar, Chim Qurghan, Deh Nahr-i Jadid, Qutlug Tepe, Qunduz (Fig. 5.5), Qurghan Tepe and Kohna Qal’a (5.3, 5.4).48 The circular town is ­ presumably an indigenous Central Asian innovation, deriving from earlier

273

274

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.8  Plan of Altin Dilyar Tepe.

prototypes such as Dashli 3 (Fig. 3.28), and marks a major development in the history of town planning broadly. New Hellenistic settlements in Bactria such as Emshi Tepe and Jiga Tepe (Figs 5.19, 5.20) adopted the form, and recent surveys of the Balkh plain based upon satellite imagery by a team from the University of Chicago have picked up many more such circular settlement plans not previously recorded.49 The circular form became a feature of later Iranian town planning such as the Sasanian circular cities of Firuzabad and Darabgerd, and satellite imagery has revealed circular town plans underneath the later orthogonal plans of Sasanian Bishapur and Jundi Shapur as well (although they are yet to be tested archaeologically).50 The form continued into the Islamic period with al-Mansur’s round city of Baghdad.

the iron age, achaemenid and hellenistic periods

Figure 5.9  General view of the site of Chashma-i Shafa.

The excavations at Tillya Tepe (discussed further below) yielded a local pottery dated to the Achaemenid period and earlier and show connections also with Nad-i Ali II.51 Pre-Achaemenid and Achaemenid ceramics have long been known at Balkh,52 but following recent researches there by the Mission Archéologique Française de Bactriane Afghane (with the support of the DAFA) led by the late Roland Besenval, a more substantial presence is now recognised.53 The most significant discoveries for the Achaemenid period in Bactria have been the results of the French Mission’s excavations at Chashma-i Shafa, also led by Besenval, to the south of Balkh. Here a substantial city of the Achaemenid and pre-Achaemenid period has been recorded, possibly the site of ancient Zariaspa, whose remains include a monumental fire temple that included a large limestone fire altar in the form of an inverted step-pyramid 2.10 m in height and with a top surface of 2.65 × 1.60. A hole in the centre is believed to have been for the sacred fire. This is one of the oldest fire temples recorded and has implications for the development of Zoroastrianism (Figs 5.9 and 5.10).54 We cannot consider the celebrated Treasure of the Oxus (discussed further below),55 now in the British Museum, as an Afghan find, since its find spot, although unproven, is most probably Takht-i Kuwad on the north bank of the Amu Darya in Tajikistan.56 The Treasure of the Oxus is to be viewed in the context of the trade or religious activity that crossed Bactria.

275

276

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.10  The excavated fire temple and altar at Chashma-i Shafa.

The Helmand-Sistan region A second area in which Achaemenid remains are exceptionally important is ancient Drangiana (Zranka of the Achaemenids), corresponding to modern Sistan; here too, in spite of the various surveys carried out in the course of several campaigns by French, US and German missions on the Afghan side of Sistan,57 most of the monuments known so far have been discovered in Iranian territory. Here an Italian Archaeological Mission has brought to light an important town (near Qala-i Nau), built near one of the branches of the Helmand river delta, with a precise and carefully planned layout. The ruins of this town are known as Dahan-i Ghulaman. Although there is evidence of restoration and rebuilding, this town was short-lived and probably died when it was abandoned by its inhabitants as a consequence of some natural event that modified the environment, maybe a change in the river bed. Exploration and excavation have revealed the existence of private houses and seven large buildings, three of them being certainly public in character, all dated by the excavators from the sixth to the fifth centuries bc.58 The large buildings probably had civil, administrative, religious and military functions. The buildings were made of mud-bricks and pakhsa, or pressed earth, and slightly carinatevaulted structures were obtained by means of two opposed rows of curved mudbrick struts. This is an important feature that seems to contrast with the principles of Achaemenid architecture to such a degree that the presence of arches has often

the iron age, achaemenid and hellenistic periods

been used as a criterion to post-date buildings which could otherwise have been placed in the Achaemenid period.59 The plans of the seven larger buildings are also noteworthy: they have a large central court, sometimes diversified by a portico, and show the adaptation to local requirements of architecture found also in the central regions of the Achaemenid empire. The territory of Afghan Sistan certainly depended on this administrative town (maybe the Zarin recorded by Ctesias), the only great provincial centre of the Achaemenids known to us. Material dating to the Iron Age to Achaemenid periods in Afghan Sistan was recorded by a French Mission led by Joseph Hackin in 1936 and two US Missions led by Walter A. Fairservis Jr in 1949–50 and subsequently by George Dales in 1968. The former executed some small-scale excavations at Nad-i Ali (Surkh Dagh). Analogies were recognised between some of the pottery from the later period, Nad-i Ali I, and the Achaemenid pottery from Dahan-i Ghulaman. The earlier period with its massive mud-brick platform, Nad-i Ali II, was initially dated to the Iron Age, and Dales’ subsequent sondages in the mound confirmed this sequence.60 However, while an Iron Age date at Nad-i Ali is still presumably present, the pottery associated with the massive mud-brick platform of Period II has since been re-dated to the Bronze Age61 and so is discussed in the previous chapter. Nevertheless, it is worth noting that such massive brick or pakhsa platforms are characteristic of Iron Age and Achaemenid architecture. A number of such platforms have been revealed elsewhere in Sistan,62 at the excavation of Site D at Kandahar63 and more recently the Iron Age platform in the deep sounding in the citadel at Herat.64 Other, more monumental, examples have been recorded at Tureng Tepe in north-eastern Iran, Tepe Nush-i Jan in western Iran and the citadel platform at Kandahar.65 In 1975, William Trousdale carried out excavations at the mound of Lat Qal’a in Sistan on behalf of the Smithsonian. It is a steep-sided mound with a late fort on top, and the only clearly multi-period site found by the Smithsonian survey in the Helmand Valley, with a relatively continuous occupation through Ghaznavid, Sasanian, Parthian, Hellenistic, Achaemenid and early first millennium bc (Fig. 5.11). Numerous sherds in spoil suggest the site extends back to at least the second millennium bc, and likely earlier.66 Iron Age and Achaemenid pottery was also collected by a British Mission in 1966 both in Sistan and further upstream in the Helmand valley, although apparently in extremely small quantity,67 as well as by surveys in the same area by the Smithsonian project between 1970 and 1974. Another US survey by Marc Abramiuk in 2011 revealed more information about this area.68 More evidence of Achaemenid remains has been recorded at the massive citadel mound of Qal’a-i Bust. Although this has never been excavated, surface collections of the pottery and the accidental discovery of a stone weight with an inscription in Old Persian confirm a major Achaemenid presence at Bust. This has also led to speculation about possible Zoroastrian associations of the site, possibly identified with the location of ancient Bigis in the sources.69

277

278

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.11  Painted Iron Age ceramics from Qala 350A in the Sar-o-Tar basin of Sistan.

The Herat region The largely Timurid citadel of Herat rests on a massive artificial mound which has long been believed to be the site of the pre-Islamic satrapal capital of Areia. The name from which Herat derives first enters recorded history as Haraiva in the Achaemenid royal inscriptions and archives, the name of the Achaemenid satrapy, becoming Areia in the Greek sources, with the satrapal capital known as Artacoana. This name reflects its huge importance in Iranian historical tradition: the name derives from the Avestan aryá meaning ‘pure’ or ‘truth’, the same root from which ‘Aryan’, ‘Iran’ and ultimately ‘Eire/Ireland’ derive,70 as well as ‘Herat’ and ‘Harirud’ – a city, a region, a river and a people. Recent excavations in the citadel by the German Archaeological Institute led by Ute Franke, as well as excavations by the DAFA at the mound of Kuhandaj in the old city led by Roland Besenval and Philippe Marquis, have revealed evidence of Iron Age settlement at Herat for the first time, confirmed by a series of radiocarbon dates ranging from the ninth to the fifth century bc. The deep sounding in the citadel revealed the edge of a pakhsa platform which, as we have observed, is characteristic of Iron Age remains elsewhere such as in Sistan and at Kandahar. Previously, the only archaeological evidence for pre-Islamic Herat was some Sasanian gems and an

the iron age, achaemenid and hellenistic periods

Old Babylonian cylinder seal picked up in or near Herat in the mid-nineteenth century.71 These investigations therefore might well have located the ‘long-lost’ Achaemenid satrapal capital.72 Arachosia Excavations at Kandahar by the British Institute of Afghan Studies directed successively by David Whitehouse, Anthony McNicoll and Svend Helms between 1974 and 1978 revealed the existence of a major city with continuous occupation from the Iron Age through to late antiquity (Figs 1.33, 5.12–5.17).73 The chronology of the separate periods recognised is based almost solely on the ceramics and (for the later periods) on coins; it is unfortunate that there are no radiocarbon dates (although the excavations were planned to continue after 1978, interrupted by the Soviet invasion). The massive urban fortifications consisting of inner and outer defences in the form of a rectangle with ramparts 14 m wide – a scale hitherto not seen in the region – were probably laid out in the Iron Age. These fortifications have been amply documented by a sectional cut through the ramparts in the north-western part of the city made by Whitehouse (5.14, 5.15), and subsequently by excavations inside the south-western ramparts carried out by Helms,

Figure 5.12  View of Kandahar from the ridge overlooking the site. In the foreground is the stupa; the ramparts are clearly visible in the background; the modern town of Kandahar is in the distance.

279

280

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.13  Map of the site of Kandahar showing excavated areas.

showing an almost continuous programme of rebuilding and strengthening from the Iron Age to late antiquity. Whitehouse initially suggested that the scale of such building might be explained by the regional centre moving from Mundigak to Kandahar, a plausible explanation given the decline of Mundigak in this period. Allchin remarks that it appears older than any other Early Historic city known in South Asia, thus marking the beginning of a tradition of large fortified urban areas that subsequently spread through South Asia during the first millennium bc.74 It might mark the arrival of new people in the region, bringing about the decline of Mundigak. Hiebert and Lamberg-Karlovsky have emphasised the key role of the Indo-Iranian borderland region in the Indo-Aryan migrations.75 The occurrence of the large-scale fortified city of Kandahar – still not accurately dated within the broad time span of the Iron Age – as possibly marking these migrations can neither be substantiated nor ignored. The city was considerably rebuilt and enlarged in the Achaemenid period, when the massive citadel (Fig. 1.33) was also probably built, as well as several inner lines of ramparts dividing the city into several quarters. This substantial programme of rebuilding and enlargement, together with the evidence of the Elamite tablets,76 confirms that Kandahar almost certainly was the satrapal capital of Harahuvatish.

the iron age, achaemenid and hellenistic periods

Figure 5.14  Section through the east rampart at Kandahar. Layer 19 marks the Achaemenid rampart.

Figure 5.15  The cut through the rampart at Kandahar.

An underlying overall urban plan dating from the Achaemenid period that determined the orientation of subsequent building has also been discerned. The only Achaemenid building excavated in its entirety was the massive Site H building, where the small size of the rooms (about 2.50 m square) contrasted with the massive size of the walls, some 1.75 m thick (Fig. 5.16). This suggests a military rather than a domestic function: an arsenal was initially suggested by the ­excavators.

281

282

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.16  Plan of the Site H building at Kandahar.

the iron age, achaemenid and hellenistic periods

The rooms were completely devoid of any occupation, as the entire building was deliberately and carefully back-filled. This might have been to form casemates for the foundations of an upper building, now gone, or perhaps for religious purposes such as the similar back-filling of the temple of Nush-i Jan in Iran.77 The massive fortifications and citadel, and careful urban layout, together with the discovery of Elamite texts at Kandahar (discussed below), leave little doubt that it was indeed the administrative capital of Achaemenid Harahuvatish – indeed, that it was one of the main centres of Persian power in the east. This is further amplified by the pottery for this period, where ‘the identifiable external influences are traced to the Iranian plateau, rather than to the Indian subcontinent’, in the words of David Fleming, who studied the Achaemenid period pottery.78 Particularly characteristic are the carinated (often deeply carinated) flared bowls, as well as the beginning of a tradition of burnishing, at first mainly radial burnishing, but a tradition that was to ‘explode’ into the spiral burnishing of subsequent periods (Fig. 5.17). Fleming further concludes his study by emphasising that ‘Kandahar is the only excavated site between Sistan and the northern Indus region that has yielded evidence of a western Iranian settlement south of the Hindu Kush in the mid-first millennium bc. Its pottery will serve as the basis for all subsequent work in the region, and has provided the first comprehensive collection of ceramics for comparative purposes.’79 The name of the satrapal capital of Harahuvatish/Arachosia survived in the Islamic sources as ar-Rukhkhaj, surviving today as the minor village of Arukh 55 kilometres north-west of Kandahar. Remains there are too minor for it to have been a satrapal capital, but three kilometres to the north is the major site of Mundigak, which does have Achaemenid material in its latest period of occupation.80 The Kabul Valley and Gandhara The Achaemenid period is less documented in the eastern regions of Afghanistan.81 Here we may recall the coins found at Chaman-i Hazuri, Kabul: they are Achaemenid, Greek and local (bent bar) coins, buried at the beginning of the fourth century bc, and evidence of trade in the Kabul valley.82 This is very little indeed, if we compare the much richer documentation on the Achaemenid period obtained from the sites beyond the Khyber Pass, in Pakistan territory: Charsada and Taxila. Charsada was probably the capital of the satrapy of Gandhara, added to the empire either by Cyrus himself or in the first years of the reign of Darius I;83 Taxila, which merged into the empire at approximately the same time and before Darius conquered the Indian satrapy, including eastern Panjab and Sind, threw off Achaemenid domination probably in the time of Artaxerxes II (404–359 bc). Achaemenid Taxila may be identified with the earliest levels of the Bhir Mound excavation,84 dated by Marshall to the sixth to fifth centuries bc. This was a particular aspect of the town, one of its richest quarters: nevertheless the plan is irregular, the building technique poor, in comparison with the upper, post-Achaemenid layers. It is to be

283

284

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.17  Achaemenid ceramics from Site D at Kandahar.

the iron age, achaemenid and hellenistic periods

observed that soak-wells existed, one in almost every house, from the Achaemenid period onwards. This is further amplified by more recent work in the Hathial area adjacent to Bhir at Taxila, revealing a substantial settlement for this period.85 Allchin speculated in 1996 that the (then) unexcavated mound of Akra in the Bannu region might also be an Achaemenid regional centre,86 confirmed since by the investigations of a substantial Achaemenid citadel mound there and identified by the excavators as the possible satrapal capital of the region.87 The Hellenistic Period88 Up until the 1960s, Alexander’s expedition, the Seleucid domination and the consequent formation of the Graeco-Bactrian kingdoms appeared to have left very few, although not meaningless, traces in Afghanistan. The problem of GraecoBactrian art was felt to be urgent because it was evident that only with its solution could the paradox of the ‘Graeco-Buddhist’ art of Gandhara (to be discussed in the next chapter) be placed in an exact historical perspective, a paradox that obviously could not be ignored. If on the one hand it was not possible to date the birth of this art back to the years of the Macedonian conquest, on the other hand no one could really believe that the only antecedents were a few toreutic works (mostly of uncertain date) and the beautiful coins of the Graeco-Bactrian kings. Discoveries since the 1960s have thrown an altogether fresh light on the history of Afghanistan in the period that immediately follows the Seleucid domination. There are now five inscriptions from Kandahar and three from Laghman which are dealt with in another section of this chapter; Hellenistic remains have been excavated at Balkh as well as at other sites in northern Afghanistan, more have been recorded in surveys in the region, and the Hellenistic presence at Kandahar has been amplified by excavations there as well as at other Hellenistic sites recorded in the Helmand region. Most important of all has been the discovery of a Greek town at Aï Khanoum: the excavations were carried out there by the DAFA between 1965 and 1978, initiated by Daniel Schlumberger and directed from then on by Paul Bernard.89 Aï Khanoum, however, will be discussed in a separate section in view of its uniqueness. Before that, we will examine the Hellenistic remains in the rest of the country. Bactria The most significant site is Balkh itself, ancient Bactra where, ever since Foucher’s disappointing initial soundings in the 1920s, the Hellenistic presence at the presumed capital of the Graeco-Bactrian kingdom has proved frustratingly elusive – Foucher’s ‘Bactrian mirage’ – although the trial trenches of 1947 present us with firmer ground for the study of the pottery in this period.90 The elusiveness of Greek Balkh was dramatically overturned by the excavations directed by Roland Besenval of the Mission Archéologique Française de Bactriane Afghane between

285

286

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.18  View of the excavations at Tepe Zargaran, Balkh, with fragmentary Classical column remains.

2004 and 2008 at the site of Tepe Zargaran within the walls of Balkh, where fluted columns and Corinthian and Ionic capitals confirmed the existence of a temple whose Hellenistic credentials could not be in doubt (Fig. 5.18).91 Further north in the Balkh oasis, the Soviet–Afghan expedition excavated several important sites dating from the Hellenistic period. The extensive urban site of Dilbarjin dates mainly from the Kushan period, but the earlier levels contained a fragmentary painting of the Dioscuri, suggesting the existence of a Hellenistic temple dedicated to the cult. At Emshi Tepe the Afghan–Soviet team excavated a completely circular town founded in the Hellenistic period and continued until the fourth to fifth century ad,92 continuing a town-planning principle from the Achaemenid period that was to have a long duration in the Iranian world (Fig. 5.19). A circular plan also characterised what appears to be a massive temple at Jiga Tepe nearby (Fig. 5.20). Both Emshi Tepe and Jiga Tepe also produced Greek ostraca.93 Other sites are promising, but either have not been thoroughly investigated or else only their existence has been recorded. Such is the case, for instance, with Shahr-i Banu, near Tashkhurgan, where in 1938–9 French excavations showed the superposition of several towns, yielding Kushan coins from the upper layers and coins of Euthydemus and Heliocles from the lower.94 Another example is the extensive urban site of Nimlik west of Balkh, remains of which extend for several kilometres and are dominated by a citadel mound, where a sherd with Greek

the iron age, achaemenid and hellenistic periods

Figure 5.19  Plan of the circular Hellenistic town of Emshi Tepe.

inscription was found.95 Most important, while the idea that it was Alexander and his successors who brought the blessings of urbanism to Afghanistan, giving rise to Bactria’s reputation as a ‘land of a thousand cities’, has long been dismissed, archaeology has conclusively demonstrated that the Hellenistic period was very much a part of a settlement continuum.96 Begram One must also mention the excavation of Begram, the ancient Kapisi of which more will be said in the next chapter, where the earliest period has been dated to the second century bc/second century ad.97 The pottery from this early period includes a grey-ware which, according to Allchin,98 seems to have its clearest

287

288

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.20  Plan of part of the excavated Hellenistic temple of Jiga Tepe.

parallels with the lowest levels of Shaikhan Dheri, Charsada, mainly in the Greek and Scytho-Parthian levels;99 and some large bowls of unpolished redware, with everted rims, that find parallels in Sirkap (Taxila) between 50 bc and ad 50, Tepe Zargaran (Balkh) immediately before and after the end of the Graeco-Bactrian kingdom and Kobadian, in Tajikistan, between the third and first centuries bc, thus providing links to other sites distributed over the whole area affected by the diffusion of Graeco-Bactrian influence.100 Kandahar and southern Afghanistan In southern Afghanistan, excavations at Kandahar (Figs 5.12, 5.13) by the British team proved elusive for the Hellenistic period despite the promise held by the Greek inscriptions (discussed below).101 To some extent this was due to the very

the iron age, achaemenid and hellenistic periods

heavy overburden of later periods that made any access to the Hellenistic problematic. Excavation was also exacerbated by very considerable robbing of the site – indeed, most of the site was heavily pitted. But the excavations did confirm continual occupation from the Achaemenid through the Hellenistic to the later pre-Islamic periods without any discernible interruption. It can be assumed therefore that it remained a major city throughout the Hellenistic period, even though nothing definably Greek (apart from the inscriptions) was found: no columns, capitals or other architectural fragments, for example, such as can be found north of the Hindu Kush (and now increasingly south as well). Kandahar has traditionally been associated with Alexander and identified with Alexandria Arachosia – indeed, the name itself was once thought to be derived from ‘Alexander’ (Afghan ‘Sikandar’), and ‘looking for Alexander’ (or at least Greeks) was one of the motives behind the British excavations there.102 ‘Kandahar’, however, has long been demonstrated to be an Arabic corruption of ‘Gandhara’, a name that was applied to no fewer than seven quite separate localities in Asia from Afghanistan to Cambodia in the Islamic source. The name is probably a corruption of the Persian qand-vihara, meaning ‘city (or fortress) of the [Buddhist] monastery.103 Nevertheless, the inscriptions, together with the archaeological urban continuity from Achaemenid through to Late Antiquity, make the identification of Kandahar Shahr-i Kohna with Alexandria in Arachosia highly probable but by no means proven.104 In this context it is worth mentioning that a separate ‘Alexandria’ (and another ‘Shahr-i Kohna’) exists in Kandahar Province: the site of Shahr-i Kohna associated with the village of Sikandarabad (‘Alexandria’) just to its north, near the right bank of the Helmand some 120 kilometres north-west of Kandahar. It is a large urban site comprising a citadel and lower town surrounded by ramparts – similar to and not much smaller than Kandahar itself (Fig. 5.21).105 No material earlier than the tenth century was noted during a very superficial observation of the surface pottery (although this does not rule out earlier material), and the site is probably identified with the medieval town of Zamindawar. The name ‘Sikandarabad’ does not necessarily have any association with Alexander: Sikandar, the modern Afghan version of Alexander, is a not uncommon name in Afghanistan.106 However, with some locating Alexandria in Arachosia as far distant from Kandahar as Ghazni, Shahr-i Kohna/Sikandarabad certainly deserves fuller investigation. More evidence of a surprising amount of – ill-defined – Hellenistic presence has been emerging from elsewhere in southern Afghanistan. At Mirwais Baba to the west of Kandahar a cylindrical marble funerary urn and a silver vessel containing gold ornaments were recovered from the remains of a Hellenistic tomb with terracotta pythons in the 1950s (the whereabouts of the objects, originally in the now defunct Kandahar Museum, are not known).107 At Mukhtar (or Mokhatar), to the north of Bust alongside the Lashkargah road, a large Hellenistic temple constructed on a high, stepped, artificial platform was reported by Trousdale.

289

290

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.21  The lower town and ramparts of Sikandarabad Shahr-i Kohna with the citadel in the foreground.

Although known since the nineteenth century, astonishingly this important site has never been properly investigated.108 At Khwaja Kanur on the south bank of the Arghandab, just above the confluence with the Helmand and a short distance from Mukhtar, both Fischer and Trousdale recorded the very eroded foundations of a sanctuary covered in limestone fluted columns and balustrade fragments, and thousands of terracotta figurines, together with fragments of a large terracotta frieze of Hellenised busts and acanthus leaves (although many of the fragments might have come from Mukhtar according to Trousdale).109 Further west near Kushk-i Nakhud on the main road to Herat a fragment of terracotta egg-anddart decoration was picked up and handed to the Kandahar excavations in 1977 (again, sadly never published).110 Further down the Helmand at Khwaja Sehyaka, the Smithsonian Helmand-Sistan Project in 1975 investigated the remains of a large Hellenistic sanctuary and associated structures. According to a good ceramic sequence as well as radiocarbon dating, the site dated from the late third century bc to the late third century ad. Objects recovered from a well included inscriptions in Aramaic and Greek, carved stucco, gold leaf probably from a wooden coffer and Classical architectural fragments in white limestone and carved brick.111 A survey by Norman Hammond in 1966 down the Helmand recorded some half a dozen more sites containing Hellenistic material.112 Clearly, while Aï Khanoum still

the iron age, achaemenid and hellenistic periods

remains (and will probably remain) unique, Bactria no longer has the monopoly on Hellenistic remains and one might expect further major discoveries from Hellenistic Arachosia in the future. Herat The identification of Alexandria in Arachosia with Kandahar might be unproven, but is at least probable given the evidence of the inscriptions and the archaeology. The identification of Alexandria Areia with Herat (or the Herat area), long assumed, is far more elusive.113 The recent Franco-German excavations in Herat revealed definite evidence for the Iron Age and Achaemenid periods (reviewed above), after which there is a gap until the Sasanian period. Even if Alexandria Areia was a new foundation away from the Achaemenid city, as some of the sources imply, not a single Hellenistic site has been recorded anywhere in the Herat region: the nearest are two sites at Farah114 which probably lie in the territory of Drangiana (the few coins from the Hellenistic period in the Herat Museum – discussed below – are unprovenanced). This appears astonishing in view of the prominence attributed to Areia in the sources, the now confirmed Achaemenid presence at Herat, and the obvious strategic position of Herat at the centre of a well-watered plain adjacent to a major river. The lack of any trace of a Hellenistic presence anywhere in the region is even more astonishing in view of surveys by the French in the 1950s, the Soviets in the 1970s and the Germans in the 2000s, when not a single Hellenistic site was recorded – material collected in the survey by Marc Le Berre and Jean-Claude Gardin in 1952 was re-examined by Gardin and Lyonnet in 1979, who could not have failed to record Hellenistic material.115 The only other early site in the Herat area which might qualify is Palgird, on the Hari Rud 65 kilometres west of Herat. It is an urban site with mounded remains covering a very extensive area approximately 1.5 km across – almost half the size of the old city of Herat itself – but the earliest material recorded there by Gardin and Lyonnet was Parthian of the first century ad.116 The Greek City of Aï Khanoum The study of the entire Hellenistic period in Afghanistan and Central Asia has, provocatively but probably justifiably, been called ‘before Ai Khanum and after Ai Khanum’,117 and one cannot exaggerate its importance.118 Aï Khanoum is a site in the north-eastern part of Afghanistan, close to the confluence of the Kokcha and the Amu Darya (Oxus). Here a Greek town has been discovered, presumably to be dated to a period from the end of the fourth century to c. 145 bc (Figs 5.22–5.24).119 The excavations have revealed the existence of a fortification system, private as well as public buildings and a necropolis. Aï Khanoum is the Turkish name of the nearby village, meaning ‘Lady Moon’. The ancient name is unknown; nevertheless, there is reason to suppose that Aï Khanoum might be identified with the

291

292

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.22  Plan of Aï Khanoum.

the iron age, achaemenid and hellenistic periods

Eucratideia of the classical sources or possibly Alexandria Oxiana, though this is placed by others at Termez, on the north bank of the Oxus. The site is usually considered to have been founded soon after the invasion of Alexander, although an indeterminate Achaemenid presence existed here before as well as an indeterminate Bronze Age presence.120 The end of Aï Khanoum has traditionally been linked to the invasions of Saka tribes in the second century bc, and the presence of Sakas is indicated in the palace after its sack by a silver ingot with runic inscription as well as by pottery.121 However, Sakas were long an element in the Bactrian population as far back as the Achaemenids and presumably continued to form a significant but integral part of the Graeco-Bactrian state: studies of the entire northern Hellenic world – particularly the Black Sea region – have shown Saka–Greek relations to be generally those of symbiosis. The overthrow, therefore, was not as simple as previously believed: it has even been suggested that the fall of Aï Khanoum was caused by rival Graeco-Bactrian elements in alliance with the Sakas.122 Furthermore, there is numismatic evidence123 for the decline of Graeco-Bactria before the fall of Aï Khanoum, with the Saka invasions a result of the decline rather than the cause, a final death blow to a city already in decline.124 Whether or not invasion was involved, the site was probably abandoned soon after 145 bc, although the date for the abandonment remains controversial and may have been much later.125 Furthermore, there is also evidence for a ‘post-Greek’ occupation, perhaps between 145 and 130, when the Temple of Indented Niches could still have been in use after the destruction of the statue in spite of its use as grain storage.126 The ceramic evidence also indicates an element of continuity after the Saka invasion.127 Suffice it to say, the end of Aï Khanoum remains controversial. The town occupies a naturally fortified position, protected by the Amu Darya, the Kokcha and an acropolis; the whole complex was encompassed by a line of defences provided in places with a moat (Figs 5.22, 5.23). In the words of Paul Bernard, ‘the site was remarkably well-suited to the implantation of a military stronghold which could eventually develop into a large city’.128 We must therefore imagine Aï Khanoum posted as a sentinel to guard the natural north-eastern gateway of Bactria, between the Amu Darya and the first slopes of the Badakhshan mountains, against the menace of nomadic invasions. The abundant natural resources of Badakhshan would also have been a factor.129 The lower town, between the acropolis and the Amu Darya, includes three well-defined parts – a habitation area to the south, the administrative quarter in the middle and an area almost devoid of structures to the north. The main street started from the main gate in the northern wall and ran parallel to the side of the acropolis as far as the Kokcha to the south (actually north-east to south-west). The predominant technique of construction is mud-brick, sometimes on a basement of baked bricks; this is a local, rather than Greek, way of building. Apart from the unmistakably Greek style of the theatre, the more local appearance of the buildings becomes more apparent in the recent Franco-Japanese 3D digital

293

294

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.23  Aï Khanoum, view from the citadel.

Figure 5.24  Aï Khanoum, photo of the site revealing extensive damage caused by robbing.

the iron age, achaemenid and hellenistic periods

Figure 5.25  Aï Khanoum, reconstructed view towards the south-west from the theatre.

Figure 5.26  Aï Khanoum, reconstructed view to the south-east: the propylaeum and courtyard of the palace.

295

296

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.27  Aï Khanoum, reconstructed view to the west: the main street, the temenos of the temple with indented niches and the courtyard of the palace.

reconstructions of the site (Figs 5.25–5.27),130 such as the flat roofs which covered certain buildings which are not in the Greek tradition. Nevertheless, ‘for everything else, the architectural techniques were Greek: stone blocks laid dry without mortar, tightly fitted by anathyroses and fastened together by metal dowels and cramps sealed by molten lead; flat Corinthian tiles with covertiles, and antefixes at the end of eaves cover-tiles’.131 Columns and some of the thresholds are made of stone, a limestone quarried 50 km south-west of the site. Bernard was able to put together chronological data deriving from epigraphic, numismatic and architectural evidence in a synoptic table,132 of which we give here only the succession of periods and sub-periods: Period I: 330–303 bc. Period II–1: first half of the third century bc. Period II–2: second half of the third century bc. Period III–1: first half of the second century bc. Period III–2: c. 150 bc. Period IV: second half of second century bc; destruction by fire, c. 145–130 bc.133

the iron age, achaemenid and hellenistic periods

The above periodisation, as well as the dates for the beginning and end of Aï Khanoum, are still a matter of debate, but Bernard’s chronology is still at least the best working hypothesis. In this context it must be remembered that (a) the excavations were meant to be ongoing until interrupted by the 1979 invasion, and (b) at the time of writing the final publications are still in progress. The administrative quarter (Figs 5.26, 5.28, 5.29) is one of the most important building complexes at Aï Khanoum and was named ‘Palace’ in the earlier

Figure 5.28  Aï Khanoum, plan of the administrative complex.

297

298

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.29  Aï Khanoum, colonnade of the palace.

reports.134 It is placed in the central area of the lower town, its north-eastern part being occupied by a large courtyard with a peristyle (108.11 × 136.77 m) consisting of 116 stone columns. These rest on Attic–Asiatic bases, have plain shafts made of drums of varying heights, and support pseudo-Corinthian capitals which probably originated in Seleucid Syria, but are now also found at Tepe Zargaran at Balkh.135 To be more precise, the bases proper show the typical profile of the Attic bases (i.e. consisting of an upper and lower torus, with a scotia between); the scotia is separated from the lower torus by a fillet, from the upper one by an astragal; the upper torus is connected to the shaft through a fillet and a cavetto. There are also variants of this type (Figs 5.30–5.31).

the iron age, achaemenid and hellenistic periods

Figure 5.30  Aï Khanoum, Corinthian capital from the south portico of the palace courtyard.

The courtyard was entered on the north-eastern side through a propylaeum with four columns (Fig. 5.26). Their capitals are similar to the capitals of the peristyle, but the bases are completely different, since they are composed of a three-stepped plinth below and a swollen torus above, and have therefore a markedly different appearance from the Classical style; indeed, they are similar to some Achaemenid bases, especially those of the Treasury at Persepolis (Fig. 5.29).136 The south-west side of the courtyard, which faces the propylaeum, is obviously the main façade into which opens a pillared vestibule (27.67 × 16.44 m) with three rows of six Corinthian columns resting on Attic-Asiatic bases. This has been related to the later Central Asian architectural tradition of the pillared ivan or portico; equally, it might derive from Achaemenid palace architecture, such as the columned porticoes of the Persepolis Apadana. The vestibule gave access to a large rectangular room (26.02 × 16.50 m) decorated with wooden half-columns; further to the south-west a block composed of two pairs of twin structures symmetrically arranged is probably part of a later extension; the main rooms in the structures to the south-east are decorated with pilasters surmounted by capitals which seem

299

300

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.31  Capitals from Aï Khanoum re-used in a nearby tea-house.

to be rather poor imitations of the capitals of the peristyle. Both these rooms (Fig. 5.28) were probably employed for cult or official purposes (audience halls?), as is suggested by the finding of fragments of stucco and clay sculptures. The two structures to the north-west, which were excavated in the campaigns of 1972 and 1973,137 do not show such large and richly decorated main rooms as do the other two structures: this probably means that they were the seat of the chancellery offices, while the other two served a more official purpose. The residential area of this administrative quarter was partially excavated in 1974 and 1975: the building discovered there shows many similarities with the large private houses of Aï Khanoum;138 the complex, or rather this part of the complex, has again been styled ‘Palace’ in the later report,139 as a consequence of these discoveries. Though the excavation of this functional complex was not completed and we cannot altogether understand the function of each part, Bernard has tried to suggest that the whole complex reflects the diarchical character of the administrative power for which it was conceived. He further expounds three possibilities for the explanation of the nature of such power. The first is that the ‘administrative quarter’ is a Palace, a basileion, one of those royal residences that Persian monarchs had built in various parts of their empires for staying during their recurrent visits; the twin structures we have described could reflect the association of the crown-prince to the throne (the joint suzerainty of Eucratides and Heliocles is a particularly reasonable hypothesis, since the date of the south-western complex

the iron age, achaemenid and hellenistic periods

Figure 5.32  Aï Khanoum, mosaic floor in the administrative building.

in the middle of the second century bc approximately corresponds to the reign of Eucratides). The second possibility is that the ‘administrative quarter’ was the residence of a governor, head of a satrapy, assisted either by a commandant or by an official of the royal administration (as, for instance, at Susa under the Parthians). A third hypothesis was suggested to Bernard by the suburban villa discovered at Aï Khanoum during a later season.140 When a private house attains the monumentality of this building extra moenia, Bernard says, one is naturally led to think that the town of which its owner was a citizen could easily avail itself of enough money for building such a complex as the ‘administrative quarter’. In a word, Aï Khanoum was a real polis with large municipal autonomy within the frame of royal suzerainty. In this case, the two twin units of the south-western complex could be the seat of the two supreme town magistracies. If this is true, one should also look for the place in which the Council was housed (one cannot imagine a Greek polis with no Council or General Assembly). Bernard points to the large room between the pillared vestibule on the south-west side of the peristyle and the architectural complex we have just discussed: its size (27.50 × 17 m) would suit well for a bouleuterion (Council Hall). The Assembly, on the other hand, could be gathered in the peristyle courtyard itself or perhaps the theatre.141 It is also interesting to note that the various parts of the administrative quarter date back to different periods: the propylaeum and the largest portion of the

301

302

THE ARCHAEOLOGY OF AFGHANISTAN

­ eristyle belong to Period II1, while the pillared vestibule was built in Period II2 p and completed during Period III1; the supposed bouleuterion also belongs to Period III1 and the south-western complex was built, as we have already said, in Period III2, c. 150 bc. Some 40 m north-east of the courtyard a funerary chapel (heroon) where four burials were found, that according to an inscription discovered in situ was known as the temenos of Kineas. The first phase of construction dates back to Period I, probably to the time of Alexander himself (329–327 bc), by whose order Bernard supposes that the otherwise unknown Kineas founded Aï Khanoum. He therefore obtained for himself and his descendants the right to be buried intra muros, a right that in the early Hellenistic period was granted only to the oikistes (‘founder of the town’). The most surprising building at Aï Khanoum is certainly the so-called ‘temple with indented niches’, south-east of the temenos of Kineas (Figs 5.33 and 5.34).142 Its name is due to the triple-stepping on the outer face of the walls; these define a row of alternating false niches. It is a mud-brick building approximately 19 m square, composed of an oblong vestibule and a smaller cella flanked by two narrow sacristies. In the vestibule, on both sides of the door which leads into the cella, were three mud-brick pedestals that supported clay and stucco statues, fragments of which were found scattered on the ground; a few fragments of stone belonging to the cult statue were found in the cella. Bernard has proved that this plan derives from Mesopotamian prototypes but was also well-known in the Seleucid empire, and the indented niches are a device known in Iron Age Iran, such as at Hasanlu and Nush-i Jan. Though so deeply Near Eastern in plan and construction (the only architectural device of Greek origin seems to be the three-stepped krepidoma), this temple housed a cult image that was purely Hellenic, at least as far as we can judge from the few existing fragments. The temple with indented niches underwent several stages of architectural modification: stage V is represented by an earlier construction which was replaced by the temple with indented niches; stages IV, III and II cover the period in which the building with indented niches was in use as a temple, while stage I marks a late re-employment of the building as a store-house. The temple probably continued in local use after the Greeks had abandoned the city.143 According to the numismatic data and the pottery, this temple seems to have been built in the first half of the third century (the pre-existing temple of stage V would therefore be as old as the beginning of the third or even the last quarter of the fourth century bc dated to Antiochus),144 while the modifications it underwent in stage II are either contemporary with or later than the reign of Diodotos.145 In this respect, one should remember that the use of pottery for the purposes of dating monuments is not always easy at Aï Khanoum, because of the high degree of stability of pottery types and quality.146 We may point out that the pottery from Aï Khanoum shows a close relationship with the Hellenistic pottery of the Graeco-Mediterranean world, from the point of view of both technique and shapes, though some types are peculiar

the iron age, achaemenid and hellenistic periods

Figure 5.33  Aï Khanoum, plan of the temple with indented niches.

to the Orient (e.g. the ‘pilgrim-flasks’). Among the Hellenistic types, ‘fish-dishes’, hemispherical bowls with ring-foot and ‘Megarian’ bowls are noteworthy.147 The scope of this book does not allow a detailed description of all the other monuments that have been brought to light at Aï Khanoum. We are only able to list them: a house in the residential area of the lower town, similar in plan to a

303

304

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.34  Aï Khanoum, the temple with indented niches.

Parthian building at Rhagae, the modern Rayy, and other buildings in Mesoptamia and Iran;148 the villa extra moenia already referred to;149 a gymnasium in the northern area of the lower town, which has been identified as such thanks to its peculiar architecture as well as to a dedication to Hermes and Heracles;150 another heroon;151 the theatre (Fig. 5.25);152 the arsenal. Nevertheless, a few words must be added concerning the necropolis, due to the enormous importance of this discovery.153 So far, a mausoleum has been brought to light, at the foot of the north-east side of the acropolis, outside the town walls. This is a mud-brick rectangular building, partly underground and with vaulted ceilings, which underwent several modifications in its structure: a door on one of the longer sides led into a corridor on either side of which opened a crypt; both the corridor and the two crypts were vaulted, but the whole building probably had a flat roof. The mausoleum housed two types of burial – mud-brick sarcophagi for inhumations, and funerary jars in which were collected and buried the bones taken from earlier graves, when these had to be destroyed or employed for new burials. As Bernard rightly points out, the architectural interest of this mausoleum (with its connections with the Parthian necropolis of Assur) is much greater than the importance of the funerary material found in it; nevertheless a schist pyxis decorated with inlaid coloured stones is a very important antecedent of some Gandharan relic-caskets, a fragmentary stone relief representing an ephebe raises

the iron age, achaemenid and hellenistic periods

some interesting iconographic problems, and some ink inscriptions on the funerary jars give us a first hint of the proper names used at Aï Khanoum, a field of research of considerable relevance for knowledge of the composition of the population of the town. Another very important field of investigation deals with the use of canals in the vicinity of Aï Khanoum: one of the water-control systems discovered belongs to the period of the Graeco-Bactrian town, and remained in use until the beginning of the Kushan period (discussed in more detail in Chapter 3).154 From the point of view of the history of art, especially sculpture, fairly rich documentation is now available at Aï Khanoum, as we have indicated.155 It clearly shows at least four stylistic trends: (1) a purely Hellenistic element to be considered alongside the Greek inscriptions, demonstrating the great attachment of the Aï Khanoum ruling class to the culture of their far-away fatherland: (2) a group of objects that, though lacking in homogeneity, reflect the stylistic trends of the contemporary Near East and Iran: (3) an element of Hellenistic derivation that shows the beginning of a development towards original solutions, such as those that gave rise to Gandharan art: (4) local Bactrian, survivals from the Achaemenid and elements from the Eurasian steppe.156 To trend (1) belong the fragments of the cult statue (an acrolith) in the cella of the temple with indented niches (Zeus Oromasdes?), third century bc; it has been suggested that the temple was dedicated to a syncretic Zeus Belos-Ahura Mazda cult.157 There is also a herm portraying a bearded old man, from the Gymnasium, probably third century bc (Fig. 5.35);158 a fragmentary stone plaque (a funerary relief) from the necropolis, which portrays a standing youth wearing chlamys and petasus with long flowing hair, third century bc(?) (Fig. 5.36);159 and a terracotta mould for a bust of (?)Demeter. Two objects at least fall into (2); they are: the silver medallion with representation of the goddess Cybele on a chariot drawn by a pair of lions (Fig. 5.37) to be dated to the beginning of the third century bc;160 and the bone figurine representing a nude standing goddess that Bernard161 compares with the figurines found in Central Asia but which seems to be rather closer to a class of statuettes of Mesopotamian tradition from Iran, from the Elamite to the Parthian period (Fig. 5.38).162 The third group of sculpture (3) includes: a bronze statuette of a beardless Heracles holding a club and putting a wreath on his head, which Bernard163 qualifies as of style rustique (Fig. 5.39); a headless limestone female statuette from the sanctuary of the temple with indented niches, leaning on a pillar, very close to Hellenistic models from a typological point of view164 but certainly provincial in style and pointing towards ‘Gandharan’ solutions (Fig. 5.40); and two heads (a female made of unbaked clay, and a male of stucco) from the vestibule of the temple with indented niches,165 which are probably the most evident link between the Hellenistic products and the later unbaked-clay sculptures of Khalchayan and Gandhara, especially Tapa Sardar and Hadda, which we discuss below.

305

306

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.35  Aï Khanoum, a herm of an elderly man from the gymnasium.

In conclusion, Aï Khanoum as an art centre appears to be closely linked to Hellenistic culture but not altogether excluding the Achaemenid tradition from its repertoire, chiefly in architecture; at the same time, the excavations at Aï Khanoum bear witness to the fact that Bactria was ready to accept the products of the Hellenised Near East and able to blend the various traditions into an original style, the Greek character of which was its distinguishing trait in relation to the other neighbouring cultures.

the iron age, achaemenid and hellenistic periods

Figure 5.36  Aï Khanoum, funerary relief from the necropolis, representing a youth with chlamys, petasus and long flowing hair (third century bc).

Bactrian art is indeed the outcrop of an élite culture. With the tradition of Greek art finding its way into everyday figural language of larger groups of population, we see the birth of Gandharan art. It is also possible that this transformation of Graeco-Bactrian into Gandharan art cannot be fully understood in the light of the excavations at Aï Khanoum alone, and that other sites in Afghanistan preserve towns still awaiting excavation that flourished from the Graeco-Bactrian period through the Indo-Scythian and Indo-Parthian, and well into the Kushan period.

307

308

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.37  Aï Khanoum, silver medallion with Cybele on a chariot; probably an import from Syria (third century bc).

Figure 5.38  Aï Khanoum, bone figurine representing a standing goddess.

the iron age, achaemenid and hellenistic periods

Figure 5.39  Aï Khanoum, bronze statuette of a beardless Heracles.

