Muslim Fortresses in the Levant: Between Crusaders and Mongols 2010007402, 0203845161, 9780415569255, 9780203845165

Citation preview

Muslim Fortresses in the Levant

During much of the twelfth century the Crusaders dominated the military scene in the Levant. The unification of Egypt and Syria by Saladin gradually changed the balance of power, which slowly begun to tilt in favor of the Muslims. This book examines the development and role of Muslim fortresses in the Levant at the time of the Crusaders and the Mongol invasion, situating the study within a broad historical, political and military context. Exploring the unification of Egypt with a large part of Syria and its effect on the balance of power in the region, Raphael gives a historical overview of the resulting military strategies and construction of fortresses. A detailed architectural analysis is based on a survey of four Ayyubid and eight Mamluk fortresses situated in what are today the modern states of Jordan, Israel, Southern Turkey and Egypt (the Sinai Peninsula). The author then explores the connection between military architecture and the development of siege warfare and examines the influence of methods of rule on the concept of defence and the development of fortifications. Drawing upon excavation reports, field surveys and contemporary Arabic sources, the book provides the Arabic architectural terminology and touches on the difficulties of reading the sources. Detailed maps of the fortresses in the region, the Mongol invasion routs, plans of sites and photographs assist the reader throughout the book, providing an important addition to existing literature in the areas of Medieval Archaeology, Medieval military history and Middle Eastern studies. Kate Raphael is a postdoctoral researcher in the Department of Earth and Sciences at the Hebrew University of Jerusalem, Israel. She received her PhD in Medieval Islamic studies at the Hebrew University of Jerusalem, and her research currently focuses on environmental disasters in the Medieval Middle East.

Culture and Civilization in the Middle East General Editor: Ian Richard Netton Professor of Islamic Studies, University of Exeter

This series studies the Middle East through the twin foci of its diverse cultures and civilisations. Comprising original monographs as well as scholarly surveys, it covers topics in the fields of Middle Eastern literature, archaeology, law, history, philosophy, science, folklore, art, architecture and language. While there is a plurality of views, the series presents serious scholarship in a lucid and stimulating fashion. Previously published by Curzon The Origins of Islamic Law The Qur’an, the Muwatta’ and Madinan Amal Yasin Dutton A Jewish Archive from Old Cairo The history of Cambridge University’s Genizah Collection Stefan Reif The Formative Period of Twelver Shi’ism Hadith as discourse between Qum and Baghdad Andrew J. Newman Qur’an Translation Discourse, texture and exegesis Hussein Abdul-Raof Christians in Al-Andalus 711-1000 Ann Rosemary Christys

Structure and Meaning in Medieval Arabic and Persian Lyric Poetry Orient pearls Julie Scott Meisami Muslims and Christians in Norman Sicily Arabic-speakers and the end of Islam Alexander Metcalfe Modern Arab Historiography Historical discourse and the nation-state Youssef Choueiri The Philosophical Poetics of Alfarabi, Avicenna and Averroes The Aristotelian reception Salim Kemal Published by Routledge The Epistemology of Ibn Khaldun Zaid Ahmad

Folklore and Folklife in the United Arab Emirates Sayyid Hamid Hurriez

The Hanbali School of Law and Ibn Taymiyyah Conflict or concilation Abdul Hakim I Al-Matroudi

The Formation of Hanbalism Piety into power Nimrod Hurvitz

Arabic Rhetoric A pragmatic analysis Hussein Abdul-Raof

Arabic Literature An overview Pierre Cachia

Arab Representations of the Occident East–West encounters in Arabic fiction Rasheed El-Enany

God and Humans in Islamic Thought Abd al-Jabbār, Ibn Sīnā and al-Ghazālī Maha Elkaisy-Friemuth

Early Islamic Spain The history of Ibn al-Qū.tīya David James

Original Islam Malik and the madhhab of Madina Yasin Dutton

German Orientalism The study of the Middle East and Islam from 1800 to 1945 Ursula Wokoeck

Al-Ghazali and the Qur’an One book, many meanings Martin Whittingham Birth of The Prophet Muhammad Devotional piety in Sunni Islam Marion Holmes Katz Space and Muslim Urban Life At the limits of the labyrinth of Fez Simon O’Meara Islam Science The intellectual career of Nizam al-Din al-Nizaburi Robert G. Morrison Ibn ‘Arabî – Time and Cosmology Mohamed Haj Yousef The Status of Women in Islamic Law and Society Annotated translation of al-Ṭāhir al-Ḥaddād’s Imra’tunā fi ‘l-sharīca wa ’l-mujtamac, with an introduction Ronak Husni and Daniel L. Newman Islam and the Baha’i Faith A comparative study of Muhammad ‘Abduh and ‘Abdul-Baha ‘Abbas Oliver Scharbrodt Comte de Gobineau and Orientalism Selected Eastern writings Translated by Daniel O’Donoghue Edited by Geoffrey Nash

Mullā Ṣadrā and Metaphysics Modulation of being Sajjad H. Rizvi Schools of Qur’anic Exegesis Genesis and development Hussein Abdul-Raof Al-Ghazali, Averroes and the Interpretation of the Qur’an Common sense and philosophy in Islam Avital Wohlman, translated by David Burrell Eastern Christianity in the Modern Middle East Edited by Anthony O’Mahony and Emma Loosley Islamic Reform and Arab Nationalism Expanding the Crescent from the Mediterranean to the Indian Ocean (1880s–1930s) Amal N. Ghazal Islamic Ethics Divine command theory in AraboIslamic thought Mariam al-Attar Muslim Fortresses in the Levant Between Crusaders and Mongols Kate Raphael

Muslim Fortresses in the Levant Between Crusaders and Mongols Kate Raphael

First published 2011 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Simultaneously published in the USA and Canada by Routledge 270 Madison Ave, New York, NY 10016 Routledge is an imprint of the Taylor & Francis Group, an informa business This edition published in the Taylor & Francis e-Library, 2010. To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk. © 2011 Kate Raphael All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data Raphael, Kate. Muslim fortresses in the Levant : between crusaders and Mongols / Kate Raphael