Gandharan art and its roots in the Graeco-Bactrian period are discussed further in the next chapter. In the 1990s, Aï Khanoum was deliberately targeted for systematic illicit pillaging, when most of the site was destroyed (Fig 4.24). Sites outside Afghanistan Several sites that are only just outside Afghanistan across the Amu Darya must be mentioned. The first is Kampyr Tepe in Uzbekistan overlooking the Amu Darya

309

310

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.40  Aï Khanoum, headless female statuette from the sanctuary of the temple with indented niches.

due north of Balkh – indeed, it probably marked the crossing point on the main route linking Bactra and Maracanda in Hellenistic times. Here, joint Russian– Uzbek excavations uncovered a third century bc Hellenistic fortified post where finds included armour similar to that from Aï Khanoum and some fragmentary Greek texts. Kampyr Tepe has been identified with Alexandria Oxiana as an alternative to Aï Khanoum.166

the iron age, achaemenid and hellenistic periods

The second is Termez, also on the Amu Darya just to the east of Kampyr Tepe. Although it is known mainly for the extensive Kushan period city and surrounding Buddhist monuments, significant Hellenistic material has been ­ recorded under the Kushan city when it was probably a fortress associated with a cult centre.167 Takht-i Sangin in Tajikistan is a major Hellenistic site near the junction of the Panj River with the Amu Darya (in fact the Vakhsh, which is the name given to the upper Amu Darya and approximates the ancient name of Oxus). This is the site of the famous ‘Temple of the Oxus’ built in the second century bc in a style similar to that of Aï Khanoum: mud-brick but with Ionic columns, altar and other elements in stone. It was dedicated to the Oxus river god, although in layout the temple resembled later Zoroastrian fire temples.168 Pits associated with the temple contained over 5,000 dedicatory objects, many of them precious. This has led many to believe that Takht-i Sangin might be the find spot of the Oxus Treasure, although Takht-i Kuwad further south on an ancient crossing point of the Amu Darya is more favoured.169 Sakhsanokhur, also in Tajikistan, lies on the Afghan border some 36 kilometres north-east of Aï Khanoum. Excavations have revealed a fortified residence and settlement of the Hellenistic period, whose remains included stone architectural elements that are suggested to have been re-used from Aï Khanoum.170 Lastly we must mention the hypothesis that Bactrian art was well-accepted outside Bactria, at the Parthian court of Nisa south of Ashkhabad in Turkmenistan, where the famous rhyta found by the Soviet archaeologists171 may be regarded as the richest group of Bactrian art objects ever found, as has been cautiously ­suggested by Bernard.172 The Nomad ‘Invasions’ Formerly a ‘dark age’ in the archaeological record, the arrival of the nomads – Sakas and Yuezhi – at the end of the Hellenistic period that culminated in the Kushans (examined in the next chapter) is now graphically illustrated by the discoveries at Tillya Tepe173 in 1978 by Viktor Sarianidi of the Afghan–Soviet Mission. This was probably the most spectacular discovery of gold artefacts in the world of the latter half of the twentieth century. Tillya Tepe is an early Iron Age site later used as a cemetery, possibly by the nearby Hellenistic to Sasanian settlement of Emshi Tepe (also excavated by the Mission), although it has been suggested that the burials were secondary, having been moved here from elsewhere, albeit intact.174 There were six undisturbed burials in raised wooden coffins of five females and one male adorned with some 20,000 gold objects: bracelets, bowls, clasps, buttons, weapons, statuary jewellery, etc. The richness and nature of the burials suggests that the male was a king or chief and the females his attendants or consorts (Figs 5.41–5.46). One of the female burials also contained narcotic plants, suggesting

311

312

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.41  Tillya Tepe, pendant with seal of Athene.

links to Scythian practice described by Herodotus as well as similar discoveries in the frozen tombs of Pazyryk. Reports of a seventh burial containing extensive amounts of jewellery on a par with the other six graves were confirmed by local officials and villagers.175 In the decades of instability and warfare in Afghanistan after 1979, the treasure was sealed for safekeeping in the vault of the presidential palace in Kabul. After the vault was reopened in 2004, the Tillya Tepe treasure formed the main focus of a worldwide touring exhibition.176 The art objects are not quite like any other treasure found in Afghanistan, representing a mixture of Indian, Central Asian, Iranian and Hellenistic styles, as well as local Bactrian, making it difficult to date. Hence, the original excavators cautiously dated the burials broadly between the first century bc and the first century ad, but in part thanks to a Roman coin of Tiberius they were later pinpointed more accurately to the later first century ad by Zeymal.177 X-ray fluorescence analysis of some of the gold objects confirmed that some at least were of local manufacture.178 Some appeared Greek, and may even have been imports from

the iron age, achaemenid and hellenistic periods

Figure 5.42  Tillya Tepe, tree.

the Mediterranean (or at least the Black Sea).179 Others were certainly of Chinese manufacture, such as Chinese mirrors, which can be dated more accurately to the mid-first century ad – indeed, the placing of Chinese mirrors over the chests of three of the female burials actually follows Chinese practice.180 Of particular interest were many that reflect the ‘Scythian’ art of the steppes – the so-called ‘animal style’ – which belongs within a broad date range of the last half of the first millennium bc and which is otherwise not represented in Afghanistan. Attention has been drawn to similarities with Saka, Xianbei and Xiongnu art and burials in Siberia and Mongolia as well as with Parthian, Pontic and Anatolian art.181

313

314

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.43  Tillya Tepe, ‘Bactrian Aphrodite’.

The eclectic nature of the assemblage, together with similarities to some of the art of the steppes, led the excavators to associate the burials with an elite group of the Yuezhi tribes (perhaps even the Kueshang tribe, who eventually became the Kushans) who originated on the western borders of China. Others have associated them with the Sakas (with whom the Yuezhi may have been affiliated, although this remains controversial). However, both the Sakas and the Kushans continued to play a major part in subsequent history of the region, but with the art styles of Tillya Tepe not continuing into the subsequent art of Afghanistan, it might be that the Tillya Tepe burials represent an isolated tribal incursion from the steppe into Bactria.

Figure 5.44  Tillya Tepe, clasp.

Figure 5.45  Tillya Tepe, dagger and sheath.

Epigraphy The Achaemenids Of all ancient Near Eastern states, the largest and latest is the least documented. For the Achaemenid Empire there is nothing like the copious indigenous source material that exists for the Assyrians or the Egyptians or even the Hittites, having to depend largely on outside secondary sources, mainly Greek (and of these, mainly Herodotus who, as a native of Caria, did at least grow up a citizen of the Persian Empire, so can in part be viewed as indigenous). Of the indigenous sources we

316

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.46  Tillya Tepe, ‘dragon king’.

are dependent almost wholly on royal proclamations – mainly Bisitun, but also Persepolis and Naqsh-i Rustam – which are propagandistic, and the occasional interpretative item that can be squeezed from a ration list (the Persepolis fortification tablets). Virtually none of the epigraphic material came from the eastern regions of the empire until the discovery of the Aramaic documents from Bactria (discussed below). The Bisitun inscription182 and the relief of Darius I that accompanies it are cut on a cliff rising high above the main road from Mesopotamia to Media. The relief shows Darius seated with the rebels he had put down bound in front of him. The inscription is in three languages, Old Persian, Akkadian (the language of Babylon)

the iron age, achaemenid and hellenistic periods

and Neo-Elamite. It sets out the deeds of Darius after he became king and the battles he fought to subdue his enemies, listing the 23 countries of his empire. These include Aria, Bactria, Drangiana, Arachosia, Sattagydia and Gandhara – satrapies that are now part of Afghanistan. A fragment of this text in Aramaic has been discovered in Egypt; other versions have been found in Babylonia and Susa and its legal terms and forms were even borrowed in Armenian and Syriac,183 giving credence to Darius’ claim that he made many copies and sent them everywhere among the provinces. A valuable source is the Persepolis fortification tablets, which are increasingly being analysed for important information on the eastern Achaemenid Empire, relating to issues of economy, tribute, administration and communications.184 The foundation inscriptions of Darius for his palace at Susa also contain valuable information on the resources of the empire, including those from the eastern satrapies.185 Elamite Elamite was also used in Achaemenid imperial correspondence, and several documents have been discovered in Afghanistan. The ‘Chaman hoard’ from Kabul included an Achaemenid silver bar with a cuneiform inscription in Elamite,186 and the excavations at Kandahar uncovered some fragmentary Elamite tablets. These are accounting tablets that are near-identical with the Persepolis fortification tablets, also Elamite – indeed, the tablets and other documents referring to Arachosia at Persepolis are the richest in the archive. Although very fragmentary, the Kandahar tablets confirm the identity of Kandahar as the Achaemenid satrapal capital and almost certainly formed part of an archive there.187 Aramaic Aramaic had become the common language of the Near East under the Assyrians. The Achaemenids used it as the official language of their administration, and presumably introduced it to their satrapies in Afghanistan and the Indus valley. The Aramaic inscription discovered on an octagonal pillar during excavations at Taxila in Pakistan in 1915188 refers to an official, when Ashoka was governor of Taxila under his father. The five Aramaic inscriptions known so far from Afghanistan, three from Laghman and two from Kandahar, also belong to the Mauryan empire which came to control the former Achaemenid satrapies in eastern Afghanistan in the third century bc, and relate to Ashoka’s religious reforms. More recently, a hoard of Aramaic Achaemenid documents, probably from Balkh, came to light and is now in a private collection in London. These have transformed our documentation of Achaemenid Bactria. They comprise some 50 texts, 48 of them in official Achaemenid Aramaic; 29 are on leather and 18 on wooden strips, all written in ink, dated to between 353 and 324 bc. Five of the

317

318

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.47  Aramaic document, probably from Balkh, concerning the campaign against Alexander.

leather documents comprise official satrapal correspondence and administrative documents relating to Akhvamazda, the Satrap of Bactria under Artaxerxes III, and his governor, Bagavant. Bagavant was governor of hlm, almost certainly modern Khulm/Tashkurghan to the west of Balkh. These refer to the events surrounding Alexander’s invasion of Bactria: one that mentions the regnal Year 1 of Bessus (under his regnal title of Artaxerxes) and another that refers to ‘15 Sivan of the Year 7’ of Alexander, which equates to 8 June 324 bc. The wooden strips are interpreted as tallies, also relating to the official administration. Although questions have been expressed about the discovery and provenance of the hoard, it is without doubt the most important official Achaemenid archive for the province of Bactria ever to be discovered. In particular, those that relate to the provisioning of Bessus’ army fighting Alexander are one of just a handful, if that, of contemporary documents for Alexander’s reign (Fig. 5.47).189 Inscriptions of the Mauryan Period190 The Aramaic inscriptions from Laghman The first Aramaic inscription to be discovered in Afghanistan was the fragmentary stone tablet found in the neighbourhood of Pul-i-Darunta in 1932 (Fig. 5.48). It was puzzling because it contained a number of unknown words, with ‘shyty’ repeated several times. Henning191 showed that it contained not only Aramaic but Middle

the iron age, achaemenid and hellenistic periods

Figure 5.48  Aramaic inscription from Darunta.

Indian Prakrit words, and that ‘shyty’ accompanied these. It seems to be an abstract from the 5th or 7th Pillar Edict of Ashoka. Such abstracts were expressly authorised by Ashoka in his Rock Edict XIV. We see that it is a bilingual IndoAramaic inscription similar in form to the one discovered at Kandahar in 1963. A second Aramaic inscription was discovered at Shalatak in the Laghman valley in 1969, 30 km from Pul-i-Darunta. It is cut on the vertical face of a rocky ridge above the river in a position that would have dominated the old road.192 Dated in year 10 of Ashoka, the year of his conversion to Buddhism, it speaks of the

319

320

THE ARCHAEOLOGY OF AFGHANISTAN

e­ xpulsion of vanity and the king’s prohibition against fishing. Dupont-Sommer’s claim that it refers to the distance to Palmyra – ‘200 arcs to Tadmor’ – cannot, however, be substantiated.193 A third Aramaic inscription of Ashoka, dated in his sixteenth year, was discovered in 1973 in the Laghman valley some 12 km from its confluence with the Kabul River.194 It refers to the King’s religious views and seems to give an indication of the distance to the next locality. The bilingual Greek and Aramaic inscription from Kandahar In 1958 a fine rock inscription was discovered in Old Kandahar.195 It was in excellent condition with a text in Greek on the upper part of the rock and one in Aramaic beneath it (Fig. 5.49). The Greek version, beginning with Ashoka’s name, gives the text of one of the pious proclamations of the king. It is complete in itself and followed by an Aramaic version of the same proclamation. Ashoka, the greatest king of the Mauryan dynasty and the first ruler to unify India, had been a hazy figure in Indian texts and Buddhist tradition until the discovery of his inscriptions – proclamations in the form of rock edicts inscribed on rock surfaces or in pillar edicts inscribed on columns. These were intended to explain the king’s concept of dharma – not formal religious belief so much as an attitude of social responsibility. Ashoka used the language and script of the localities in which he set up his inscriptions – middle Indian with Brahmi or Kharoshthi script in India, Aramaic in Taxila and Laghman and Greek and Aramaic here at Kandahar. Although the Indian texts are clear enough in outline, there have been difficult problems of interpretation, which the versions in non-Indian language have helped to solve. For example, dharma was long translated as ‘law’; but the bilingual Kandahar inscription uses an Aramaic word that Dupont-Sommer translates as ‘truth’ and a Greek word that must be ‘piety’. The Ashokan inscription in Greek from Kandahar The second Greek inscription from the ruins of old Kandahar (Fig. 5.50) was discovered in 1963 and presented to the National Museum.196 It is cut on a block of porous limestone some 45 × 70 cm. Its text gives the end of edict XII and the beginning of edict XIII of King Ashoka. The stone is a rectangular block 12 cm thick, which must have been part of a much larger monument – a building of stone on which the fourteen major rock edicts of Ashoka were inscribed – and this suggests that we should in due course find more inscriptions from this building in old Kandahar. Edict XIII is about ‘the sects’. Ashoka insists they must respect each other and accept the lessons of others. It is interesting to see how ‘the sects’ are translated by ‘schools of thought’ in Greek. Edict XIV, as in other versions, describes his con-

the iron age, achaemenid and hellenistic periods

Figure 5.49  The Graeco-Aramaic bilingual inscription of Aśoka from Kandahar.

quest of Kalinga when 100,000 of its inhabitants were killed and 150,000 deported; the king’s remorse for this was his order to abstain from eating the flesh of any living creatures and his zeal for piety. The Indo-Aramaic inscription from Kandahar About the same time, in the latter part of 1963, a fragmentary Aramaic inscription, now in an Italian collection, was bought in the bazaar at Kandahar.197 The

321

322

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.50  The Greek building inscription from Kandahar.

stone is some 24 × 18 cm and contains a text of seven lines in Aramaic letters. The fragment has no name but is clearly Ashokan because of its contents. It contains the strange word ‘shyty’, which is used several times as in the Aramaic inscription from Pul-i-Darunta. Each time it occurs, it comes immediately before an Indian and after an Aramaic word. Although the script is Aramaic throughout, it is in fact a bilingual inscription involving two languages – Aramaic and Prakrit. The two are mixed – each group of words in one language is followed by a paraphrase or translation in Aramaic of the Indian text and the change is marked by ‘shyty’. It is an interesting presentation, which gives the original Indian text transliterated into Aramaic, followed word by word with an Aramaic translation. It contains part of the seventh Pillar Edict of Ashoka. The end of the Kandahar inscription appropriately has, in Aramaic alone, ‘[These orders] were set to writing on pillars’. The Sophytos inscription from Kandahar Another Greek inscription on a stele appeared on the international antiquities market in the eary 2000s and is now in a private collection. However, Paul Bernard, one of the scholars who first examined and then fully published the stele, was satisfied that it is not a fake and that it originally came from Kandahar.198 It is on a square block of white limestone measuring 62 × 62 × 12.5 cm, and contains 22 lines in perfect Greek (Fig. 5.51). It is dated broadly to the second century bc. The inscription is a funerary epitaph commemorating one Sophytos son of Naratos, who, after leaving his home (presumably Kandahar), returned having made his fortune and restored the family home and mausoleum. Its peculiarity lies in the

the iron age, achaemenid and hellenistic periods

Figure 5.51  The Sophytos inscription, purportedly from Kandahar.

nature of the names Sophytos and Naratos (or Narates), neither of which are Greek despite the linguistic quality of the inscription. Both have been interpreted as well-known Indian names, the Prakit forms of which are Subhuti and Narada. This adds to the increasing evidence of Kandahar being identified with Alexandria Arachosia, but as being of mixed Indian and Greek identity. Historical significance The discovery of this important group of inscriptions of Ashoka from Laghman and Kandahar gives us a clear picture of the western extent of the Mauryan empire, and its control of Arachosia. The Greek inscriptions, in form and style, belong to

323

324

THE ARCHAEOLOGY OF AFGHANISTAN

the high Hellenistic period and are the same as one would find elsewhere in the Greek world at this date – impressive evidence for the unity of Greek culture in the third century bc. Greek colonists at Kandahar still constituted an important nucleus of culture under the Mauryans, in touch with the main Greek world, and it is to them that the preaching of Ashoka’s edicts in Greek are directed. But there is equal interest in the Indian and Iranian population of the Mauryans – in particular the Kambojas mentioned with the Yonas (i.e. Greeks) in edict XIII as peoples to whom Ashoka sent missionaries.199 Seleucid and Graeco-Bactrian Inscriptions and Epigraphy200 In discussing the Greek inscriptions from Afghanistan, Georges Rougemont writes that they ‘really are exceptional documents which would also be considered as such, had they been found by the Aegean sea; in fact, these texts would not have made sense anywhere else than in Central Asia’.201 The Aristonax inscription from Kandahar The only inscription to come from controlled excavations at Kandahar was an inscribed statue base on white marble recovered in 1978.202 However, it was from a secondary context re-used in a wall, hence of minimal value stratigraphically. It contained four lines of badly worn text and seems to commemorate the escape from a wild beast attack by a certain ‘son of Aristonax’, probably dating between 300 and 250 bc, with a preference towards the earlier date. It might therefore support the view that Kandahar remained in Seleucid hands in the early third century, although the suggestion that it implies ‘the other regular features of Greek civic life – other sanctuaries, an agora, public buildings, a theatre, and the corresponding civic institutions associated with them’203 – could not be substantiated (and indeed is unlikely). Inscriptions from Aï Khanoum The excavations at Aï Khanoum have produced three well-preserved inscriptions – important evidence for the pure Greek character of the city in its language, culture and system of education in Bactria. 204 The inscriptions of Klearchos Inside the walls of Aï Khanoum in the pronaos of the funerary monument of Kineas, who seems to have been the founder of the city, the base of a stele was discovered in situ with two texts of the third century bc.205 One in cursive script describes in two elegiac couplets how Klearchos had erected in the temenos of Kineas a transcript of the precepts at Delphi. He had gone to copy them carefully

the iron age, achaemenid and hellenistic periods

at Delphi himself. To the right in a different hand is the text of five maxims: ‘As a child be moderate, as a young man be self-controlled, in middle age be just, as an old man of good counsel and at death without regret’ – an exhortation to acquire the chief qualities of man appropriate to each age of life. The importance of this philosophy to the city is seen from the location, the temenos of its founder. Robert comments aptly on the impressive fidelity to the most authentic form of Hellenism represented by the wisdom of Delphi and on the community of race, language and culture fostered by these remote Greek colonists in a strange environment. The dedications in the gymnasium In the north part of the lower town a dedication to Hermes and Hercules, protectors of the gymnasium,206 was discovered in the wall enclosing a large court, the centre of a traditional Greek establishment for physical and intellectual education. The dedication was made by two brothers, Triballos and Strato, both sons of Strato. Triballos is the name of a tribe in northern Thrace, suggesting that the family had had some link with the Macedonian army. Finds from the necropolis The chance find of a fragmentary Greek inscription of funerary character led to the excavation in 1971 of a Greek mausoleum in the necropolis outside the walls. In its north-west vault were three funerary jars inscribed in ink with the names of the deceased whose remains they contained207 – a small boy and small girl, Lysanias and Isidora, and Kosmas. Here again we have important evidence for the ethnic character of the population. Lysanias is a Macedonian name and Isidora is a theophoric name – interesting evidence for the worship of Isis in this remote north-east of the Greek world. On the approach to the mausoleum were discovered fragments of two funerary inscriptions. One was part of a stele containing the words ‘Kings’ – presumably the tombstone of some dignitary in the royal administration. Economic and other documents A series of storage vessels, found mainly in the palace treasury, contained brief inscriptions in ink of the names of city officials and texts pertaining to the contents deposited there. They relate to the contents of the vessels and their value, presumably a form of taxation. These were initially interpreted by Claude Rapin as records of the city’s chief officials just prior to the abandonment of the city.208 A later study by Jeffrey Lerner interprets them as records of the depositors rather than city officials.209 There was also a fragment on parchment concerning tragedy and another fragment on papyrus concerning philosophy from the treasury.210

325

326

THE ARCHAEOLOGY OF AFGHANISTAN

The Aramaic ostracon Evidence for the coexistence of a more local culture is to be seen in the ostracon in Aramaic script discovered in the sanctuary of the temple of indented niches in 1970.211 It contains a number of Iranian names, but the absence of any syntax makes it difficult to decide whether it is Aramaic or middle Persian. The central text is an economic document recording the state of a series of payments of the type known from the archives of Nysa. It is probably not so much Aramaic official but the local Iranian language – Bactrian transliterated into Aramaic script. Miscellaneous Greek Documents from Northern Afghanistan The first Greek inscription (other than those on coins) from Bactria was the potsherd with atpoc discovered by Schlumberger at Tepe Nimlik 35 km west of Balkh in 1946, and a number of other brief texts on potsherds have been recovered from excavations at Emshi Tepe and Jiga Tepe since.212 Several Greek documents written on skin have recently appeared, originating from northern Afghanistan. In the early 1990s a leather document was found from a cave in the Sangcharak District in Sar-i Pul Province and is now in the Ashmolean Museum. It is a third century bc tax receipt from a place named Asangorna by Menodotus, presumably a government official, from payment by a certain Dataes during the reign of Antimachus and his (presumed) two sons Eumenes and Antimachus.213 Two further texts dating from the mid-third century bc were discovered from Yusufdarra near Balkh and are now in a private collection, one of which also dates from the reign of King Antimachus and refers to the employment of Scythian mercenaries in the (unlocated) city or town of Amphipolis. The second records a payment to a certain Archises.214 Kuliab Although not in Afghanistan – Kuliab is in Tajikistan about fifty kilometres north of Aï Khanoum – the inscription of Heliodotos purportedly from there can also be discussed with this group. It appeared on the antiquities market in the early 2000s and is only known from photographs, but, like the hoard also supposedly from the Kuliab region (discussed below), it bears many similarities with the Aï Khanoum inscriptions. It is a temple dedication to Hestia dated c. 200–195 bc by Heliodotos in the name of ‘Euthydemos, greatest of all kings, and his outstanding son Demetrios’. Both the inscription and the hoard have major implications for the extension of Greek influence from Aï Khanoum.215 Early Kharoshthi Inscriptions Bühler216 has shown that Kharoshthi letters are derived from Aramaic and that its alphabet was elaborated with the help of Brahmi, although this view has since been

the iron age, achaemenid and hellenistic periods

modified.217 Kharoshthi was used in the versions of Ashoka’s edicts at Mansehra and Shahbazgarhi in north Pakistan, while Aramaic was used to transliterate Middle Indian (Prakrit) in the Laghman valley and at Kandahar in Afghanistan. It was widely used with Greek on the coins of the Graeco-Bactrians struck south of the Hindu Kush; but most of the earlier Kharoshthi inscriptions belong to the period of the Indo-Scythian empire of Maues and the dynasty of Azes, who marked the establishment of their empire on the Indus with a new era – the Old Saka era. This is probably the era beginning 58/7 bc, although 46 bc, 84/3 bc and 88 bc have also been suggested.218 The inscription of Tiravharna the Satrap The earliest Kharoshthi inscription from Afghanistan seems to be that of the Satrap Tiravharna in year 83, discovered while digging an irrigation channel near Jalalabad.219 It has early letter forms (Fig. 5.52), and belongs to an Indo-Scythian satrap, apparently in ad 25, that is, during the reign of Azes II as King of Kings. This inscription provides proof of the western extension of Indo-Scythian rule to Jalalabad at this time – reinforcing the evidence from some hoards of copper coins of Azes II discovered in the locality.

Figure 5.52  The Kharoshthi inscription of Tiravharna from the neighbourhood of Jalalabad, now in the National Museum.

327

328

THE ARCHAEOLOGY OF AFGHANISTAN

The Bimaran vase The other early Kharoshthi inscription from Afghanistan was also found near Jalalabad. When Masson excavated the relic chamber of Stupa No. 2 at Bimaran, 12 km west-north-west of Jalalabad, he recovered a steatite vase containing pearls, beads, a gold casket for relics (discussed further in Chapter 5) and four billon coins of Azes with the tamgha of Kujula (Fig. 6.75). The vase has two Kharosththi inscriptions, one on the lid and one round the body of the vase relating to the relics.220 To judge by the coins, the relics were deposited soon after the collapse of Azes II’s empire, after the first invasion of Kujula, and before the Indo-Parthian conquests, and the inscriptions will then belong to this period of transition. Numismatics Evidence for Currency and Circulation With limited evidence in other forms, the plentiful coinage which circulated in Bactria during this period must be studied almost in isolation. Evidence for the coinage current in different parts of Afghanistan during the Achaemenid and Greek periods comes from a variety of sources. We must distinguish isolated coin finds, that is, coins that have been accidentally lost and so indicate the relative frequency of coins current in that locality at the time of their loss, from hoards, that is, groups of coins that have been collected by someone in antiquity, deliberately concealed and for some reason not recovered. Some hoards may have been collected over a period of time and are evidence for the currency of the locality over a period rather than for current circulation at the time the hoard was secreted. This will only become apparent on studying the composition of the find. Much hoard material unfortunately does not come from controlled excavations, and information about the discovery is often incomplete and only some of the coins originally discovered may be available for study. The Achaemenids Role of the siglos The official currency of the Achaemenid satrapies in Afghanistan consisted, as one might expect, of the royal Achaemenid sigloi, but as in other parts of the Achaemenid empire the royal silver sigloi are very heavily outnumbered by Greek coins and their copies, some of a much earlier date from Athens and other Greek cities. Schlumberger has shown that prior to the Macedonian conquest the circulation of silver throughout the Achaemenid empire depended heavily on Greek imports and was normally accepted as bullion, not at its nominal value.221 The traditional

the iron age, achaemenid and hellenistic periods

Achaemenid relation between gold and silver was 1:13, and this was the basis of the standard of the gold daric, silver siglos and their official exchange. In Greece the relative value of gold to silver was 1:12 and then 1:10 after Philip opened the gold mines at Pangaeus in 358/7 bc. This is a strong economic reason why older Greek coins were preferred to the overvalued Achaemenid siglos, until Alexander and the Seleucids reformed the currency. Although recent evidence from the second Mir Zakah hoard (see below) has suggested that there may have been an Achaemenid mint at Babylon, there is no evidence of the production of darics or sigloi in Bactria.222 Indeed, the evidence for the circulation of darics and sigloi in Central Asia is extremely limited.223 The Oxus Treasure The Treasure of the Oxus, containing rich Achaemenid objects of gold and jewellery, reached Europe with some 1,500 coins ranging from the early fifth century to around 200 bc.224 The coins, although their identification is not entirely secure, probably fall into two distinct periods – that of the Achaemenid empire and that of the kingdoms of Alexander the Great and his successors. Schlumberger has studied the pre-Hellenistic coins of the Oxus Treasure that remain in the British Museum and has shown that, in containing old Greek silver and its imitations from Athens, Acanthus, Byzantium, etc., it follows the pattern of other Achaemenid hoards. Among the silver hoards of Achaemenid date known to him from all parts of the Empire, five contained Achaemenid sigloi only, ten contained both sigloi and Greek silver and more than 44 contained Greek silver but no sigloi. The Miho Treasure In the early 1990s, a similar treasure from Afghanistan came to light and reached the Miho Museum in Japan; it has been suggested that it is a second part of the Oxus Treasure.225 It is completely unprovenanced, but seems likely to be a separate hoard. It comprised about 1,200 gold and silver objects, mostly exhibiting Achaemenid style, including many human and animal figures as well as jewellery, seals, gems and other precious objects. The treasure also included around sixty Achaemenid, Seleucid, Graeco-Bactrian and Indo-Scythian coins ranging from the fourth century bc to the first century ad.226 The al-Sabah collection There is a large collection of Achaemenid to Indo-Scythian treasures in the alSabah Museum in Kuwait. It mainly comprises precious metalwork, including human and animal figures, bowls, drinking vessels, jewellery, coins and other precious objects. All are unprovenanced, but most have been traced stylistically to Bactria or to Afghanistan broadly. They are probably not from a single hoard

329

330

THE ARCHAEOLOGY OF AFGHANISTAN

(although some might be); it is possible that some might be from the second Mir Zakah discovery (discussed below).227 The 1966 Balkh hoard The hoard of more than 170 Greek coins found in a pot in the neighbourhood of Balkh in 1966 conforms to a similar pattern as the coins found with the Oxus Treasure.228 It contained 150 Athenian tetradrachms and coins from some thirteen other Greek cities that suggest a burial date c. 380 bc – but no sigloi. The Chaman-i-Hauzuri hoard The Chaman-i-Hauzuri hoard from Kabul was discovered in 1933 by workmen digging foundations for a house.229 Of the 1,000 or so coins it contained, Schlumberger was able to recover and publish 115. The striking feature of the hoard is the large number of Greek silver coins: 34 from Athens and 30 from other Greek states, and the small number of Achaemenid coins, merely eight sigloi – reflecting the recurrent pattern in Achaemenid hoards throughout the empire. Most of the Greek coins can be dated fairly closely, and there is no reason to place any of them later than 400 bc, although one particular imitation may be derived from an Athenian prototype of 394/3. This, with the absence of Macedonian and Hellenistic coins, suggests a burial date in the mid-fourth century bc. The remaining silver pieces are bent bar silver coins – mostly of about 11.7 g – and local silver punched coins of a completely new type with a full denomination ranging from 9 to 12 g. Bent bar coins provided the silver currency of Taxila and the Achaemenid satrapies of north-west India before the invasion of Alexander of Macedon. Examples are recorded from Charsada and the Bhir Mound excavations at Taxila in Pakistan. Two hoards containing them have been reported from Bhir Mound, and a third from Bajaur, also in Pakistan. From Afghanistan, in the first deposit from Mir Zakah near Gardez, there were 50 bent bar silver coins and 550 of the round (sometimes scyphate) single-type silver that seemed to constitute the three-quarters and one-eighth denomination of the series. Other hoards of bent bar coins Two hoards of bent bar silver coins are known to have been found in recent years in Jalalabad province. Some 50 bent bar silver coins were found at Khugjani near Jalalabad in 1962; and a second hoard of some 100 bent bar coins, all of the same type, was found on the outskirts of Jalalabad by workmen digging the foundations for a building in 1970.230 Bent bar coins are double the weight of the Achaemenid royal siglos; chronologically they belong to the fourth century bc, the last century of Achaemenid rule; and their presence at Kabul, Gardez and Jalalabad, as well as at Taxila and Charsada, suggests that they were made for currency in the eastern-

the iron age, achaemenid and hellenistic periods

most Achaemenid satrapies. The new-type silver punched coins in the Chaman hoard are very puzzling. They have the thick roundish fabric of Greek coins – not the flat square or oblong flans of Indian currency. Their general appearance, incuse punch, concave reverse and convex obverse again seem Greek although there is little Greek about their types. Their full denomination ranges in weight from 9 to 12 g and they are less worn than the bent bar coins in the hoard. We should therefore regard them as a local product subsequent in date to the long bent bar variety in the Chaman hoard, but owing more in fabric and appearance to the worn Greek silver present in the hoard. Finally, the Chaman hoard231 contains fragments of silver bracelets and jewellery – one with two cuneiform characters that are Elamite letters. We should therefore regard the Chaman hoard as a treasure valued for its bullion like other Achaemenid hoards but reflecting also the silver readily obtainable in south-east Afghanistan at the time. Changes Due to Alexander Alexander’s currency reform The conquests of Alexander led to a major change in the pattern of the currency of the near east. In a major currency reform, he introduced throughout the empire a new coinage based on the realistic value of gold to silver of 1 to 10 that Philip II had introduced to Macedon instead of the archaic ratio of 1 to 13 retained by the Achaemenids from an earlier period. He struck a gold stater of Attic weight and a silver drachm on the Attic standard that was tariffed at 20 drachms per gold stater. His purpose was clearly to establish the use of imperial coined money (not bullion) through the empire, and the weight standard was well-chosen in view of the wide popularity of Athenian silver coin as bullion through the Persian empire. The success of this policy in the far east of the empire is questionable, given extremely limited evidence of the circulation of Alexander’s coins in the region. The new imperial currency rapidly became the standard and was followed by the successor state of the Seleucids and later the Bactrian Greeks. In the numerous hoards of Hellenistic date, coins of the period prior to Alexander are hardly ever found. When the state issued a plentiful currency and gave a realistic value to its own silver coinage, avoiding the overvaluation that had developed under the later Achaemenids, there was no occasion to prefer old coins traded as bullion. Later element of the Oxus Treasure The later stratum of the coins of the Oxus Treasure reflect this phenomenon, with around a hundred tetradrachms and a hundred drachms in the name of Alexander the Great, probably struck posthumously, followed by coins of Alexander’s successors, Seleucus I, Antiochus I and Antiochus II among the Seleucid kings of Syria and Diodotus I among the first independent rulers of Bactria.

331

332

THE ARCHAEOLOGY OF AFGHANISTAN

After Alexander Although coins of Alexander are rarely found in Bactria, and were certainly not produced in the region, a number of examples in the name of Alexander, struck posthumously by his successors, were found in a hoard discovered at Aï Khanoum in 1973 (see below). Many of the coins in circulation in Bactria after the death of Alexander seem to have been imitations of Athenian ‘owl’ tetradrachms. In regions unused to coinage and with no central minting authority, the imitations were popular since they copied the types of Athenian coinage, the most prolific currency before Alexander. The 1990 ‘Afghanistan’ hoard An important group of 65 of these imitation coins appeared on the market in 1990 with a general provenance of Afghanistan only.232 Some of these coins copied the well-known types of earlier Athenian coinage: a helmeted head of Athena on the obverse and an owl on the reverse with the legend ΑΘΕ. Others retained Athena on the obverse, while the reverse showed an eagle with no legend. The publishers of the hoard date these coins to the period after the death of Alexander, as part of an irregular issue produced by a local ruler. Alongside the coins from the 1990 hoard a particularly enigmatic series should also be considered. These coins have an obverse helmeted male head, while on the reverse is a cock with the legend ΣΩΦΥΤΟΥ. It has been assumed that the head on the obverse is that of a certain ‘Sophytos’, a ruler with a non-Greek name for whom there is no literary evidence (Fig. 5.53). The acrostic inscription discussed above provides a tantalising possibility of identification, but any connection between the numismatic and epigraphic evidence should be treated with great caution in this case. Although it is very difficult to establish details about Sophytos with any

Figure 5.53  Sophytos silver coin of unknown denomination. Diameter 16.5 mm.

the iron age, achaemenid and hellenistic periods

certainty, it seems likely that the coins were struck under his authority in Bactria (none is known from south of the Hindu Kush) at some point after the death of Alexander and the later establishment of mints producing Seleucid coins in the region under Antiochus I. Under the Seleucids, production of coins in Bactria seems to have been centred on one or two mints. The assimilation of these mints with geographic locations has been the subject of much numismatic debate. It is likely that the majority of Seleucid coins produced in Bactria were minted either at Aï Khanoum or at Bactra.233 Coins were produced under Antiochus I and Antiochus II in the usual pattern of Hellenistic royal coinage, with a portrait of the king on the obverse and a divinity on the reverse along with the king’s name as the legend. Mauryan and Graeco-Bactrian Coinage234 The overall pattern For the Mauryan and Graeco-Bactrian periods the pattern of currency reflects the successive stages of political suzerainty. Gandhara, Arachosia and the Paropamisus were initially Mauryan provinces, and used the Mauryan silver and copper punchmarked coinage. North of the Hindu Kush mountains in Bactria, we find first a Seleucid then a Bactrian currency following the Attic weight standard with fine Greek portraits and reverse types using Greek legends only (Figs 5.54–5.57). When the former Mauryan provinces were captured by the Bactrian Greeks, we see a new bilingual Indo-Greek coinage with legends in Greek on the obverse and Kharoshthi on the reverse (Fig. 5.58), struck to a new reduced Indian weight standard, as well as square copper coins (Fig. 5.59).

Figure 5.54  Gold stater of the Graeco-Bactrian king Diodotus (note test cut). Diameter 19 mm.

333

334

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 5.55  Silver Attic tetradrachm of the Graeco-Bactrian king Antimachus (early second century bc), with reverse type of Poseidon holding a trident. Diameter 30 mm.

Figure 5.56  Square bronze coin of the Graeco-Bactrian king Antimachus (early second century bc), with the figure of an elephant on one side and a thunderbolt on the reverse. 18 × 22 mm.

Figure 5.57  Silver Attic standard tetradrachm of Eucratides I. Diameter 35 mm.

the iron age, achaemenid and hellenistic periods

Figure 5.58  Square silver drachm of reduced Indian weight struck by Apollodotus I, with an elephant and Greek legend on the obverse and a humped bull and Kharoshthi legend on the reverse. 15 × 14.5 mm.

Figure 5.59  Square copper bilingual coin of Pantaleon, with non-Greek female figure on the obverse and the figure of a leopard in an incuse square on the reverse. 28 × 23 mm.

Mir Zakah I The first Mir Zakah Treasure,235 consisting of more than 11,000 Indian, GraecoBactrian, Saka and later coins, was discovered in 1947 in a village 53 km north-east of Gardez on one of the old routes linking Ghazni and northern Arachosia with Gandhara. There were 50 bent bar coins and 563 round scyphate and minuscule punched coins of ancient India, 4,820 punch-marked silver coins, 2,012 GraecoBactrian drachms (compared with six tetradrachms) and 3,335 Saka drachms (against 13 tetradrachms). The find included much smaller numbers of copper coins of these periods and of the Indo-Parthians and Kushans. The French excavations of 1948236 established that the place of discovery had been two sacred tanks or basins, into which offerings, notably coins, had been thrown. This explained the presence of a variety of votive offerings and items of jewellery, the enormous

335

336

THE ARCHAEOLOGY OF AFGHANISTAN

chronological range of the coins extending over five centuries, the excellent state of preservation of some of the oldest coins present in the hoard and the very heavy predominance of smaller silver denominations. The coins from this excavation showed the same general pattern as the coins recovered the previous year for the National Museum. The treasure is therefore a deposit, not a currency hoard, and provides extremely important evidence for monetary circulation in Gardez and its locality over the whole period. We see the substantial role played by the punchmarked silver (the silver currency of the provinces of the Mauryan empire) and its replacement by the bilingual Greek and Kharoshthi silver drachms struck on the Indian standard when the Graeco-Bactrian kings conquered the former Mauryan provinces. This in turn was replaced by the bilingual Greek and Kharoshthi Saka coinage of Azes I, Azilises and Azes II, as at Taxila. On the other hand, the silver drachms of the Parthian and early Indo-Parthians are conspicuously absent, with only three examples in all. Copper coins are less numerous than the silver, but may be an even better guide to political suzerainty in the area as their circulation was much more restricted than silver. There were 78 Mauryan copper coins of the square Taxila type in the deposit and 54 bilingual square Graeco-Bactrian coins, compared with three bilingual round copper coins, and three Greek monolingual coppers from Bactria. The coppers of Apollodotus I (22) are most heavily represented, followed by Pantaleon (9) and Eucratides (10). Mir Zakah II In 1992, a second deposit of coins was discovered at Mir Zakah. This time the size of the hoard was enormous with reports of 4 tonnes of coined metal, about 550,000 silver and bronze coins, as well as 120 kg of objects in gold. The hoard was never fully investigated and the only information about the composition comes from the work of Osmund Bopearachchi, who was able to examine rapidly a mere 300 kg (approximately 38,000 coins) in the bazaar of Peshawar in 1994.237 Although many of the items from the hoard were said to have shared a particular patina due to the wet conditions of their deposition, it is very difficult to know whether coins that appeared on the market were initially part of the find. It is certain, however, that the number of coins that have been offered for sale for the first time in recent years has increased dramatically, and some numismatists have suggested that this is the result of a combination of hoards rather than a single enormous find.238 The composition of the second Mir Zakah deposit, so far as it can be determined, was similar to that of the first hoard. The earliest coins were said to be Achaemenid darics and ‘bent bar’ issues, while Graeco-Bactrian and Indo-Greek coins were heavily represented, but more than 40% of the hoard apparently consisted of coins of the Indo-Scythian king Azes II and posthumous imitations of Hermaeus. The hoard included a wide variety of objects covering some four hundred years, including many Achaemenid and Hellenistic gold and silver vessels, statuettes, jewellery and gold strips. Vessels included a gold censer and a silver bowl with

the iron age, achaemenid and hellenistic periods

a hippocampus impressed on its base. Sculptures included Zoroastrian priests, figurines and Hellenistic intaglios, including some depicting Hermes and Athena. Bopearachchi further writes: The most sensational numismatic discovery was a coin of Nasten, a hitherto unknown Iranian ruler in India. On the obverse, within a bead-and-reel border, the coin carries a bust of the diademed king to right wearing a helmet with a long, flowing crest and a mantle. The reverse shows the king on a prancing horse riding to the right. He wears a helmet with a long, flowing crest. The Greek legend reads Nastenes/Xatrannou, ‘Nasten, son of Xatran’. Judging by his name, Nasten was presumably not a Greek, but an Iranian, probably a Bactrian Iranian.239 The hoard was not associated with a city or temple, so is presumed to have been plunder deposited in a temporary hiding place. The hoard had been subject to damp, so much of the silver that Bopearachchi examined was heavily oxidised. The eventual destination of the treasure remains unknown (although some might have reached the al-Sabah collection in Kuwait; as late as 2010 there were still rumours of several more tons from Mir Zakah having reached Switzerland and awaiting sale).240 Finds from Begram During his residence in Afghanistan, Charles Masson discovered that large numbers of coins were constantly being found on the plain of Begram near the confluence of the Ghorband and Panjshir Rivers in the Kohistan 60 km north of Kabul. During 1833 he purchased 1879 ancient coins, mostly coppers, and in the following four years he collected many more. He described his 1833 finds in some detail241 and gave an enumeration of the total collected from Begram in 1833, 1834 and 1835.242 This important evidence is reinforced by the much smaller number of coins from the French excavations at Begram in 1941, 1942 and 1946.243 These excavated coins show the same basic pattern as the Masson surface finds. We can therefore check details from the excavated coins and use their evidence in conjunction with the statistical evidence of the Masson finds. More than 80% of the Graeco-Bactrian copper coins from the French excavations prior to the collapse of the silver denominations under Hermaeus are square, bilingual Greek/Kharoshthi copper coins. Among Masson’s finds there are some six hundred Graeco-Bactrian copper coins of this period. The denomination of the 78 Mauryan square copper coins is copied by the 43 square coppers of Agathocles and seven of Pantaleon, and the later stages of the same square denomination is seen in the 268 square coins of Eucratides, 73 of Apollodotus I and 153 of Menander, followed by the 37 of Antialcidas and 14 of Lysias. We see how the square copper bilingual copper denomination in this period was initially derived from the Mauryan and then progressively modified.

337

338

THE ARCHAEOLOGY OF AFGHANISTAN

The silver currency of the Parapamisus is seen from the hoard of 120 GraecoBactrian silver drachms of the Indian standard purchased at Charikar.244 It consisted of seven square coins of Apollodotus I, 5 round coins of Antimachus and 108 of Menander – similar in composition to the 1926 Gang hoard and the 1942 Bajaur hoard – both from modern Pakistan. The 1942 Bajaur hoard is remarkable for the presence of some 700 bent bar and punch-marked coins with 800 Indian drachms of Apollodotus I, Antimachus and Menander – a treasure in which Indian standard drachms are hoarded with the Mauryan punch-marked silver, from which their metrology is eventually derived. A later stage in the silver currency of the Upper Kabul valley is to be seen in the 1923 hoard of 97 Graeco-Bactrian drachms of the Indian standard of later kings down to Hermaeus.245 Excavated coins from Aï Khanoum The discovery of two non-struck coin flans of Seleucid or early Bactrian Greek fabric in the French excavations at Aï Khanoum in 1968246 suggests that there may have been a local mint for copper coins – either official or unofficial – in the city and adds weight to the identification of Aï Khanoum as an important mint under the Seleucids.247 The excavated coins provide clear evidence for the currency of eastern Bactria.248 The copper coins of the Seleucid kings Seleucus and Antiochus I (37%) are succeeded by coins of the independent Graeco-Bactrian kings Diodotus I and II (14%), Euthydemus I (26.5%) and Eucratides I (6.5%), mostly monolingual Greek coins of the Attic/Seleucid system, but with occasional bilingual Greek/ Kharoshthi square copper coins of Eucratides found. There is evidence of trade with provinces south of the Hindu Kush in the two punch-marked Mauryan silver coins found (in addition to the Graeco-Bactrian Attic standard silver coins one would expect), in the 1970 hoard of Mauryan punch-marked silver and in the presence of three bilingual tetradrachms among the 63 silver coins in the 1973 hoard (see below). The first hoard from Aï Khanoum, consisting of 679 Indian and Indo-Greek silver coins found in the 1970 excavations, has been fully published by Audouin and Bernard.249 The hoard was discovered in a traveller’s water flask buried in room 20 of the so-called administrative quarter. It had been hurriedly buried by the latest inhabitants of Aï Khanoum at the time of an invasion. Punch-marked silver coins and the Indian standard bilingual Indo-Greek silver drachms were the currency not of Bactria but of the provinces of the Mauryan and former Mauryan empire. There is no doubt, therefore, that this treasure represented an import by way of trade from Gandhara. The six silver coins of Agathocles were of a unique type and are bilingual – with a figure perhaps to be identified as the god SamkarshanaBalarama and Agathocles’ name in Greek on the obverse and an image perhaps of Vasudeva Krishna and a Brahmi legend on the reverse.250 Whatever the correct identification of the deities, the shift away from Greek iconography is clear. Their square shape and method of manufacture is closely modelled on the square

the iron age, achaemenid and hellenistic periods

punch-marked silver with which they were found. The weights of the coins (four of which are struck from the same obverse and reverse dies) range from 2.3 to 3.3 g, the same broad weight range as the punch-marked silver coins in the hoard. In this hoard we see the same process and the origin of the Indo-Greek bilingual drachm, perhaps copying the denomination of the Mauryan punch-marked silver that the Graeco-Bactrians found current in their new territories south of the Hindu Kush. Apollodotus I, while retaining the square shape of the bilingual silver Indian drachm, subsequently stabilised its weight at the bottom of this weight range at about 2.45 g, and this was acceptable for parity of value with the punch-marked silver because of the consistently high silver content of the Indo-Greek coins. Subsequent kings retained this standard for the bilingual drachms but reverted to the round silver flan normal for Attic standard issues. The 1973 Aï Khanoum hoard, also found during the French excavations, contained seven tetradrachms in the name of Alexander and seven Seleucid tetradrachms with 49 Attic tetradrachms of the Graeco-Bactrian kings who ruled north of the Hindu Kush up to Eucratides (but excluding Heliocles).251 The hoard included a hitherto unknown Attic tetradrachm of Apollodotus I. It has been suggested that the hoard was buried around the time Eucratides I took control of the city from Antimachus I.252 A third hoard was discovered north of Aï Khanoum in the spring of 1974 and made its way to the bazaar of Kabul, from where it was sold onto the international coin market. As it passed through New York in 1975, the hoard was examined by Nancy Waggoner and was later published by Frank Holt. The hoard was reconstructed from trade, a process that was accurate thanks to the initial examination.253 Holt was able to identify 139 coins that had originally been part of the find. Of these, 81 were coins of Euthydemus I, with smaller numbers of issues of kings down to Eucratides II. The increased number of coins of Eucratides I, in comparison with the excavated hoard, may suggest that the 1974 assemblage was buried somewhat later than the second Aï Khanoum hoard, the inclusion of Eucratides II providing a terminus post quem. A final hoard from Aï Khanoum (often referred to as Aï Khanoum IV) was apparently discovered in the mid-1990s as the product of the systematic clandestine excavations that now cover the site. The find was said to have consisted of 1,500 coins, but our knowledge is limited to only 500 examples, the information again coming from Osmund Bopearachchi.254 The composition was apparently similar to that of the second and third hoards to have been discovered at Aï Khanoum, although the final coins in this hoard were those of Eucratides I. The absence of the issues of later kings led Bopearachchi to suggest that the hoard was buried around the time of the destruction of the city, which he dated, following Bernard, to the end of Eucratides’ reign around 145 bc.

339

340

THE ARCHAEOLOGY OF AFGHANISTAN

The Kuliab hoard Clandestine excavations in the 1990s were also the source of a hoard of coins from the region of Kuliab in Tajikistan. Once more, the work of Bopearachchi was essential in publishing details of the find, but of the reported 800 original coins in the hoard only 205 have been published.255 The composition of the hoard was said to be similar to that of the second, third and fourth hoards from Aï Khanoum: the earliest coins in the hoard were those in the name of Alexander, while Eucratides I examples seem to have been the latest. Once again the terminus of Eucratides was important, suggesting that the hoard was buried late in his reign or soon after his death. The Qunduz hoard The ‘Qunduz hoard’ was discovered during 1946 in excavating the foundations for an extension of the barracks at Khisht Tepe on the south bank of the River Oxus 90 km west-north-west of Qunduz, on the ancient caravan route crossing the river to northern Bactria and Sogdiana.256 It contained three Seleucid tetradrachms of Seleucus I, Alexander Hierax and Antiochus I Balas (150–145 bc), five magnificent double decadrachms of Amyntas, 17 drachms of Heliocles and 602 tetradrachms of other Graeco-Bactrian kings. All the coins were struck on the Attic standard with legends in Greek only. Most of the tetradrachms were of Demetrius II (5), Eucratides I (144), Eucratides II (130) and Heliocles (204). The hoard contained Greek tetradrachms on the Attic weight standard of several Graeco-Bactrian kings previously known solely from their bilingual Greek/Kharoshthi currency on the Indian standard. The Qunduz hoard is important for the evidence it contains regarding the question of mint marks. Most Graeco-Bactrian coins have Greek monograms, usually composed of two or more Greek letters, in the reverse field that have been the subject of extensive discussion. Cunningham argued that they stood for the mint of issue, but Tarn maintained that they represented the initials of magistrates or mint masters on the Seleucid pattern. Fussman’s die study of the coins in the Qunduz hoard has shown that several coins struck from the same obverse die have different reverse monograms, so the monograms cannot be mint marks. He shows that several mint marks found on these Attic tetradrachms obviously intended for circulation in Bactria are identical with mint marks of the same kings found on their Indian bilingual tetradrachms intended for circulation south of the Hindu Kush. He suggests that they are therefore not the marks of moneyers, but of engravers who served more than one mint. Further die links between coins with different reverse monograms have since been discovered. The debate concerning the correct interpretations of the monograms will doubtless continue for some time, and it is highly likely that the meaning of these symbols will never be resolved satisfactorily. What is certain, however, is that they played some role

the iron age, achaemenid and hellenistic periods

in the production process, whether geographical or chronological. It is most likely that their purpose was to allow some sort of quality control over the minting process. The latest Seleucid coin in the hoard was that of Antiochus Balas (ruled 150–145 bc), but many of the Graeco-Bactrian coins should be dated even later. One of the latest is probably the tetradrachm of Hermaeus. The date of the hoard’s concealment can probably be placed towards the date of the invasion of Bactria south of the Oxus, c. 100 bc. It was under Hermaeus that the Graeco-Bactrian silver currency in the Kabul valley collapsed and was replaced by a copper currency retaining the types that had been used for the silver. The rich silver mines at Panjshir in the Hindu Kush mountains had no doubt supplied silver for the Graeco-Bactrian coinage. This further analysis of the Qunduz hoard suggests that some of the later Graeco-Bactrian kings may have struck at a common mint both on the Attic standard for their remaining territories north of the Hindu Kush and on the Indian standard for the Paropamisus; also, the final loss of the northern territories under Hermaeus made it impossible to continue working the Hindu Kush silver mines and so led to the debasement of the southern coinage. The evidence from the hoards is one that seems to be strongly biased towards the end of the Graeco-Bactrian kingdom. Hoards are, however, only one part of the ways in which these coins can be studied. There is potential in iconographic studies. Metal analysis holds many intriguing possibilities; in particular, with the ability to analyse trace elements accurately, it may be possible to begin to link monograms to geographic locations. Such an approach is unfortunately some way off and will be expensive. The future of Graeco-Bactrian and Indo-Greek numismatics is currently the die study. By learning more about the way in which the coins were produced and their striking organised, and about the overall size of the output, we will finally have some objective data on which to base historical reconstructions. Work by Sergei Kovalenko, Frank Holt and Brian Kritt has been based on a sound analysis and reconstruction of the system of production of the coins of the Diodotids.257 In the meantime, the large numbers of coins that have appeared for sale in recent years make the likelihood of obtaining a representative sample for further studies highly likely, and it is to be hoped that this approach will begin to bear further fruit soon. The currency of western Afghanistan The currency of western Afghanistan during this period can be reconstructed from the evidence of the Tate collection of ancient coins found in Sistan258 and from the local museum at Herat. It consisted at first of Seleucid coins then of GraecoBactrian coins with Greek legends struck on the Attic weight standard. Like the currency of Carmania259 it included some local copies – possibly coins of the Sakas. This contrasts, as one would expect, with the currency pattern of Kandahar, where the local museum and finds from the British excavations suggest that the Mauryan

341

342

THE ARCHAEOLOGY OF AFGHANISTAN

coinage was replaced by Indo-Greek bilingual Greek and Kharoshthi coins on the Indian standard.260 From the first century bc we find Parthian silver drachms at Herat and Kandahar as at Mazar (sometimes countermarked by later rulers), for example with the tamgha of Gondophares’ dynasty by the Indo-Parthians or the small helmeted head like the figure to be seen on the ordinary coins of Sapaleizes. Yuezhi, Sakas and Indo-Parthians261 The Yuezhi currency of Bactria The currency of Bactria after the Yuezhi conquest in the first century bc is far from clear. When the nomads issued coins they copied the denomination and types of preceding Graeco-Bactrian kings, though usually in a debased form. The commonest coins of the Yuezhi in northern Bactria during the period were the copper tetradrachms and drachms of the barbarous Heliocles which occur in stratified finds immediately prior to those of Soter Megas and the early Kushans.262 These copper coins are also found in northern Afghanistan. The Kushan Heraeus, who seems to have ruled in Bactria at an early date, issued tetradrachms and obols in base silver following the denominational pattern of Eucratides and the Sakas who succeeded him. The later stages of the Kushan currency are discussed in the next chapter. Coins of the Azes dynasty and Su Hermaeus Silver coins of Azes I are found at Mir Zakah and in the hoard from Chaman between Kandahar and Quetta.263 The Saka empire at this period was centred on the Indus provinces of Pakistan, but during the reign of Azes II they extended their rule to some parts of eastern Afghanistan. A hoard of copper coins of Azes II is reported from a stupa near Jalalabad264 and a further hoard from Jalalabad was acquired by an Afghan collector in 1970. Silver coins of Azes II are very common at Mir Zakah and a few are reported from other localities, but not from Begram. Masson explicitly states that he discovered no moneys of the genuine Azes kings at Begram, and his experience is borne out by their extreme rarity among the Begram excavation coins. The currency of Begram and Kandahar at this period consisted of the long series of copper tetradrachms and drachms, copying in copper the obverse and reverse types of the earlier silver coins of the last Graeco-Bactrian ruler Hermaeus. It is a long series that gives little clue about the identity of its issuers, but probably represents the currency of the Pahlavas who conquered Arachosia in the first century bc.265 Arachosia is included in Isidore of Charax’s list of Parthian provinces at the beginning of the Christian era. There was a major debasement of the Saka silver currency at the end of the reign of Azes II, when his empire began to break up. The old silver denominations

the iron age, achaemenid and hellenistic periods

were now struck in billon instead of silver and the copper denominations proper largely disappeared; but the principal impact of this was in the coinages of the Indus valley. The distribution of Indo-Parthian issues Around ad 30 the Indo-Parthian king Gondophares established an independent empire that controlled the Panjab, much of the Indus valley, Arachosia and adjacent parts of eastern Afghanistan. His currency in Arachosia consisted of Niketype copper tetradrachms, based on the copper tetradrachm struck in the name of Su Hermaeus, and these coins are found at Begram, Hadda and Mir Zakah. His successors lost these territories and the Indus valley to the rising power of the Kushans, but later copper tetradrachms in the Nike sequence are known from Sistan for Orthagnes, Pacores, a second Gondophares and Sanabares266 ruling territory in the west during the period of the main Kushan dynasty. The latest issue in this sequence has a Pahlavi legend which has been tentatively read as Arda Mitra. It has a Sasanian fire altar of the type used by Ardeshir I and seems to have been a provincial currency for Sistan at the time of his eastern campaigns. For several kings of the Indo-Parthian dynasty very rare silver drachms of Parthian type are known. The series eventually suffer a serious debasement, and later kings have debased drachms with the same types struck in copper – a coinage which is common in Merv and sometimes encountered in Herat.

343

CHAPTER 6

From the Kushans to the Shahis Warwick Ball, Olivier Bordeaux, David W. Mac Dowall, Nicholas Sims-Williams and Maurizio Taddei

Historical Background The Kushans1 The migration of the Yuezhi into Bactria (discussed in the previous chapter) marks a new stage in the archaeology of Afghanistan. The discoveries at Tillya Tepe have

Figure 6.1  Map to illustrate main sites discussed in Chapter 6. Inset 1: Fig. 6.67; Inset 2: Fig. 6.19; Inset 3: Fig. 6.2.

from the kushans to the shahis

been associated with these migrations. The Yuezhi have been identified with the Tocharoi (or Tocharians) of the classical sources, the name surviving as the region of Tukharistan, first mentioned on a silver dish from the time of Kanishka I2 and continuing as the name for eastern Bactria in the medieval Muslim sources (becoming the modern province of Takhar in north-eastern Afghanistan). At this stage according to Chinese sources the Yuezhi were divided into five Xihous or Yabghus, usually interpreted as regional principalities. Eventually, the dominant principality of the Guishuang attacked and destroyed the other four, towards the end of the first century bc. The Guishuang are almost certainly identified with the Kushans. The Chinese sources describe how the Kushan king Qiujiuque (Kujula Kadphises) invaded Anxi (Indo-Parthia), took control of Gaofu (Kabul) and destroyed Puda and Jipin (probably Paktia and Gandhara).3 The early Kushan king Kujula Kadphises is known from his coins to have ruled in several provinces, but seems to have lost control of most of the Indus valley to the Indo-Parthian kingdom of Gondophares and Abdagases in the first half of the first century ad. He was succeeded by the ‘Nameless King’, known from his titles on coins and inscriptions as the King of Kings, the great, the saviour (Soter Megas), who seems to have been the first Kushan ruler of the empire that stretched from Bactria across eastern Afghanistan and northern Pakistan to the upper Ganges in India and even briefly into the south-western Tarim Basin in Xinjiang. Since the discovery of the Rabatak Inscription, Soter Megas has been identified with Vima Takto.4 Vima Takto’s successor, Vima Kadphises, consolidated the empire, and was in turn succeeded by the three great Kushan kings, Kanishka I, Huvishka and Vasudeva, who between them ruled the Kushan empire for about a century (Table 6.1).5 Although he was not the first Kushan king, Kanishka introduced a new era. Its reference date has been disputed (see below), but the year of Kanishka’s accession is now widely agreed to be ad 127/8. In a series of bold campaigns he succeeded in enlarging the Kushan empire; his successor Huvishka lost the most distant provinces of Xinjiang and eastern India; but his successor Vasudeva still ruled a powerful kingdom which controlled major sections of important trade routes. When overland trade through Iran was interrupted by the Parthians, the Kushans were able to provide a safe route from Balkh through Kabul, Peshawar and the Indus valley to Broach on the Indian Ocean. From here sea traders would carry merchandise to Alexandria and the Roman Empire.6 During this period there is striking evidence of strong Graeco-Roman influence in the development of Gandharan sculpture and Buddhist art and in the Kushan coinages. Under the patronage of the Great Kushans, Buddhism spread westwards into Afghanistan and Central Asia. The Later Kushans, Sasanians and Kushano-Sasanians7 The history of the later Kushan empire remains obscure. A chronology and the relationships of the Kushans with adjacent and later dynasties has been expressed

345

346

THE ARCHAEOLOGY OF AFGHANISTAN

Table 6.1  Kushan king list King

1

Kujula Kadphises

2

Wima Takto

3

Wima Kadphises

4 5 6 7 8 9 10 11 12 13

Kanishka I Huvishka Vasudeva I Kanishka II Vasishka Kanishka III Vasudeva II Mahi Shaka Kipunadha

Dated inscriptions

ad

Azes era year 1 = 46 bc Kanishka era year 1 = ad 127 Azes era 122 Azes era 126 Azes era 136 Greek era 270 = Azes era 142 Greek era 279 = Azes era 151 Greek era 287 = Azes era 159 Greek era 299 = Azes era 171 Kanishka era 1–23 Kanishka era 25–60 Kanishka era 64–98 Kushan era [4–18] Kushan era 22–30 Kushan era 41

76 80 90 96 105 113 125 127–149 151–186 190–224 230–244 248–256 267

by Joe Cribb in Tables 6.2 and 6.3.8 In the third century, there seem to have been some minor kings named Kanishka II, Vasishka, a Kanishka III and possibly a Vasudeva II. The Kushan kingdom came into conflict with the powerful Sasanian Empire of Iran, which was established early in the third century ad and which seems to have conquered some western Kushan provinces. The Sasanians are traditionally viewed from a western viewpoint, focusing on their relationship with the Roman Empire and on Sasanian remains in western and southern Iran. However, newer discoveries in the east now reveal the Sasanians to have been at least as (or more) concerned with their eastern empire, and recent studies have emphasised the eastern viewpoint.9 Discoveries in Afghanistan include the Bactrian documents, probably the most comprehensive indigenous sources for the Sasanian period, the Ghulbiyan painting and the rock relief of Shapur at Rag-i Bibi, revealing Sasanian art forms far to the east of their traditional homeland. The conquest of the Kushan Empire was probably the greatest conquest of the Sasanian era, and their wars with the Huns probably the greatest wars ever fought by the Sasanians.10 The evidence from the east generally and Afghanistan in particular is increasingly altering our understanding of the Sasanian period as a whole. There is late evidence in Tabari for an eastern campaign by the Sasanian king Ardeshir I (212–241). The trilingual inscription of the Sasanian emperor Shapur I at Naqsh-i Rustam dated to 262 lists part of the Kushan empire among the eastern Sasanian provinces, and the Rag-i Bibi relief appears to depict Shapur I on campaign on the Indus; and Shapur II campaigned against the ‘Cuseni’ (i.e. Kushans) in 356/7. The beginning of the Kushano-Sasanian period is generally thought to be the date of Ardeshir, 230–245,11 but there are problems with the chronology, order

347

from the kushans to the shahis

Table 6.2  Approximate framework for rulers of Gandhara and adjacent region

bc/ad

Bactria

Begram/Kabul

Gandhara

Taxila

ad 10  20  30  40  50

Da Yuezhi

Indo-Scythians

Indo-Scythians

Kushans Kujula Kadphises

Kushans

Indo-Scythian Indo-Parthian Gondophares

Abdagases

Kushans

 60  70  80  90 100 110 120 130 140 150

Sasan

Wima Takto

Kushans

Wima Kadphises Kanishka I

Indo-Parthians Kushans

220 230

Kushanshahs

Kanishka II

240 250 260

?/ Ardashir Peroz I

Vasishka

270 280 290 300 310 320 330 340

Hormizd I

Kushanshahs

Hormizd II Peroz II Varahran Kidarite Hunsii Kirada/ Peroz/ Sasanian Kidaraiii Shapur II

Orolano Pidoko

Bimaran Casket (Cribb 2015)

year 318 (Konow 1929: 106–7) Kanishka coins (Cribb 1999) and reliquary (Errington 2002)

Huvishka

Vasudeva I

390 400 410 420 430 440 450 460 470 480

Satraps

i

160 170 180 190 200 210

350 360 370 380

Dated and datable Gandharan images

Ardashir II/ Shapur III Alchano Huns

year 384 (Konow 1929: 117–19) year 89 (Konow 1929: 171–2) year 399 (Konow 1929: 124–7) year 5 (Fussman 1974: 54–8; Harle 1974: 128) disputed by Kushans and Kushanshahs

Kanishka III/ VD II Vasudeva II Mahi Shaka

[year 89]

Kidarite Huns

Kipunadha Kidarite Huns

[year 5]

Alchano Huns

Alchano Huns

Tobazino Okilano(?) Sasanian Peroz Hephthalites

Notes: i Allied with the Apracarajas ii Initially with Kushanshah as puppet iii After initial period adopting title Kushanshah Source: Cribb 2018: 26–9, Table 6.

348

THE ARCHAEOLOGY OF AFGHANISTAN

Table 6.3  Kushan, Kushanshah and Sasanian links Kushan kings

K1 = ad 127

Kanishka

K1= ad 127 K 25 = ad 151 K 64 = Ardashir I ad 190 K [1]05 = Unknown king 224–240 ad 231 Ardashir

Huvishka Vasudeva I Kanishka II

Vasishka

Sasanian Sasanian Kushanshahs emperors and Kidarite Hun successors

K [1]20 = Peroz I ad 246

Kanishka III K [1]41 = Hormizd I Vasudeva II ad 268 Hormizd II

Shaka Kipunadha

Peroz II Varahran Kirada Peroz Kidara

Orolano Pidoko

Tobozino Okilano(?)

Source: Cribb 2018: 26–9, Table 5.

Numismatic links

Vasudeva sends embassy to China in ad 230

Issue of Kushan gold coins in Bactria stops. Sasanian imitations of Vasudeva I gold and Kanishka II coppers. Copper coins issued by unidentified Kushanshah showing goddess Anahita offering king Kushan crown. Copper coins issued by Ardashir Kushanshah inscribed in Bactrian, copying Kushan coins of Kanishka II. Shapur I Peroz I Kushanshah issues coins copying the imitation 240–270 Kushan coins and with new coin design based on Record of Ardashir’s throne/altar type. conquest Examples of both types overstruck on copper coins of ‘as far as Shapur I from Merv mint. Peshawar’ Gold coin shows Peroz being offered Kushan king by Kushan goddess Ardochsho in style of Kanishka II coins. Bahram II Early Vasudeva II copper coin overstruck on Peroz I’s 276–293 copper coin. Brother of Late Vasudeva II copper coins overstruck on Hormizd I’s Hormizd I? early coinages. Narseh Hormizd II Sasanian emperor (wearing similar winged 293–303 headdress to his Kushanshah predecessor) adopts bust on Hormizd II altar type from Kushanshahs’ copper coinage. 303–309 Shapur II Shapur II takes direct control of part of Kushanshah 309–379 domain, issuing Kushanshah-style copper coins and Sasanian silver coins in Kabul region. End of Kushan coinage in Gandhara. Kirada imitates Kipunadha coins in Gandhara. Ardashir II Peroz issues coins with ram horns in Gandhara and with 379–383 same crown in name of Varahran in Balkh. Kidara issues coins in Gandhara and Balkh, replacing Varahran’s name with his own and giving himself title Kushanshah. Coins issued in Gandhara in the Kushan style with image of Kidara, but acknowledging Samudra[gupta] (c. ad 330–380). Kidara coins from Bactria in Tepe Maranjan hoard with coins of Shapur II, Ardashir II and Shapur III. Silver coins of Ardashir II issued in Kabul area. Shapur III Silver coins of Shapur III issued in Kabul area. 383–388 Orolano coin from Bactria in Tepe Maranjan hoard with coins of Kidara and Shapur II, Ardashir II and Shapur III. Varahran IV Alkhano silver coinage begins in Kabul region. 388–399 Pidoko issuing gold coins in Bactria. Yazdgard I 399–420 Varahran V Tobozino issuing gold coins in Bactria and imitating 420–438 silver coins of Varahran IV and overstriking coins of Varahran IV and Yazdgard I. Yazdgard II Okilano(?) issuing gold coins in Bactria. 438–457 Peroz Peroz imitating gold coins in Kidarite style of Okilano at 457–484 Balkh 467–484.

from the kushans to the shahis

and names of the Kushano-Sasanian kings.12 There are consequently divergent views about the political status of the Kushan kings of the fourth century ad – particularly those who have Sasanian names or wear Sasanian-type headdresses. They may be Sasanian viceroys in the old Kushan empire, or simply viceroys of some of its western provinces lost to Sasanian control; or they may be independent Kushan sovereigns, ruling a much-diminished territory, influenced by Sasanian art and culture and linked at times by marriage alliances with the Sasanians. Most of Afghanistan probably reverted to direct Sasanian rule in the early seventh century. A king with Kushan royal titles is mentioned in the inscription of Samudragupta (c. 335–380) from Allahabad in India; and the Chinese sources indicate that during the period of the Wei dynasty (386–556) power in Kabul and Gandhara was exercised by a dynasty founded by Ji-duo-lo, prince of the Great Yuezhi, who built up the Kingdom of the Little Yuezhi with its capital at Peshawar.13 One branch of the Yuezhi – known in Chinese sources as the Lesser Yuezhi – never left the original homeland in Gansu and may have founded the Northern Liang Dynasty at the end of the fourth century. At the time of the capture of Samarkand by the Arabs (c. ad 712), a Chinese source refers to its Yuezhi rulers fleeing to the Pamirs.14 Hunnic Groups In the latter part of the fourth century ad a succession of Hunnic tribes appeared in Bactria and established control over the Hindu Kush mountains. Formerly these were all lumped together under the general term ‘Hephthalites’, but in newer studies of the coins and historical sources several groups of Huns are now referred to: the Chionites, the Kidarites, the Alkhan, the Nezak and the Hephthalites proper, generally termed ‘Iranian Huns’. Much of the chronology and the relationship between the various groups are still controversial, and there is also confusion between some of them. There is no broad agreement about their origins, although they are generally regarded as of Inner Asian origin and belonging to a broad ‘Hunnic’ group, related to the Huns who invaded Europe, and probably descended from the Xiongnu of Chinese sources.15 According to Greek sources it was the Chionites who first invaded Bactria in about ad 350, but it is controversial whether this group is recorded in subsequent numismatic evidence. Of these, the Kidarites were the first of the Hunnic groups to appear (although there is dispute as to whether this name is appropriate), emerging soon after the midfourth century at the end of the Kushano-Sasanian period (and the Kidarites might have maintained a subordinate Kushano-Sasanian state). Their earliest coins are associated with the coins of the Kushano-Sasanian king Peroz II, so date to before 379 – probably beginning 371 according to Göbl, although dates are disputed. A king called Kidara, with his capital at Balkh, ruled north and south of the Hindu Kush between about 312 and 337. The Kidarites continued into the fifth century, when they gave way to the Alkhans south of the Hindu Kush and the Hephthalites (or possibly the Sasanians) to the north (Table 6.2).16

349

350

THE ARCHAEOLOGY OF AFGHANISTAN

Alkhan (also spelt Alkhon or Alchona) Hun coins appear from about ad 390 onwards in the Kabul-Begram area. The Alkhan were probably at first subordinate to the Kidarites until they eventually replace them after the mid-fifth century and, under their leader Mihirakula, extended their conquests into northern India. Coins of the Hephthalites appear from the late fifth century north of the Hindu Kush.17 To these we can possibly add a fifth Hunnic group known from the early Islamic sources ruling south of the Hindu Kush from some time in the seventh century: the Ratbils of Zabulistan, probably a branch of the Hephthalites (although they may have been related to the Turki Shahs of Kabul: see below). Of these, the Hephthalites have left the most record in the sources. After defeating the Sasanians and killing their Emperor Peroz in 484, the Hephthalites established a major empire which extended from Central Asia to the Indus Valley, replacing the Alkhans (although the evidence of the Taluqan scroll suggests that a rump Alkhan state in Bactria remained under Hephthalite overlords).18 The Chinese annals of the period assert that the Hephthalites belong to the Yuezhi or Kushans, and it seems clear that they did mix with the local Iranian population of Bactria from whom they adopted the use of the Bactrian script. The Hephthalite ‘empire’ did not, however, last very long. In India Yasovarman, King of Malwa, led a confederacy to defeat the ‘Hunas’ in 528 and obliged them to withdraw to Kashmir; and in the mid-sixth century the Oxus empire of the Huns was overthrown by the Turks allied to the Sasanians, but a rump Turco-Hunnic kingdom, the Zunbil, hung on in Arachosia until the ninth century.19 Soon after the rise of the Hephthalites, a separate group of Huns with distinct coinage, the Nezak Shahs, appear to rule south of the Hindu Kush from about 560. The Nezak Shahs become absorbed by the Turki Shahs in about 748.20 The Western Turks21 With the collapse of the Hephthalites, the Western Turk Empire (or Kaganate) expanded rapidly into northern Bactria. By 616/7 a Turk army had penetrated into Sasanian Iran as far as Rayy. In 619 the most powerful of the Western Turk Kagans succeeded to the throne, Tong yabgu Kagan. Under Tong yabgu the empire reached its greatest extent, incorporating the Tarim Basin, Ferghana, Bactria and parts of Afghanistan and northern Pakistan, with Tong yabgu himself advancing to the Indus in 625. Tukharistan – ancient Bactria – and its capital at Qunduz became the centre of the empire. There was an increase in Turk immigration into these areas, particularly in and around the Hindu Kush. Tong yabgu was favourably inclined towards Buddhism, and the Buddhist art of Afghanistan underwent a revival. Many of the mountain valleys of the Hindu Kush as well as the plains surrounding them came to be ruled by Turk princes, and there was a consequent upsurge in Buddhist art in the Hindu Kush area.22 The best-known of these Buddhist centres patronised by the Western Turk Empire was Bamiyan. However, Bamiyan was only a principality in the Western Turk federa-

from the kushans to the shahis

tion, the capital of which was at Qunduz. It has been suggested that Bamiyan may have been a dynastic centre for the Western Turks,23 in much the same way that Surkh Kotal functioned for the Kushans. Tong yabgu was killed in a revolt of the Karluks in 630. Following his death the empire disintegrated, with the last Kagan killed in battle against the Chinese in 659. However, petty Turk principalities also managed to survive for another century or so, particularly in the Hindu Kush.24 The Progress of Islam The Arab armies under Islam defeated the Sasanians in ad 642. During the latter half of the seventh century they first raided the western Afghan provinces of Sistan and Herat and then controlled them with Islamic governors. When subsequently the power of the caliphate declined, local Islamic rulers came to power in the ninth century – the Saffarids in Sistan and the Samanids in Bukhara (Chapter 7). During these centuries the mountains of eastern Afghanistan formed, with northern Pakistan, a powerful non-Islamic Kingdom of the Turki and Hindu Shahis. Turki and Hindu Shahis25 At the time of the visit of the Chinese pilgrim Xuanzang in 630 the Turk Kaganate was in decline, and there were a series of states dependent on Kapisa (Begram). Laghman had become tributary and Gandhara had a commandant from Kapisa, but subsequently the Kabuli Shahis became the powerful rulers of east Afghanistan. The Turki dynasty was founded by Barhatigin, one of the petty rulers who survived the Kaganate collapse and who seized Kabul in about 666 after an Arab raid from Sistan. Barhatigin routed another Arab army in 683 and again in 698–9. Other kings were Vrahitigin, Tigin Shah and Khinjil. The kings were Buddhists and excavations at Tepe Narenj and Qol-e Tut in Kabul are evidence of an expansion of Buddhist religious building under the Turks,26 but the Hindu gods, especially Shiva and Durga, were also worshipped. Al-Biruni describes them as Turks who were said to be of Tibetan origin, tracing their descent in sixty generations from ‘Kanik’ (i.e. Kanishka) and so descendants of the Kushan dynasty. The Zunbil Hunnic dynasty of Zabulistan were related, but the relationship is uncertain. The last Turki Shah of Kabul was Lagartuman, who was forced in the end to buy off the Arabs of Sistan with tribute. This prompted his chief minister, the Brahman Kallar, to overthrow Lagartuman in about 880, bringing about the end of the Turk dynasty and starting his own, the Hindu Shahi. Inevitably they came into conflict with the rising power of Islam. Yaqub Ibn Laith captured Kabul in 870. The Shahis subsequently recovered the city, but transferred their capital at this period to Hund, a town on the Indus 25 kilometres above Attock in modern Pakistan. The position of the Shahis worsened when a line of strong Islamic rulers became established in Ghazni, after Alptegin, a

351

352

THE ARCHAEOLOGY OF AFGHANISTAN

Turkish slave commander of the Samanid army, became master of the fort there in 962. His general and successor Sebuktegin (977–997) annexed Kabul in 977 and repeatedly raided the territory of Jaypal the Shahi king. Finally, in a pitched battle near Laghman in 990 Sebuktegin decisively defeated Jaypal and annexed the province of Jalalabad and the remaining Afghan territories of the Shahis and founded the Ghaznavid state (Chapter 7). Settlement, Material Culture, Architecture and Art The archaeology of Afghanistan for the period following the Graeco-Bactrians and preceding the Muslim conquest suffers from the fact that most of the field researchers who have worked in the area have been more interested in the discovery of monumental architecture and objects d’art than in the reconstruction of everyday life and material culture. This attitude has brought about a somehow misleading picture of Afghanistan in the Kushan and post-Kushan period as one of intense religious and artistic activity, and often thwarts our efforts to understand the economic and social background of that activity. Such efforts are indeed based chiefly on epigraphy, coinage (often unstratified) and literary sources; only to a much lesser degree are they based on the results of digging in habitation areas. Actually, this kind of excavation, digging out a town with all its successive building horizons and its extension over a large area, calls for a greater financial effort, and often appears less rewarding in the eyes of the public, than does the more certain success of the excavation of a tepe concealing the remains of a stupa, with rich and attractive sculptural decoration. Unfortunately, even the understanding of the ideological level is difficult if not impossible while the socioeconomic level of a civilisation remains unknown. This is why the many Buddhist architectural complexes that archaeological excavations have brought to light are mute when we try to understand which social group had them built, for whose use they were intended, who was actually employed in their construction, what social rank or status was occupied by those who planned them and those who decorated them with sculptures and paintings, and who was entitled to accept a project and choose the subjects to be represented on a stupa or in a chapel. The successive Kushan, Kushano-Sasanian, Hunnic and Turk periods are marked above all by cultural continuity, characterised in particular by the Buddhist religion and its associated architecture and art – the stupas and art of Gandhara – the study of which has been plagued by problems of chronology. For this reason, these periods are considered together in the following discussion and not necessarily in chronological order, even though some – Begram, Surkh Kotal – are generally earlier while others – the Bamiyan, Kabul and Ghazni sites – are generally later.27 It must also be emphasised that the focus is almost exclusively on eastern Afghanistan: roughly east of a line drawn from Balkh to Kandahar. This imbalance is by no means an indication of the lesser importance of western Afghanistan, but merely of the nature of discoveries. The Ghulbiyan painting, for

from the kushans to the shahis

example (discussed below), has done much to correct this imbalance, and equally important discoveries in western Afghanistan – Sistan in particular – might yet do more. In this connection mention must be made of the possible remains of a large fire temple of the Sasanian period at Sna Qal’a on the right bank of the lower Helmand recorded by the Smithsonian Helmand-Sistan Project.28 This assumes added significance in view of the recent discoveries and excavations in the eastern Iranian world of the most intact fire temples so far known, such as Bandiyan in north-eastern Iran, Sarakhs in Turkmenistan and Tash Kirman in Uzbekistan.29 Begram (Figs 6.2–6.11) Among the few town sites identified and partially explored, Begram first demands attention. This site, which is about 45 kilometres north of Kabul, was noticed in 1833 and then made known to scholars by Charles Masson;30 Alfred Foucher identified it with Kapisa, the capital town of several Indo-Greek sovereigns and summer residence of the Kushans, from when most of the remains date.31 The DAFA carried out several limited excavations from 1936 to 1946, covering only a very small portion of the town area: trenches were laid down in the Bazar or the ‘Royal City’, with a gateway and a qal’a almost certainly belonging to a period when the town had lost most of its vitality, in the western section of the town, and extra muros.32 The fame of Begram nevertheless rests on the discovery of two rooms (10 and 13) of what appears to be the ‘palace’ which were found to have been filled with objects of enormous value from the point of view of art history, including plaster models and glass and bronze objects of Western (Hellenistic) origin, Indian ivories and Chinese lacquers (Figs 6.5–6.11).33 A detailed description of the objects found in these ‘store-rooms’ is beyond the scope of this book; nevertheless, we discuss this exceptional discovery because of its importance for understanding the nature of the traffic which was carried on, in or through Kapisa, even if the Begram finds can only provide information concerning a luxury level of trade. It is noteworthy that at least one of the doors leading to the two rooms had been blocked up in antiquity and that the wares found are far from being homogeneous, not only with regard to the origin of the various objects (from Alexandria to India and China, as we have seen) but also as regards their functions and dates, and therefore the reasons that apparently led to their being stored or hidden. There are some objects that must have had great commercial value, such as the Indian ivories or the Mediterranean glasses, and may therefore have been put in a safe place to preserve them from the hands of enemies, as Ghirshman suggested; but others – especially the plaster models – have no intrinsic value and it is very difficult to imagine that they were kept as precious things in a palace, or that an invading army could look on them as a desirable booty.34 Others suggest that the two rooms were ‘a Customs depot for the receipt of dues in kind collected by the kings or viceroys of Kapisa from the caravans which traversed the adjacent

353

354

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.2  Map of sites associated with Begram on the Kuh-i Daman plain north of Kabul.

from the kushans to the shahis

Figure 6.3  The excavations at Begram.

highway in the luxury traffic of Orient and Occident’.35 But this explanation too does not find support in the peculiar nature of the objects: principally in the lack of intrinsic value in the case of the plaster models. We may perhaps solve the problem if we think of some particular purpose in collecting so many and such peculiar art objects, connected with the production of other objects. The two rooms at Begram probably contained wares taken from the ‘palace’ in a moment of danger, together with objects belonging to a royal atelier: the models for silver ware and possibly also for stucco decorations, even if made of worthless plaster, were certainly precious for an art workshop. Neither the Hellenistic nor the Indian and Chinese objects have been unanimously attributed by scholars to a definite period. Differences concern the Hellenistic wares,36 the Chinese lacquers37 and the Indian ivories.38 In general, the time span within which the various components of the whole hoard were placed was originally dated from the first century bc to the beginning of the third century ad. More recent studies, however, place the treasure in the first century ad to early second century,39 although another study based on a re-examination of the original excavation records dates the deposition to the mid-third century.40 Ghirshman41 tried to establish a stratigraphic succession, including three phases: Begram I Indo-Greek, Begram II corresponding to the Great Kushan, Begram III – following a violent destruction – the period of the Kushano-Sasanians, up to the Hephthalite invasion that caused the town to be abandoned. The building

355

356

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.4  Plan of the site of Begram.

from the kushans to the shahis

Figure 6.5  Ivory from Begram in the Musée Guimet.

Figure 6.6  Ivory from Begram in the Musée Guimet.

357

358

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.7  Ivory from Begram in the Musée Guimet.

structures found by Hackin would belong, according to Ghirshman, to the second and third phases; the ‘palace’ would have been destroyed in about the middle of the third century, when the Sasanian emperor Shapur I conquered the Kushans. We must confess that the whole picture is not altogether clear and several inconsistencies have been pointed out. Following comparisons with the Japanese excavations at Tepe Skandar, together with re-evaluation of the excavations, the coinage and the sources (notably the visit by Xuanzang), Kuwayama now dates Begram III to the sixth to seventh centuries: that is, there is a long gap between

from the kushans to the shahis

Figure 6.8  Glass from Begram in the Musée Guimet.

Figure 6.9  Glass from Begram in the Musée Guimet.

359

360

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.10  Glass from Begram in the Musée Guimet.

Begram II and Begram III.42 Morris, on the other hand, places the abandonment of Begram III at the same time as the deposition of the hoard, at ‘a conservative terminus post quem’ of c. ad 260.43 Even if it is indisputable that the Begram hoard is a proof of the cosmopolitanism of the Kushan sovereigns to be ranked along with the documents of local artistic activity which will be discussed in the following pages, Begram itself, a site where, in the words of Wheeler, ‘so little work has yet been done though with such dramatic results’, remains ‘a challenge to the explorer’.44 Both the chronology of Begram and the reasons behind the deposition of the hoard – merchant cache, royal collection, taxation – still remain highly debated.

from the kushans to the shahis

Figure 6.11  Plaster plaques from Begram in the Musée Guimet.

Kandahar (Figs 5.12–5.15; 6.12–6.13) The importance of the ancient town of Kandahar (Shahr-i Kohna) was emphasised as a consequence of the discovery of Ashokan inscriptions and several discussions dealing with the identification of Kandahar with the metropolis Arachosias of Isidore of Charax, as well as with the ancient topography of Arachosia (discussed in Chapter 5). Excavations since by a British Mission have added much to our knowledge.45 What is visible on the surface has been subject to extensive natural erosion, robbing and – in recent years – over-building. However, excavations have made it possible to clarify the main elements of the pre-Islamic town: (1) the impressively thick fortifications rebuilt and strengthened from the original Iron Age ramparts encompassed by a large ditch (Figs 5.12–5.14); (2) the huge citadel, continued from

361

362

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.12  The stupa-monastery at Kandahar.

probable Iron Age foundation (Fig. 1.34); (3) a Buddhist monastery with a stupa dating from the Hun period (Fig. 6.12). A provisional chronology, with many a caveat, was made on the basis of Whitehouse’s initial excavations and largely followed by McNicoll, with ten main cultural divisions (Periods I–X).46 This was subsequently discarded in Helms’ final publication of his excavations, where the chronology was divided into more cautious ‘Epochs’, correlated with the Whitehouse–McNicoll chronology in an Appendix.47 For greater ease, the Whitehouse–McNicoll periodisation is followed here. The period at the turn of the first millennium bc/ad, Period V (Indo-Parthian/ Kushan), is marked by a major new upsurge in large-scale building activity throughout the site: a major rebuilding of the city’s defences, a possible monumental building at Site D and massive brick walls at Site F. The method of construction, too, marks a sharp break from previous periods: the pakhsa that characterised the previous periods (especially Period II, the Achaemenid) is replaced by substantial mud-bricks generally measuring about 45 cm square. This means that at Kandahar there seems to be the same variety of square bricks the Italian archaeologists have noticed at Tepe Sardar, Ghazni, between the Kushan and the Shahi period, the succession being 43 × 43 × 9/10 (early), 46 × 47 × 9 c. (middle), 40 × 40 × 10 or even

from the kushans to the shahis

Figure 6.13  Spiral-burnished pottery from Kandahar.

38 × 38 × 10 (late). A similar contrast marks the pottery repertoire, characterised by a massive upsurge of spiral-burnished wares (Fig. 6.13). This remains the most dominant pottery type until Period VIII, the beginnings of Islam – virtually the whole of the first millennium ad. The international links that marked the pottery of the previous Achaemenid and Hellenistic periods seem to recede, with the ‘native’ spiral­-burnished wares predominating. Although spiral-­burnished ware is often taken as the ‘hallmark’ by which Kushan sites are recognised, it seems clear that it is not a ware that can be associated exclu­sively with the Kushans. This makes any dating difficult, as the spiral-burnished wares show very little variation throughout most of the first millennium ad. The use of spiral-burnished ware for dating purposes – let alone dynastic attributions – is, therefore, clearly fraught with difficulties. Period V therefore marks the end of one major era at Kandahar and the beginning of a new one that was to remain without any major breaks down to the Islamic conquest. In the preliminary report, following Fussman and Whitehouse, McNicoll applied the term ‘Kushan’ to the period at Kandahar down to the Period VI Sasanian conquest with a due caution to the reader.48 In the light of Helms’ subsequent excavations in 1976–8 as well as other studies, it appears that the Kushans

363

364

THE ARCHAEOLOGY OF AFGHANISTAN

may not have held sway at Kandahar at all;49 the area of southern Afghanistan was more likely under the control of the Indo-Parthians in the first two centuries ad. It must be noted, on the evi­dence so far available, that Helms’ strictures on the inaccuracy of the term ‘Kushan’ are based on negative evidence (i.e. lack of stratified coins or inscriptions) as much as on positive evidence. However, Helms’ remarks are salutary and the application of historical labels to the material of different periods needs to be treated with caution. The end of Indo-Parthian/Kushan rule at Kandahar and the Sasanian conquest in the early third century ad (Period VI) are hardly marked by changes in the pottery, although there appeared to be some hurried fortification, which may suggest an invasion. This could be the Sasanian invasion at the beginning of the third century ad or that of the Huns in the mid-fifth – or it may have been nothing more than a purely local incident not recorded in the sources. But making archaeological phases fit historical events is never easy: cultural changes usually occur well after the event. There is in any case strong continuation of the pottery types throughout. There is no major new building activity or particular change apparent, but similar types of buildings were rebuilt on the same alignments, and no rebuilding of the city defences is apparent until the Islamic period. This implies that as well as considerable continuity, the period saw little external threat. The gradual abandonment of the city defences might mean that the city was open in military terms, with the population of Kandahar attenuated and agriculture flourishing intra muros. This might be seen as evidence for external rule of Kandahar, such as Sasanian. But many ancient cities had large tracts of arable land within their walls, and yet were still stoutly defended if attacked. We may also note the interesting circumstances that among the burials there were four in which the skeletons had silver Sasanian coins (probably fifth century ad) in their mouths. The period immediately before the Islamic conquest in the ninth century ad corresponds to the period of rule by a local Hunnic dynasty known in the sources as the Ratbils (Period VII). The stupa-monastery complex overlooking the site remained in use during this period, confirmed by a hoard of Hunnic Nezak coins as well as a single Umayyad coin found in the excavations,50 confirming the late upsurge of Buddhism in Afghanistan noted elsewhere (Fig. 6.12). Zur, or Zun (a variation), was the name of the main deity of the Huns in this area.51 Zur or Zun, therefore, refers to the cult as a whole as well as the specific deity.52 It gave its name to the place where the cult centred: the Jabal az-Zur where the main temple was located, figuring prominently in the early Islamic accounts. Jabal az-Zur is probably located in modern Zamindawar sub-province, probably at the hill and associated ruins near the modern village of Deh Zur.53 It is worth noting, however, that the site of Old Kandahar is today known as Zur Shahr – ‘City of Zur’ – as well as the more commonly used Shahr-i Kohna.54 The Ratbils ruled in the Kandahar area for nearly 250 years until the Islamic conquest – indeed, Kandahar probably became their winter capital under the name of ar­-Rukhkhaj, deriving

from the kushans to the shahis

from Harahuvatish, the Achaemenid region and city. This seems confirmed by the first Islamic sources in the ninth and tenth centuries, which refer to a city of ar-Rukhkhaj for Period VII, as well as a re­gion.55 The end of the Hunnic period seems marked by some hurried re-fortification. This might well mark the final collapse of Kandahar against the Muslim forces, presumably the Samanids in the tenth century. Wardak A large town on a plateau in the Wardak valley was first recorded in the nineteenth century. The town comprises inner and outer enclosures surrounded by ramparts, and appears deliberately laid out according to a plan. To the south are a monastery and four stupas, from one of which an inscribed reliquary was excavated on Masson’s instruction (Figs 6.14–6.16).56 Gérard Fussman noted that some analogies are to be recognised with the plan of Aï Khanoum (although this might be illusory). The pottery collected by Fussman nevertheless is to be attributed to the Kushan period, most of the comparative material being provided by Begram II. An important problem concerns the route connecting Arachosia (Kandahar) with Kapisa:57 it is not impossible that this route or one of the alternative routes passed through the town at Wardak, certainly rich and flourishing, as it appears from the Buddhist monasteries built in its immediate vicinity and the dam that probably gave water to the town by means of a canal. Fussman believes that he has found traces of this canal but cannot say whether it brought water only to the ditch or also into the town.58 Jaghatu Although later, another town site may be recorded here, since it was probably on the same route connecting Arachosia to Kapisa, but closer to Ghazni. It is now a group of mounds in the Jaghatu-i Wardak, where some very limited trial trenches were dug by Umberto Scerrato in 1958, which ‘may probably be considered a village born as a resting place on the caravan-road, protected by the fortress of the Bad-i Asya and dependent on the settlement of Tabak-sar’.59 The importance of the site is also documented by the presence of two inscriptions in Graeco-Bactrian cursive script, but these, like the coins and sherds, point to a date after the Great Kushans: actually, sherds from big storage jars bearing the impressions of large medallions with ‘Hephthalite’ (but probably Hindu Shahi) Graeco-Bactrian cursive inscriptions have accidentally been found on or near the tepe of Dubakh Sar (Scerrato’s ‘Tabak-sar’). The Dubakh Sar tepe is certainly an imposing fortified centre dominating what was probably an important caravan route, and one might reasonably expect to find more such strongholds in the region – as well as further minor centres similar to that excavated by the Italian Mission. Indeed, among the many tepes in the Ghazni

365

366

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.14  Plan of the site of Wardak.

region, one may recall the imposing Takht-i Jamshid at Sajavand (or Sakavand), in the valley connecting Wardak with Logar (Lohgar), where huge ‘Gandharan’ walls of schist slabs and blocks are still standing along with mud-brick structures representing a major Hindu Shahi temple;60 and the many small mounds scattered on the plain between Moqur and Dila, north of the Ab-i Istada.

from the kushans to the shahis

Figure 6.15  Stupa 1 at Wardak.

Figure 6.16  Remains of paintings in the monastery at Wardak.

367

368

THE ARCHAEOLOGY OF AFGHANISTAN

Kharwar One of the most important Buddhist sites in Afghanistan is Kharwar, an extensive area of Buddhist remains reportedly covering many square kilometres in the upper Logar valley about forty kilometres north-east of Ghazni (Figs 6.17, 6.18). It was first reported in the nineteenth century, when many coins were found. The site has been the scene of some of the heaviest fighting in recent years, and following extensive looting, investigations revealed the remains of numerous stupamonastery complexes that contained many stucco Buddhist sculptures and other important art objects. Reports state that a number of the structures are several storeys high, and include a fortified gateway, a ‘cylindrical tower’ – presumably a stupa – up the side of a mountain, and what appear to be several collapsed cave openings.61 Following a brief survey by an Italian mission in 2003, a walled town was mapped, and several stupa-monastery complexes were recorded as well as the remains of a fortress or palace. Discoveries include wooden fragments dated to the second to third century ad, seven ‘clay’ heads from the fifth to the seventh century (now in the National Museum) and parts of the feet of an enormous (standing) statue, possibly of Buddha.62

Figure 6.17  Photo of the site of Kharwar.

from the kushans to the shahis

Figure 6.18  Plan of the remains at Kharwar.

The Kabul Sites63 The large number of remains in and around the old city of Kabul attest to its importance as a major urban site since earliest history (Fig. 6.19). Much of the evidence has been due to accidental discoveries, which has meant that each discovery has usually been discussed in isolation. However, when all are put together, in particular in the light of recent excavations on the hills overlooking the city, Kabul emerges as a pre-Islamic religious and urban centre second to none in Afghanistan. It is a large urban site of uncertain character and extent due to modern overbuilding, between the three hills of Shir Darwaza, Asma’i and Maranjan. In Kabul itself the only extant remains are the Bala Hisar and city walls on Shir Darwaza hill, a series of stone and mud fortifications of possible Hunnic origins with extensive later rebuildings (Fig. 6.20). The walls originally extended across Asma’i hill and over the plain around Deh Afghanan as well, and had six gateways. The Bala Hisar is mainly nineteenth century, although it sits on a partly artificial mound and three sondages on the northern side by the DAFA in 2008 reportedly revealed sherds going back to the Graeco-Bactrian period.64 At Chaman-i Hauzuri, approximately 1,000 Greek and Achaemenid coins were discovered in 1933 (discussed in the previous chapter). Between the Qul-i

369

370

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.19  Map of the sites in and around Kabul.

from the kushans to the shahis

Figure 6.20  The city walls of Kabul.

Hashmatkhån and the Kuh-i Takht-i Shah, three kilometres south-west of the Bala Hisar, is a large area of mostly modern cemeteries where many antiquities have been found. The modern Ziyarat-i Panjashah is built over a Buddhist stupa-monastery complex of the Kushano-Sasanian to Shahi period, excavated in the 1830s by Masson and again by Gérard. It consisted of some caves, a cistern and many cells, one of which was domed. Construction was of mud-brick. Finds included an arched niche supported by pillars, several tuz-leaf manuscripts, frescos and much Buddhist and Hindu sculpture. Modern building in the old city obscures the nature of earlier remains. Far more evidence for ancient Kabul can be seen in the hills surrounding the old city (although modern uncontrolled building is increasingly encroaching on the slopes as well), particularly in the light of recent excavations. At the eastern end of the Maranjan hill, four kilometres east of the Bala Hisar, is the Buddhist monastery complex of Tepe Maranjan, of which almost nothing remains today. It was excavated in 1933 by Jean Carl, and finds include sculpture, frescos and a hoard of 368 Sasanian silver drachmae, dating from ad 383–8. In 1981, a second monastic complex, which includes a stupa of diaper masonry, was discovered by chance. A thick ash layer may mark the Hun invasion. At the foot of the hill at the south side are twelve artificial caves with low parallel openings. Immediately to the east of these caves are many mounds.65

371

372

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.21  Isometric view of the Tepe Narnenj remains.

Between 2004 and 2012, a very large Buddhist monastic complex was excavated by Zafar Paiman at Tepe Narenj on the hill of Kuh-i Zamburak (formerly Kuh-i Takht-i Shah) overlooking the old city of Kabul, south of the Bala Hisar.66 It consists of five groups of buildings on a series of nine artificial terraces, possibly the main Buddhist centre for the Kabul region for the Hunnic and Turk periods that remained active until the Muslim conquest in the ninth century (Fig. 6.21). The terracing recalls Surkh Kotal, but no earlier date than the fifth century was recorded. It comprises one large and five small stupas. The associated chapels contained many votive stupas and large numbers of Buddhist sculptures of clay coated with stucco. Just 400 metres to the north and probably associated with it is a smaller monastery, Qol-i Tut, excavated during the Tepe Narenj excavations (Fig. 6.22). This

from the kushans to the shahis

Figure 6.22  Photo of the excavated stupa of Qol-e Tut.

is possibly the same as Takht-i Shah, excavated by Masson, who recorded preMuslim remains underneath a Timurid shrine, but the identification is disputed. It comprises a main stupa and eighteen chapels on ten terraces containing painted Bodhisattvas and a monumental seated Buddha. It was constructed in the sixth century and remained in use as late as the end of the eleventh century, ­confirming – like Tepe Narenj – the continuing practice of Buddhism until well into the Islamic period.67 On Shir Darwaza Hill overlooking the old city of Kabul to the west of the Bala Hisar is a site similar to Tepe Narenj at Khwaja Safa. This is an Islamic shrine built over a series of older terraces on the side of the hill. Several third century Buddhist sculptures were discovered here in 1905. Excavations in 2004 revealed traces of a Buddhist monastery complex, although much is obscured underneath the Muslim holy site. The excavations produced many more Buddhist sculptures, mainly fourth to fifth century, and a coin of Vima Kadphises. Like Tepe Narenj, the monastery continued in active use until the ninth century.68 On a spur of the Shir Darwaza Hill elsewhere, overlooking the Kabul River, is the site of Tepe Khazana, the site of remains of a fifth to seventh century Buddhist stupa, now destroyed. In the 1930s a series of fifty stucco and terracotta heads and other sculptural fragments were discovered here by chance.69 Fussman locates the ancient city at the site of Bagrami to the east of the Bala Hisar, which was recorded by Masson but never excavated (and now ­increasingly

373

374

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.23  The stupa at Shiwaki.

Figure 6.24  The Minar-i Chakri.

from the kushans to the shahis

overtaken by modern urban spread).70 This is roughly midway between the Buddhist monuments of the city of Kabul described above and those skirting the Monaray Ghar hills to the south-east: the stupas of Yakhdarra, Shiwaki, Kamari and Seh Tupan, with the ‘minars’ of Surkh Minar and Minar-i Chakri on top of the hills (Figs 6.23, 6.24). In this case, these monuments might be considered a part of ancient Kabul. Mes Aynak Another site in the Kabul region that has revealed considerable new information on the Kushan through to the Shahi periods has also been the subject of new excavations (although little has been published at the time of writing). This is Mes Aynak in Logar Province to the south of Kabul, where occupation has been traced back to the Bronze Age. Although this was first examined briefly in 1977, threats from major copper mining at the site led to excavations by Philippe Marquis and an international team between 2009 and 2014, as well as considerable international publicity. Scattered over the hills are diaper masonry terrace walls, and there are remains of mud-brick structures lower down marking the remains of a substantial

Figure 6.25  Aerial view of Mes Aynak.

375

376

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.26  A stupa at Mes Aynak.

monastic settlement and associated town covering some forty hectares. Some four hundred Buddha statues have been recovered, and excavations have revealed several Buddhist monasteries, substantial Buddhist sculpture and paintings, and gold and semi-precious objects (Figs 6.25, 6.26).71 Surkh Kotal A place of particular importance in the archaeology of the Kushan period in Afghanistan is occupied by the royal sanctuary of Surkh Kotal, set in the very centre of a fortress, which we know from epigraphic evidence to have been founded by Kanishka himself.72 It consists of two parts (Figs 6.27–6.32). At the top of the hill (west), in a court surrounded by a portico there is a temple with a cella encompassed by a corridor; on the east side of the hill the area is divided into four terraces connected to each other and to the upper court by monumental flights of steps (Figs 6.27, 6.28). Daniel Schlumberger observed that the building technique at Surkh Kotal, as well as the plan of the temple, is to be placed in the Iranian tradition as it is represented chiefly by Achaemenid architecture: mud-bricks being used along with stone, the latter for the stairways, some plinths and the column bases, the plan of the main building, the temple, being quite similar to that of an Achaemenid temple near Susa.

from the kushans to the shahis

Figure 6.27  The excavations at Surkh Kotal.

It has been suggested that this is a fire temple, based as it is on the evidence of the excavation (platform in the centre of the cella, accessible by a flight of steps; Fig. 6.29) and on the comparison with Iranian presumed prototypes,73 but this identification is now not so widely accepted.74 As Schlumberger himself remarks, nothing leads us to believe that it was a Zoroastrian fire temple; it is much more probable that it was a ‘dynastic’ sanctuary, as is suggested both by the epigraphic evidence and by the comparison with the sanctuary at Mat near Mathura, on the Ganges plain. The evidence of both the Rabatak inscription and the Bactrian documents suggests that while the Kushans incorporated Zoroastrian elements, they were essentially eclectic in their religious beliefs and incorporated other cults and religions as well.75 The sculptural decoration of the temple is an unparalleled mixture of different artistic trends. It includes stepped merlons reminiscent of Ancient Near Eastern tradition (Fig. 6.30), pseudo-Classical pilasters in the Hellenistic tradition (Fig. 6.31), a Gandharan stone frieze, a series of unbaked and painted clay figures, a very badly damaged stone relief which Schlumberger suggested was to be compared with some of the enthroned figures of Nemrut Da©, Commagene (middle of the first century bc) and three stone statues representing Kushan kings (or gods?) (Fig. 6.32). These latter may be compared with the Kushan images of Mathura,76 not only in some details of the dress (which has certainly nothing in common with either Classical or Indian costume, and is rather Iranian or ‘nomadic’) but

377

378

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.28  Plan of Surkh Kotal.

from the kushans to the shahis

Figure 6.29  The ‘altar’ at Surkh Kotal.

also for stylistic and technical reasons: chiefly because they do not reproduce the volumes of human figures but are rather slabs on which folds and ornaments are superficially carved. Rabatak In 1993, a new Bactrian inscription of the early Kushan period was recovered from the Kushan site at Rabatak Kafir Kala about 40 kilometres east of Haibak and about 15 kilometres north-west of Surkh Kotal, and is now in the National Museum (Fig. 6.90).77 The site itself was another dynastic sanctuary, probably similar to Surkh Kotal in that it consisted of a series of terraces, dedicated by Kanishka I to his ancestors and probably containing statues of the kings and deities (Fig. 6.33). Rabatak has a perimeter wall with a well-defined gateway on the north which leads directly onto the processional way(?) to the upper terrace(s). The ancient road to Balkh ran past this site to the west (not the east as the modern road does). In the area where the inscription was found (at the apex of the site) there were several Kushan-style column bases as well as other monumental fragments.78 However, it was almost completely destroyed before it could be recorded, with just the one inscription and a few architectural fragments (parts of a lion sculpture, some lotus decoration) saved and taken to Kabul. The inscription engraved on a block of

379

380

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.30  One of the merlons from Surkh Kotal in the Musée Guimet.

limestone records that King Kanishka gave orders to Shafar the Karalang to make a sanctuary and images of a range of gods79 for King Kujula Kadphises his great grandfather, the hitherto unknown King Vima Takto, his grandfather, King Vima Kaphises his father, and himself (discussed further below). In this context, mention must also be made of the site of Chahrshanbeh Tepe, some 25 kilometres north-east of Surkh Kotal near the road to Qunduz. An impressive and extensive series of ruins surrounding a central, elevated mound consisting of a series of three level, artificial terraces was recorded by Jonathan Lee in

from the kushans to the shahis

Figure 6.31  Pseudo-classical pilaster and cornice from Surkh Kotal in the Musée Guimet.

2004, when the Historic Monument Department of Baghlån Province recovered two ‘Kushan’ column bases and other fragments of monumental architecture from the site.80 Rag-i Bibi Five kilometres south of Surkh Kotal and not far off the main north–south road between Kabul and northern Afghanistan is the Sasanian rock relief of Rag-i Bibi,

381

382

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.32  Statue of Kanishka from Surkh Kotal in the National Museum.

first brought to international attention by Lee in 2003 and subsequently published by him and Frantz Grenet (Fig. 6.34).81 Although much damaged, it is a scene depicting a Sasanian emperor, probably Shapur I, so dated to the middle decades of the third century. Shapur is depicted hunting rhinoceros, hence commemorating the Sasanian advance to the Indus (where there were rhinoceroses). It also depicts a captive Kushan, presumably a Kushan king, and so marks the Sasanian conquest of the Kushan Empire. In this way it recalls Shapur’s other more famous triumphalist reliefs at Naqsh-i Rustam and – most especially – at Bishapur, depicting captive (or slain) Roman emperors. However, unlike the captive Roman prisoners of war depicted in, for example, the famous Bishapur 3 relief, the captive Kushans

from the kushans to the shahis

Figure 6.33  The sanctuary site of Rabatak.

Figure 6.34  The Sasanian rock relief of Rag-i Bibi.

383

384

THE ARCHAEOLOGY OF AFGHANISTAN

at Rag-i Bibi are depicted still bearing their arms. In the words of Grenet, ‘Taken together, the message addressed to the Kushans is one of reconciliation rather than of humiliation.’82 The importance of the Rag-i Bibi relief cannot be overestimated: it has in one stroke extended our knowledge of this distinctive Sasanian art form over 1,500 kilometres further east than all other known Sasanian reliefs. The Bactrian documents (see below) further suggest that the region was a royal enclave known as ‘Kadagstan’ that appears to be particularly associated with the early Sasanian kings. It also adds considerably to our knowledge of the nature of the Sasanian presence in the region, adding to an increasing amount of new evidence emerging from the eastern fringes of the Sasanian Empire that appears to be re-orienting the previously western bias of Sasanian studies.83 It also reopens the vexed question of the elusive ‘Ferrier relief’, first reported by a French soldier of fortune, General Ferrier, in the mid-nineteenth century deep in the mountain regions south of Sar-i Pul in north-western Afghanistan. He described a rock relief of an enthroned king before his court with a kneeling captive imploring justice and another captive already executed. Ferrier’s description seemed to tally with known Sasanian triumphal reliefs further west. However, several attempts since to locate and record this ‘relief’ proved unsuccessful, and it has been assumed to be, at best, a conflation with known reliefs that Ferrier observed in Iran or (also known as Sar-i Pul), and at worst, a figment of his imagination.84 The discovery of Rag-i Bibi, however, proves that the existence of the ‘Ferrier relief’ might after all be likely. Balkh The vast site of Balkh has always been known to be one of the key urban sites of Central Asia, but the combination of heavy over-burden and high water table has made any investigation of its early history problematic (Fig. 6.35). However, discoveries in the 1990s led to excavations at Tepe Zargaran that confirmed a major Hellenistic presence at Balkh (discussed in Chapter 5) as well as Kushan. These included a stupa, which contained coins of Soter Megas, and the remains of a stone-lined channel that was part of an extensive Kushan period irrigation network of the Balkh oasis.85 Outside the walls of Balkh are the mounds of Tepe Rustam and Takht-i Rustam, probably representing the remains of the major Buddhist stupa and monastery complex of Nava Vihara (new monastery’) or Nau Bahar (Figs 6.36, 6.37).86 This was the main Buddhist religious centre in Bactria down until the Islamic period, and was in the hands of a hereditary priestly family known as the Barmakids, who became a hugely influential family of viziers to the early Abbasid caliphs in Baghdad. This may have been a determining factor in the location and construction of the Mosque of Nau Gunbad by one of the Barmakid family members in the eighth century (discussed in the next chapter).

from the kushans to the shahis

Figure 6.35  The ramparts of Balkh.

Figure 6.36  Tepe Rustam at Balkh.

385

386

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.37  Takht-i Rustam at Balkh.

Zadiyan and the Balkh Oasis Wall New investigations by French archaeologists, as well as older surveys by Soviet archaeologists, have documented Kushan fortifications in the Balkh Oasis on a truly gigantic scale, indeed, among the most extensive in the ancient world. These investigations, enhanced by satellite imagery, have revealed a vast military camp near Zadiyan, over four kilometres square surrounded by pakhsa ramparts preserved up to seven metres high in places (Fig. 6.38). In the exact centre of the enclosure is the large square citadel mound of Zadiyan Kafir Qal’a measuring approximately 200 metres square, with walls standing up to twelve metres high surrounded by a ditch. Construction is large-size mud-bricks. There is a bent entrance ramp in the centre of the west side. This faces a monumental entrance gateway in the exact centre of the western wall of the enclosure (Fig. 6.39). Radiocarbon analysis has confirmed a date between 61 bc and ad 7687 for the construction of the outer ramparts, and between ad 222 and 394 for its reconstruction.88 It was built, therefore, in the early Kushan period, and the French archaeologists have interpreted it as a Kushan military camp, noting that it lies exactly midway on a straight line between Balkh and the Kushan site of Kampyr Tepe on the northern bank of the Oxus.89

from the kushans to the shahis

Figure 6.38  A digital elevation model of Zadiyan from satellite imagery. (Image prepared by the Center for Ancient Middle Eastern Landscapes at the Oriental Institute of the University of Chicago. Image courtesy Esri and CNES/Airbus DS.)

Figure 6.39  The citadel of Zadiyan.

387

388

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.40  The Balkh oasis wall.

In the 1970s Soviet archaeologists recorded, incorporated into the Zadiyan enclosure, a long pakhsa wall stretching for over sixty kilometres from a few kilometres to the east of Zadiyan right across the northern edge of the Balkh oasis to just short of Aqcha (Figs 6.41, 6.42). The eastern and western limits of the wall were not determined, so the total length would have been longer; the eastern end might have extended to and been incorporated into the Kushan fortified urban site of Toprakkale, 16 kilometres beyond Zadiyan, and the western was probably incorporated into the citadel of Aqcha (of indeterminate date). Its best-preserved part is at Kam Pirak, where it is up to three metres high for four kilometres, pierced by brick arrow slits capped by bricks in the form of a triangle in the Central Asian tradition (Fig. 6.40). The Soviet archaeologists initially dated the wall to the Hellenistic period on the basis of ceramic evidence; radiocarbon tests by French archaeologists showed a date between ­ ad 134 and 334.90 Although much work still remains to be done on this clearly linked and very intriguing complex of fortifications, it is one of the most important – and certainly the largest – work of Kushan architecture, with major implications for the history of Bactria. Following the absorption of Bactria into the Sasanian Empire, the Zadiyan–Kam Pirak complex might well have been the model for similarly massive military constructions in the Sasanian world, such as the Gurgan Wall,91 the defences of the Caucasus passes and the immense fortification of Qal’a-i Gabri near Tehran.92

from the kushans to the shahis

Figure 6.41  Approximate alignment of the Balkh oasis wall in relation to other Kushan fortifications in the Balkh oasis.

The East Bactria survey of the 1970s has further recorded large numbers of settlements, generally dated by ceramic evidence to broadly the ‘Kushan to Hephthalo-Turk’ period. Many of these are substantial settlements, often fortified and often in continuous occupation from older periods. Notable is the vast fortified enclosure of Qal’a-i Zal measuring 1,800 × 900 metres, a scale comparable to that of Zadiyan above.93 This and many of the other fortifications recorded in Bactria seem to belong to a broad Central Asian style, such as those recorded in Chorasmia.94 However, current studies of the vast Sasanian fortified enclosure of Qal’a-i Gabri to the south of Tehran might throw new light on the Bactrian fortifications.95 The complex and extensive irrigation system (discussed in Chapter 3) from earlier periods was also maintained.96

389

390

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.42  Elevation, section and plan of the Balkh oasis wall.

Buddhist Art and Architecture From the point of view of religious architecture, the archaeology of Afghanistan of this period shows us a picture that at first appears almost exclusively Buddhist. Apart from Surkh Kotal there are some interesting exceptions that will be discussed when we deal with the artistic products of Tepe Skandar97 and Khair Khana,98 both near Kabul, and Chaghan Sarai in the Kunar Valley.99 Cave Monasteries A typical monument of Buddhist Afghanistan is the cave monastery. It is hardly necessary to recall here the world-famous caves of Bamiyan with their colossal Buddhas and ‘Indo-Iranian’ paintings,100 to which others have been added in the Foladi valley.101 For most people, Bamiyan (Figs 6.43–6.51) is somehow synonymous with Afghanistan,102 particularly after the notorious destruction of the two Buddhas in 2000, such is the impressiveness of its rock-cut monastic caves and the two huge Buddhas which aroused the admiration of the Chinese pilgrim Xuanzang (seventh century). He wrote about the bigger (and later) of the two, 53 m high: ‘On the declivity of the hill to the north-east of the capital was a standing image of Buddha made of stone, 140 or 150 feet high, of a brilliant golden colour and resplendent with ornamentation of precious substances.’103 Both the Buddhas and the caves were finished and coated by means of stucco and clay plaster: for instance,

from the kushans to the shahis

Figure 6.43  General view of the main cliffs at Bamiyan with the two colossal Buddhas.

the folds of the bigger Buddha’s cloth were obtained by modelling the plaster on cores of ropes which were nailed to the image (i.e. to the rock) by wooden pegs. This larger Buddha was initially assigned to the fifth to sixth centuries on account of both the paintings inside the niche, which reflect themes from the Ajanta paintings, and its own style, which seems rather to be reminiscent of the Gupta sculptures of Mathura, but radiocarbon analysis now dates it to the middle of the seventh century.104 A discussion of the style and chronology of the Bamiyan paintings would take us beyond our present scope (Figs 6.49, 6.50). Influences from several regions have been pointed out, Iranian (Sasanian), Gandharan and Indian. These provide a general picture of the art centre which clearly points towards Central Asian solutions:105 an attempt at a classification into four styles has been made by a team from Kyoto University.106 There are related cave groups in the Bamiyan area, including those at Kakrak (Fig. 6.46) with the wall painting of the ‘hunter king’,107 datable to the sixth to seventh centuries. The brick fortress of Shahr-i Zohak nearby was probably constructed in the same period, although renovated in Muslim times (Figs 6.52, 6.53).108 This, together with Sarkhushak further north on the Bamiyan valley and Shahr-i Barbar and Chehel Burj (Figs 6.54, 6.55) to the west of Bamiyan beyond Band-i Amir,109 were a series of equally substantial fortresses which, along with hundreds of smaller forts and towers recorded by Marc Le Berre,110 ringed Bamiyan in late antiquity and the

391

392

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.44  The 53-metre Buddha at Bamiyan. Note the bench at the foot for scale.

from the kushans to the shahis

Figure 6.45  The 38-metre Buddha at Bamiyan.

393

394

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.46  The standing Buddha at Kakrak.

from the kushans to the shahis

Figure 6.47  Seated Buddha at Bamiyan.

395

396

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.48  Wooden architecture imitated in stucco in one of the caves at Bamiyan.

Figure 6.49  Painting from Bamiyan in the Musée Guimet.

from the kushans to the shahis

Figure 6.50  Wall painting in the niche of the 38-metre Buddha.

397

398

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.51  Plan of the caves at the foot of the 54-metre Buddha (top) and 38-metre Buddha (bottom) at Bamiyan.

from the kushans to the shahis

early Islamic period and which appear to be a single complex fortification system. At Chehel Burj a series of paintings in the Sasanian style that appears to depict scenes resembling the Shah Nameh was recorded by Lee in 2000.111 Other cave complexes are known, but they show neither the complexity of Bamiyan nor its richness in sculpture and painting decorations: we may recall that of Haibak, close to the famous rock-cut stupa of Takht-i Rustam (Fig. 6.56),112 that of Hazar Sum, near Haibak, the date and even function of which are still somewhat obscure,113 the Buddhist caves of Fil Khana near Jalalabad,114 those of Basawal,

Figure 6.52  Plan of the fortifications at Shahr-i Zohak.

399

400

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.53  The fortress of Shahr-i Zohak.

Figure 6.54  The fortress of Chehel Burj.

from the kushans to the shahis

Figure 6.55  Detail of the decoration and arrow slits at Chehel Burj.

Figure 6.56  The rock-cut stupa at Haibak.

401

402

THE ARCHAEOLOGY OF AFGHANISTAN

between Jalalabad and the Khyber Pass,115 those at Humay Qala near Qarabagh-i Ghazni and others in the same zone of the Ghazni province, discovered in 1976 by the Italian Mission.116 Deborah Klimburg-Salter regards the Buddhist centres of Bamiyan, Kabul and Ghazni as part of a single complex Buddhist communications network in eastern Afghanistan between the seventh and tenth centuries.117 Stupa-monastery Complexes Afghanistan is very rich in monumental Buddhist stupas built of stone, chiefly preserved in the Kabul Valley and Kapisa (Figs 6.2, 6.19). These are usually attributed to the Kushan period, and a chronological sequence is considered in Fussman’s work on the Monuments bouddhiques de la région de Caboul.118 One of the best-known and best-preserved (partly because of two programmes of conservation work in the 1960s and again in the 1970s) is the stupa-monastery complex of Guldarra in Logar (Fig. 6.57).119 It was initially attributed to the third to fourth century on the basis of a gold medallion of Mokadphises found by Masson in the relic chamber, but a more recent study by Fussman has favoured a fifth to sixth century date.120 For the time being we may mention the groups of Jalalabad (Hadda) and Kabul, surveyed by the Japanese Mission,121 the group of Wardak,122 the stupas of Tepe Sardar,123 that of Kuh-i Muri,124 and that of Kandahar.125 The Kandahar stupa has been regarded as marking the westernmost spread of this type

Figure 6.57  The stupa at Guldarra.

from the kushans to the shahis

of monument, but the Smithsonian Helmand-Sistan surveys in the 1970s revealed a stupa and monastery complex with artificial caves at Khana Gauhar on the east bank of the Helmand seventeen kilometres south of Lashkargah, and another stupa at Khwaja Hasan some thirty kilometres further south dated (with many a caveat) to the Parthian period, although Sasanian material was also found at both sites.126 Sculpture The problems concerning sculptural and pictorial works of art are also many and many-sided. We have already briefly discussed the ‘dynastic’ images of Surkh Kotal (Fig. 6.32); the fragments of unbaked-clay sculptures are totally different from the point of view of style and show clear influence of the Hellenistic tradition.127 We can also recognise a similar influence in the unbaked-clay sculptures from Khalchayan, Transoxiana, that Pugachenkova128 considers one of the earliest expressions of art commissioned by the Kushans (first century bc) and of the greatest importance for the understanding of later developments.129 We are in a phase of transition from the Greek art of Bactria to Gandharan art, the documents of which are neither so numerous nor so well-known as those of Gandhara. Nevertheless, excavations have increasingly provided clear definition to this tradition, which flourished in a period when Afghanistan was not yet culturally Indianised. At the same time, an artistic output of Classical tradition is known also from the Khyber Pakhtunkhwa Province of Pakistan: here the excavations of the University of Peshawar130 have shown that, with very few exceptions, this region had no Hellenised products in the time of the Graeco-Bactrian kingdoms but only a traditional terracotta art represented by the so–called ‘Baroque Ladies’131 and other mainly female figurines of a very simplified shape. It is only during the following periods, called ‘Scytho-Parthian’ and ‘Early Kushana’ by Dani, that a Hellenising taste spreads over this area. The excavations of the Italian team at Butkara in Swat between 1956 and 1962 have provided a more refined framework in terms of dates and style.132 Here, then, is documented a Hellenising artistic phase that precedes Gandharan art and differentiates itself from it by the absence of Indian elements. It can well be considered as the expression of social strata broader than those that commissioned the purely Greek art of Aï Khanoum and obviously also broader than the groups, closely connected with the dynasty, that caused the unbaked-clay sculptures to be made in the official building of Khalchayan and in the sanctuary of Surkh Kotal. The Italian excavations in the earlier layers of Tepe Sardar have thrown some fresh light on this period (Figs 6.58, 6.59). These layers are simply a thick filling that derives from the destruction of a rich decorative complex made of unbakedclay sculptures (Figs 6.60–6.62) (most of them accidentally burnt by the fire that destroyed the sanctuary) that are to be connected stratigraphically with stupas similar in technique to those of the Kabul–Kapisa and Jalalabad areas.133

403

404

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.58  Plan of the stupa-monastery at Tepe Sardar.

Figure 6.59  The main stupa at Tepe Sardar.

from the kushans to the shahis

Figure 6.60  Tapa Sardar: male head from earlier period, Bodhisattva (?); originally unbaked clay, height 18 cm.

Figure 6.61  Tapa Sardar: bearded head from earlier period, Bodhisattva Vajrapani (?); originally unbaked clay, height 5.5 cm.

The clay sculpture of the earlier phase at Tepe Sardar, in its manifold aspects, is undoubtedly to be included in the tradition of Bactrian Hellenism and shows affinities with the clay images from Surkh Kotal, with those from Tepe Maranjan, near Kabul, a Buddhist sanctuary, probably of the sixth to seventh century,134 and also with the sculpture of Transoxiana, later than Khalchayan, found at Dalberjin Tepe, attributed by Pugachenkova135 to the second century ad, as well as with Gandharan art.

405

406

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.62  Tepe Sardar: colossal gilded foot from Room 100, Antique Period 2.

It is therefore evident that when Gandharan art was flourishing in northern Pakistan and in the eastern region of Afghanistan, the other parts of this country had a cognate though different art of Hellenistic tradition, which is nevertheless documented also in the eastern region itself, at Hadda as the Afghan excavations have shown. Other promising sites from this point of view are Basawal,136 Tepe Narenj137 and Mes Aynak.138 Even Gandharan art proper calls for a thorough re-examination of its Afghan products. This is usually considered as being characterised by the almost total absence of schist reliefs and by the predominance of stucco in an output which is placed between the third and fifth centuries ad. The only well-known Gandharan schist reliefs from Afghanistan are those from Paitava, Shotorak and Begram (Kapisa) (Figs 6.62, 6.63, 6.64), and are closely linked to a dynastic environment.139 Hadda, better known for its stucco sculpture, also produced schist reliefs (Fig. 6.65).140 Further Gandharan schist material was found at Kham-i Zargar (Kuh-i Muri).141 Other, still little-known, stylistic trends in the Gandharan style from Afghanistan have been located in the region of Qunduz, which yields typical limestone reliefs (6.66).142 The Afghan excavations at Tepe Shotor, Hadda (Figs 6.70–6.73) revealed the presence in the same sanctuary (and contemporarily in use if not contemporary

from the kushans to the shahis

Figure 6.63  Schist Buddha statue from Paitava in the Musée Guimet.

407

408

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.64  Schist relief from Shotorak in the Musée Guimet.

Figure 6.65  Schist reliefs from Chakhil-i Ghundi, Hadda, in the Musée Guimet.

from the kushans to the shahis

Figure 6.66  Relief from Qunduz in the National Museum.

in execution) of purely Gandharan stucco sculptures143 and of unbaked-clay sculptures that rather recall those of Tepe Sardar and Tepe Maranjan and those from the earlier French excavations at Hadda (Figs 6.76, 6.77).144 An examination of the stratigraphic and structural connections between the stupas decorated with stucco images and the niches containing unbaked-clay sculptures (some of them in the round, as in the case of the so-called ‘aquatic niche’) was never carried out before they were destroyed in the fighting;145 for the time being we must content ourselves with the group of much-worn Sasanian bronze coins found in one of the latest stupas, that seem to belong to Shapur III (ad 383–388), and with the possible comparisons with other sculptural complexes in the area. It is quite probable that

409

410

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.67  Map of the sites in the Hadda-Jalalabad area.

from the kushans to the shahis

Figure 6.68  Hadda: plan and front view of Stupa 121 at Tapa Kalan.

411

Figure 6.69  Plan of Tepe Shotor at Hadda.

Figure 6.70  Aerial view of Tepe Shotor.

from the kushans to the shahis

Figure 6.71  The ‘Heracles’ relief from Tepe Shotor at Hadda.

413

414

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.72  The ‘fish porch’ from Tepe Shotor at Hadda.

the unbaked-clay sculptures of Tepe Shotor do not cover a short period but are the result of additions over a considerable span of time; this would explain the great differences between some of the most Hellenising images, such as the HeraclesVajrapani Fig. 6.71),146 and the ‘aquatic niche’ (Fig. 6.72)147 that even recalls a clay relief from Panjikent.148 Classical Connections and Chronology The study of Gandharan sculpture presents many problems. The story of its rediscovery since the nineteenth century is sad tale of despoliation of sites, looting, accidental discoveries and poorly recorded excavations. The majority of objects in museum collections are isolated from their contexts, with very few from controlled excavations. The problem is further exacerbated by a flourishing industry of faking Gandharan art. Consequently, it has not always been possible to date much of the sculpture closer than to the first to sixth centuries ad, and chronology is the biggest problem that continues to plague the study of Gandharan art.149 Too many objects are divorced both from their sculptural and archaeological contexts and from the broader social and historical contexts that might provide us with the information with which to fill the many gaps in their evolution. Gandharan art and the academic questions which surround it, perhaps more than most other art styles, is a victim of its own intrinsic collectability and value.

from the kushans to the shahis

Figure 6.73  Stucco bust from Tepe Kalan at Hadda in the Musée Guimet.

415

416

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.74  Stucco head from Tepe Kalan at Hadda in the Musée Guimet.

Even when images are dated, the problem is not necessarily solved. To begin with, it is generally regarded that only five dated Gandharan images are known, but the dates themselves are controversial. In one such dated image, for example, a seated Buddha in the Peshawar Museum, the inscribed date of ‘89’ has been interpreted as equivalent to dates ranging from ad 216 to ad 432, a difference of more than two centuries, depending upon which calendrical system it refers to (Table 6.2).150 Such disparities are almost unknown in any other field of art history. Considerable discussion has revolved around the origin of the Buddha image, thought to derive from Classical figural art. Before about ad 100 (the date is subject to dispute), the Buddha was depicted in Indian art in the abstract, for example as a footprint. Central to the argument has been the date of the Buddha images depicted on the gold Bimaran casket from Hadda, now in the British Museum (Fig. 6.75). The current consensus is for the second half of the first century to the first half of the second century ad for the deposition of the casket, on the basis of a combina-

from the kushans to the shahis

Figure 6.75  Composite image of the gold reliquary of Bimaran in the British Museum.

tion of stylistic analysis and the dates of the coins buried with it. However, arguments have ranged from an older date (i.e. the casket might represent a reburial of a precious object) to a much later date, the coins merely providing a terminus post quem. Suffice it to say, a precise date still remains elusive.151 We do know that by the time of Kanishka, images of Buddha were beginning to appear on Kushan coins,152 that is, after ad 127/8 according to the most commonly accepted date (reviewed above). The undoubted Classical appearance of much of the Buddhist art of Gandhara in the first few centuries ad has prompted wide speculation.153 Gandharan art – usually sculpture – is almost entirely Buddhist in subject matter but displays essentially Classical styles. The term ‘Romano-Buddhist’ was first coined to describe this in 1876. The characteristic style consists of relief panels adorning virtually every wall surface of stupas, as well as associated chapels and monasteries. These wall surfaces are often divided horizontally by Classical entablatures, with relief panels separated by pilasters, usually in the Corinthian order. More Classical in style is the sculpture itself. The closest similarities are with the art of the Roman Near East of the mid-second century, particularly sarcophagus art. Recurring motifs include mythical beings from Greek mythology, such as ichthyocentaurs, tritons, gryphons, Atlas figures – often depicted supporting Classical entablatures – cupids, garlanding and other details of Classical ornament, such as tendrils, bead and reels, egg and darts, palmettes, and the almost universal use of the Corinthian order. Many Indian deities are depicted in well-known Classical guise: Vajrapani, for example, is depicted as Heracles and Hairiti as Tyche, and Dionysiac imagery constantly recurs. There is even a puzzling Trojan Horse relief that could only have been inspired by Virgil rather than by Homer. Such works bear almost no relation to contemporary or earlier Indian sculpture, and almost as little with the Hellenised art of Palmyra or Parthia.154 Perhaps most striking are the large numbers of portrait busts in Gandharan sculpture, particularly the stucco sculpture from Hadda in Afghanistan (Figs 6.73, 6.74). Some, such as the Heracles relief from Tepe Shotor Fig. 6.71), appear almost identical to Roman art in Asia Minor or Syria of the first to second centuries ad.

417

418

THE ARCHAEOLOGY OF AFGHANISTAN

Most of the Hadda stuccoes are generally thought to be later, occurring after the third century, although there are problems with dating. One authority sees these stuccos as ‘essentially Greek figures, executed by artists fully conversant with far more than the externals of the classical style’.155 They are usually heads of Buddhas or Bodhisattvas which form part of a relief narrative, but now occur mainly in isolation in museum collections. Such busts closely resemble Roman portrait sculpture – the resemblance of one Bodhisattva to the ‘Antinous’ sculptures of the Hadrianic period is a particularly famous example (Fig. 6.73).156 Much of the ‘Roman’ style of Hadda appears to anticipate European Gothic art a thousand years later: just as late Roman art in the West evolved into Gothic, late RomanoBuddhist art in the East evolved into the ‘so-called Gothic art of Hadda’.157 There are three channels from which this art might be derived: first, as a natural evolution of the Hellenistic art of the same area (Bactrian Greek), a parallel evolution to the development of Roman art itself in the West; second, as direct influence by itinerant groups of artists from the Roman Empire; and third, as indirect influence stemming from Romanising elements in Iranian art.158 Its evolution from Greek art of the same region – the Hellenistic successor states of Bactria and India – at first appears the most plausible. After all, the Hellenism was undoubtedly there, and the Roman appearance of the later art seems as logical an evolution as Roman art in the West, which evolved similarly from Hellenistic art.159 The main objection, however, is that the Greek kingdoms disappeared long before the appearance of the first syncretic Gandharan art style, with the Hellenism of Bactria dwindling almost to ‘vanishing point’; there seems no smooth, artistic transition such as occurred between Hellenistic and Roman art in the West. Consequently, there are too many missing links to enable a convincing evolution to be documented, and only the arrival of entirely new contacts with the West can explain the style. Discussion, therefore, has revolved around efforts to date many of the pieces with demonstrable Western influence as early as possible. After the initial discovery of Gandharan art, these arguments gradually gave way to those who favoured a more direct Roman origin for the style. But the argument still has much to recommend it. The archaeological record indicates far stronger continuity in the region than previously believed, with the invaders promoting and continuing Hellenism rather than extinguishing it: the Classical elements at Surkh Kotal, for example (Fig. 6.31), or the continued use of the Greek alphabet. The absence of artistic links, therefore, does not mean that such links never existed. Finally, relatively recent studies have focused on a hitherto neglected aspect of Gandharan art. These are the so-called palettes or ‘toilet trays’, dated between the late second century bc and the first century ad, belonging to a Hellenistic tradition that originated in Anatolia or Alexandria. The art of these stone palettes has done much towards filling gaps in the evolution.160 Most discussion, however, has centred on the second hypothesis, that the art must have been a product of journeymen craftsmen from the Roman east after the first century ad.161 Support for such hypothetical Roman sculptors roaming the

from the kushans to the shahis

East is seen in the ‘Yavana’ (i.e. Greek, presumed Roman) artisans mentioned in southern Indian literature at the time, or the apocryphal story of the journey of St Thomas – a carpenter – to Taxila in the first century ad. The collection of such a travelling artist’s stock of finished and unfinished Graeco-Persian gems was found at Taxila.162 The spectacular discovery of the hoard of Roman glass at Begram in Afghanistan certainly demonstrated links with the art of the Roman world (Figs 6.8–6.10). While these were merely trade items, and not Roman works of art produced locally, Begram and the other trade links reviewed above in this chapter may provide a background for the possibilities of artists arriving in the wake of the objects. Other speculation has emphasised the importance of Hellenistic elements in Iranian art of the same period. The sculpture at Palmyra, Hatra and Shami in south-western Iran, as well as the paintings at Dura Europos, belongs to this ‘IranoHellenistic’ style. The stucco medium of, for example, the Hadda busts, which are generally late, is also thought to be inspired by Iranian art which favoured the stucco medium, particularly after the Sasanian occupation of Gandhara in 241.163 But while such connections undoubtedly exist, the Roman element is far more pronounced than the Parthian or Palmyrene, requiring more complex explanations which suggest more direct links. The similarities between Classical and Gandharan sculpture are self-evident – and overwhelmingly persuasive enough to be convincing of artistic links. Most of all, it must be pointed out that the controversies over Graeco-Bactrian versus direct Roman versus Irano-Hellenistic origins for Gandharan art are not in conflict: all hypotheses must be substantially correct. To argue for one hypothesis over the others is to miss the point.164 Pictorial Art Unfortunately, very little or nothing is known of pictorial art in Afghanistan during the Kushan period, but we can easily imagine that the Gandharan school of sculpture had its counterpart in painting, just as it had in Central Asia, for instance at Kara Tepe, near Termez165 and even at Miran, in Xinjiang.166 Only a few remains survive from Hadda that might be roughly attributed to the Kushan period (Fig. 6.76).167 The importance of Afghanistan from the point of view of the history of painting rested entirely on the later Buddhist wall decorations from Bamiyan and the nearby cave complexes of Kakrak and Foladi (usually attributed to the fifth to seventh centuries ad) and a few other examples from Dukhtar-i Nushirvan,168 Fondukistan,169 Tepe Sardar and Basawal,170 datable to the seventh to eighth centuries ad. In this context the Sasanian painting at Ghulbiyan assumes considerable importance. It was while searching for the ‘Ferrier relief’ south of Sar-i Pul in 1978 that Jonathan Lee discovered a large painting in a rock-shelter not far from the purported location of Ferrier’s relief (Figs 6.77, 6.78). It depicts two groups

419

420

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.76  Painting from Bagha Gai at Hadda in the Musée Guimet.

from the kushans to the shahis

Figure 6.77  The Sasanian painting at Ghulbiyan.

Figure 6.78  Interpretive drawing of the Ghulbiyan painting.

421

422

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.79  Dilbarjin Tepe: wall painting in the temple depicting Shiva and Parvati.

of worshippers, a king with four standing male donors to one side and their female consorts on the other worshipping three enthroned deities, with a hunter aiming at an ibex. At first attributed broadly to the third to fifth century, a more precise fourth or early fifth century date is now favoured for Ghulbiyan, and it is interpreted as a mountain sanctuary like Dukhtar-i Nushirvan dedicated to the god Tish.171 Let us also record the discovery of a polychrome painting (homage to a hero?) at Dilbarjin Tepe (Fig. 6.79), forty kilometres north-west of Balkh.172 It is to be compared with the paintings from Balalyk Tepe (southern Uzbekistan) which the Soviet archaeologists date to the end of the fifth/beginning of the sixth century ad.173 The dating proposed for this painting of Dilbarjin is first half of the fifth century, on the basis of coin evidence. But this painting is not the earliest among those found at Dilbarjin; indeed, it appears to be one of the latest. The temple of Dilbarjin, in its first phase (attributed by Kruglikova to the Graeco-Bactrian period), had its façade decorated with a painting representing the Dioskouri with their horses (Chapter 5), while another painting on a later wall of the same temple depicts Shiva and Parvati

from the kushans to the shahis

seated on Shiva’s vahana, the bull, between devotees.174 Kruglikova does not commit herself on the chronology of this Shiva painting, nor does she propose any definitive chronology for the other paintings of Dilbarjin. Iconography does not seem to be very useful in the dating of the Shiva panel, especially because the two deities’ heads are lost; moreover, we have to rely chiefly on line drawings and watercolours (which, even when accurate, are obviously interpretations) since neither colour nor black-and-white photographs have been published. We may only point out that the devotee standing on Uma-mahesvara’s left side wears a long kaftan, bound at the waist, that looks much more like some figures in the wall paintings from Balalyk Tepe or Panjikent than any known representations of the Kushan period.175 Late Buddhist Sculpture Deborah Klimburg-Salter has emphasised the massive increase in late Buddhist art and archaeology between the seventh and ninth centuries, perhaps as a result of the expansion of the Turk Empire in Central Asia.176 Strictly connected with all these classes of sculptures are the unbaked-clay images from Fondukistan (Figs 6.80–6.82), a sanctuary excavated by the DAFA and attributed to the seventh century on the ground of some Arabo-Sasanian coins,177 and those from the later phase of Tepe Sardar (Figs 6.83–6.85), presumably to be dated to the seventh to eighth and even into the ninth century ad.178 These two sites have provided documents of exceptional value from the point of view of both iconography and style, such as the ‘royal couple’ (Fig. 6.82) and the bejewelled Buddha179 from Fondukistan, the two Nagas supporting the stem of a lotus on which the Buddha was seated, a subject found both at Fondukistan and Tepe Sardar, and the colossal parinirvana Buddha at Tepe Sardar180 (Fig. 6.85) that finds a precise counterpart in the parinirvana of Ajina Tepe (southern Tajikistan), a site that has yielded sculptural and architectural material very similar to that of Afghanistan.181 Another Central Asian Buddhist site quite close to the sanctuaries of Afghanistan is Kuva, in ancient Ferghana.182 For the history of religion, the most interesting find is certainly the image of Mahishasuramardini (a form of the Hindu goddess Durga) found at Tepe Sardar.183 This is the first instance of a Hindu deity placed in an otherwise purely Buddhist context in Afghanistan. It has been observed184 that the sanctuary of Tepe Sardar probably belonged to the upper classes: we have but scant evidence, for instance, of those ‘popular’ cult objects such as miniature stupas and inscribed tablets of clay (Fig. 6.87) that have been found in numbers at other sites near Ghazni and belong to the same late period.185 Also, the introduction of a Hindu image into the sanctuary seems to support this view, since the marble sculptures of the ‘Shahi’ period, which were certainly produced for the upper classes and the court (as it is shown by their small number and comparatively previous medium), almost exclusively represent Hindu deities.

423

424

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.80  The so-called ‘royal couple’ from Fondukistan in the Musée Guimet.

This is, therefore, one more point that archaeology is able to make clear in the religious history of pre-Muslim Afghanistan, a subject that already owes much to field research.186 Nevertheless, we do not have sufficient evidence to place these cultural data in a more precise historical context, such as we do have in Central Asia, where the farm-palace of Balalyk Tepe,187 for instance, tells us so much about the social and economic organisation and therefore allows us to understand more fully the ideological background of the paintings found there and, to some extent, elsewhere in Central Asia. Most of such ideological background we can surmise for Afghanistan and Central Asia, though historical-religious data often point to different directions.188 Non-Buddhist Art The fifth century ad is indeed a crucial point for the North West of India and Afghanistan, and it is to this century that the marble image of Surya from Khair

from the kushans to the shahis

Figure 6.81  The bejewelled deity from Fondukistan in the Musée Guimet.

425

426

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.82  The bejewelled Buddha from Fondukistan in the Musée Guimet.

from the kushans to the shahis

Figure 6.83  Tepe Sardar: row of clay stupas and thrones on the eastern side of the main stupa; later period, c. eighth century.

Khana was first attributed (Fig. 6.88).189 Scholars are now rather inclined to date this and the other marble sculptures of Hindu subjects from north Pakistan and east Afghanistan to the period of the Shahi dynasties (Turki Shahis and Hindu Shahis; only Turki Shahis according to Kuwayama),190 usually between the seventh and tenth centuries.191 Nevertheless we must confess that very little stratigraphic evidence is available and that the soundest data are provided by comparisons with the bronzes, mainly Buddhist, produced in Swat and Kashmir between the eighth and tenth centuries ad.192 Although these bronzes do not appear to have enjoyed a wide diffusion in Afghanistan, some of them are reported to have been bought there and one was certainly found in Hazarajat, central Afghanistan, but was later lost.193 Nor do we get much help from another almost contemporary non-Buddhist religious monument, which in any case deserves a special mention because it appears to be unique in Afghanistan, although it is reasonable to think that further excavations and surveys will provide parallels: the site is Chigha Sarai in

427

428

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.84  Tepe Sardar: detail of the unbaked-clay sculpture in chapel 37: left hand naga; later period, c. eighth century.

the Kunar valley, where some ancient fragments incorporated in the tombs of a Muslim cemetery were noticed by members of a Danish expedition between 1947 and 1954.194 Van Lohuizen de Leeuw put them in the right historical perspective by stating that the fragments from Chigha Sarai indicate the existence at that site ‘of a temple belonging to the middle phase of the mediaeval architecture of North West India of about the eighth or the ninth century’, that the shrine was possibly dedicated to the linga cult, and that ‘the stones in question prove for the first time that the North West Indian style of medieval architecture extended as far as eastern Afghanistan, a fact which the previous finds of images belonging to the contemporaneous school of sculpture had already made highly probable’.195 A Hindu temple dating back to the Turki Shahi period has been discovered in Swat, at Barikot.196

from the kushans to the shahis

Figure 6.85  Tepe Sardar: detail of the unbaked-clay Parinirvana Buddha in shrine 63; later period, c. eighth century.

Figure 6.86  Tepe Sardar: multiple mould (also other side) used to obtain decorative plaques for clay stupas, thrones, etc.; plaster, 14.4 × 22 × 3.5 cm. From left: pilaster with shaft made of amalakas; standing Buddha in varadamundra; Yaksa/Atlas; seven-petalled rosette. Later period, c. eighth century.

429

430

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.87  Gudul-i-Ahangaran; Ghazni: an inscribed clay tablet from inside of miniature stupa; Buddhist profession of faith.

The Japanese excavations at Tepe Skandar have provided very good chronological data for the ‘Shahi’ marble sculptures: a group of Shiva and Parvati (Umamahesvara) was found there that cannot be assigned to a period before the seventh to eighth century, or even later, on grounds of both palaeographical and archaeological evidence.197 Both these classes of objects, Hindu marble sculptures and bronze images, are to be considered, along with the temple of Chigha Sarai and some at least of the paintings of Bamiyan, as documenting a widespread Indianisation. This phenomenon is to be understood in terms of at least two Indian trends, one being north-western (temple of Chigha Sarai and Buddhist bronzes), the other rather Gupta in tradition (especially the ‘Indian-style’ paintings of Bamiyan). We may recall another class of objects, small in size, made of a very compact ‘schist’ stone (casket lids, portable sanctuaries, small images, etc.) that were probably produced in the north-west of India at the same time as the bronze images and were certainly also introduced into Afghanistan, although there is no evidence that they were also made there; these objects are characterised by a peculiar technique of very flat relief and a very clear ‘Sasanian’ influence.198

from the kushans to the shahis

Figure 6.88  Marble statue of Surya from Khair Khana in the Musée Guimet.

431

432

THE ARCHAEOLOGY OF AFGHANISTAN

Everyday Life Lastly, we cannot disregard the importance of digging in the few cemeteries to be found from pre-Muslim Afghanistan. One of these, at Said Qala Tepe, about twenty-five kilometres west of Kandahar, has yielded very few funeral offerings, but the study of the skeletons could lead to some tentative conclusions about the social structure of the people buried there.199 The graves have been attributed to the ‘Kushano-Sasanian’ period200 and are probably to be connected with the puzzling cave settlement of the nearby Shamshir Ghar.201 The exploration of the latter site, and of Aq Kupruk (Balkh Province, north Afghanistan), has led Dupree to point out: While the monumental religious sites and urban secular centres were in full swing, and elaborate commercial routes criss-crossed the area, the peasanttribal society dominated the socio-economic scene; it is thanks to these cave sites that we have now some evidence of the daily, annual, and life cycles of the common people, the bulk of the population, the perpetuators of the fundamental cultural patterns. The occupations of the caves during this period were probably temporary (as in modern times), the occupants being either nomads, semi-nomads, or semi-sedentary groups.202 The discovery and assessment of the Bactrian documents (see below) from northern Afghanistan have complemented the less monumental evidence from archaeology that Dupree discusses, and have provided us with copious details about daily life in the period.203 The cave of Shamshir Ghar,204 possibly a refugee site, contained a variety of Kushano-Sasanian objects: Indo-Sasanian and Sasanian seals; Red Streak-Pattern Burnished pottery; iron and bronze horse trappings; and bronze (particularly the trilobate type) projectile points, etc.205 Two rock-shelters at Aq Kupruk206 yielded specimens of peasant-tribal art and utilitarian objects of the period under discussion. The excavators uncovered two levels at Aq Kupruk I (Snake Cave), with the following ranges in C-14 dates: Early period, ad 200–450; Later period, ad 450–600. Important finds of the former included: flint and bone implements; many unidentified iron fragments; several bronze trilobate projectile points, socketed projectile points, bracelets, etc.; glass, terracotta, carnelian beads. An extensive painted pottery series was found, including black on buff surface, but red on buff dominated. Designs included: free-flowing, repeated spirals; wavy lines; chequer motifs; naturalistic and stylised faunal and floral designs, etc. Undecorated red and buff utilitarian wares occurred in great profusion. Faunal remains have been identified as domesticated sheep, goat, cattle and horse. The later period finds in Aq Kupruk I included a series of livestock-retaining walls of pakhsa and sun-dried bricks and several large storage jars. A series of fragmented, defaced Buddhist paintings occurred in an upper chamber. The pottery

from the kushans to the shahis

consisted of Red Streak-Pattern Burnished wares (designated ‘spiral-burnished ware’ at the Kandahar excavations), a variety of painted wares (red or black on buff surface being dominant, repeated spirals the dominant motif) – plus a buff ware with wavy, comb-marked striations, punctation and appliquéd sheep/goat horns. Aq Kupruk IV (Skull Cave) contained an intensive burial area (10–11 human skeletons) and probably dated from the fifth to sixth centuries ad. Grave furniture (richer than that found at Said Qala Tepe) included: two complete Red StreakPattern Burnished plates; a pottery lamp, a cup and an unguent(?) jar; a bronze mirror and other bronze ornaments and projectile points; iron weapons and horse trappings; a silver ring with a lapis setting; carnelian and lapis lazuli beads. Another site of the period, Tepe Shahidan,207 just east of Khulm (Samangan Province, northern Afghanistan), has yielded an important ceramic sequence ‘and should illuminate the daily lives of the peasant farmers who furnished the economic base for the Hephthalite satrapies north of the Hindu Kush and the Kushan satrapies of the south’.208 One may further mention a sondage in the mound of Pol-i Zak, near Qala Shaharak in the western Hindu Kush mountains about 260 kilometres east of Herat,209 where the C-14 dates of the earlier stratigraphic period fall into the Sasanian (possibly late Hunnic) period; and the results obtained by Leshnik at Qala Ahingaran, also in central Afghanistan,210 that seem to agree with the dates from Qala Shaharak, ‘although many of the painted pottery motifs superficially resemble those of the late Indus valley ceramics’.211 This ‘minor’ culture of pre-Muslim Afghanistan does not seem to share many of its characters with the much richer and better documented nomadic burial sites of the ‘Kushan’ period in the Kafirnigan valley northern Bactria (southwestern Tajikistan).212 These are actually older (end of the second century bc to the beginning of the first century ad) and seem to be the remains of the nomad tribes that crushed the Graeco-Bactrian kingdoms. Nomad burial sites in Inner Asia are characterised by the distinctive kurghans that can be found from Ukraine through to Mongolia. The only such burial site that might be dated to the Kushan period in Afghanistan was recorded near Rustaq in Takhar Province (a name associated with the Tocharians/Kushans as we have noted). This comprised ‘a well-defined zone of [about ten] kurghans … The configuration of the site suggests a necropolis, in the Kushan tradition (kurghans).’ However, the only sherds on the surface were Islamic.213 Only two ‘kurghan’ burial sites have been excavated in Afghanistan. Le Berre excavated nine such tumuli at Shakh Tepe in Qunduz Province, a flat plateau with some one hundred tumulus burial mounds, stretching for about two kilometres. Finds included a gold Byzantine coin of Anastasius (491–518), gold jewellery set with semi-precious stones, and joint human and animal burials, so might be related to the various Hunnic incursions.214 The other was at the extensive ‘barrow cemetery’ at Kandahar of several hundred tumulus burials to the south of the old city, larger than the walled city itself (Fig. 6.89). Here Whitehouse excavated two burials, dated conclusively to

433

434

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.89  The barrow cemetery at Kandahar.

the Ghurid period, despite the notable paucity of Ghurid material recorded in the excavations of the old city itself.215 Problems of Dating The study of town sites in Afghanistan is often based on criteria that are not agreed by all scholars. Such is the case with square and round towers in the town walls; Fussman, dealing with the dating of the ancient town in the Wardak valley (foundation before or in the time of Huvishka), summarises the problem as follows, starting from the fact that the town walls of Wardak are strengthened by round towers.216 Central Asian town walls, at least since the Graeco-Bactrian period, have square, often massive, towers or bastions (e.g. at Aï Khanoum, Balkh, Shahr-i Banu, Dalberjin in Bactria).217 South of the Hindu Kush, this technique is known at Begram, Tepe Narenj and Sirkap.218 These square towers continued to be in use in the Kushan period, even when the structures underwent works of enlargement or modification. At Surkh Kotal, built under Kanishka, the outer temple wall and the town walls have only square towers.219 The same is true of the Gandharan reliefs that reproduce fortification walls, most of which are to be dated in the

from the kushans to the shahis

Kushan period. That is why, Fussman says, Schlumberger thought that Sirsukh (Taxila) with its semi-circular towers220 was not a foundation of Vima Kadphises or Kanishka, as Marshall and Ghirshman221 believed, but belonged to the postKushan period. Since the exclusive use of round towers in a complex system of fortifications is not known in Parthian Iran and first appears probably under Shapur I,222 Schlumberger thought that Sirsukh could not be earlier than ad 260. If the first year of Kanishka is to be placed in the early second century, both Sirsukh and the town in the Wardak valley are necessarily much later than Kanishka and his successor Huvishka. Nevertheless, as we have already mentioned, the pottery sherds collected by Fussman in the town of Wardak are all Kushan, which again makes plausible the dating of Marshall and Ghirshman. It seems therefore that north-west India and Afghanistan in the second century ad employed semicircular and probably also round towers in their fortifications.223 This subject is dealt with also by Kuwayama,224 who begins by stating that the bastions in the Amu Darya valley are traditionally square, covering the time span from the period of the Greek rulers to that of the Sasanian hegemony. According to him, the round bastion at Kohna Masjid, near Surkh Kotal,225 which is a late addition (the site being contemporary with the Sasanians and later than the Great Kushans), is ‘an exceptional intrusion and repercussion from the neighbouring countries’. He adds that the homeland of this kind of bastion is presumably the region south of the Hindu Kush. In a different approach, Francfort analysed 120 fortifications in the Amu Darya basin as a whole (including northern Afghanistan) from the second millennium bc to the fourth century ad. In the 52 fortifications dating from the Kushan period he recognised a reassertion of the indigenous Central Asian styles of the Bronze and Iron Ages, characterised by hollow ramparts, after an intrusion of foreign styles during the Macedonian interregnum.226 Kuwayama also suggests that there is a close interrelation between the round bastions and the use of pottery with a stamped medallion decoration, peculiar to the regions south of the Hindu Kush. He recalls that this kind of pottery was recognised as a new device in Begram III by Ghirshman227 and was found at the Begram Bazar,228 the Saka fort,229 Tepe Maranjan,230 all sites where round bastions are also found. In his attempt at establishing a date for Begram III, Kuwayama gives great importance to ‘the simultaneous existence of both cultural elements, the round bastions and the stamp decorations of the medallion types, at Kohna Masjid’ – a simultaneity which is ‘against the tradition of the region’. This ‘should be a reflection of the phenomena’ that occurred on the south side of the Hindu Kush after the fifth century ad. The existence of round bastions and of that particular type of pottery on both sides of the Hindu Kush ‘should not be regarded as a chance event’. Therefore, Kohna Masjid is taken by Kuwayama as ‘a counterpart of the monuments south of the Hindu Kush’ that can give Begram III (when round ­bastions make their appearance) a date later than the one suggested by Ghirshman.

435

436

THE ARCHAEOLOGY OF AFGHANISTAN

This strict cultural relationship between the round bastions and the medallionstamped pottery does not seem to find further support elsewhere. For instance, Fussman remarks that his surface sherding in the ancient town of the Wardak valley has given practically no specimen of that kind of pottery.231 He thinks that the medallion-stamped pottery makes its appearance some time after the reign of Vasudeva, that is, towards the end of the second or the beginning of the third century ad.232 The same date is probably to be attributed to the miniature fortress with square, round and semi-octagonal bastions, found in the Buddhist complex of Tepe Sardar near Ghazni,233 which is also connected with medallion-stamped pottery.234 The same kind of pottery was also found at Guldarra in Logar (Lohgar),235 Jaghatu-i Wardak,236 Chanwar, near Gardez,237 Chaqalaq Tepe, near Qunduz238 and elsewhere. Epigraphy Bactrian Inscriptions and Texts239 According to the Hou Hanshu, the Chinese chronicle of the Later Han dynasty, the Kushans (Guishuang 貴霜) formed a part of the Yuezhi (月氏), a people formerly resident in Gansu in northwest China, who had settled in Bactria by the second century bc.240 The role played by the Yuezhi in Chinese sources seems to be comparable to that of the people known as Tokharoi or Tochari in western sources.241 It is still a matter of debate whether there is any connection between the Tochari and the Indo-European languages known to modern scholarship as ‘Tocharian A’ and ‘Tocharian B’, which were spoken in Agni (Karashahr) and Kucha to the north of the Taklamakan desert. There is no doubt, however, about the significance of Tukharistan, ‘the land of the Tochari’, a term commonly used as a name for Bactria in sources of the Islamic period but first attested in an inscription on a silver dish from the time of Kanishka I.242 Nothing definite is known about the ethnicity of the Tochari/Yuezhi/Kushans, although it has been claimed that the names of some of the Kushan rulers contain elements which may derive from an Iranian ‘Saka’ language.243 Be that as it may, when the Kushans became masters of Bactria they took over an administration for which Greek was the primary written language, as is clear from the numerous Greek documents and inscriptions found in Bactria.244 Most of these belong to the centuries preceding the arrival of the Kushans, but some may be of early Kushan date, for instance the text on a cauldron mould from the ‘Temple of the Oxus’ at Takht-i Sangin, which contains the Bactrian loanword molr, ‘seal’.245 The coins issued in Bactria by the early Kushans also bear inscriptions in Greek, a practice changed abruptly early in the reign of Kanishka I, when Greek was replaced by Bactrian, the local Iranian language, written in Greek script. The earliest Kushan inscription from Afghanistan, that of Dasht-i Nawur, has a date in Greek followed

from the kushans to the shahis

Figure 6.90  The Rabatak Inscription in the National Museum.

by a text in Bactrian. Near it is a Middle Indian inscription in Kharoshthi script, bearing the same date and possibly representing a version of the same text,246 and another inscription in an undeciphered script and language.247 These inscriptions date from the time of Vima Taktu, now known from the Rabatak inscription (see below) to be the grandfather of Kanishka I. Somewhat later, probably from the time of Kanishka, is the ‘Palamedes’ inscription found at Surkh Kotal, so called from the Greek ‘signature’ dia Palam≠dou, ‘by Palamedes’, which stands under the fragmentary text in Bactrian.248 Apart from the examples just mentioned, all of the Kushan inscriptions found in Afghanistan are in Bactrian alone. One of the earliest and the most important is that of Rabatak, which refers to Kanishka I as the current ruler and seems to mention events of his first, third and sixth years (Fig. 6.90).249 Primarily an account of the foundation of a dynastic temple on the orders of the king, the inscription is extremely informative, listing the numerous gods to be worshipped in the temple, Kanishka’s ancestors as far back as his great-grandfather Kujula Kadphises, and the cities of northern India as far as Campå in eastern Bihar, all of which, it is claimed, have submitted to Kanishka. An interesting but not quite clear phrase near the beginning of the inscription seems to allude to a policy of replacing Greek by Bactrian, as can be observed on the coinage of Kanishka. A recently discovered inscription on a silver dish contains a number of close parallels to the Rabatak inscription and mentions Kanishka’s return from India to ‘Tukharistan’ in the year ten.250 The author of this text is the official Nukunzik or Nukunzuk, who is also

437

438

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.91  The well inscription from Surkh Kotal in the National Museum.

named in the Rabatak inscription and who appears yet again as the author of the best-known Bactrian inscription, the well-inscription of Surkh Kotal (Fig. 6.91). This inscription, which exists in three copies (with minor, mostly orthographic differences), is concerned with the renovation of the sanctuary at Surkh Kotal in or soon after the year 31 of the era of Kanishka, that is, during the reign of his son Huvishka. From a historical point of view, the most significant information seems to be a reference to certain unnamed enemies, whose actual or threatened attack had led to the removal of the gods from the temple.251 Most other Bactrian inscriptions of Kushan times are too short or too badly preserved to convey much useful information. A six-line text engraved on the base of a statue found at Ayrtam in southern Uzbekistan was interpreted by Harmatta

from the kushans to the shahis

as containing the name of Huvishka and a reference to ‘year 30’,252 but nothing of the kind is legible on the stone itself. Even more fantastic are Harmatta’s reconstructions253 of other fragmentary Bactrian inscriptions such as those of Dasht-i Nawur and Dilberjin,254 or the minor inscriptions of Surkh Kotal. The latter include the façade inscription or ‘inscription pariétale’, of which only the last few words survive in situ,255 and the so-called ‘unfinished inscription’, which begins with a date, possibly the same date as that found at Dasht-i Nawur, but is otherwise totally illegible. In the Indian territories of the Kushan empire, inscriptions were naturally composed in Middle Indian languages and written in Kharoshthi or Brahmi script. These inscriptions include some on monuments erected on the orders of the Kushan kings themselves and others which refer to the rulers and members of their families.256 The Kushan Bactrian inscriptions are almost exclusively written in what has been referred to as a ‘monumental script’, in which each letter is written separately and (in principle) unambiguously, ligatures and cursive forms being quite rare. After the Kushan period the monumental script ceases to be used and is superseded by a ‘cursive script’ which suffers from several ambiguities and which continued to develop over the course of the following centuries. An early example of this cursive script is a bulla in the Ashmolean Museum, Oxford, which bears the impression of a seal of Kanishka II or (most probably) III, the text of which describes him as ‘king of the Indians’.257 Excavations at Buddhist monastery sites near Termez on the north bank of the Amu Darya have unearthed a number of Bactrian wall inscriptions (graffiti) in cursive script,258 mostly in a poor state of preservation, as well as fragments of pottery with short ownership inscriptions, some of them bi- or even trilingual (cursive Bactrian, Kharoshthi, Brahmi).259 This material belongs to the KushanoSasanian period, as do several inscriptions on silver or silver-gilt vessels.260 The most important of these bears a circular inscription announcing the dedication of the vessel to the god Mana, the Manao bago of Kushan coins, and referring to the ‘year 43’. Apart from one of the Kara-tepe wall inscriptions, dated in ‘year 35’, this is the earliest text bearing a date in a new era which is used in all dated Bactrian texts from this time onwards. It is convenient to refer to this era, which is never named in any text, as the ‘Bactrian era’. François de Blois has argued that its starting-point was Nowruz ad 223, the official date of the foundation of the Sasanian empire, which shortly afterwards took Bactria from the Kushans;261 according to an alternative proposal, the new era was instituted to mark the beginning of the second century of the Kushan era in ad 227 or 228.262 This is not the place for a discussion of the pros and cons of the two solutions, but it is worth observing that they differ by no more than five years, a margin of error which is insignificant for most historical questions. The Bactrian era is best attested in a series of documents on parchment or leather, wood and cloth, which began to come to light in the early 1990s and

439

440

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.92  A polyandric text from the Bactrian documents in the Hirayama Collection.

from the kushans to the shahis

Figure 6.93  A mention of Shahanshah Peroz in a text from the Bactrian documents in the Nasser D. Khalili Collection of Islamic Art. (Accession number doc 0002.1.)

whose number is now approaching 200. Unfortunately, these documents have no reliable provenance, but it is clear from the places mentioned that most of them were written in northern Afghanistan, chiefly in three regions: Rπb, modern Rui in the northern foothills of the Hindu Kush;263 Kadagstan, which probably lay to the east of Rπb, in the valley of the Qunduz River; and Guzgan, in north-west Afghanistan. One document is unique in being written in a place named Khesh, which seems to have lain to the south of the Hindu Kush, between Bamiyan and Kabul.264 Among these documents, about forty are dated in years ranging from 110 to 549 of the Bactrian era (= ad 332–771/2 if the era began in ad 223 as proposed by de Blois). With them are associated some 32 Arabic documents dated in the years 138–160 hijri (= ad 755–777),265 most of which refer to members of the family of Mir son of Bek, who are also known from the latest of the Bactrian documents. Virtually all of the Bactrian documents have now been edited and translated, mostly in the three volumes of Sims-Williams’ Bactrian Documents from Northern Afghanistan.266 Since this edition is chiefly focused on establishing the correct reading and literal meaning of the documents, there remains much work to be done on their interpretation as historical documents. A step in this direction is the volume Studies in the Chronology of the Bactrian Documents from Northern Afghanistan by Sims-Williams and de Blois, which examines not only the chronology of the dated documents but also all available data that may help to determine the approximate dates of the undated documents.267

441

442

THE ARCHAEOLOGY OF AFGHANISTAN

The Bactrian documents are principally letters and legal contracts (deeds of sale, gift, manumission, etc.) as well as economic documents such as receipts and lists of produce. The only ones with religious content are a few Buddhist texts, but many personal names are Zoroastrian and a couple of legal texts refer to the possibility of land being set aside for a dakhma as well as for a vihara. Many attest to the dramatic historical changes which occurred in the region, mentioning the Sasanian Kushan-shahs and later the Hephthalites, the Turks and the Arabs. Among the highlights, it is worth mentioning: the earliest dated document (year 110 = ad 332?), a polyandric marriage contract (Fig. 6.92), which shows that fraternal polyandry was customary in the region even before the arrival of the Hephthalites, to whom it has usually been attributed;268 two letters which refer to the shahanshah Peroz as overlord (Fig 6.93); other letters which contain the first known occurrences of the term ‘Afghan’ in any language (ca. ad 460; Fig 6.94); and the very last dated document (year 549 = ad 771/2?), which begins with a translation of the bismillah (Fig. 6.95). The documents, whose full potential is yet to be realised, are thus a rich source of information for the political history of Afghanistan as well as for the everyday life of the ordinary people between the fourth and eighth centuries.

Figure 6.94  The oldest mention of ‘Afghans’ in a text from the Bactrian documents in the Hirayama Collection.

from the kushans to the shahis

Figure 6.95  A translation of the Islamic invocation ‘bismillah’ in a text from the Bactrian documents in the Nasser D. Khalili Collection of Islamic Art. (Accession number doc 0117.)

443

444

THE ARCHAEOLOGY OF AFGHANISTAN

Bactrian inscriptions from the centuries covered by the documents are surprisingly few. Of the two inscriptions of Jaghatu, one is incomprehensible, while the other consists of a Bactrian transcription of the Buddhist triratna formula 269 ‘Homage to the Buddha, homage to the dharma, homage to the sam · gha’. Two further inscriptions found at Uruzgan seem to name a local ruler who claims to be ‘under the protection of the Sun and Moon’.270 More significant is the inscription of Tang-i Safedak near Yakaolang, which commemorates the establishment of a Buddhist stupa in the year 492 (= ad 714?) ‘when there was a Turkish ruler and an Arab ruler’.271 Finally, we must mention a group of inscriptions in Bactrian, Sanskrit and Arabic found in the Tochi valley in Pakistan and now preserved in the Peshawar Museum, which belong to the mid-ninth century. At one time these poorly preserved inscriptions were regarded as bilinguals and their dates (in the Bactrian, Laukika and hijrÈ eras respectively) were taken as providing synchronisms on the basis of which year one of the Bactrian era could be fixed as ad 232 or 233. However, they have lost much of their importance since a new edition of the texts shows clearly that even where the texts in two languages are engraved on the same stone they do not have the same content.272 The Era of Kanishka From the various provinces of the Kushan empire come an important series of inscriptions dated in the era of Kanishka. Those in Brahmi from Mathura and the upper Ganges valley are dated between years 2 and 98, and there is a second series with later dates from 1 to 57. Among the Bactrian inscriptions from Surkh Kotal, the Great Inscription, discussed above, has a date of 31 in the era of Kanishka. Most of the Kharoshthi inscriptions associated with this era come from the Indus valley – in particular the northern provinces of Pakistan and the western province of Afghanistan. Recorded dates range from year 2 to year 89, and this gives a firm relative chronology. There has been considerable dispute about the initial year to which Kanishka’s era should be referred.273 Many Indian scholars assert that it is identical with the Saka era of the western satraps in ad 78. Most Western scholars support a date in the early second century ad between ad 120 and 144, while Göbl274 argued for ad 230, Zeymal275 for ad 278 and Schindel for ad 227.276 Following the new evidence of Rabatak, Cribb narrowed the date to between ad 107 and 120277 and Falk argued for the date ad 127/8, which holds widespread support.278 Inscriptions Connected with the Kanishka Era The Afghan Kharoshthi inscriptions dated in Kanishka’s era are associated with Buddhist stupa deposits. Among the antiquities sent from Kabul to the East India Company by Masson was a brass casket which had an inscription on its lid: ‘in year 18 … the Gotama’s relic was enshrined.’ From a stupa at Hadda, eight

from the kushans to the shahis

kilometres from Jalalabad, Masson recovered a jar which contained a Kharoshthi inscription ‘written with a pen but very carelessly’. It was published by Thomas from a copy he found among Masson’s papers at the India Office in London. He read: ‘in year 28 … a relic was deposited in the king’s grove in a stupa by the architect Samghamitra.’ From a stupa near Wardak, 50 kilometres west of Kabul, Masson recovered a bronze reliquary some 25 cm high and 14 cm broad, now in the British Museum, and this is matched by another inscribed reliquary from the same stupa at Wardak in a private collection. The reliquary has a long inscription in four lines round its shoulder and circumference. The inscriptions dated in year 51 provide the name of the site, Khavada, together with the name of the donor and his father, Vagamarega and Kamagulya, the first a Bactrian and the second an Indian name, suggesting a Bactro-Indian family governing Wardak/Khavada on behalf of the Kushans. It records the establishment of the relic by Vagramarega in a stupa for the Maharaja Huvishka, and for the honour of the dedicator’s relatives, friends and associates.279 To these can be added several more inscriptions from stupa deposits from Jalalabad, Wardak and other unlocated regions in Afghanistan, now mainly in private collections. Undated Kharoshthi Inscriptions From Lalpura near Jalalabad comes a small stone relief of two wrestlers now in the Peshawar Museum.280 It has a short Kharoshthi legend in later Kushan script: Minandrasa, the Greek name Menander. Also of Kushan date in the second or third century ad, because of its letter forms and the pot on which it was written, is the Kharoshthi inscription of Sihusada from Hadda, now in the National Museum in Kabul.281 The inscription is written in ink on the shoulder of a vase which was found in a larger jar with earth and bones. It seems to have been a funerary jar and inscription. Three of the small terracotta implements used by potters to thin the walls of their vessels that were discovered in the Begram excavations have Kharoshthi inscriptions of the Kushan period – a name in the genitive indicating their owner. One of them refers to a Buddhist – ‘he who is protected by the Samgha’.282 Numerous sherds with inscriptions in Kharoshthi and Brahmi have been discovered at Hadda in the excavations of Mustamandi and Tarzi. The French excavations at the Buddhist monastery of Guldarra 15 kilometres south-east of Kabul between 1963 and 1965 produced nine fragmentary Kharoshthi inscriptions, two fragmentary inscriptions in cursive Bactrian and one in an unknown language, found on sherds from vases that had once belonged to the Buddhist monastery in Kushan times.283 Other fragmentary Kharoshthi inscriptions on sherds have been discovered at Basawal284 and in southern Bactria. These sherd inscriptions in ink on vases that belonged to monastic communities are known from a series of Buddhist sites such as Takht-i-Bahi, Palatu Dheri and Shahr-i-Bahlol in Pakistan, and Kara Tepe by Old Termez in Uzbekistan on the Amu Darya frontier.285

445

446

THE ARCHAEOLOGY OF AFGHANISTAN

The Jaghatu Inscriptions Two later Bactrian inscriptions have been discovered at Jaghatu, twenty kilometres from Ghazni on the arid plateau crossed by an ancient caravan track from Ghazni towards Kabul.286 The first inscription is carved on the flat surface of a granite boulder by scratching on the thin dark surface of the stone leaving the inscription standing out. It contains the Buddhist triratna formula (i.e. reverence to the Buddha, to the Dharma and to the Sangha). The second inscription is on a large rock at the side of the caravan track, and used the same technique. But it is poorly preserved and difficult to interpret. Humbach287 suggests it contains a reference to Tigin Shah and uses uluγ, a word borrowed from Turki. Both inscriptions probably belong to the seventh or eighth centuries ad in the Turk period. The Inscriptions of Uruzgan There are two other Bactrian inscriptions of this period at Uruzgan some 280 kilometres north-west of Kandahar on an ancient caravan route to the north. The inscriptions are carved on a boulder by scratching with a pointed punch, as at Jaghatu.288 The second inscription, also rock-cut, two miles to the north of the first, is more or less the same but has a division of words not found in the first. Bivar interprets both inscriptions as ‘the divine and glorious King of Zabul Mihira’ and connects them with the Hephthalite dynasty of Toramana and Mihirakula in the sixth century ad. Humbach offers a different transliteration and sees in both inscriptions a reference to the Tegin and his sun and moon radiance.289 The letter forms seem to be somewhat later than those of the Jaghatu inscriptions and this makes a date during the seventh or eighth century ad in the Turkish period likely for the Uruzgan inscriptions also. However, not much is understood of these texts. Indian Inscriptions in Sharada Script The use of Brahmi during the Hephthalite period is well-attested in Gandhara both from coins and from inscriptions such as the Wartir image and the Wano stone inscriptions, both from the north of modern Pakistan.290 The subsequent development of Brahmi into early Sharada script is seen in two inscriptions from Afghanistan. The inscription from Gardez, now in Kabul, is engraved in two lines on the base of a marble image of the Hindu god Ganesa.291 It refers to a Maharajadhiraja Sahi Khingala and dates to his sixth year. From its early letter forms Tucci suggested a date in the sixth or early seventh century ad. The second is the inscription of the pedestal of the Uma Mahesvara image discovered in the Japanese excavations at Tepe Skandar.292 The statue portrays Mahesvara, his consort Uma and the child Skanda. The inscription refers to the Hindu Triad, Brahman the creator, Shiva the destroyer and Visnu the preserver. Yamada, by

from the kushans to the shahis

analysing the differences between the letter forms here and in scripts in ordinary acute angled Brahmi from India, has shown that both the Skandar and Ganesa inscriptions represent transition scripts – probably in the eighth century ad. The decline in the use of Bactrian script in the Shahi period seems to be related to the growing loss of territory in Bactria and western Afghanistan to the advance of Islam, which brought with it the use of Arabic. The corresponding growth in the use of Indian script in eastern Afghanistan – Proto-Sharada being derived from Brahmi and Gupta script – was the natural consequence. The Shahis looked increasingly to Gandhara, Kashmir and the Panjab. The clear evidence of the important role of this script in Afghanistan during the eighth and ninth centuries comes from the long series of Shahi coins from the later eighth century ad, in which both obverse legends and mint control marks use Sharada, while the Bactrian legend that had been found on the first issue of Spalapati is completely ousted from the coinage,293 Foucher294 noted nine Sharada inscriptions on the River Alishing twenty-five kilometres from Tagarhi, and the Bourgeois295 have reported four more between Qargha’i and the town of Laghman on the granite rocks at the side of the valley. A defaced rock inscription of Shahi date from Jalalabad, now in the Lahore Museum,296 adds little to the Afghan material, and the other published inscriptions of the Shahi period come from Pakistan. Numismatics The Kushan Coinages297 The Kushan coinages are among the most numerous and typologically diversified of Central Asia and India. These coins give instrumental data in front of often cryptic and conflicting written sources and epigraphic evidence. From Kujula Kadphises’ bronze coinage to his successors’ multiple-denomination gold coins, the Kushans show a remarkable aptitude to build up on pre-existing monetary systems and their coins may be found almost anywhere from Western China to the Afghan-Iranian border, and from Marakanda (Samarkand) to Pataliputra (Patna). The Kushan monetary system The Kushan kings used several weight standards, whether for monetary continuity’s sake, or as a tool to control their own economy through the metal purity after almost a whole century of silver debasement which started with Hermaeus’ first posthumous issues. These coins, sometimes labelled as ‘imitations’, copied the types from Graeco-Bactrian (Heliocles I, Eucratides I) and Indo-Greek (Hermaeus) coins, and are characterised by a progressive degradation of the style and corruption of the legend.298 Their attribution to the Yuezhi seems clear, while their relative chronology in regard of Kujula Kadphises’ ascent to power, and his first issues, remains to be clarified.

447

448

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.96  Copper didrachm of Soter Megas, with the rayed head of Mithra and the reverse type of the king mounted on horseback (late first–early second century ad). Diameter 21 mm.

The bronze coinage of the first Kushan king Kujula Kadphises can be divided into two categories: anonymous and in his own name. Typically, the first is struck on copied denominations and types circulating in each of the localities he brought under his control (Indo-Parthians, Indo-Scythians), while the second always innovate at least on one side of the coin. The attribution of the anonymous coins to Kujula rather to the Yuezhi is mainly based on the interpretation of the word KOPPANOY (and its variants) as equivalent to the ‘Kushan’:299 the ‘Heraios’ silver coins have thus been identified with Kujula Kadphises,300 but this attribution has been challenged many times by numismatists. Depending on the series, several standards are used: a reduced Attic standard for the Heraios coins, the Indian (= Indo-Greek) standard for the Heracles-type coins, etc. Kujula Kadphises adopts the tamga 1on a single series (the ‘Roman emperor’ coins), and still uses monograms on several others, doubtless depending on monetary practices. The coins of the so-called ‘Nameless King’ (i.e. the Soter Megas coins (Fig. 6.96)) are a matter of deep controversy among numismatists and historians, even more since the discovery of the Rabatak inscription in 1993.301 Since then, this enigmatic coinage has indeed recurrently been associated with Vima Takto, whose name appears in the inscription as grandfather of Kanishka I (and son of Kujula Kadphises).302 This attribution is, however, far from being unanimous.303 Whoever the issuer was, most likely a Kushan, the Soter Megas coinage establishes a standard currency from Northern Bactria to Mathura: the general type (bust of Mithra304/ king on horseback) follows a slightly reduced Attic standard with didrachms of c. 8.5 g and hemidrachms of c. 2.1 g.305 This standard coinage was intended to replace the multiplicity of local coinages that the Kushans had inherited. Numerous variations on the coins (solar rays around Mithra’s head, tamga with four (2) then three ( ) prongs, square and cursive Greek letters, etc.) point to a huge monetary production. There are also several less numerically important series that have been associated with specific parts of the Kushan empire (Gandhåra, Qunduz and

from the kushans to the shahis

Figure 6.97  Kushan gold dinar of Vima Kadphises, with the bearded and diademed king emerging from rocks or cloud, and Shiva naked and standing on the reverse (early second century ad). Diameter 20 mm.

Mathura), with specific weight standards. The Soter Megas coins are particularly common in Afghanistan, and Charles Masson acquired 695 examples from Begram in three years. As for Vima Takto, the only coins that can be attributed with some certainty are the bronze bull/camel one apparently bearing his name in Kharoshthi on the reverse (‘Vema Tak[h]’).306 Vima Kadphises is the first Kushan ruler to strike gold coins (Fig. 6.97), while increasing the range of denominations, notably with a new heavy bronze coin of c. 16 g, and a gold dinar (c. 8 g). His coinage deeply modifies the Kushan coins design through standardisation: from his reign forth, the obverse/reverse types always follow the king/divinity model. The king is represented in several ways: seated, on a chariot, emerging from a mountain top, inside a window, etc., although without variation on the bronze coinage (making an act of worship to a fire altar), always with the tamga (). Recent coin discoveries have confirmed the genealogy detailed in the Rabatak inscription (i.e. that Vima Kadphises is the son of Vima Takto).307 The coins display a tamga on the obverse (sometimes also on the reverse) and a Nandivarta on the reverse. On the gold as well as on the bronze coinage, Oesho, here closely associated with √iva, is depicted standing with or without the bull Nandi behind him, and shows some similarities with Kujula Kadphises’ Heracles design. However, Vima Kadphises’ first (commemorative) issues are characterised by some variation in the position of the legend (6), most likely influenced by the design on Agathocles’ and Antimachus I’s commemorative coins. Overall and minus a few exceptions, the coinage is struck with Greek and Prakrit with a very high quality standard.308 Vima Kadphises uses a great deal of epithets on his coinage, especially in kharoshthi: maharajasa, rajadirajasa, (sarva)loga iªhvarasa,309 mahisvarasa, tratara. Finally, a recent die study has divided Vima Kadphises’ coinage into five discontinuous phases.310 Kanishka I introduces several major changes on his coinage: Greek is replaced

449

450

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.98  Kushan gold dinar of Kanishka, with the king standing at an altar and the reverse type of Nana (mid-second century ad). Av. diameter 20 mm.

by Bactrian, both for the king’s royal titulature and for the (now indicated) names of the divinities he used as reverse types (Fig. 6.98); a wide pantheon is depicted on his coins (more than a dozen Greek, Iranian and Indian gods)311 struck with the tamga 4; finally, kharoshthi on the reverse is abandoned, while the letter Ϸ (‘sh’) is introduced in the Greek alphabet. The Rabatak inscription has added much to our understanding of the depiction of such numerous divinities: their role is to confer royal authority to Kanishka I, Nana(ia) being then of first importance.312 Kanishka I uses fewer gold denominations than his father did, while the bronze ‘tetradrachm’ of c. 16 g becomes a standard unit which divides itself into halves, quarters and eights. Gold and bronze together, the divinities depicted are as follows (Fig. 6.99): Ardochsho, goddess of good fortune; Ath(o)sho (= Hephaistos), god of fire; Boddo, the Buddha; Lrooaspo, god of animals; Manaobago, god of good mind; Mao (= Selene), goddess of the Moon; Miiro/Mioro (= Helios), god of the Sun; Mozdooano, Nana/Nanashao (= Nanaia), goddess of the Moon(?); Oado (= Anemos), god of the wind; O(o)esho (= Heracles), god of the wind on high; Orlagno, god of victory; Pharro, god of good fortune. At this period, there was probably a mint south of the Hindu Kush at Begram, close to the rich copper deposits of the Ghorband valley; Joe Cribb has also hypothesised the existence of two mints striking gold coins at Bactra and in Gandhåra.313 Huvishka retained the weight of Kanishka I’s gold dinar unchanged, while his coinage shows a slight decrease in gold fineness (down to 91%)314 (Fig. 6.100). His early copper coins followed Kanishka I’s weight system but had three obverse types – that of the king riding an elephant, the king seated cross-legged and the king reclining on a couch (in rajalila asana, the royal ease pose). In Masson’s finds from Begram the couch lounger type is commonest and probably was the type of the Afghan mint at Begram – but the other two types are also regularly found. On gold coins, however, the king is always represented emerging from a mountain top. The place of Huvishka immediately after Kanishka I is clearly demonstrated

from the kushans to the shahis

Figure 6.99  Kushan copper tetradrachm of Kanishka, with the reverse type of Mioro (Mithra) radiate. Diameter 26 mm.

Figure 6.100  Kushan gold dinar of Huvishka, with the reverse type of Miiro (late second century ad). Av. diameter 21 mm.

by a common reverse die used to strike a Kanishka I’s late phase coin and a Huvishka early phase coin;315 Huvishka also inherits his predecessor’s tamga ( ) before using his own (8). The diversification of divinities depicted which started under Kanishka I continues: almost thirty of them are engraved on the reverse of his coins. In Huvishka’s later issues there was a sharp reduction in the weight standard of the main copper weight standard from c. 16 to c. 13 g, then c. 6 g, before stabilising at c. 10–12 g at the end of his reign with the introduction of a new weight standard (Fig. 6.101) The experiment was not a success: there was an extensive series of local unofficial imitations, and Huvishka’s successors eventually reverted to the earlier pattern of a full value copper denomination. Apart from a few rare exceptions, Huvishka’s official mints had not struck subdivisions of the standard copper denomination. Local copies, often very crude and barbarised in style, supplied this need, and are well-represented in the local museum collection at Kandahar and among Afghan finds in the National Museum; Cribb considered

451

452

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 6.101  Kushan copper tetradrachm of the reduced standard of Huvishka, showing the king riding an elephant and a figure of Mao. Diameter 24 mm.

them as a response to the monetary crisis and the consequent dropping in the copper weight standard.316 Vasudeva I is likely to be the last Kushan king to rule in the Balkh region,317 while his coinage is characterised by the almost exclusive use of Oesho on the reverse (with some rare coins depicting Vasudeva-Krishna and Nana),318 most of the time with a bull. Vasudeva I uses the tamga . The multiple royal representations introduced by Huvishka are also abandoned, being replaced with the king-sacrificing type of Vima Kadphises and Kanishka I (with the later addition of an altar trident). The weight standard for gold coins remains the same, with a dinar weighing c. 8 g, but Vasudeva introduces a new main copper standard of c. 8–10 g and a lighter one of c. 4–6 g. In the later issues of the gold coins bearing the titles of Vasudeva, we see the emergence of two sequences.319 The northern series, which retain the Shiva and the bull reverse while adding a trident and a Nandivarta on the obverse, evolves into the distinctive scyphate or saucershaped gold dinars on which eventually we find cursive Bactrian inscriptions, and kings like Hormizd and Vahran with a Sasanian headdress. The southern series introduces Brahmi letters in the field replacing the Nandivarta, and the type is eventually changed by Vasudeva’s late Kushan successors to the reverse type of an enthroned goddess Ardochsho, a type copied by the local chiefs of the Panjab and the Guptas. Vasudeva’s coins are often quite worn, especially the copper ones, which might be explained by their use beyond the sole reign of Vasudeva, whose types were borrowed by some of his successors. Finally, interestingly enough, Vasudeva is the first Kushan king to have a name of Indian origin. Foreign influence In these monetary reforms and experiments we see foreign practices adapted to local Kushan needs. Soter Megas’ standard coinage is characterised by its

from the kushans to the shahis

­ onetary reform and the use of the (slightly reduced) Attic standard introduced in m Central Asia by the Seleucids. Thus, this coinage certainly succeeded in creating a currency that was not liable to fraud through debased imitations. While Vima Kadphises retains the Attic standard for his gold coins, the introduction of the lighter gold dinar (c. 8 g) during Kanishka II’s reign was probably inspired by the current Roman denarius aureus.320 The presence of Roman coins in Central Asia and India is well-attested through hoards finds and Kujula Kadphises’ ‘Roman’ series, on which coins the obverse depicts a Julio-Claudian bust. The use of reverse types derived from Indo-Greek, Indo-Parthian, Indo-Scythian and Roman coin designs321 and the rapid adoption of a localised production system under Kujula Kadphises, followed by a possible centralisation under Vima Kadphises,322 reflects a unique mix between standard Hellenistic monetary practices and Kushan habits in the context of the early empire. Coins in Buddhist stupa deposits A large number of Buddhist stupas in Afghanistan were examined by C. Masson and J. M. Honigberger in the early nineteenth century.323 They often found among the treasures deposited in the relic chamber coins of the period in which the relics were deposited and the stupa was built. At the Bimaran stupa, from which the Bimaran reliquary was recovered, there were four billon coins with the tamga of Kujula Kadphises. In Tope 3 at Bimaran there were 27 coins of Soter Megas. In Tope 4 at Chahar Bagh there were 28 copper coins of Kanishka I. Particular interest attaches to the stupa deposits that also contained Roman coins. From the Ahin Posh Tope near Jalalabad, three Roman aurei of Domitian, Trajan and Sabina, the wife of Hadrian (ad 119–138), were found with ten gold dinars of Vima Kadphises, six of Kanishka I and one of Huvishka;324 and gold coins of Vima Kadphises and Kanishka were found with an aureus of Trajan (ad 98–119) at Shiwaki near Kabul. These associated Kushan and Roman finds provide important prima facie evidence for the context and date of Kanishka.325 Recently, a gold stupa deposit, consisting of two parts containing 23 gold Kushan coins (Vima Kadphises, Kanishka I and Huvishka) and two Roman aurei (Nero and Domitian), has been discovered in or near Begram.326 Other hoards and site finds Because there was no Kushan silver currency327 following the dramatic debasement of this metal in the first century bc, most substantial payments had to be made in copper coinage, and this explains the commonness of Kushan copper coins in Afghanistan. Excavations at sites of Kushan date such as Begram, Surkh Kotal and Balkh have provided substantial numbers. In Begram, following the French excavations in 1936–7, the Kushan coins have been published in detail by Osmund Bopearachchi:328 apart from Vima Takto, all the Kushan kings from Kujula

453

454

THE ARCHAEOLOGY OF AFGHANISTAN

Kadphises to Kanishka II or Vasishka are represented (including Eucratides I and Hermaios’ posthumous imitations). Particularly important for the interpretation and the chronology of the later coinage, as well as for understanding their relations with the Kushano-Sassanians, are the hoards of late Kushan copper coins in Central Asia museums.329 Particularly interesting are the coins in the 1946 hoard from the Begram excavations which contain coins of the earlier Shiva type deliberately cut at the edge to reduce the amount of copper so as to assimilate the coins to the lower weight of the next group. Recent excavations led by French archaeologists in Balkh and Mes Aynak (Afghanistan), as well as in Termez (Uzbekistan), have yielded still unpublished Kushan coins.330 However, Kushan gold coins were highly valued internationally, as the hoard of 104 gold coins depicting just four Kushan kings found at the monastery of Debra Damo in Ethiopia demonstrates.331 The Late Kushan coinages The seven Late Kushan kings are distributed over a c. 120-year period, and succeed each other in the following order: Kanishka II, Vasishka, Kanishka III, Vasudeva II, Mahi, Shaka and Kipunadha. Their coinages are characterised by a progressive drop in gold fineness and weight standards.332 Parallel to the high-quality gold issues, the copper coinage is often crudely struck, with part of the type off flan, and rarely has any legible legend, especially from Vasudeva II onwards, and the appearance of whole words in Brahmi. The obverse type remains the same for all the Late Kushan kings (i.e. the king sacrificing with a trident) with the exception of Vasudeva II (on whose coins the king is sometimes seated). Among the later issues, Oesho and Ardochsho are both depicted, but the latter becomes predominant after the reign of Vasudeva II.333 One can distinguish three major stages for the bronze coins: 1. Dumpy Oesho and the bull coppers, sometimes with a Nandivarta symbol, struck to a weight standard of 6 g (Vasishka) (Fig. 6.102). 2. Dumpy Ardochsho coppers with an increasingly arched termination of the King’s dress – from 5 to 8 g (Kanishka II, Vasishka) (Fig. 6.103). 3. Crude Ardochsho coppers with very crude form for the king – from 3 to 6 g (Kanishka III, Vasudeva II, Shaka, Kipunadha) (Fig. 6.104). Some of the later series are cast local copies and moulds for their manufacture have been found in Pakistan. Although some Ardochsho coppers have traces of a Brahmi letter in the field and were thus possibly issued in the mint responsible for the Ardochsho gold, all the three later Kushan copper series are represented in quantity at late Kushan sites across Afghanistan. Copper weight standards seem to be reduced under each king’s reign and are often a determining feature of the identity of the king.334 They are represented in a series of overlapping hoards in the National Museum which

from the kushans to the shahis

Figure 6.102  Kushan copper unit (c. 4.5–6.5 g) of Vasishka, with the king standing at an altar, and Oesho in front of a bull on the reverse (mid-third century ad). Diameter 19 mm.

Figure 6.103  Kushan copper unit (c. 7.5 g) of Kanishka II, with the king standing at an altar, and Ardochsho on the reverse (first half of the third century ad). Diameter 20 mm.

Figure 6.104  Kushan copper unit (c. 1.6–3 g) of Shaka, with the king standing at an altar, and Ardochsho on the reverse (early fourth century ad). Diameter 18 mm.

455

456

THE ARCHAEOLOGY OF AFGHANISTAN

contain coins of successive issues and reinforce the evidence of metrology, type development and stratigraphy for the sequence. Finally, the last Late Kushan kings overstruck Kushano-Sassanian coins, aggravating the reduction in weights, before the demise of the empire c. ad 350. Sasanian Influences The Kushano-Sasanian coinages Whatever may have been the political status of the later Kushan empire, there was certainly a sharp growth in Sasanian influence, which is seen most clearly in the so-called Kushano-Sasanian coinage.335 The Kushano-Sasanian gold scyphates have been found on the Amu Darya, at Qunduz, in Badakshan and in hoards from Charikar and Kabul.336 Kushano-Sasanian copper coins have a more localised distribution. The small series with neat Bactrian legends are well-represented from the Balkh excavations and in the Mazar-i Sharif Museum, and may have been intended to provide fractional denominations for the later Kushan copper currency. The larger series with a dumpy fire altar begins in the Indus Valley with an issue of Shapur II c. ad 365. Coins derived from this coinage provided the later copper currency at Begram and Hadda and constitute a further stage in the development of the standard denomination. The Tepe Maranjan hoard and other Sasanian finds The hoard from Tepe Maranjan near Kabul337 containing 368 Sasanian silver drachms – 326 of Shapur II, 28 of Ardeshir II and 14 of Shapur III (ad 383–388), with 12 gold Kushano-Sasanian scyphates – provides important evidence for the previous metal currency of eastern Afghanistan, the dating of the Kushano-Sasanian gold coinage and the chronology of Kidara, who became king of the southern Kushan provinces. Sasanian silver coins from Shapur II onwards are commonly found in Afghanistan. They were reported from Sistan.338 Hackin notes that they are common at Herat, Maimana and Shahr-i-Banu near Tashkurgan, and they are frequently found in the neighbourhood of Kabul.339 Masson encountered them at Begram and they were well-represented among the silver coins found in stupas near Hadda. In the absence of an indigenous silver coinage, Sasanian silver obviously filled the gap, even in territories beyond the Sasanian empire. Coinage of the Huns340 The earliest coins of the new Hephthalite kingdom in Bactria consisted of silver drachms of Sasanian type with the usual obverse of Shapur II (309–379) but struck from dies which had been meant to replace Shapur’s name with the

from the kushans to the shahis

Hephthalite title Alchono.341 In the following century we see the development of the Hephthalite currency of silver drachms based on the Sasanian denomination retaining a fire altar on the reverse but with a distinctive Central Asian bust, the Hephthalite tamgh and Bactrian legends.342 Other coins with the same types have Brahmi legends, but these seem to have been struck for circulation in the Indus valley and Kashmir. The Hephthalite coins of this period from Tope 10 at Hadda, found with five Roman gold solidi of Theodosius, Marcian and Leo (ad 457–474), seem to have been mostly Hephthalite coins with Bactrian legends. Several of the later silver drachms of Sasanian kings found in Afghanistan have Hephthalite countermarks and seem to have been coins paid in tribute by the Sasanians after their defeats under Firuz, countermarked to serve as Hephthalite currency. The coins of the Nezak In the succeeding years, silver drachms with the legend Nezak Shah, and the copper denominations that accompanied them, provided the currency of Kabul, the Kohdaman and Kohistan districts.343 The copper coins are common at Begram, and a noteworthy hoard of silver coins of this type was discovered at Gardez in 1957.344 The variety of styles and scripts used in this coinage suggests that there may have been a number of issues by different kings over a broad spread of time; but there is serious disagreement about their attribution. Earlier scholars associated them with the Hephthalites, but there is much to commend the view that they are coins in the Hephthalite tradition but of the Turki period of the later sixth and early seventh centuries ad. The issues of Vrahitigin Demonstrably later, because they are sometimes overstruck on the drachms of Napki Malka, are the silver drachms of Vrahitigin.345 They have a bust threequarters frontal of the king with legends in both Bactrian and Proto-Sharada, and the facing head of a divinity crowned by flames with a Pahlavi legend on the reverse. The king’s crown has a wolf’s head design – the wolf being the legendary ancestor of the Turki race. The deity is copied from the type on coins of Khusru II (ad 591–628). Vrahitigin’s coins have been found in the Indus valley and from the stupa at Manikyala in northern Pakistan; Lord acquired 40 specimens from north of the Hindu Kush mountains and Cunningham received around thirty from Kabul. These rare coins seem to be issues of the Turki Shahis of the later seventh century ad. From this period is the small group of coins, buried in a cinerary urn below the princely couple in Niche E at Fondukistan. It includes countermarked coins of Sri Sahi and two silver drachms of Khusru II. One of these was struck in year 37 (i.e. ad 657), but the two countermarks can be dated to ad 682 and 689.346

457

458

THE ARCHAEOLOGY OF AFGHANISTAN

Gadhaiya paisa The base silver Gadhaiya currency of Rajputana and Gujarat (from the eighth century ad),347 derived ultimately from a Sasanian prototype, is sometimes found in Afghan collections, and a hoard discovered near Kandahar was offered for sale in the bazaar at Kabul during 1973. Problems with the Shahi Coinage The pattern of the coinage Between ad 750 and 1000 the currency of eastern Afghanistan and Gandhara was provided by the extended issues of the silver coinage of the Shahis, at first the kings of Kabul and subsequently the rulers of Ohind.348 Throughout the whole of this period the Shahis used standard obverse type of a recumbent Indian humped bull with the reverse type of a horseman. Most of the coins have an obverse legend in Sharada script – either Sri Spalapati Deva or Sri Samanta Deva, but these are titles, not personal names, Spalapati being a Sanskritised version of a Bactrian title for commander-in-chief and Samanta having some significance in Sanskrit. Bull and horseman coins of this series in both silver and billon are all of approximately the same weight and were clearly intended to pass as the same denomination. We can, however, distinguish the successive issues of the coinage by the letters and other symbols serving as privy marks in the reverse field and can put them into their chronological sequence by a study of the developing changes in the type, and from the small but progressive reduction in the real silver content of the denomination and the weight standard to which successive issues were struck. Silver coins of Spalapati Deva The earliest group of the Shahi silver coins is distinguished by its obverse legend – the titles Sri Spalapati Deva. Coins with this legend are struck between 3.1 and 3.5 g with a remarkably uniform content of 70% silver. In the first issue of this group the reverse legend repeats in cursive Bactrian script the Sharada of the obverse. This is copied but progressively misunderstood in subsequent issues. At one stage in the series a Brahmi legend, Shahi Deva, is substituted, but this in turn is misunderstood and becomes a decorative scroll that has been mistaken for an Arabic date. In issue there is a series of small letters serving as privy marks below the horseman on the reverse. Silver coins of Samanta Deva The second major group of this coinage has the obverse legend Sri Samanta Deva. It is still in good metal, but now with a wider variation in silver content of between

from the kushans to the shahis

60 and 70% and is struck to a slightly lower weight range of between 2.9 and 3.9 g. Again we can distinguish a series of issues by the use of Sharada letters and other control marks. Associated chronologically with this group are two comparatively rare issues, one with Sri Khudavayaka Deva struck to the distinctly lower weight standard of the Arabic dirhem of the reformed currency introduced by ‘Abd alMalik, possibly to be associated with the Muslim capture of Kabul in ad 870, and the other with Sri Bhima Deva, perhaps the King Bhim known from inscriptions to have been the ruler in Hund around ad 950. The Shahi copper denominations The copper issues constituted a complementary denomination that provided the small change for the Bull and Horseman silver. They too suffered a progressive reduction in weight and have privy marks, sometimes echoing those on the silver, that enable us to establish their chronological sequence. The Bull and Horseman coppers of Spalapati are succeeded by the Lion and Elephant coppers of Vakka and, finally, the Lion and Elephant coppers of Samanta. The billon currency The third major group – coins in billon with the Samanta Deva legend that have a silver content of 25 to 30% – only shows a sharp reduction from the last silver issue, with no progressive debasement that could bridge the gap. The reason seems to have been that the Shahis had exhausted their reserves of bullion and were deprived of the important silver mines of Panjshir which seem to have passed permanently under Muslim control by the middle of the tenth century ad. This was the billon currency that was copied extensively by the Islamic rulers of Ghazni, and by the dynasties of Kanauj, Ajmir and Delhi. Hoards and site finds The principal evidence for the distribution of Shahi coins in Afghanistan has so far been drawn from collections made in Kabul and its locality. The arrangement set out above has necessarily been based on internal numismatic analysis; but the sequence of the copper issues has now been tested and confirmed by the stratified finds from the excavations at Damkot near Chakdara in Pakistan.349 A hoard of 199 copper Shahi coins, all of Vakka Deva found in Jalalabad in 1971, a silver hoard all of the last silver issue of Samanta Deva discovered at Shiwaki in 1970, a silver hoard of unknown but Afghan provenance containing three worn coins of Spalapati and three issues of Samanta, offered for sale in Kabul bazaar in 1972, and a silver hoard of some fifty coins of Samanta Deva from Qunduz offered for sale in Kabul bazaar in 1976 will produce important new material for study.

459

CHAPTER 7

From the Rise of Islam to the Mongol Invasion Warwick Ball and Klaus Fischer

Historical Background With the overthrow of the Sasanian empire, Khurasan and Sistan became provinces under Arab governors, but the early Muslim invasions effected no permanent control over the Kabul region, which remained semi-independent under native princes entitled rutbil (or zunbil), even after the transfer of the Caliphate from the Umayyads to the Abbasids in 750. In the reign of the Caliph Ma’mun (198/813–218/833) one of these local rulers accepted Islam and his territory was joined to the imperial postal service (barid); but in the tenth century his successors appear to have been pagan once more (see Chapter 6, ‘Turki and Hindu Shahis’). The Tahirids, Saffarids and Samanids Ma’mun had obtained power with the aid of his lieutenant Tahir b. al-Husain, a Persian, whom he rewarded with the governorship of Khurasan and whose family continued to rule the province after his death. Under the Tahirids (205/821–59/873), Khurasan was virtually independent of the central authority; but a far greater threat to the Caliphs’ power arose in Sistan. Here, in 253/867, another native, Ya‘qub b. Laith, whose surname al-Saffar (‘the coppersmith’) betrays his plebeian origins, seized the province and began to extend his sway over the neighbouring territories, occupying Kabulistan (871) and destroying the Tahirids two years later. Although both Ya‘qub and his brother and successor ‘Amr (265/879–287/900) extracted recognition from the Caliph, their attempts to establish their power in Khurasan met with constant opposition. It was Amr’s efforts to assert his authority over Transoxiana, now under the hereditary governorship of the Samanids (204/864–389/999), however, which brought about his downfall. He was defeated and captured by the Samanid Isma‘il b. Ahmad (279/892–295/907), and sent as a prisoner to Baghdad (900). But the Samanids, who thus acquired Khurasan, were unable to exercise more than a tenuous control over Sistan, which even during the tenth century revolted under scions of the Saffarid dynasty and from 1002 was

from the rise of islam to the mongol invasion

Figure 7.1  Map showing early Islamic sites described in the text. Sistan inset: Fig. 7.32.

to remain continuously under the rule of a line of princes who may or may not have been of Saffarid extraction; while to the east the native rulers of Kabul again became independent.1 The Samanids were a dynasty of native Central Asian Iranians. Their most notable ruler, Isma’il Samani, ruled from 279/892 to 925/907 and established a brilliant court at his capital of Bukhara, nominally under the Caliph in Baghdad but in reality independent, ushering in a renaissance of Central Asian arts and the Persian language. Much of present-day north-eastern Iran came under Samanid rule, as well as most of Uzbekistan and Afghanistan. More importantly, Isma’il established the use of Persian as the court language rather than Arabic, which had hitherto enjoyed a virtual monopoly throughout the Islamic world in both Arab and non-Arab courts. Indeed, the word used for the Persian language in Afghanistan today is still Dari, which means ‘language of the court’, rather than Farsi, which means ‘language of Fars’ (i.e. Persia in southern Iran). This was the birth of modern Persian, now written in the Arabic alphabet, soon to spread westwards back into the area of modern Iran. As well as creating a revival of Persian literature, this made Bukhara one of the great centres of learning in the Islamic world.2

461

462

THE ARCHAEOLOGY OF AFGHANISTAN

The Ghaznavids The Samanid period witnessed the beginnings of a new phenomenon, the appearance in the eastern Islamic world of dynasties of Turk origin. The Caliphs themselves maintained a corps of Turkish slave guards at least as early as the ninth century, when their decline enabled the Turks for some decades to exercise a stranglehold upon the government. In the east, the pattern was the same among the provincial governors. The Samanids continued the practice, both as suppliers to the Caliphate and for their own army. In 351/962, Alptegin, a Turkish slave general of the Samanid ruler, was ousted from the governorship of Khurasan after a palace revolution, and made for Ghazna, where he defeated the native chief and established his own base. This may be regarded as the date of the definitive Muslim conquest of the region. His immediate successors were of little significance, but in the following decade power was seized by a slave of Alptegin’s, Sebuktegin (366/977–388/998), who founded the dynasty properly known as the Yaminis, more usually as the Ghaznavids (366/977–582/1186) after their capital at Ghazna, 145 km south-west of Kabul.3 Initially, Sebuktegin and his still more energetic son Mahmud (388/998–421/1030) ruled Khurasan nominally as governors for the now moribund Samanids; but when in 999 the latter were overthrown by an invasion of the Turkish Qarakhanid dynasty from further east in Central Asia, this de jure subjection came to an end. The Samanid territories were now divided between Mahmud and the new power, the Qarakhanids, with the Amu Darya as the frontier between them. The overthrow of the Samanids marks a major linguistic shift: it ushered in rule by Turkish speakers, which continues to this day, and the end of several thousands of years’ rule by Iranian speakers in Central Asia. Under Ghaznavid leadership there was a second burst of Muslim expansion eastwards. The Arab campaigns of the early eighth century had conquered Sind for Islam, but beyond it the Panjab and the rest of the subcontinent had remained Hindustan – ‘the land of the Hindus’. The reigns of Mahmud and of his son Mas‘ud (421/1031–432/1041) witnessed the beginnings of the Muslim drive into India proper. Even by the time the Hindu Shahi rulers of Kabul had been finally eliminated (1019), a series of brilliant campaigns into the heart of northern India had brought great quantities of plunder back to Ghazna, and had extended Muslim rule over the western Panjab. The achievement was all the more remarkable in that Mahmud was simultaneously engaged in sporadic warfare with the Qarakhanids beyond the Amu Darya and in the conquest of large areas of eastern Iran, where he appeared as a Sunni champion of the Caliphs against the regime established in Iraq by the Shi‘a Buyid dynasty. In the event, however, this function was to be appropriated, together with the Ghaznavid possessions lying west of Afghanistan, by the Seljuks. Mahmud became known as a great patron of the arts and science, and under his successors architecture and literature flourished, especially under Mas‘ud I

from the rise of islam to the mongol invasion

(421/1031–432/1041) and Mas‘ud III (492/1099–507/1114). Then followed a slow decay of the dynasty; Bahram Shah (512/1118–547/1152) ruled under the protection of the Seljuk Sanjar. In 544/1149 the ruler of Ghur, ‘Ala al-din Jahan-suz (i.e. ‘Destroyer of the World’), burnt Ghazna. The dynasty survived for a time in India, but without political or cultural importance. The Qarakhanid invasion of Transoxiana had constituted merely the vanguard of a wave of inroads by free Turkish tribes into western Asia that were to be spread over a hundred and fifty years. In the first decades of the eleventh century the Seljuks, an Oghuz tribe who had been quartered on the lower Syr Darya (possibly an offshoot of the Khazars further west), began to press on Mahmud’s province of Khurasan. At first they were accepted as Ghaznavid clients, but in Mas‘ud’s reign they could no longer be contained. In 1037 their leaders took over western Khurasan and assumed the insignia of sovereignty. Mas‘ud took up arms against them, only to be routed decisively at Dandanqan (431/1040). Within the next ten years the Seljuks, who went on to conquer the rest of Iran and to occupy Baghdad in 1055, even threatened Ghazna, but were repulsed. Ghaznavid power in Khurasan and Sistan was now a thing of the past; but the abandonment of these western provinces at least enabled Mas‘ud’s successors to concentrate on the reduction of Hindustan. Under Ibrahim (451/1059–492/1099) peace was made with the Seljuks and lasted through the reign of his son Mas‘ud III (492/1099–507/1114), thereby affording the Ghaznavid empire a period of stability and consolidation. Under Bahram Shah (512/1118–547/1152), however, who had obtained the throne with the assistance of the Seljuk Sultan Sanjar and was obliged in consequence to become his vassal, the decline of the dynasty set in. In 1141 the Seljuks under Sultan Sanjar were defeated in a battle to the north of Samarkand by the Qidan, a Mongol state to the north of China, known in Arab sources as the Qara Khitai (from which the name ‘Cathay’ derives) and Chinese sources as the Western Liao. Both Samarkand and Bukhara were absorbed into the Qara Khitai Empire, but most of Afghanistan came under the rule of a new Iranian-related dynasty. The Ghurids4 The remote mountainous region of Ghur east of Herat was inhabited by an Iranian group of tribes only recently converted to Islam who spoke a distinct dialect of Persian. The mountainous nature meant that it was relatively disunited, ruled by petty chiefs from individual fortified strongholds in remote valleys, until experiencing a number of invasions by Mahmud. Although never fully subjugated, they had recognised Ghaznavid overlordship. With the growth of Seljuk influence in Afghanistan Ghur became subject for a time to Sultan Sanjar, but its rulers united under the Shansabani family and were already powerful enough to challenge the Ghaznavids. After a series of engagements Ghazna itself was ruthlessly sacked in 1151 by the Ghurid ruler ‘Ala al-Din Husain (544/1149–557/1161), who thereby

463

464

THE ARCHAEOLOGY OF AFGHANISTAN

earned the sobriquet of Jahan-suz (‘the world-burner’). He was subsequently defeated in an attempt to throw off Seljuk suzerainty; but this proved to be only a temporary reverse. In the period 1125–1140, Central Asia had been invaded by the non-Muslim Qarakhitai, a Mongol-related people from north China (where they are known as the Western Liao) who reduced the Qarakhanids to the status of subordinate rulers. Their arrival set in motion once more the Oghuz tribes settled on the lower Syr Darya, and Khurasan was overrun a second time. Sanjar, who tried to check them, was defeated in 1153 and carried off into humiliating captivity, and in 1163 the Oghuz occupied Ghazna. But this burst of energy soon dissipated itself, leaving the Shansabanids, or Ghurids as they are more usually known, as the more powerful element in the eastern Islamic world. Junior members of the dynasty ruled in Firuzkuh and Bamiyan, but the main line was represented by the brothers Ghiyath al-Din (602/1206–609/1212) and Muizz al-Din (569/1173–602/1206), who were able to recover from the Oghuz not merely Ghazna (1173) but Herat also (1175), and in 1186 they finally extinguished the remains of the Ghaznavid principality at Lahore. Even before this, Muizz al-Din had assumed direction of the Muslim drive into India and had given it fresh momentum: his campaigns and those of his generals may be regarded as the foundations of Muslim domination in the subcontinent. Nevertheless, the dynasty’s military activities were not restricted to India, and this was to prove its undoing. The Khwarazm-shahs The eclipse of Seljuk power had created a vacuum in Khurasan which was in part filled by the rulers of Khwarazm to the north. Themselves descended from a Turkish slave of the Seljuks, the Khwarazm-shahs had remained dissident vassals of Sanjar until the end. Then they began to expand southwards at the expense of local Oghuz leaders and so clashed with the Ghurids. Initially, the balance in the conflict was against them, and by 1200 they had been driven out of more of Khurasan, but Muizz al-Din, who by the death of his brother in 1203 was to become sole head of the dynasty, overreached himself. His invasion of Transoxiana, now under the Khwarazm-shah’s influence, in 1202 was a disaster of the first magnitude; and his assassination in 1206 while preparing to avenge it left his empire with no effective ruler. While his generals in India assumed practical independence, his weak and ephemeral successors in Ghur and Ghazna were gradually reduced by the Khwarazm-shah, Muhammad b. Takash (596/1200–618/1221), who occupied both territories in 1215–16. Muhammad, whose father had destroyed the last of the Seljuks in western Iran in 1194, was now the master of a dominion that extended from Hamadan to the Indus and from western Transoxiana to the borders of Sistan. Yet within five years his ambition had led him to challenge the new power founded in Mongolia by Chingiz Khan and his vast empire had been swept away by the Mongol invasion.

from the rise of islam to the mongol invasion

Architecture of the Early Islamic Period Background The lonely deserts and mountain ranges, as well as populous rural and urban settlements of present-day Afghanistan, have always been famous for preserving outstanding monuments of Islamic art: either marking the final phase of an age-old civilisation, as, for example, the shrine of Khwaja Abu Nasr Parsa (Fig. 8.54–8.56) near the massive ruins of ancient Balkh, or constituting the beginning of a stylistic sequence important in the artistic evolution of Eastern architecture, for example the minarets of Ghazni (Fig. 7.18). After the Second World War archaeological surveys in Afghanistan revealed remarkable buildings contributing to our knowledge of the development of Islamic art: the very early Mosque of Noh Gunbad near Balkh (Figs 7.2–7.4), the tower of Jam (Figs 7.48–7.51), the fortified posts in Ghur (Figs 7.53–7.60), the abandoned city of Gul-i Safed (Figs. 7.36–7.40) and the Timurid dome of Kuhsan (Figs 8.45–8.47), to name just some. These discoveries led, among others, to renewed discussion of the naming of art styles after Islamic dynasties, for instance ‘Seljuk’,5 ‘Ghurid’6 or ‘Timurid’.7 Twice ancient Turanian, Iranian and Indian traditions led builders in Afghanistan to the creation of new national dynastic styles that in their part influenced artistic developments in neighbouring eastern lands, during the Ghaznavid and Timurid periods.8 When dealing with architecture and minor arts produced in Afghanistan under the rule of these dynasties we shall first examine the main monuments, of court as well as provincial art, and secondly try to give an idea of their salient features. In Chapter 7 we have the Samanid prelude, the spread of Ghaznavid court art across Afghanistan, and the persistence of Seljuk, Khwarazmian and north-west Indian art forms in the Ghurid dominions; Chapter 8 comprises the art of eastern Iranian lands under Mongol rule, during Kart and Ilkhanid rule, the high-water mark of architecture and minor arts in Timurid Afghanistan, and a few later works executed under the Mughals and Safavids, who were great patrons of culture in neighbouring India and Iran respectively. The Earliest Period The earliest surviving major Muslim religious monument in Afghanistan is an eighth century mosque in the suburbs of Balkh, the Noh Gonbad Mosque, among the remains of the earliest known Islamic habitation in this pre-Islamic and Islamic city.9 A new study by Chahryar Adle of the Mission Archéologique Française de Bactriane Afghane, made possible by the recent programme of conservation at the site, dates it earlier than previously thought to 794–5 and relates it to the building activity of Fazl the Barmakid. This in turn is possibly related to the ‘sacred space’ of the great Buddhist monastery of Nau Bahar at Balkh, of which the Barmakids were hereditary guardians before Islam.10

465

466

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.2  Plan of the Mosque of No Gunbad at Balkh.

The Mosque occupies a square area measuring about 20 m on a side (Fig. 7.2). The ruins consist of four large pillars of brick standing in the centre of a square, two fallen pillars on the north-east, three curtain walls, and arches springing from the pillars and the coupled columns which are attached to the walls (Fig. 7.3). In the middle of the south-west wall there is a semi-dome which served as the hood of the mihrab, the prayer niche. Deeply carved stucco ornamentation occurs on the capitals, imposts and bases of the columns and on the spandrels and soffits of the arches. The bricks in the masonry of the curtain wall measure 30 × 30 × 6 cm. The original form of the mosque can easily be reconstructed from these remains. The existing curtain walls mark the south-east, south-west (qibla) and north-west limits of the building. The four supports (Fig. 7.3) were linked to

from the rise of islam to the mongol invasion

Figure 7.3  Balkh, mosque of No Gunbad, view of interior.

each other by an arcade. We may assume that the roof was composed of brick vaults, presumably a series of domes. The columns supported arcades which rose into the walls of the barrel vaults covering the corridors. We have also to consider an oriental tradition of pre-Islamic origin. The so-called kushk was a small square building of dimensions comparable to those of the Abbasid mosque. Its interior was divided up into nine squares of equal size, each covered by a cupola. It would appear reasonable to seek the model for the architectural design of the Balkh mosque at the same time in Mesopotamia, Egypt and North Africa, where we have a tradition of small cubical buildings divided internally into nine sections; this type, of a mosque of nine domes, was also known to the historian Maqrizi (1364–1442). A small building like the mosque at Balkh belongs to the type of hypostyle architecture on a square base.11 A characteristic vocabulary of motifs includes grape leaves, vine-scrolls, palmettes and fir-cones (Fig. 7.4), grouped so as to fill almost completely the surface occupied by the design, and separated from one another only by narrow, deeply cut lines. As a result, the background against which the relief appears is reduced to a linear pattern of deep shadow, undiminished in its effectiveness even on close viewing. The surface of the design is varied by the drilling of holes and the i­ ncising of striated and hatched patterns, rings of pearls, feathering and other devices. Given the earlier dating, this stucco belongs to a pre-Samarran tradition.

467

468

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.4  Balkh, mosque of No Gunbad, detail of stucco decoration.

from the rise of islam to the mongol invasion

Parallels for the vine ornament at Balkh are to be found in Sasanian stucco from Mesopotamian Kish and early Abbasid monuments of Central Iran at Nayin, Yazd, Buzan and the earliest parts of the Friday Mosque at Isfahan.12 We observe geometric grid designs of the soffits, girths and plinths, repetitive friezes of the impost blocks and the palmette frieze of the capitals. The surface is divided into a series of compartments by a network of intersecting bands (Figs 7.3–7.4). The compartments are then filled with vegetal ornament. The strap work at Balkh shows patterns derived from grids of touching and intersecting circles, circles inscribed in squares and star-and-cross arrangements. The arrangement of motifs within the major compartments of the strap work designs are symmetrical, for example in the soffits (Fig. 7.3), or concentric compositions on the girths of the pillars and columns (Fig. 7.4). The style of the ornament may be realistic, in the variety of vine ornaments, or abstract, in the palmette. Ghaznavid Buildings Monuments recalling Iranian, Central Asian and Indian styles were erected under Ghaznavid rule, often described under the broad category of ‘Seljuk’.13 Nowadays the most extensive and impressive ruins are preserved not in the city bearing the name of the dynasty (i.e. Ghazni), but at Lashkargah on the confluence of the Helmand and Arghandab Rivers, to the north of the citadel of Bust. Lashkari Bazar14 Sultan Mahmud and his nobles built their palaces and villas along the banks of the Helmand River. Most of them were concentrated along a seven-kilometre stretch between the citadel at Bust and the modern town of Lashkargah (Fig. 7.5). The court was accompanied by a large military escort who lived in barracks and cantonments near the palaces. The three most important palaces were built on a bluff overlooking the Helmand River and of these three the southern palace, built at the bend in the river, is the largest and most elegant. Extending for half a kilometre along the bank, the palace is built around a central court with four ivans or arched doorways (Fig. 7.6). Passing through the northern ivan one enters a spacious rectangular audience hall once bordered by columns and decorated with frescos and intricately sculptured stucco. Large panels with epigraphic borders surrounding a welter of sculptured stucco and interlacings of cut brick were found in the debris of the southern ivan. Mud-brick structures still preserve architectural and decorative features, for example large vaulted halls in both the south and central palaces and the horseshoe arch (Figs 7.7, 7.9, 7.10). The geometrical decoration includes angular interlacing strapwork, radiating from six-, eight-, tenor twelve-pointed stars and including polygons of various types. Panels displayed in the National Museum give an idea of the elegant decoration which once faced these barren walls. The excavation revealed also distemper paintings on the walls

469

470

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.5  Lashkari Bazar, general plan.

from the rise of islam to the mongol invasion

Figure 7.6  Lashkari Bazar, ground plan of the south palace.

depicting richly garbed guardsmen, weapons at the ready in brocaded belts, standing against a background of flowers, fruit-laden trees, birds and other animals. The palace guard show Central Asian features in the heads rendered in three-quarter profile (Fig. 7.11). In the centre of the great hall was a rose-petalled water basin fed by a canal running from east to west, indicating that the palace had running water. To the south of the great audience hall was discovered a small mosque elaborately decorated in sculptured stucco with borders of Qur’anic inscriptions, reconstructed in the National Museum. To the east of the central palace there was a large garden with a central pavilion, and a platform with an octagonal centre (Fig. 7.7). A contemporary observer, Baihaqi, mentions gazelles rounded up and herded into this garden, and also the great outings on the plains when the Sultan crossed the river on a canopied barge hung with silks.

471

472

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.7  Lashkari Bazar, forecourt of the central palace.

Figure 7.8  Lashkari Bazar, entrance to the central palace.

from the rise of islam to the mongol invasion

Figure 7.9  Lashkari Bazar, south palace, wall decoration.

Figure 7.10  Lashkari Bazar, south palace, squinches.

Evidence of some remodelling of the southern palace parallels political changes within the empire: new walls are to be observed, and then comes evidence of the great fire when the Ghurid ruler Ala al-Din burned the residence. There are, however, also archaeological traces of a reoccupation and restoration under the Ghurid Sultans. The final destruction came through the armies of the Khwarazmshah around 1215 or from Chingiz Khan in 1220.15

473

474

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.11  Lashkari Bazar, fragment of wall painting from pillar.

Bust The patrons of Lashkari Bazar or of Ghazna, in either case the early rulers of the Ghaznavid dynasty and their nobles, erected palaces and holy shrines also in the surroundings of Bust, which may have been re-used during the later Ghurid occupation. Here we illustrate a palace ruin at the southern end of the mud-brick buildings south of Lashkari Bazar and north of Bust (Fig. 7.82), an especially large structure with a sparsely but expressively decorated façade. Palaces of this type were to become the models for Islamic mud-brick architecture in adjoining Sistan (Figs 7.37, 7.38, 7.40). Nearby, we find the ruins of the arch of Bust, which is now generally ascribed to the post-Ghaznavid period (Figs 7.79–7.81). To this Ghurid period we ascribe also certain terracotta panels from Ghazna, to be dealt with later.

from the rise of islam to the mongol invasion

Figure 7.12  Ghazni, palace of Mas‘ud III, ground plan.

Ghazni16 The Ghazni plain is full of tombstones and marble graves of Ghaznavid nobles bearing beautiful inscriptions.17 Italian excavators revealed the palace of Mas‘ud III (492/1099–507/1114) (Fig. 7.12). They were led to associate a ruin field with this ruler by an inscription in his name on a piece of stone used on the keel arch of a mihrab (Fig. 7.13), and the proximity of the site to the minaret erected by the same Sultan (Fig. 7.18) suggested this connection. Another inscription bearing the date 505/111218 confirmed the connection of the site with the reign of Mas‘ud III. Contemporary sources describe the sumptuous palaces, bejewelled with booty from India; nowadays the heart of the complex is a large open rectangular court paved with marble, 50 × 31 m; even the footpath around this courtyard is paved with this precious stone. The palace underwent five quite distinct construction phases, with a mosque inserted in Phase 4. In the centre of each of the four walls

475

476

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.13  Ghazni, polylobated arch with inscriptions bearing the name of Mas‘ud III.

Figure 7.14  Ghazni, palace of Mas‘ud III, fragments of marble decoration. a: dado with benedictory inscription. b: panel with octagon pattern.

from the rise of islam to the mongol invasion

Figure 7.15  Ghaznavid copper dish. National Museum.

surrounding the court was an ivan, an architectural form much favoured by the Ghaznavids. The ivan on the north, which included a large imposing vestibule, functioned as a monumental entrance or propylaeum. Opposite, the southern ivan contained the throne room. On the east and west there were small rooms, on either side of the central ivans, including a pillared mosque in the north west corner. The walls of the court were gorgeously decorated, the upper portions embellished in terracotta and stucco in sculptured geometric patterns which were painted yellow, red and blue. The lower section of the walls preserves a spectacular example of carved marble (Fig. 7.14), containing an inscription 250 m long, not in Arabic as was usual for this period, but in Persian, representing one of the oldest examples of Persian epigraphy. Remains of marble decoration show geometric patterns familiar to us from Islamic art (Fig. 7.14), and further rare examples of animals and men, among others in lively hunting scenes. More carved and decorated stone was used in embellishing the palace than in surviving Islamic palace architecture elsewhere for the period.19 In the National Museum in Kabul are preserved fine specimens of Ghaznavid handicrafts: ewers, stirrups, copper dishes with mythical figures and Kufic inscriptions (Fig. 7.15, 7.17) and glazed bowls (Fig. 7.16). In the vicinity of this palace is the minaret of Mas‘ud III (1099–1115), mentioned above, on a round socle (Figs 7.18, 7.19). This tower and the minaret of Bahram Shah (1117–1153) situated nearby on a modern octagonal socle are conceived on the plan of an eight-sided star. Today only the ground storeys remain, but sketches

477

478

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.16  Ghaznavid glazed bowl with lustre decoration. National Museum.

of the early nineteenth century and older photographs show that the prismatic basements carried cylindrical shafts; according to these records the third storey of Mas‘ud’s minaret began with a plain circular wall with flat segmental projections above, then carved niches like a kind of negative form of the preceding pattern. The terracotta decoration of the towers corresponds to these rich and novel architectural inventions. The minaret of Mas‘ud III includes a fragmentary inscription of the Victory sura (XLVII) from the Qur‘an, probably to commemorate victory over non-Muslims in India. There is less evidence for the minaret of Bahram Shah, but both minarets are interpreted as victory towers.20 In the village of Rauza outside Ghazni is a modern building housing an extremely beautiful carved marble cenotaph marking the tomb of Mahmud, although it probably dates from after his reign (7.20).21 Baba Hatim Ziyarat At Imam Sahib in northern Afghanistan is a splendidly decorated tomb known to the local population as Salar Khalil and described as a Ghaznavid mausoleum.22 The building rises on a square plan which is transformed into the octagon carrying a squinch dome. This Iranian and Turanian type of construction can be compared with monuments of the tenth to twelfth centuries from Bukhara, Termez, Merv, Uzgand and Sangbast, and the architectural decoration in baked brick links the ruin with well-known buildings, especially of Khurasan art of the eleventh century. Rich

from the rise of islam to the mongol invasion

Figure 7.17  Ghaznavid copper and gold dish. National Museum.

stucco ornament again recalls eastern Iranian art. The calligraphy of inscriptions on the walls, recording the memory of the martyrdom of Salar Khalil Sayyid (Fig. 7.21), belongs to the best examples of the foliated Kufic script of the eleventh century. Charkh-i Logar Another fine specimen of Ghaznavid decorative art survives in a rare example of wood carving; in the upper Logar valley, in a mosque named after Shah Muhyi al-Din, in the village of Charkh-i Logar, exists a wooden mihrab with Qur‘anic inscriptions in Kufic script (Fig. 7.22).23 This piece of wooden architecture is said to have been brought from an old mosque in the village of Kachari, situated some

479

Figure 7.18  Ghazni, minaret of Mas‘ud III.

from the rise of islam to the mongol invasion

Figure 7.19  Ghazni, minaret of Mas‘ud III, detail of the carved brick-work.

481

482

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.20  Ghazni, marble cenotaph on the tomb of Mahmud. a: inscription on the face of the marble cap on top of the cenotaph. b: lobed niche on the rear of the marble cap. c: niche in the form of a mihrab on the base of the cenotaph.

distance to the north. In this perishable material we observe architectural innovations well-known to us from stone, burnt brick or mud-brick structures of the age, and can study the transformation of the rectangular mihrab by trabeate beams into a semi-circular base for a cupola. The latter is embellished by arabesques (Fig. 7.23). The mihrab opens in a cusped horseshoe arch (Fig. 7.22). The mihrab with conch, column and side panels represents a fine example of angular interlacing decoration. Further studies of this extraordinary specimen may also enlighten our understanding of the wood-carving in the doors from the tomb of Mahmud of Ghazna (now preserved in the fort at Agra) with their bold composition of postSamarra pulvin leaves interlacing with a heart-shaped pearled braid. The leaves

Figure 7.21  Baba Hatim Ziyarat, façade of monument to the memory of Salar Khalil.

Figure 7.22  Charkh-i Logar, wooden mihrab in local mosque.

484

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.23  Charkh-i Logar, zone of transition in mihrab.

have spiral tendril tips and include one pair of long-lobed fleurons. These muchdebated wooden doors are of a period later than Mahmud of Ghazna.24 Daulatabad In Bactria, to the north of Balkh, stands a minaret of pure cylindrical form (Figs 7.24, 7.25). This monument of Daulatabad carries a Kufic inscription naming the artist Muhammad Ali and the year 502/1108–09, and was interpreted as an important work in the Seljuk style.25 The style of the brick ornaments belongs, after Herzfeld and others,26 to the so-called hazar baf variety. Its essence is a highly

from the rise of islam to the mongol invasion

Figure 7.24  Daulatabad, minaret.

485

486

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.25  Daulatabad, minaret detail.

from the rise of islam to the mongol invasion

pronounced contrast of light and shade produced by all-over patterning of raised bricks in complicated meanders, the angles mostly being right-angles. Minor monuments The northern Afghan town of Sar-i Pul27 contains ten ziyarats, some of which are of considerable architectural importance. The most significant are the Seljuk Ziyarat-i Imam-i Kalan and the Ziyarat-i of Imam-i Khurd, ‘the Greater’ and ‘the Lesser’ Imam, both small simple dome structures of mud-brick with elaborate stucco decoration inside (Figs 7.26, 7.27). The latter is situated about 1∙.5 km south east of the centre of the town, and consists of a simple domed chamber about 4.8 metres square. The outer walls were covered in natural mud plaster; the inside, however, has walls with a spectacular decorative inscription in carved stucco. Its rich floral decoration belongs to the best Seljuk tradition.28 Two larger inscriptions in the mihrab are written in a highly decorative variety of foliate Kufic script. The text of the inscription informs us that Sar-i Pul corresponds to the medieval town of Anbir. Also in the north, to the south of Bilchiragh at Darra-i Shakh, are the remains of a Seljuk mosque whose decorated stucco mihrab includes floral designs and a kufic inscription (Fig. 7.28).29 Near a small cemetery south-west of the inner wall of ancient Balkh is the mausoleum Baba Rushnai, a massive baked brick mausoleum with a shallow pointed dome of the first half of the eleventh century. It has two entrances. Inside are the remains of a suls inscription recording the restoration of the building under the Timurids. There is a simple zone of transition of arched squinches (Fig 7.29).30 Also in Balkh Province, not far from the Daulatabad minaret, is the mud-brick mausoleum or khåniqåh of Abu Huraira, dated stylistically to 1000–1050. It has an ivan entrance opening onto an octagonal dome chamber with simple arched squinches, containing a cenotaph and mihrab (Fig. 7.30).31 The site of Shahr-i Gholghola in the Bamiyan valley includes the remains of two small Ghaznavid mosques, and accidental finds have included an elaborately engraved Ghaznavid brass bowl and an early thirteenth century carved wooden door (Fig. 7.31).32 Sistan Mapping mud-brick ruins and remains of burnt brick architecture of Islamic strongholds and cities in Afghan Sistan, British boundary surveys in the late nineteenth and early twentieth centuries, followed up by French, US and German archaeological surveys, came across a wealth of structural and decorative forms inherited from the Parthian, Sasanian, Seljuk, Ghaznavid and Ghurid art repertoire and heralding the subsequent evolution of Timurid times.33 Desert conditions have preserved extensive areas of standing urban remains at Chehel Burj, Chigini, Diwal-i Khudaidad, Gul-i Safid, Khwaja Siah Push, Kurdu, Nad-i ‘Ali, Nishk,

487

488

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.26  Sar-i Pul, stucco decoration in the Ziyarat-i Imam-i Kalan.

from the rise of islam to the mongol invasion

Figure 7.27  Sar-i Pul, stucco inscription in the Ziyarat-i Imam-i Kalan.

Figure 7.28  Darra-i Shakh oratory, stucco decoration.

489

490

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.29  Plan and elevation of the Baba Rushnai mausoleum.

Peshwaran, Sar-o-Tar, Tarakun and elsewhere (Fig. 7.32). All produced ceramics dating from the Ghaznavid and Ghurid periods, although it is not always possible to distinguish the standing remains from the Timurid and post-Timurid. The name Sistan, or Sijistan, derives from Sakastan (or Sakasthana), ‘Land of the Sakas’ or Scythians, and was also known to the Iranians as Nimruz, Persian for ‘A Half Day’, or ‘South’, that is, the land south of Khurasan. It possessed in the course of time various capital cities. Extensive Islamic ruins exist at Zaranj, a town erected on earlier Sasanian settlements and referred to frequently as simply Shahr-i Sistan, ‘the [main] city of Sistan’. The name Zaranj is derived from Zranka, the name of the Achaemenid province located in Sistan, which became Drangiana in Greek. From Islamic sources, Sistan or Shahr-i Sistan was well-known as the home of Islamic poets and religious teachers.34 In 1383 the city was destroyed by Timur, but according to literary sources it was rebuilt immediately after the

from the rise of islam to the mongol invasion

Figure 7.30  Plan and elevation of the Abu Hurayra mausoleum.

491

492

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.31  Plans of the two mosques at Shahr-i Gholghola, Bamiyan.

from the rise of islam to the mongol invasion

Figure 7.32  Map of early Islamic sites in Sistan.

493

494

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.33  Plan of Nad-i Ali.

catastrophe. Zaranj was on a branch of the Helmand River, and 48 km above the city the stream was confined by a series of dams and drawn off into five great canals which flowed towards the city. One of these irrigated the gardens adjacent to the town and another, the Sana Rud, constituted the city water supply, while the remaining three were for farm irrigation in the vicinity. We can compare this documentary evidence with archaeological evidence supplied by air photographs and field surveys.35 The ancient city of Zaranj is at Nad-i Ali to the north of the modern city of the same name, the capital of Nimruz Province, a large urban site whose origins go back to the Bronze Age as we have seen (Chapter 4).36 It was built on the concentric plan with an outer and an inner town, both walled (Fig. 7.33). The chief approaches were from Khurasan on the north, Nishak and Bust on the

from the rise of islam to the mongol invasion

east, and Fars to the south-west, and to each of these corresponded a gate, with two to the latter in the inner wall. Between the two Fars gates were the inner town markets, the old government house, the prison and the Friday Mosque, the minaret of which can still be recognised in the ground plan, while the structure itself has collapsed. Zaranj was connected with the capitals of the Islamic East by caravan tracks leading in all directions. An old route carried the traffic from Central Afghanistan to Sistan along the Helmand River. Near the fort at the present ruin site of Kordu37 a line of mud-brick pillars marks the site of a settlement that, due to wind erosion, has vanished, like many other medieval Islamic towns of this area. The same is true of a dilapidated castle, ‘Qala Hauz’, still known by this name to the Baluchi nomads, situated in the heart of the Registan desert, reminding us of the fact that traditional land routes passed through arid countries. All these caravan tracks that may be reconstructed from archaeological remains such as Bust, Kafar Qala, Rudbar and Qala Hauz had one common goal: the villages and towns of fertile Sistan on both sides of the lower Helmand, nowadays divided into Iranian and Afghan Sistan. In trying to combine historical evidence with the extant ruins of Sistan, we have the itineraries of Ibn Rusteh and al-Istakhri.38 The stages from Juwayn to Zaranj, for instance, were Basher, and the site of an old fire temple, Karkuyeh. Basher is a group of ruins known in the present day to villagers and nomads as Peshwaran. At one time there was a town of some size here and part of the south-west wall and a huge fortress can still be recognised (Fig. 7.41). Semi-round towers and threequarter-round bastions protected the citadel, of a multi-towered type well-attested by other Sistan strongholds such as Chehel Burj.39 Here we shall review the contribution of Sistan to the development of the minaret.40 One of the rare baked brick monuments of the country is the tower at Khwaja Siah Push (Fig. 7.34), in the centre of a vast field of heavily decayed mud and mud-brick houses full of slip-painted and incised pottery of the early Islamic period. They played an important role in the evolution of the Islamic prayer towers, leading to masterpieces like the Qutb Minar at Delhi. We have to compare purely stellate (Ghazni: Fig. 7.18) or cylindrical (Daulatabad: Fig. 7.24) forms of tower, the sequence of such units in various storeys, as in the cylindrical tiers of Jam (Fig. 7.48), and finally the combination of straight and rounded forms, as for instance in the original sequence of the upper part of the tower of Mas‘ud III at Ghazni. In Sistan, the Mil-i Kasimabad, dated by Tate from inscriptions to the twelfth century, consisted of a square base which formed the foundation for a round tower;41 this building collapsed only recently. While this tower seems to have followed a simple model, Sistan also preserved more complicated structures. The master builders of the Ghaznavid and Ghurid ages had at their disposal a variety of technical and decorative solutions. In the vertical elevation of storeys they combined prismatic and cylindrical units; at the same time, they used projecting cornices in the horizontal section of the respective storeys. At Nad-i Ali

495

496

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.34  The minaret of Khwaja Siah Push.

from the rise of islam to the mongol invasion

Figure 7.35  The stellate plans of the minarets at Ghazni, Nad-i Ali, Khwaja Siah Push and the Qutb Minar in Delhi.

(Zaranj) there existed until recently a minaret with a first storey of octagonal plan decorated in the middle of each face by semi-circular buttresses (Fig. 7.35).42 These decorative elements may be studied as structural units at the recently discovered baked brick minaret at Khwaja Siah Push: here eight rounded and angular buttresses alternate (Fig. 7. 34).43 The tower was decorated with bricks laid to form a series of repeating patterns, lozenges and semi-circles; lines of serrated bricks divided the lower and upper parts of this first storey. We have here various elements of Islamic brick decoration well-known from the rich repertoire in the centres of Muslim religious art44 – at Khwaja Siah Push in rather provincial

497

498

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.36  Gul-i Safid, ground plan of a courtyard house in the centre of the city.

e­ xpression. We do not know what forms may have crowned the minaret of Khwaja Siah Push, but we recognise clearly in the ground plan of this twelfth century construction the direct architectural model for the ground plan of the first storey in the Qutb Minar, Delhi, erected towards the end of that century.45 The second storey of the Delhi monument is closely related to a structural type to be deduced from the lost upper portion of the Ghazni minaret,46 while the third, star-like storey of the Qutb al-Minar is modelled on the prismatic forms of the still existing Ghazni tower (Fig. 7.18). Thus, artistic creations in the heart of Ghaznavid Afghanistan as well as in the far-off region of post-Ghaznavid Sistan contributed to further architectural development in the eastern Islamic countries. Gul-i Safid is a very large unfortified area of remains that includes some ninety houses, usually of the courtyard ivan type, with horseshoe arch decoration (Figs 7.36–7.40).47 The position of all buildings at Gul-i Safid – secular as well as ­religious – relates to the north-west wind. The main area of ruins – about ninety houses – consists of a type prevailing both in the smaller rural settlements and in the townlike units: to the centre, on the north-west of an open rectangular courtyard, rises one ivan, a rectangular hall, open to the south-east, covered by continuous tunnel

from the rise of islam to the mongol invasion

Figure 7.37  Gul-i Safid, decorated wall of ivan in the same courtyard house as Fig. 7.36.

Figure 7.38  Gul-i Safid, the same courtyard house as Fig. 7.36.

vaults or, in one case (Figs 7.37, 7.38), by a series of cross arches with transverse filler vaults of barrel profile – one of the typical vaulting systems of this period. A small door leads from this ivan out of the house towards the north-west. The ivan opens by a magnificent keel-shaped arch to the courtyard, flanked on both sides by mud-brick walls screening the entrances to additional chambers (Fig. 7.37). The walls are decorated by a series of niches decorated by horseshoe arches, by chequer-board patterns and other devices. Round the open courtyard are situated dwelling chambers covered by varieties of dome: squinch, ­pendentive, Turkish

499

500

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.39  Detail of the decoration at Gul-i Safid. Compare with Lashkari Bazar, Fig. 7.9.

from the rise of islam to the mongol invasion

Figure 7.40  Gul-i Safid, double-storeyed mud-brick tower to the north of the ruin-field.

501

502

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.41  The citadel at Peshwaran.

triangle. Half-round and three-quarter-round bastions may in earlier times have possessed a defensive function, but were used by the Islamic builders of Gul-i Safid as mere enrichment and decoration. In the north-west of the city a line of multistoreyed towers seems to have protected the arable land and irrigation system. Further, air photography allows us to identify a central hill with the ruin of a fortress, several cisterns, a series of tombs, two windmills, one mosque and one ziyarat. As we observed, the cultivable area and the open settlements of Islamic Sistan were protected by a line of fortification towers and fortresses. In the agricultural zones between we also encounter unique buildings like the rectangular enclosure of Qala-i Chegini among isolated dwelling houses, rectangular fields and abandoned gardens. No inscriptions or literary sources inform us of the function of this mud-brick complex, which may have been a caravanserai or the seat of a governor. Spacious ivan halls with lofty tunnel vaults and baked brick revetments supporting mud-brick friezes of typical horseshoe arched niches (Fig. 7.37) belong to an architectural tradition going back via Ghaznavid Lashkari Bazar (Fig. 7.9) and early Islamic palaces to Sasanian models. At Peshwaran is a vast area of extensive remains, covering an area of c. 9.3 × 13 km. It includes a huge circular citadel some 200–300 m in diameter, with baked brick walls 17 m in height with a 2 m wide gallery around the top (Fig. 7.41). There is also a large mosque and adjoining madrasa that was recorded in better condition by the first explorers of the site,48 the entrance ivan of which still preserves part of a floriated Kufic script in moulded plaster (Fig. 7.42). In the present state we can recognise the remains of magnificent ivan structures, wall decoration in horseshoe-arched niches and traces of inscriptions frequently alluded to because of their extremely fine workmanship.49 Inside, some painted plaster decoration is still preserved and there is the design of a moulded star over the mihrab.50

from the rise of islam to the mongol invasion

Figure 7.42  Peshwaran, façade of the mosque.

In the construction of the ivan-hall of Peshwaran and in the wall decoration of this mosque with horseshoe arches we may observe the evolution of art motifs that originated under the Ghaznavids, and also the four-ivan type of layout as at Lashkari Bazar and Ghazni which was to find further elaboration in the provincial art of Sistan.51 Extensive irrigations systems have been recorded at Diwal-i Khudaidad, Qal’a-i Chigini and elsewhere. The most prominent feature of the latter is a large fortresspalace enclosed in a low wall with a decorated façade of blind keyhole-shaped arches (Fig. 7.43).52 One of the largest areas of remains in Sistan is Sar-o-Tar, a vast area of remains and ancient cultivation stretching for several kilometres across the dunes. The main part, known as Shahr-i Gholghola, is a large fortified urban site covering about a square kilometre consisting of successive square, circular and pentagonal enclosures marking a probable palatial complex (Figs 7.44, 7.45). This includes the remains of a mosque and over a hundred rooms, many of them domed and containing stucco decoration. Most of the fortifications are mud, although there is extensive baked-brick construction as well. A hoard of 406 copper coins dated 1167–1221 was found in the mosque.53 At Qal’a-i Nau is a large, well-preserved fortress of early Islamic date connected to a probably pre-Islamic tower (Figs 7.46, 7.47).54 The landlords in the rustaqs of Sistan seem to have been in close contact with the leading cultural centres of eastern Islam. In the façade of a large ivan courtyard

503

504

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.43  House façade at Chigini II.

Figure 7.44  Citadel and walls of Shahr-i Gholghola in Sar-o Tar.

from the rise of islam to the mongol invasion

Figure 7.45  Plan of Sar-o Tar.

house at Chegini II (Fig. 7.43)55 are again reflected architectural and decorative inventions from Sasanian Ctesiphon and Abbasid Ukhaidir. The spacious vaulted ivan, the monumental keel-arch door and the distribution of horseshoe-arched blind niches in the walls lead us to conclude that master builders were engaged to create magnificent buildings for local chieftains in cheap mud-brick, and that they worked on the same lines as in other centres where architecture and decoration were executed in burnt brick or cut stone.

505

506

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.46  Plan and elevation of the fortress of Qal’a-i Nau.

Outside Afghanistan Outside Afghanistan we find some buildings in eastern Iran that may belong to an extension of Ghaznavid artistic influence, like the Mausoleum of the Ghaznavid governor Arslan Jadhib at Sangbast56 or the caravanserai of Ribat-i Mahi with its early Iranian vaulting system.57 However, the ornaments of interlacing frame work and other motifs of the early twelfth century still lack definite proof of their connection with Ghaznavid dynastic art.58 In Sind the so-called mausoleum of Khalid b. al-Walid might also be Ghaznavid.59 Summary of Ghaznavid Art and Architecture The art which flourished in Zabulistan under the dynasty of the Ghaznavids collected and transmitted motifs of Sasanian and other origins and was a source for Islamic art in India. The cradle and centre of this culture was Ghazna; before

from the rise of islam to the mongol invasion

Figure 7.47  The fortress of Qal’a-i Nau.

Sebuktegin and his son Mahmud it was a Buddhist centre, a trading city on one of the old routes between Iran and India. It became suddenly one of the leading centres of Asia when Mahmud made it a cultural centre using wealth acquired during his Indian expeditions. He endowed a madrasa with a rich library attracting famous scholars and poets like Firdausi and al-Biruni. In order to gain an overall idea of Ghaznavid art we must use the rich literary sources to interpret the archaeological evidence collected during recent years. Before Mahmud, the historian al-Muqaddasi mentions wooden structures and mosaic art, while al-Utbi describes various mosques; there had been great builders before the Ghaznavids and Samanids on the soil of Afghanistan, as Ibn Hauqal reported. Gigantic monuments which have now disappeared once distinguished these vanished cities. The dome of the Naubahar at Balkh, more than a hundred cubits high, built in the Sasanian period by the Barmakids, was greater than the very palace of Mansur at Baghdad.60 As for Ghazna, there can be no doubt of its splendour in the early eleventh century. Here the poets confirm the historians: How many a palace did great Mahmud raise, At whose tall towers the Moon did stand at gaze, Whereof one brick remaineth not in place.61

Of some of Mas‘ud’s buildings more specific accounts have survived. When he came to the throne he already had at his disposal a series of royal residences:

507

508

THE ARCHAEOLOGY OF AFGHANISTAN

in addition to his father’s great estate, the Bagh-i Mahmud (Mahmud’s Garden), and his old palace, Gawshak-i Kuhan-i Mahmud, there were the garden of the Hundred Thousand, the Victory Garden, the State Palace and the White Palace, but all these and more besides did not suffice. And so there came into being the Gawshak-i Mas‘udei, dedicated with a fabulous festival. The main royal buildings formed a symmetrical group, in the centre the audience hall, on the right the summer palace, on the left the winter palace which was domed. Nowadays we may visualise in these architectural groups great ivans such as those cleared by archaeologists at Ghazni (Fig. 7.12) and Lashkari Bazar (Figs 7.6, 7.7). The latter palace has also traces of wall paintings such as those vividly described by contemporaries in the great palace paintings of Mahmud of Ghazna (Fig. 7.11). We know that his victorious armies and his elephants were depicted and in one hall were portraits of Mahmud himself. In other apartments were the combats of the Sasanian kings.62 Historical and literary texts attest to the building activity of Mahmud, Mas‘ud I and Mas‘ud III; to a lesser degree we know also of the works sponsored by Ibrahim and Bahram Shah. Ghaznavid architecture tends to grandeur and opulence, a contrast with the building material of mud-brick. One of the great innovations of Ghaznavid art was the grouping of ivans. The dome and the ivan are the two major spatial forms which give the buildings of Iran their impressive monumental character. The word ivan can be used for an enclosed hall with flat roof, but in Islamic architecture it means a space, whether portal or hall, which is surrounded on three sides, open to the fourth side and covered by a barrel vault. The ivan opening on a court can be followed by a square domed room, which in turn opened into a series of porticoed courts. Archaeological work at Lashkari Bazar and Ghazni showed that compositions of four ivans on the axes of a courtyard were a prominent feature in Ghaznavid architecture before four-ivan compositions became commonplace in Khurasan as ground plans for mosques, madrasas and caravanserais. Another invention or adaption of earlier forms, of far-reaching consequence for later eastern architecture and decoration, was the use of the horseshoe arch (Fig. 7.9). The round horseshoe arch may have been derived from Persia on the strength of details of construction observed at Firuzabad, built by Ardashir in the third century ad. Here the arch was set back at the springing so as to provide a support for the centring. Some believe that the motif came from India, where a number of rock-cut caves or Chaitya halls have an entrance arch of this form. Another possible origin lies in the late antique architecture of northern Syria. In the history of Islamic architecture the horseshoe arch is first met with in the great mosque at Damascus, where the arches of the transept and the lower arches of the arcades are slightly larger than a semi-circle. In Ghaznavid art, Near Eastern and Indian origins may have been combined; they were enriched by decorated pointed forms and led to a regional development in Sistan (Figs 7.37, 7.38, 7.43).

from the rise of islam to the mongol invasion

Palaces These provided the most imposing architecture. Four-ivan compositions ultimately go back to Parthian prototypes at Assur, Sasanian structures at Ctesiphon and Abbasid imperial buildings at Samarra and Ukhaidhir. We learn from Farrukhi that Mahmud’s brother Yusuf resided in a pavilion of four columns, from where the view was directed through four gates into four regions of the world.63 Gardens Gardens surrounded all palaces, and were also conceived alone with temporary structures of tents and pavilions. Mahmud’s Bagh-i Nau at Balkh was embellished by a small lake with a marble surround and at Ghazni water pipes with silver mouthpieces spouted fresh water on the parks, in a desert climate. This is now further amplified by the complex hydrological networks installed in the Balkh oasis after the arrival of Islam.64 The three main palace complexes at Lashkari Bazar had vast garden enclosures associated with them (Fig. 7.7). The Ghaznavid tradition probably originates with the Sasanian garden and hunting parks (paradaisa, hence English ‘paradise’). Mosques Mosques had, after al-Utbi and especially according to his description of the mosque, ‘The Bride of Heaven’, perfect proportions. In the construction trees from India were used; the walls were polished and sparkled ‘like a maiden’s face’. One special room was reserved for the Sultan. While no remains have been discovered of this magnificent building at Ghazni, we know the small prayer room from Lashkari Bazar, serving as a palace mosque, situated in an angle of the huge enclosure near the palace entrance. A central room in front of the mihrab had four pilasters that may have carried a dome. Minarets Minarets in baked brick (Figs 7.18, 7.33, 7.34) marked the sites of mud-brick mosques. They sometimes functioned at the same time as victory towers.65 During Ghaznavid (and Ghurid) times the knowledge of Hindu towers in the conquered parts of northern India may have inspired Muslim builders to copy jayastambhas or kirtistambhas.66 Decoration Decoration is best preserved in the Ghazni baked brick minarets (Fig. 7.19), where the tendency to encrustation has been sustained and realised. Its very form, with

509

510

THE ARCHAEOLOGY OF AFGHANISTAN

the salient angles, is decorative, and the division of each of the great facets into a succession of panels, decisively separated by repeated horizontal bands, provides a handsome scheme of ornaments in the patterning of the brickwork. Their stellate plans also occur in the minarets of Nad-i Ali and Khwaja Siah Push in Sistan, the Qutb Minar in Delhi, Jar Kurghan in Uzbekistan and some of the Seljuk tomb towers in Khurasan in Iran as well as some temples in India (Fig. 7.35).67 The plan – at Ghazni at least – possibly derived from the plans of miniature Buddhist stupas at Tepe Sardar near Ghazni.68 Each successive panel is larger by 50% than the one below, so that they increase in size as they recede upwards from the spectator. Each unit is filled with a different all-over geometrical design. Ghaznavid sculptors evolved a noble decorative repertoire. At the beginning of the dynasty, the carvers were evidently still modest in their attempts, judging from the tomb of Sebuktegin, for here the ornament is limited to bands of Kufic inscription with the simplest possible interstitial motifs. The technique of cutting in two levels only, with very low relief and firm regular outlines, whereby the background is removed from around the pattern leaving it slightly raised but flat, had long been practised in wood. Essentially the same technique was continued at least through the twelfth century, but it was progressively enriched in order to execute the increasingly complex designs organised in multiple systems that implied superimposed levels (Fig. 7.19). Wood and stone alike are carved into delicate foliate interlacements in which many old forms are retained, but enriched and supplemented with new elaborations; thus, in the border of each panel on the doors of Mahmud’s tomb the undulating stem with foliation on alternate sides carries leaves in the form of a halberd with a recurvate tip. The arches are characteristically trefoil, the form long used in this region for Buddhist art, and a cusped interior outline likewise has an Indian flavour (Fig. 7.20). Painting Painting was inspired by Abbasid as well as by Central Asiatic Sogdian and Turkic prototypes. Literary sources tell of the lavishly embellished private quarters of Mas‘ud I at Herat. The remains of wall paintings from Lashkari Bazar prove once more the mediating role of Ghaznavid art between ancient Oriental or Central Asiatic themes and later Islamic developments. The walls of the ivan hall must have been decorated by a continuous frieze of male figures in isocephalia. From the surviving fragments of 44 figures, distributed on the walls between the door openings rhythmically in the relation 14:8:8:14, there would have been a procession of about sixty persons, oriented towards the throne of the ruler, and rendered in frontal view, but with the feet in profile (Fig. 7.11). The fragment of a head on a round pillar suggests that the faces were in three-quarter profile before a nimbus. The facial expression suggests a Turkic origin for the palace guard – precursors of which may be sought in the Central Asian Sogdian painting of Afrasiab and

from the rise of islam to the mongol invasion

Panjikent69 or even among the ‘Tocharian princes’ on the walls of buildings in the Turfan oasis or in the processions of Achaemenid reliefs. If we possessed Ghaznavid miniatures, we could check the information supplied by a Ghaznavid treatise, Bayan alt-Adyan, that among the treasures of the Ghazna library there was a copy of the Arzang, the legendary book of Mani, with illustrations, and also whether the influence of Manichaean book art was felt in Ghaznavid monumental and miniature painting. Sculpture Best preserved on marble slabs, this shows palace life, dancers, animal fights and hunting scenes. The persons depicted wear Central Asian costumes and have Mongol visages. The figures are clumsy, their movements rather awkward; the relief is flat and scarcely modelled. There is no archaeological evidence of bronze sculpture such as the four bronze warriors around the throne of Mas‘ud I guarding the crown, described by Baihaqi. Ceramics Samanid and Ghaznavid ceramic styles were recognised in the slip-painted wares of Lashkari Bazar70 and of Ghazni. Kilns and kiln-wasters at Lashkari Bazar prove the local manufacture of the red or buff sandy fine-grained earthenware, covered with white, cream, manganese-purple, brown or even, rarely, tomato-red slip. In the early period the main decorative pattern was Kufic script. Palmettes and floral motifs, rosettes, scrolls and three-petalled flowers are frequent, while geometric patterns and cross-hatchings are rare. Some slip-painted vessels are decorated with dots outlining the main decorative themes, such as birds or other animals. The ‘House of Lustre-ware’ at Ghazni71 supplied rare examples of white-bodied ware with lustre decoration (Fig. 7.16). Lustre is formed by painting on the glazed and otherwise finished pottery a pigment derived from metallic salts, which, when fired at a low temperature in a special kind of kiln, deposits a thin film of metal on the glaze. Slip-painted and lustre ceramics dateable to the Ghaznavid period were surface finds in neighbouring Sistan.72 Ghurid Buildings The Ghurids were an eastern Iranian dynasty which flourished as an independent power in the twelfth and early thirteenth century, based on the mountainous region of Ghur, the hills and valleys of the classical Paropamisadae, now western central Afghanistan. The family name of the Ghurid Sultans was Shansab, and at the time of the inflorescence attempts were made to attach their genealogy to the ancient Iranian epic past. Within the empire of Sultan Mahmud of Ghazna, Ghur remained an unabsorbed enclave, and during his reign at least three expeditions

511

512

THE ARCHAEOLOGY OF AFGHANISTAN

were sent by him into Ghur. In 1011, the Shansabani chief Muhammad b. Suri was captured in his stronghold of Ahangaran. He was deposed and his pro-Ghaznavid son, Abu Ali, set up as the Sultan’s vassal in 401/1011. Abu Ali is said to have erected mosques and madrasas for the newly introduced Islam. With the accession of Izz al-Din Husain (493/1100–540/1146), Ghur became a buffer state between the truncated Ghaznavid empire and the empire of the powerful Seljuks, then, with the relative decline of the Ghaznavids after Ibrahim’s death in 1099, Ghur was drawn into the Seljuk sphere of influence. After ‘Ala al-Din Husain Jahan-suz, the ‘World-Burner’ (544/1149–556/11161), however, the Ghurids became an imperial power by the middle of the twelfth century. ‘Ala al-Din earned his name with the sack of Ghazna in 1150 destroying Ghaznavid power for good, enabling his empire to expand both eastwards – Lahore was captured in 1186 and Delhi in 1192 – and westwards at the expense of the Seljuks: Nishapur was captured in 1200. Under Shams al-Din (later Ghiyath al-Din) Muhammad of Ghur (558/1163–599/1203) and his brother Shihab al-Din (later Mu’izz al-Din) Muhammad of Ghazna (569/1173–602/1206) the Ghurid empire reached its apogee. These two brothers maintained a partnership and amity rare for their age. Broadly speaking, the first was concerned with expansion westwards and the checking of the Khwarazm-shahs’ ambitions in Khurasan, while the second attacked India. These Indian expeditions had manifold cultural consequences: north-west Indian art styles led to the erection of the unique mosque of Larwand (Figs 7.65, 7.66) and Afghan conquerors raised the Indo-Islamic masterwork of the Qutb Minar at Delhi. Ghurid cultural influence extended also towards the west, where in Sistan the Saffarid Amir, Taj al-Din Harb (562/1167–612/1215), acknowledged Ghurid suzerainty, but within a decade of Muizz al-Din’s death the Ghurid empire fell apart, passing for a brief while into the hands of the Khwarazm-shahs in 1215 before coming under Mongol rule. A separate branch of the Shansabani family ruled in the Bamiyan region from Shahr-i Gholghola between 1145 and 1215, where extensive fortifications bear witness to their rule, and for a while even extended their rule as far as the Amu Darya, but they too succumbed to the Khwarazm-shahs until they in turn were conquered by the Mongols (indeed, Bamiyan was the scene of one of the more savage destructions by Chingis Khan, hence the modern name of its capital, Shahr-i Gholghola or ‘City of screams’). However, the Ghurid legacy was to enjoy a post-Mongol ‘Indian summer’: quite literally, as Mu’izz al-Din appointed his Turkish general Qutb al-Din Aybak as governor in Lahore of the Indian provinces in 602/1206. Qutb al-Din thence initiated the line of the Mu’izzi Sultans or ‘Slave Kings’ which ruled until 1290. In the ruined sites of the Ghurat and in ruins of Ghurid age from Sistan towards north Afghanistan, from Herat to eastern Afghanistan, we find sgraffiato and carved wares of pottery.73 The sgraffiato wares, with incised lines of geometric or free patterns on green, brown or yellow splash, bridge in the Islamic period, the gap between the early Iranian slip-painted wares on the one hand and the ‘Fine Seljuk’

from the rise of islam to the mongol invasion

wares on the other. Archaeological evidence is present from the Ghurid levels at Bamiyan and Lashkari Bazar/Bust74 and from Herat and Jam.75 The minaret of Jam Islamic historical accounts inform us of a mighty capital of the Ghurid rulers at Firuzkuh; after the discovery in 1956 of the minaret of Jam (Figs 7.48–7.51), with subsequent surveys of the region and sondages in the vicinity of the minaret, this centre was identified with Jam.76 The minaret stands 65 m high with a basal diameter of about 9 m; it is constructed of fired brick. Above the octagonal base rises the first cylindrical tier. On the first balcony projecting wooden beams are still to be seen, while the fragile brickwork has fallen away. The whole tower consists of four slightly tapering cylinders narrowing inwards at the stages marked by the corbels of balconies now gone. The entire exterior of the shaft of the minaret is decorated with carved brick relief ornamentation laid over the plain structural bricks (Fig. 7.51). The most intricately decorated tier is the first cylinder, its surface being divided into eight vertical segments. Each vertical zone is subdivided into smaller areas by a narrow band of inscription which moves in an unbroken line around each panel and from section to section. The text is the entire Sura of Maryam, the nineteenth chapter of the Qur‘an.77 Just below the corbels of the first balcony are three floral bands (Fig. 7.50, upper portion), the fourth row down comprising a trefoil or stylised tulip motif. Between the first and second balconies is an undecorated area to the height of the second-level doorway. The lower of the two Kufic inscriptions on this tier rests above a band composed of a network of geometric designs. Above the second balcony is another plain area rising to the height of the third doorway. On this level there is another band of Kufic inscription. The walls are composed of interlocked layers of plain fired bricks about 20 cm square and 5 cm thick. Inside the base and the first cylindrical tier there is a double spiral staircase. These staircases, one over the other, terminate at the top of the central core. Inscriptions on the first cylinder from the bottom refer to Ghiyath al-Din Muhammad ibn Sam, who ruled Firuzkuh from 1153 to 1203. The style of the Kufic is traditional.78 On the lowest section of the north face an inscription gives a date that has been interpreted either as 570/1174–579 or 590/1193–4.80 If erected as a victory monument, as is commonly interpreted, the first date would celebrate the Ghurids’ conquest of Ghazna, the second their conquest of Delhi. The first cylinder of the minaret utilises ‘panel architecture’.81 This technique is also seen in early Islamic art, where structural sections are emphasised by the decoration. The three-dimensional ornamental brickwork within the wall panels creates a pattern of light and shade. Decorative brickwork technique of the same kind is found in Ghurid buildings of Herat (Figs 7.67, 7.68), Shah-i Mashhad (Figs 7.70–7.72) and Chisht (Figs 7.63–7.64). Among the motifs used, bands of pearls

513

514

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.48  Jam, the minaret, with fortifications in the background.

from the rise of islam to the mongol invasion

Figure 7.49  Jam, the minaret: view showing horseman at the foot for scale.

Figure 7.50  Jam, detail of the tile decoration on the minaret.

515

516

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.51  Jam, detail of the brick decoration on the minaret.

from the rise of islam to the mongol invasion

(Fig. 7.50) are known from Sasanian times, and are also found on the Tower of Bahram Shah, at Ghazni (Fig. 7.18). A survey of the area in 1973 by Werner Herberg did much to contextualise the minaret. On the hill of Kush-Kak, to the west of the Jam-Rud, halfway between the village of Jam and the tower on the confluence, are some tombstones with Hebrew inscriptions indicating the cemetery of a large Jewish community in this area of Ghur from 1149 to 1215.82 To the south of the Hari-Rud were mapped mud-brick watch-towers constituting a fortification line. To the north of this river is a fortress opposite the tower of Jam (Fig. 7.48). The natural rock was connected by inner and outer stone walls that are preserved up to a length of 200 m. On the summit was found a rock-hewn and stone-built water-reservoir. In and around these ruins were sherds of sgraffiato ware; to the west of this fortress, in the valley of the Hari-Rud, immediately north of the tower, some ruins were still visible years ago, but they have been cleared away; local tradition called them a ‘bazaar’. Finally, in the angle between the Hari-Rud and the Bedan-Rud there are to be seen remains of a smaller fortress, and on the bank of the Bedan River was found an Islamic rock inscription. The fort was probably destined to protect the valley of the Bedan-Rud, that historically important route towards Ahangaran, a place mentioned by Islamic writers, preserving both pre-Islamic and Islamic ruins. Further intensive investigations of the immediate environs of the minaret in 2003 and 2005 by David Thomas and his team further amplified the context of the minaret.83 Investigations and clearances of alluvium immediately to the east of the minaret revealed fragments of a large baked brick building with a paved brick courtyard, almost certainly the remains of the mosque or madrasa associated with the minaret. A series of further sondages, plus the examination of some of the large numbers of robber pits that had appeared on either side of the river valley in recent years, revealed traces of many more buildings and ceramics, leaving no doubt that a sizeable town existed here in the eleventh and twelfth centuries. The minaret of Sakhar In the nineteenth century, persistent reports that ranged from the remains of a ‘Ghurid palace’ to an ‘ancient city’ in the remote mountain region of Sakhar in Ghur reached members of the British Afghan Boundary Commission, remains that reportedly included a brick minaret 30 m high.84 Investigations in 2013 confirmed the existence of the minaret, somewhat less than 30 m high, with a tapering shaft of mud-brick but with the top decoration of interlaced lozenges and zigzags in baked brick (Fig. 7.52). Local reports confirm the one-time existence of a mosque, now gone. The reported ‘palace’ might the fort of Qal’a-i Sultan nearby; the ‘ancient city’ is presumably exaggeration.85

517

518

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.52  Sakhar, the minaret.

Shahr-i Arman and Naraiman In the mountains south-east of Qal’a-i Nau in Badghis Province in north-western Afghanistan are the remains of a small town, Shahr-i Arman, consisting of a ruined stone fort surrounded by mounds, ruins and ancient gardens and irrigation, possibly the site of the Ghurid city of Shurmin. Many Seljuk ceramics have been found at the site.86 Further to the west, just off the main road from Qal’a-i Nau to Herat, is the massive stone fortress of Naraiman or Naratu occupying the whole of a rocky plateau, separated from a mountainous range by a narrow pass, surveyed by Marc Le Berre and Jean-Claude Gardin in 1952. The remains appear early Islamic on the basis of the pottery collected, but Achaemenid and Timurid pottery was also present. The north and south faces are mostly sheer and thus impregnable. The main gate is formed by a pointed arch of sandstone voussoirs joined with lime mortar. The few sections of walls which remain are built on the rock itself directly above the cliff, consisting of an exterior facing and a continuous interior archway, which probably carried a covered way. The north and north-west walls include very eroded towers. Inside the fortress, at the highest point, are the remains of two monuments. Of the first, only one room remains, flanked by an ivan on the north and another on the south, pierced by a door on both the east and west sides. The construction is of the same type as that of the rampart, with in addition some brick elements. Of the second monument, there remains only the central part, octagonal in plan. Each face contains a niche, with a pointed arch faced with bricks. The whole is constructed in bricks on a rubble foundation. In the centre is a ziyarat. The remains were tentatively identified by Le Berre with early Islamic Dehistan.87

from the rise of islam to the mongol invasion

Ghur fortifications We may now survey the other main archaeological sites of the Ghurid period. In the north is Ahangaran, with an old fortress in the middle of a valley on an earthen mound some 60 m from the left bank of the Hari-Rud.88 From a distance, the earthen ramparts, blocks of stone and fragments of wall are still impressive. The centre of the castle was a tower and the outer walls were reinforced by towers also. It was because of such forts that Sultan Mahmud could not force a passage in spite of his great army. Isolated towers and smaller castles connected Ahangaran in a line of fortifications among which fortresses like Guzarpam and Chehel Gazari stand foremost. South of Jam is the valley of Shaharak89 with numerous tepes, mud-brick ruins and towers marking pre-Islamic and Islamic settlements. Several surveys have recorded many ruins in the area between Hari-Rud and Farah-Rud, predominantly fortified sites. We know that in the Ghurid period the land route through these ranges was protected by watch-towers and castles of both military function and artistic importance. On both sides of many of the valleys on this route, for example, lines of square or round mud-brick towers dominate the crests of gently rising hills (Figs 7.53, 7.57). A cylindrical tower at Nili rises on a stone basement (Figs 7.54d, 7.55); the military architecture is embellished on the

Figure 7.53  Qal’a-i Chahar Baradar, line of square and round towers.

519

520

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.54 Comparative plans of towers in Ghur: (upper left) Chahar Baradar 1; (upper right) Chahar Baradra 3; (lower left) Chahar Baradar 4; (lower right) Nili 2.

from the rise of islam to the mongol invasion

Figure 7.55  Nili, mud-brick tower on stone base. Note remains of two more towers on the other side of the valley.

Figure 7.56  Qal’a-i Chahar Baradar 4, detail of the decoration.

521

522

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.57  Muna ‘Ala tower decoration.

from the rise of islam to the mongol invasion

Figure 7.58  Qal’a-i Qaysar fortress. Note strategic position overlooking the pass.

upper part of the mud and mud-brick walls by simple but expressive geometric patterns (Figs 7.56, 7.57).90 The mud-brick fortress of Qal’a-i Qaysar (‘Caesar’s Castle’), one of the largest and most elaborate fortifications in Ghur, consisting of inner and outer defensive walls reinforced by circular and hexagonal towers, is perched upon steep rocks in a dominant position blockading a pass on one of the medieval trade routes through the Ghurat (Figs 7.58–7.60).91 The present village of Yahan is situated in a valley around which are rising terraces studded with mud-brick ruins among which excavation might reveal dwelling houses or even residential areas. Again, towers of pakhsa or mud-brick are placed upon small hillocks to protect an old urban site and connect with a line of watch-towers in the vicinity. Large numbers of similar towers have been recorded in Ghur. A study of these fortifications confirms the historical accounts of the Ghaznavid campaigns in the early eleventh century and the information by the geographer Yaqut that Ghur had no towns of note, but only agricultural settlements and – most typical of

523

524

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.59  Qal’a-i Qaysar fortress.

the landscape – many fortified places and towers (qasr, qala, hisar, kushk) where the freedom-loving people could defend themselves.92 These strongholds were also favoured places for immuring political prisoners. The commanding positions of the fortresses (Fig. 7.58) reflect the fragmentation of political power in Ghur during the early medieval period, with no obvious centre from which the whole mountain region could be controlled by a single ruler. The sheer quantity of the remains means they add up to one of the most extraordinary architectural complexes in the eastern Islamic world. Furthermore, due to the impenetrability of the terrain the recorded remains probably only represent the tip of the iceberg, and the 1:1,000,000 maps based on satellite images show substantially more unrecorded towers and other remains (indicated on the maps). The largest concentration of remains are certainly to be found in the region to the south-east of Jam – the Firuzkuh–Zamindawar route, broadly – and there are also a large number of uninvestigated remains recorded on maps to the northwest of Zamindawar.93 The fortifications appear to defend the routes leading out of Ghur towards the south and south-east, to Kandahar and the Ghaznavid centres of Lashkari Bazar and Ghazna. As an overall pattern, they appear to be a defensive shield sheltering the Ghurid heartland from Ghaznavid threats from the south-east. If so, the Ghur fortifications are one of the most impressive systems of fortifications of the Islamic world.94

from the rise of islam to the mongol invasion

Figure 7.60  Qal’a-i Qaysar fortress detail.

Bamiyan fortifications The nearest comparable archaeological complex to the Ghur fortifications is in the Bamiyan region, where similar quantities of fortifications occur and have been subject to more detailed study. Like the Ghur fortifications, those in the Bamiyan region cluster along the valleys and routes, in this case the routes leading in and out of the main centre at Bamiyan. They too usually consist of small forts, fortlets and individual towers, with the occasional larger more complex fortification such as Shahr-i Zohak near the confluence of the Bamiyan, Hajjigak and Qunduz Rivers (Figs 6.52, 6.53; also described in the previous chapter). A four-ivan courtyard house was recorded by Godard before it was totally destroyed.95 In the deserted

525

526

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.61  Bamiyan, tower decoration.

town of Shahr-i Gholghola surface collection of sgraffiato pottery indicates a Ghurid occupation and two small mosques were recorded (Fig. 7.31).96 Surface finds of incised ceramics allow us to date to the Ghurid period the mountain fortress of Kafir Qal’a97 on the Shibar Pass dominating one of the main routes from the north towards the south. The decoration of impressed triangles in the form lozenge or diamond patterns is a pronounced feature at Bamiyan (Fig. 7.61) and is similar to the towers in Ghur (Figs 7.56, 7.60). However, none of the more curvilinear plaster decoration recorded in Ghur has been recorded at Bamiyan (Figs 7.56, 7.57). The Bamiyan fortifications have been broadly dated to the fifth to sixth centuries through to the early thirteenth century, largely on the basis of (admittedly few) ceramics. The earlier part corresponds to the period of the ascendancy of the Turkish Kaganate and its dependencies, with its offshoot in eastern Afghanistan, the Turki Shahs. The later period corresponds to the Ghurid Shansabani kingdom of Bamiyan in the twelfth to thirteenth centuries, an offshoot of the main Sultanate. However, all one can say for certain is that both the Ghur and the Bamiyan fortifications systems belong to the same broad architectural traditions, and that the Ghurid dynasty of the twelfth to thirteenth centuries is the only state structure that is common to both.98 All pre-Islamic and Islamic settlements were destroyed by the Mongols in 1221. To the east of the Bamiyan valley are buildings of the Ghurid age, fortifications in the Kunar valley and ruins in the Logar valley.99

from the rise of islam to the mongol invasion

Chisht The western frontier of Ghur is marked by the village of Chisht-i Sharif on the Hari-Rud, situated in a lovely valley with large trees and a small stream. Here are remains of a building still not adequately explained, but possibly a school (madrasa), including a tomb built by Sultan Ghiyath al-Din ibn Sam, according to the traces of an inscription on the mosque (Figs 7.62–7.64). The monuments were subject of a study by a German team in 2006 (whose publication is forthcoming at the time of writing).100 The remains now comprise two widely separated dome chambers, but traces of walls plus the scatter of building debris connecting and surrounding them show that they were originally part of a single complex. Both chambers are richly decorated in elaborate brick decoration, recalling that of Shah-i Mashhad (below). The structural technique of the squinch dome as well as the delicacy of the wall decoration are typical of Ghurid art. Fifteen kilometres west of Chisht the German team also surveyed a Ghurid octagonal mausoleum of baked brick surmounted by a dome 11 m high. It is surrounded by a cemetery that includes inscribed marble.101 Larwand Near the southern frontier of Ghur we find the ruin of Larwand that throws further light on medieval connections between the heart of Afghanistan and

Figure 7.62  Chisht, view of the two dome chambers.

527

528

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.63  Chisht, plans and sections of the two dome chambers.

north-western India.102 The name of the locality is Ziyarat-i Malikan and refers to a pilgrimage thought to have been customary among Ghurid sovereigns; it is situated about halfway along the track running from Larwand to Parjuman. The building is fashioned in blocks of dark stone and square in plan (Figs 7.65, 7.66). Each side measures 2–5 m in length on the inside and the height of the walls is 3 m. The front is designed as an arched screen placed in front of an inner core,

from the rise of islam to the mongol invasion

Figure 7.64  Chisht, details of the brick decoration on the east dome chamber.

set back about 40 cm, and structurally and stylistically independent, although forming a unified whole, as is clear from the well-jointed blocks connecting the two parts. At the bottom of the screen there is a plinth, the lower section of which consists of a fillet (or foundation rebate?) and the upper of a small moulding made up of a course of hanging palmettes; the plinth is surmounted by a projecting fillet that ends abruptly at a point corresponding to the arch. From the plinth there rise four pilasters reproducing ‘columns’ with an octagonal shaft and capital consisting of four brackets with volutes. The pilaster bases are of two different types and are not arranged symmetrically, for the two on the left are ‘vase-shaped’ while the two on the right are octagonal, each of the faces being decorated with a characteristic elongated triangle motif with the apex uppermost. The four pilasters support the upper part of the screen, which is divided into three panels by four large ribs corresponding to the pilasters. The central panel is of larger size and filled by a slightly pointed horseshoe arch which in its lower, narrowing section is absolutely straight. The screen is surmounted by an S-shaped eave, the upper surface of which is ribbed. The doorway is flanked by two elaborate pilasters: low down, the shafts are square but with rebated edges, then they become octagonal and finally round at the top, where they are decorated with the Indian Kirtimukha symbol. The jambs and lintel are diversified by a wavy scroll. The panels above the door are decorated by, upwards: (1) a lozenge motif inside small rectangular panels alternating with dentils; (2) small trefoil arches alternating with small columns with large capitals;

529

530

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.65  Larwand, façade and entrance of mosque.

(3) a motif of intertwined arches; (4) a moulding with a series of hanging palmettes; and (5) a band of lozenges inside panels. In the narrow and dark interior are remains of a dome construction.103 At each of the four corners of the room, a kind of beam fashioned in stone blocks, equal in length to roughly one third of each side, transforms a square into an octagon, which was probably the base of the cupola. Taddei has convincingly interpreted the architecture and decoration as of northwest Indian origin.104 According to historical accounts of the Ghurid period we may surmise that among slaves, prisoners and various rarities dispatched from the Indian campaigns with the rich booty to Ghazni and the Ghurat were also craftsmen skilled in the art of building and sculpture according to the canons of Hindu art. During Ghurid times India was no longer simply a land for raiding, but had become a settled conquest. Temples were destroyed at the conquest, but their parts were used in the building of mosques and their decorations adapted to the requirements of the new faith. From Ghurid expeditions to north-west India, parts of Hindu or Jaina temples could have been transported to the centre of the Ghurat and re-used in buildings for the Islamic religion on a very small scale; or master builders and masons from India may have been brought to Afghanistan, converted to Islam and constructed tombs and mosques for the Muslim rulers in Indian style. In any case, the ruin of Larwand gives a vivid picture of east–west cultural interrelations under Ghurid rule.

from the rise of islam to the mongol invasion

Figure 7.66  Larwand, plan and elevation.

531

532

THE ARCHAEOLOGY OF AFGHANISTAN

Herat At the same time, the traditional links with Iran were maintained. Situated to the west of the Ghurat, Herat looks back to pre-Islamic history and can boast of possessing a masterpiece of Ghurid art nowadays in the midst of a shrine dominated by Timurid features, to which we shall return in Chapter 8. Here we deal with the Ghurid parts of the Friday Mosque (Figs 7.67, 7.68).105 Three parts of the mosque

Figure 7.67  Herat, Friday Mosque, the Ghurid portal.

from the rise of islam to the mongol invasion

Figure 7.68  Herat, Friday Mosque, detail of the Ghurid portal.

533

534

THE ARCHAEOLOGY OF AFGHANISTAN

are of pre-Timurid date. The first is the low vaulting leading right and left from the western ivan; it has retained a brick pattern of a distinctly twelfth century appearance. Further, on the left-hand side of the eastern façade a large portal was decorated in late Timurid style until 1964, when an earlier glazed brick decoration with a large Kufic inscription was uncovered (Fig. 7.67).106 Finally, an older part with an inscription was still to be seen about forty years ago at the top of the walls enclosing the so-called tomb of Ghiyath al-Din.107 Shah-i Mashhad To the north-east of the ancient capital of Herat we find the ruins of a madrasa and tomb in a lonely position and reflecting the best tradition of Ghurid court art. The ruins stand on the left bank of the Murghab, about 2 km downriver from its confluence with the Kucha. The inhabitants of the region call both the ruin and the river valley between the mouth of the Kucha and the ruin ‘Shah-i Mashhad’.108 The building is constructed of baked bricks 25 × 25 × 5 cm. The surfaces are simple brick walls, with more or less complicated brick work mosaics set into the plaster and plaster-and-stucco areas. The ground plan is almost square, 44.2 m north–south axis and 44 m east–west. The building is oriented at 269° (Fig. 7.69). Today, parts of the north and south tracts remain, with the eastern part of the south façade (Fig. 7.70), the main ivan of the entrance, the remnants of two originally domed rooms and a small fragment on the north-east. On the south side there is a typical nichefaçade with a rhythmical sequence of five pointed arches which are of almost equal height (except for the ivan), but of different widths. In the original dome-chambers are squinch constructions. The zone of transition in the larger room consists of a rhythmic series of corner squinches and niches – this special decorative system is used coevally in mud-brick ivan courtyard houses of Sistan.109 There are at Shah-i Mashad fifteen bands of inscriptions, of which ten are Kufic and five Nashki. The architecture of the niches, the decoration and the inscriptions, as, for example, the Kufic of the Sura 48, form a perfect artistic unity (Figs 7.71, 7.72). For the most part we find the usual Iranian keel-arch. In the north tract, however, a horseshoe arch was noticed, indicating a relationship with Ghaznavid art, especially in connection with the beginning of the cusped arch. The Nashki inscriptions have been compared with an inscription at the undated mosque at Peshwaran in Sistan (Fig. 7.42) and with the above-mentioned Ghurid inscription in the mausoleum of Ghiyath al-Din in the Friday Mosque of Herat. Structure, inscriptions and decoration point to the Ghurid period. The ornamentation is especially rich on the southern façade. The arrangement of the brick and terracotta right-angled ornaments is characteristic, with square fields often framed by special decorative friezes. The discoverers compared the architectural and ornamental style with structures of the twelfth century in west Khurasan, Central Asia, Ghazni and Sistan, and suggested a Ghurid foundation and a designation as Khurasanic/Late Seljuk. The presence of sgraffiato pottery on the site corroborates their stylistic dating.

from the rise of islam to the mongol invasion

Figure 7.69  Shah-i Mashad, ground plan.

535

536

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.70  Shah-i Mashad, south façade.

Danestama Further north, at Danestama, Ghaznavid slip-painted and Ghurid incised pottery was collected during the excavation of a ruin in the Surkhab valley of the Baghlan province by Le Berre.110 A stone basement carried a mud-brick structure that can be reconstructed as a rectangular building 40.4 m × 36.35 m, protected in each corner by a three-quarter-round bastion, and reinforced on the northern flank by one, and on the east and west by two, semi-circular bastions (Fig. 7.73). In the south a gate flanked by huge walls leads into a vestibule and an open court. The excavator found specimens of wall decoration in stucco of geometric designs. The function of the building is not yet absolutely clear; the many chambers, regularly distributed along the north, west and east walls might indicate a madrasa. Equally, the plan indicates a pre-Islamic Buddhist vihara. However, the twelfth century date seems secure; this in turn has led to speculation of the Buddhist vihara being the architectural inspiration behind the Islamic madrasa.111 Ghazni Finally, Ghurid rulers destroyed buildings in Ghazna and later constructed a residence of their own, which in turn was razed to the ground by the Mongol attack. This short period of reconstruction under the Ghurids can be recognised archaeologically in an extensive series of small glazed tiles,112 mostly square (Fig. 7.74),

from the rise of islam to the mongol invasion

Figure 7.71  Shah-i Mashad, detail of decoration.

but also polygonal, for the most part regular hexagons and in rectangular strips, and star-shaped. The dimensions of the square tiles range from 5 × 5 to 10 × 10 cm, the thickness from 0.7 to 1.2 cm. All the tiles have received a monochrome lead oxide glaze, of variable quality but constant for each type, green, yellow, brown, red and turquoise. The decoration is theriomorphic, vegetal or epigraphic. The great majority of the square tiles have a beaded frame enclosed by a double fillet. The gazelle is the animal most frequently represented, with a large tail turned up and ending in a floral device. The neck is drawn in sinuous lines. These tiles have been found mainly in the upper layers of the palace of Mas‘ud III, that is, from post-Ghaznavid layers, and from the ‘House of lustre-ware’ destroyed in 1221,

537

538

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.72  Shah-i Mashad, detail of decoration.

the year of the Mongol invasion. They are associated with sgraffiato pottery of the Ghurid period. We have here some comparatively rare documents of animal decoration on secular monuments, also known from animal scrolls and carved marble slabs with a border of paired sphinxes in the palace of Mas‘ud III.113 Thirty-three kilometres west of Ghazni, Fussman recorded a small polygonal tomb tower of baked brick at Tala Begum consisting of nine upright, rectangular baked brick fields decorated by chequer boards, crosses, bonds and serrated bricks ending in a beehive summit (Fig. 7.70). Relatively common in Seljuk Iran, this is the only tomb tower recorded in Afghanistan. Dated to the Ghurid period by Giovanni Verardi, it has some simple decoration and a slightly pointed dome.114

from the rise of islam to the mongol invasion

Figure 7.73  Danestama, ground plan of ruin.

539

540

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.74  Ghazni, moulded terracotta from the Ghurid phase of the palace of Mas‘ud III.

Figure 7.75  Tala Begum, plan.

from the rise of islam to the mongol invasion

Figure 7.76  Bust, so-called Ziyarat of Ghiyath al-Din.

Figure 7.77  Bust, Ziyarat of Ghiyath al-Din, interior.

541

542

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 7.78  Bust, Ziyarat of Ghiyath al-Din, inscribed tombstone.

Bust Traces of Seljuk culture under Ghurid rule are to be observed in many ruins round Bust, for example in the baked brick ziyarat of one Ghiyath al-Din or Husain Shah (Figs 7.76, 7.77).115 In this building, still venerated, are several marble tombstones

from the rise of islam to the mongol invasion

Figure 7.79  Bust, the arch.

Figure 7.80  Bust, the arch detail.

543

544

THE ARCHAEOLOGY OF AFGHANISTAN

with Kufic inscriptions. On one of them (Fig. 7.78) four lines on the edge of the rectangular slab praise one Najm al-milla wa’1-Din, the support of Islam, friend of the Sultan, Mufti of the east and the west. In the central part under the cusped arch the date of 595/1199 is given. The arch itself is filled by the beginning of a Qur‘anic text (XXIX, 57 or III, 182/185) and the field above this arch contains the Shahada, the Muslim creed. Bricks with inscriptions of Ghurid age were recently discovered during American explorations in the Helmand area.116 The arch near the citadel of Bust may have belonged to a Ghurid mosque or served as a ceremonial arch on the principal approach to the citadel.117 A photograph (Fig. 7.79) shows the technique by which the well-known horseshoe arches to the left and right of the huge keel-arch were constructed. The decoration of the main arch was comparable with that of the minaret of Daulatabad (Figs 7.80, 7.81) in the diaper work of carved terracotta tiles sunk in the hexagonal and star-shaped interstices of elaborate brick interlace. On top of the citadel are the remains of many structures, both in mud and baked brick, most notable of which is a sevenstorey galleried ‘well’ down through the middle.118 A palace or mansion was also constructed (or perhaps restored from a Ghaznavid mansion) at Bust in the Ghurid period (Fig. 7.82).

Figure 7.81  Bust, the arch detail.

from the rise of islam to the mongol invasion

Figure 7.82  Bust, façade of mud-brick palace.

545

CHAPTER 8

From the Mongols to the Mughals Warwick Ball and Klaus Fischer

Historical Background The Mongols reached Afghanistan in 1221. The Khwarazm-shah had abandoned his northern territories and fled into the mountains south of the Caspian, where he soon died. His son Jalal al-Din Mankobirti offered resistance to the invaders in the Ghazni and Hindu Kush region, but was defeated on the Indus and escaped

Figure 8.1  Map showing main sites discussed in Chapter 8. Sistan inset: Fig. 8.2; Herat inset: Fig. 8.39.

from the mongols to the mughals

across the river: the Mongol detachments sent in pursuit never succeeded in overtaking him, and after a stay in India of less than two years he passed on to Iraq by way of Baluchistan and southern Iran, eventually succumbing to assassination in 1231. Meanwhile Chingiz Khan’s armies sacked Ghazni, which had been Jalal al-Din’s headquarters. The conqueror himself soon returned to the east, dying in Mongolia in 1227. The slaughter in the cities of Khurasan was enormous. One local chronicler sets the numbers massacred at Herat at 1,600,000;1 and the province as a whole was subjected to a devastation from which it has never recovered. The region of western Afghanistan around Herat also suffered in the sixteenth and seventeenth centuries when it became a battleground between Safavid and Shaybanid forces.2 Underlying these ruthless tactics, in addition to the motive of discouraging further resistance, may have been a longer-term policy of converting well-­populated areas into pasturage. We know that the Mongols effected such measures in Central Asia;3 and the pasturelands of Badghis, north of Herat, and of Shibarghan, immediately south of the Amu Darya, were highly prized by the nomads, later becoming an object of competition between the rulers of Iran and Transoxiana. Nevertheless, under Chingiz Khan’s third son and successor Ögedei (1229–41) contrary influences made themselves felt, and Herat was restored around 1236. When fresh Mongol forces were sent to Iraq and Azerbaijan in 1229 to eliminate Jalal al-Din, Khurasan was allotted a civil and military administration of its own, though to the east Ghazni and Kabul became the sphere of Mongol detachments which were regarded as the reserve for the main army in Iraq and were directly responsible for intensifying the pressure on India. On the accession of Möngke (1251–9), further expeditions were launched against those regions of Asia which remained unconquered. In 1254 the Great Khan’s brother Hülegü arrived in Iran with an army which included contingents commanded by princes representing the various branches of the imperial family, especial prominence being given to the descendants of Chingiz Khan’s eldest son Jochi, whose territorial base lay in Khwarazm and the steppes of southern Russia. This army, which overthrew the Assassins in Kuhistan (1256) and sacked Baghdad (1258), extinguishing the Caliphate, never in fact operated in Afghanistan; but its importance for our purposes lies in the dissensions which broke out within its ranks soon after the Baghdad campaign. In 1261–2 Hülegü executed the Jochid princes and massacred their troops. Many of the survivors fled east and joined the Jochid contingent operating in Afghanistan under a certain Negüder. The Negüderis, as they came to be known, maintained their independence around Ghazni and in the Indian borderlands for some decades. Consequently, Afghanistan did not at this stage form part of the empire founded by Hülegü in Iran and Anatolia and ruled by his descendants, the ‘Ilkhans’, down to about 1350.4 Even the westernmost part lay under the influence of a native Iranian dynasty, the Kart rulers of Herat (c. 1250–1383), who claimed descent from the Ghurids, and who behaved as highly unreliable vassals of the Ilkhans. To the south, Sistan, still ruled by the line of

547

548

THE ARCHAEOLOGY OF AFGHANISTAN

princes who had been in power since 1002, proved still less amenable to the Ilkhans’ control. Around the turn of the century, the Negüderis were reduced to obedience by the Mongols of Transoxiana, ruled by the descendants of Chingiz Khan’s second son Chaghatai. Their position here was only temporarily disturbed by two invasions on the part of the Ilkhan’s forces, in 1312–13 and 1326, and when the Moroccan traveller Ibn Battuta passed through Ghazni in 1333 he found the region securely under the sway of the Chaghatai khan’s lieutenants. In 1352 even the Karts were obliged to become Chaghatayid vassals. But the authority of the Khan himself was already being appropriated by members of the nomad aristocracy; and in 1369 it passed to Timur, a member of the Turkish tribe of the Barlas, who proceeded to nominate a puppet Chaghatai sovereign but who remained ruler of the empire in everything but name. Timur embarked on a career of conquest that destroyed the various local dynasties who had supplanted the Ilkhans, crushed the Jochids in Russia, and even achieved what his Mongol predecessors had never accomplished, the sack of Delhi. In Afghanistan he captured Herat in 1380, executing the last Kart ruler a few years later: Kabul, Ghazni and Kandahar were all incorporated in Timur’s empire. Although after his death in 1404 they initially fell to a grandson, Pir Muhammad, they were soon annexed by his fourth son Shah Rukh (807/1404–850/1447), whose capital was Herat and who was recognised as the head of the dynasty. After the brief reigns of Shah Rukh’s son Ulugh Beg (850/1447–853/1449) and grandson Abd al-Latif (853/1449–854/1450), Khurasan was disputed among various scions of the Timurid line until Husain Baiqara (873/1469–912/1506) finally established his power in Herat and re-created the empire of Shah Rukh: during their reigns the city reached the zenith of its fame as a centre of culture. Kabul, however, now constituted a separate principality under another branch of the dynasty, while the western provinces of Iran fell away, to be ruled first by the Turkomans, and subsequently by the Safavid dynasty (c. 1500).5 The end of the fifteenth century witnessed further upheavals in Central Asia. Here a descendant of Jochi, Muhammad Shaibani or Shahi Beg, who had welded together the nomadic confederacy known as the Özbegs, conquered the Timurid principalities of Samarkand (1499) and Ferghana (1504) and began to press on Khurasan, finally wresting Herat from Husain Baiqara’s ineffectual sons in 1507. For a time the refugee prince of Ferghana, Babur, held out against the Özbegs in the Kabul region. But he had many enemies. The fastnesses of the Hindu Kush were held by former Timurid lieutenants, while to the south Babur had to maintain himself also in the face of attacks from the Arghuns. Dhu’l-Nun Arghun had been governor of Kandahar for Husain Baiqara, and had declared his independence. The Özbegs played off against Babur his son and successor; and even after Shaibani’s death in battle with Ismail, founder of the Safavid dynasty in Iran (1510), Babur was unable to make any headway. At first he was distracted into an attempt to recover his patrimony north of the Oxus; but when this failed, he again took up

from the mongols to the mughals

the struggle with the Arghuns, who were endeavouring to create a new dominion in Sind. Babur’s capture of Kandahar in 1522 was soon followed by the definitive occupation of Herat by the Safavids (1528), thus establishing the pattern which was to last for nearly two centuries and which was not materially affected by Babur’s own invasion of India and conquest of Delhi from its Afghan rulers, the Lodis, in 1525. His successors, the ‘Mughals’, retained their hold on Kabul from their new base in Hindustan, though Kandahar periodically passed to the Safavids, whose sway extended over western Afghanistan until the rise of the Durranis. Sistan During archaeological fieldwork in Sistan (Fig. 8.2) we have recognised in the mud-brick ruins of Sistan the persistence of Ghaznavid, Seljuk and Ghurid art forms. Monuments probably erected mainly under Ilkhanid rule contributed to the evolution of Timurid architecture.6 Reliable evidence from stratigraphic excavations with associated coins and pottery is still lacking and we have also only sparse information from inscriptions and coins.7 Survey in the vast ruin fields, however, helps us to understand the Arabic and Persian historians and geographers who describe rich settlements in the lower Helmand area. People living in semi-deserts and lands of moving sand dunes depend upon artificial irrigation: the Helmand and to a lesser degree the Khash Rud fed an extensive canal system. According to Islamic writers, such as al-Muqaddasi, Sistan contained few cities, but numerous rural estates, rustaq. In antiquity Sistan was known as a fertile land, and consequently was a constant temptation to conquerors. The inhabitants seem to have protected themselves by lines of fortresses and watch-towers (Fig. 8.3). After every destruction the irrigation systems and settlements were rebuilt. The native dynasty of the Saffarids ruled, independently or as governors and vassals, from 867 until c. 1495, under the Ghaznavids, Seljuks, Ghurids, Ilkhans and Timurids. In archaeological exploration, Sistan has even recently quite rightly been regarded as an almost total blank.8 Pottery of all these Islamic periods lies strewn together on the desert surface due to heavy wind erosion, caused especially by the so-called bad-o sad-o bist ruz or ‘wind of the hundred-and-twenty days’ blowing with great force from June to September (e.g. in the foreground: Figs 8.3, 8.6, 8.12);9 thus we find in the same ruin-fields side by side East Persian slip-painted wares, lustre-wares and sgraffiato wares, of the tenth to thirteenth centuries, as well as Iranian pottery of the Ilkhanid and Timurid periods from the fourteenth to the sixteenth century. From the Ilkhanid period onwards throughout the fourteenth century we find blue and white pottery, closely resembling Chinese porcelain of the fourteenth century but probably of independent Near Eastern origin. The Timurid period has well-known wares painted in blue, greenish-black and green under a clear glaze with floral and animal designs.10 Early slip-painted wares predominate in the Ghaznavid ruins of Lashkari Bazar and Bust, connected with later developments in Sistan, and the

549

550

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.2  Map of late Islamic sites in Sistan.

from the mongols to the mughals

Figure 8.3  Characteristic fortifications in Sistan.

same holds true of Ghurid sgraffiato pottery from the Bamiyan valley, Bust and other places. This early Islamic pottery is found in the vast ruin field of Khwaja Siah Push (Fig. 7.34) on the surface, while Ilkhanid blue and white and Timurid black and blue wares occur in ruins that are still standing up to the vaults, like Gul-i Safid (Fig. 7.38), and in the Timurid settlements of Herat and Balkh. Under the Ilkhanids typical features of Iranian baked-brick architecture were developed.11 In Sistan buildings were mainly constructed of mud-brick; they displayed a great variety of vaulting techniques, some perhaps of local origin. The vaults of the period can be classified into major categories: barrel vaults, vaults on a square or rectangular base, half-domes and stalactite vaults. Like the builders of the Seljuq period, the masons of the Ilkhanid period strove to erect their vaults with a minimum of centring and scaffolding. The key to the builders’ ability to construct vaults rapidly and with this minimum of technical means was the standard use of struts made of gypsum plaster stiffened with reeds. Liquid plaster and reeds were combined in a mould, the curvature of which reflected the profile of the planned vault. Several such plaster struts, placed in position upon the bearing vaults, divided the envelope of the proposed vault so that the masons had only to fill in the segments between the planks. An important factor is the consistent use of gypsum quick-setting plaster, which meant that if the bed of a row of bricks in one segment of a vault was inclined from the vertical the mason could hold the bricks in place until the mortar had set. In this manner the vault

551

552

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.4  Characteristic zone of transition used in Sistan.

was completed, segment by segment, without the need for support other than the hands of the mason. As a result of this method of construction, these Iranian vaults tended to be unitary constructions.12 The first of the major categories of vaults is the barrel vault (Fig. 8.8). This was extensively employed in the Ilkhanid period, either for simple, continuous vaults, or in more elaborate variants. One variant form may be described as the covering of a rectangular area by a series of cross arches, with each cross arch then joined to its neighbours by transverse filler vaults of modified barrel profile. Other problems lay in the construction of the cupola, that is, placing a dome upon a square plan. Ilkhanid master builders continued to employ devices common to the Seljuk period. Squinch arches, spanning the corners of the domed chamber, establish the size of the octagon upon which the dome rises. There are also zones of transition with squinched and shield-like pendentives (Fig. 8.4). During this period the double dome, which developed in Seljuk Iran, gains increasing popularity from this period into Timurid times, resulting from a union of symbolic, aesthetic and practical considerations. Islamic writers describe the rural settlements and cantons of Sistan, the socalled rustaq (plural: rasatiq), and we find numerous deserted villages of about 15–30 houses in close vicinity to abandoned ancient irrigation systems; from the latter were fed rectangular fields behind small earthen walls, a protection against wind erosion, still used by Afghan peasants. The houses of these settlements

from the mongols to the mughals

Figure 8.5  Plan of Diwal-i Khudaydad.

Figure 8.6  Diwal-i Khudaydad, courtyard house.

553

554

THE ARCHAEOLOGY OF AFGHANISTAN

were not oriented north–south or east–west, but with one corner towards the north protecting the interior of the habitation site with one continuous side to the north-west, towards the direction of the ‘wind of the hundred-and-twenty days’. The plan of the northern part of a ruin-field nowadays known as Diwal-i Khudaydad shows the arrangement of the houses of the landlords towards the north-west (Fig. 8.5). The house ruins represent the so-called ivan courtyard house (Fig. 8.11), details of which are discussed in Chapter 7. While these village-like settlements are numerous,13 we only rarely find larger agglomerations of about a hundred houses; this corresponds to Muqaddasi’s statement of the tenth century that in Sistan there were no, or few, cities. Although only excavations can give us details of the distribution and function of the buildings, in our present state of knowledge, Peshwaran with its mosque (Fig. 7.42), Khwaja Siah Posh with the minaret indicating a mosque (Fig. 7.34) and an aggregation of about one hundred ruined houses known as ‘Gul-i Safid’ may reasonably be called towns. In the Tarikh-i Sistan, the rasatiq of Sistan are enumerated. While the majority of the places cannot be located, we may put the rustaq of Taq at the ruins of Shahr-i Gholghola (or Sar-o Tar) in the south of Sistan and we may recognise in northern Sistan the rustaq of Nishk, represented by the vast ruin-field of this fortified post itself, or Gul-i Safid and Khwaja Siah Push. We shall now consider the archaeological remains of the city of Nishk (Figs 8.7–8.9), connected in the past with the river of Nishk, probably the present Khash Rud. Islamic writers name Nishk as a populous district east of Zaranj which gave its name to the eastern gate of the capital. Khash was described as the largest town of this district and was famous for its date palms. The whole district along the Khash River was known as Nishak, and modern maps record the name of Nishk or Nishak for the most conspicuous remains of city and fortress in the northern part of Afghan Sistan. Pottery collected on the surface belongs to a period from the eleventh to the thirteenth century. There is a multi-storeyed castle with four corner bastions (Fig. 8.7) dominating this part of Sistan.14 Here we illustrate for the first time details of the gate of the walled city (Fig. 8.9), flanked by doublestoreyed semi-circular towers, a remarkable piece of eastern Iranian military architecture. City walls are informative about the military character of medieval Iranian settlements: portals were always admirably defended, several imposing barbicans guarding narrow entrances with a secondary protection afforded by their close-coupled towers. The walls were further strengthened by passages in the interior. Despite their utilitarian character the walls were in part beautifully decorated with ornamental devices. In contrast to central Iranian cities, the ornamentation of the city wall of Nishk is reduced, as might be expected in this provincial area. In the Mongol wars many stone and brick buildings in Asia perished, while in the mud-brick ruins of Sistan, creations such as ivan halls or decorative motifs – such as cross and star patterns – survived. After the destruction of Sistan by Timur at the end of the fourteenth century, the Timurids from the fifteenth century

from the mongols to the mughals

Figure 8.7  Nishk, castle with corner bastions.

Figure 8.8  Nishk, interior of castle.

555

556

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.9  Nishk, fortified gate of city wall.

Figure 8.10  One of the ruin fields of Sistan with the University of Bonn mission in the background.

from the mongols to the mughals

Figure 8.11  Typical ivan house in Sistan, showing characteristic vaulting, squinches and niches.

Figure 8.12  The scouring effect of the ‘wind of 120 days’ on the ruins of Sistan.

557

558

THE ARCHAEOLOGY OF AFGHANISTAN

onwards revived building activities in new centres in western and northern parts of Afghanistan. After the death of Timur in 1405 the arts flourished for a century under the Timurids at Samarkand in Central Asia and at Herat in Afghanistan. Shah Rukh, Timur’s son and successor, who moved the capital to Herat and ruled until 1447, effected this change. His wife was Gauhar Shad, whose mausoleum at Herat is one of the masterpieces of Timurid architecture. Their sons were Ibrahim, Ulug Beg (famous for building the observatory at Samarkand) and Baysunghur, who was a great patron of miniature painting and calligraphy. A second period of Timurid renaissance began in 1469, when Husain Baiqara, a descendant of Timur’s son Umar Shaikh, assumed power in Khurasan and initiated a peace that lasted until his death in 1506. It was the Herat of this period which became the centre of artistic life, which Babur saw and which this founder of the Mughal dynasty described afterwards as the scene of a golden age and the source of civilisation.15 Herat Herat possesses monuments of outstanding beauty in which we can study the imperial style of the Timurids, but monuments of equal sophistication also existed in the provinces of Afghanistan (Fig. 8.16).16 City plan and citadel During the Timurid period, the population increased and new problems arose. While for tenth century Zaranj we have mainly to rely upon sources of Arab and Persian geographers and historians, the present city of Herat gives us an idea of a town of the fifteenth century. Herat was a perfect example of the concentric four quarters plan, with radial thoroughfares cutting the enclosed city into quadrants. In the tenth century the standard arrangement already prevailed: a citadel with four gates set on the main axes, which corresponded with the four gates in the city wall. This had 149 towers and was surrounded by a double moat. Inside each gate was a market. By the fifteenth century, although the city had in the interval been rebuilt by the Karts, it had essentially the same outlines (Fig. 8.13). The city wall, already in disrepair by the end of the fifteenth century, was still guarded by towers and encircled by moats. The fortresses and city walls of Herat have been frequently illustrated in surveys of the twentieth century, shown still in military use17 or in a ruined state.18 The walls and gates were still in a reasonable condition, as can be seen in drawings and photos taken in 1885 by members of the Afghan Boundary Commission (Fig. 8.16). The old city – and indeed the Herat plain – is dominated by the immense fortresspalace of Qal’a-i IkhtiyaruddÈn, a mainly fifteenth century citadel of baked brick standing on an artificial mound comprising archaeological strata going back to the Iron Age (see Chapter 4; Figs 8.15, 8.17, 8.18). It is divided into upper and lower

from the mongols to the mughals

Figure 8.13  Herat, map showing main monuments.

enclosures, with the treasury and royal residence in the upper part and barracks and stables in the lower (Fig. 8.17). The main tower on the north-western side is decorated with glazed tiles (8.18). The circuit is protected by a glacis paved with sandstone slabs. The glacis and the towers along the northern side have been dated to a restoration by Shah Rukh in 1415. An extensive international programme of excavation, conservation and restoration in recent years has now documented this important building in detail, and it is now a museum.19 The ‘musalla’ complex This complex has caused considerable confusion ever since the nineteenth century, not least because it was demolished in 1885 in order to provide a clear line of vision to defend the walled city against an expected Russian attack.20 To begin with, there is no evidence that it was a ‘musalla’: this misnomer emerged in the nineteenth century and has been used since to describe two quite distinct but adjacent buildings: the mosque and madrasa of Gauhar Shad. The later and quite separate madrasa of Sultan Husain nearby is also often confused under this term.

559

560

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.14  Herat, distant view with the two minarets and dome of the Gawharshad complex and the four minarets of the Husain-i Baiqara complex clearly visible.

Figure 8.15  Herat, distant view of the city in 1974: in the left foreground is the Shrine of Shahzadeh Abulqasim; in the distance is the Friday Mosque; on the right is Qal’a-i Ikhtiyaruddin.

from the mongols to the mughals

Figure 8.16  Herat, the city walls photographed in 1885 by Lt Griesbach of the Afghan Boundary Commission.

Figure 8.17  Herat, plan of Qal’a-i Ikhtiyaruddin.

561

562

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.18  Herat, tile decoration on the northern tower of Qal’a-i Ikhtiyaruddin.

from the mongols to the mughals

A great building project was initiated by Queen Gauhar Shad, wife of Shah Rukh, in about 1417 to build a madrasa, or place of learning, and a mosque, or place of worship as parts of a single complex. She commissioned one of the foremost architects of the day, Qavam al-Din (who also built Gauhar Shad’s mosque in Mashhad). The complex was completed in 1432. Although only fragments survive, the two buildings can be reconstructed with a reasonable degree of certainty from contemporary descriptions and sketches as well as from analogies with surviving Timurid buildings elsewhere (Figs 8.19–8.21). The mosque, of which nothing now

Figure 8.19  Herat, reconstructed plan of the Gawhar Shad complex.

563

564

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.20  Herat, sketch of the Gawhar Shad mosque by Durand.

Figure 8.21  Herat, sketch of the Gawhar Shad madrasa by Durand.

survives, was the southern of the two buildings. Four minarets stood at each corner, of which two still stood in the early twentieth century (Figs 8.23, 8.24). The entrance was formed by a high portal, leading into a courtyard with ivans in the centres of all four sides. That at the western end, facing the entrance, was the largest, flanked by two smaller ivans, leading to the central dome chamber of the prayer hall (Fig. 8.20).

from the mongols to the mughals

Slightly more survives of the madrasa. One entered through a lofty portal flanked by graceful two-balconied minarets. The minaret which stands to the east of the mausoleum (Figs 8.22) was one of a pair which stood on either side of the portal. The shaft of the minaret belongs to the simple circular variety mentioned earlier. Two balconies, the muezzin’s platforms, ring the shaft and each one is heavily ornamented with deep stalactite brackets similar to, but more ornate than, the corbels on the mausoleum dome. The square courtyard has four ivans on each side. The main open ivan faced the entrance, which was flanked on either side by two dome chambers; that on the right, still extant, was built for Gauhar Shad’s mausoleum. The interior of the mausoleum is based on a square of 9.5 m. Four recesses are roofed by four great arches, from the haunches of which spring concealed squinches (Figs 8.26–8.28). Inside, four big interlacing arches form elongated squinches. Above the four intersecting points of the interlacing arches, four shallow, fanshaped squinches support an octagon above which a zone of sixteen small stalactite niches leads to the ceiling dome. The whole system is enriched with stalactites, in full relief or flattened, with fan-vaulting in the squinches and with a general multiplication of the architectural pattern by polychrome painting. The unique, and at the same time typical, Timurid feature of the building is that, as well as a ceiling dome, it possesses an intermediate constructional dome which is visible from neither outside nor inside. The exterior decoration of the mausoleum is again of special beauty (Fig. 8.28). The ribbed dome was popular with the Timurids and this one is very similar to the Gur-i Amir mausoleum at Samarkand. Above the base a tall drum ending in a bulbous fluted dome rises high above the ground. The dome is of Persian blue; the flutes are patterned with royal blue lozenges centred with red, yellow or white. These flutes are supported by stalactites. Imposing minarets rose on the corners of the building project. Later, in 1492–3, Sultan Husain Baiqara built a madrasa to the east on the other side of the canal. Here there were lofty gateways, arcades, domes and, again, majestic minarets. Only the minarets and mausoleum remain, with corbels encircling the drum (Figs 8.29, 8.30). Unlike the Gawhar Shad monuments, there are few surviving descriptions on which to base a reconstruction, although a standard madrasa design can be assumed. The tile work on the minarets is particularly fine, with each panel framed by marble, although most has been lost due to wind erosion.21 The Friday Mosque We have already referred to the Ghurid parts of this building (Chapter 7: Figs 7.67, 7.68) and mention below the large bronze cauldron preserved in the courtyard. Here we sketch the history of this famous mosque, which reached its greatest extension during the Timurid age (Figs 8.31, 8.32). In the tenth century, Herat’s great mosque was already an important centre of Islamic thought. This building was destroyed by fire and reconstructed on a

565

566

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.22  Herat, surviving minaret of the Gawhar Shad madrasa.

from the mongols to the mughals

Figure 8.23  Herat, surviving minaret of the Gawhar Shad mosque.

567

568

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.24  Herat, detail of the tile-work inscription on the minaret of the Gawhar Shad mosque.

magnificent scale by Sultan Ghiyath al-Din of the Ghurid dynasty in the year 1200. Mongol raids destroyed much of the Ghurid mosque, which was again repaired and extended by the Kart kings, whose devoted attention was continued by the Timurids. The period of greatest magnificence came during the reign of the Timurid Sultan Husain Baiqara in the fifteenth century. Mir Ali Sher Nawai, minister, poet and patron, personally directed its redecoration in 1498. Badly damaged during the sixteenth century, it was later repaired by successive rulers. A corridor leads to the main courtyard, which is paved with brick. Directly ahead, on the west, is the principal ivan of the mosque, its arched entrance flanked by minarets. On either side of this hall there are smaller ivans, their walled entrances pierced with doors and windows. The inner face of the courtyard is covered with a profusion of delicate floral motifs. Arabic and Persian calligraphy gracefully interrupts these designs. Scattered here and there throughout the mosque one may see fragments of tile work from the fifteenth century. The tomb of the Ghurid king accredited with the building of this Friday Mosque lay under a dome situated behind the north ivan (demolished in the 1950s).22 Mosque of Hauz-i Karboz In the suburbs of Herat, the mosque of Hauz-i Karboz lies hidden behind a high mud-brick wall near a streamlet with a reservoir, from which the mosque has received its local name. This sanctuary constitutes a rare example of the ‘Guzar

from the mongols to the mughals

Figure 8.25  Herat, plan of the Gawhar Shad mausoleum.

Mosque’ (i.e. a mosque of that particular city quarter). In Timurid Central Asia this type of building was widespread, but is now rare. They are of small dimensions, but exhibit all the splendour of their age. Fortunately, the mihrab of this mosque has been preserved (Fig. 8.33). Its mosaics contain Qur‘anic inscriptions, while another wall decoration carries the date of construction, 845/1441–2, in the reign

569

570

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.26  Herat, interior decoration of the Gawhar Shad mausoleum.

from the mongols to the mughals

Figure 8.27  Herat, interior dome of the Gawhar Shad mausoleum.

Figure 8.28  Herat, exterior dome of the Gawhar Shad mausoleum.

571

572

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.29  Herat, the minarets of the Sultan Husain madrasa.

of Shah Rûkh. The exquisite Timurid calligraphy is among the best examples of that style.23 Mausoleum of Shahzadeh Abdullah On the northern outskirts of the city, opposite the modern garden with Gauhar Shad’s tomb, we find the mausoleum of the Shahzadeh Abdullah, a highly honoured man who died in 134/751–2). The present architecture belongs to the fifteenth century. The main room of the building complex is an octagonal structure (Fig. 8.34) covered by eight interlaced arches forming pendentives that carry a dome on a twelve-sided drum. This building preserves typical eastern Iranian features. The octagonal ground plan is of a type of which the history has recently been established by excavation of an Ilkhanid building at Ghubayra in Iran:24 the interior of the building has a complex pendentive structure and a high dome.25 Gazurgah The most celebrated shrine in Herat is situated to the east of the city and is known as Gazurgah. It is built around the tomb of the famous eleventh century Sufi poet and philosopher Khwaja Abdullah-i Ansari, born in Herat in 1006. A long inscription explains that the building was restored by Shah Rukh in 1428. In the interior

from the mongols to the mughals

Figure 8.30  Herat, tile detail on the Sultan Husain madrasa.

573

574

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.31  Herat, plan of the Friday Mosque.

Figure 8.32  Herat, Friday Mosque.

from the mongols to the mughals

Figure 8.33  Herat, mosque of Hauz-i Karboz, mihrab.

575

576

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.34  Herat, plan and section of the mausoleum of Shahzadeh Abdullah.

Figure 8.35  Herat, plan of the Gazurgah shrine.

Figure 8.36  Herat, shrine complex of Gazurgah.

578

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.37  Herat, tile decoration on the façade of Gazurgah.

we face a tall arched ivan in a high wall crowned with an arcade of five arches topped by two cupolas. Tall ivans in high walls, which emphasised their towering effect over their surroundings, were extremely popular with the Timurids. Fortunately, remains of the splendid interior decoration have been preserved (Figs 8.35–8.38). The ivan at Gazurgah offers one of the most varied samples of Timurid decoration. It is composed of a series of squares and rectangles. The entire surface is covered with inscriptions repeating tributes to the greatness of God. The whole complex has been interpreted as a ‘Hazira-Compound’. In its primary sense this is an enclosure, most commonly in stone or wood, and very early used for the prophet’s tomb in the Umayyad mosque at Medina. It was another solution for tombs instead of the domed variety mainly used in Iranian countries.26 Herat Region With the move of the Timurid capital from Samarkand to Herat by Shah Rukh, the entire region experienced an upsurge in building activity. Some of the finest buildings are now across the border in Iran, so are not included in this survey, but mention must be made of the Mosque of Gauhar Shad in Mashhad, the Shrine of Torbat-i Jam, the Shrine of Taybad and the magnificent Madrasa at Khargird.27 Large numbers of buildings of the Timurid can also be found in the Herat region, of which we discuss just the more significant ones here (Fig. 8.39).

from the mongols to the mughals

Figure 8.38  Herat, detail of tile decoration inside the main ivan at Gazurgah.

579

580

THE ARCHAEOLOGY OF AFGHANISTAN

Figure 8.39  Herat Province, map showing the number of sites, most of which are Timurid, in the Hari Rud catchment.

Azadan The Shrine of ‘Abd al-Walid is a complex consisting of a tomb, a mosque, an inn and a cistern. The tomb is an original Kart structure with extensive modification under Amir ‘Ali Shir Nawai and w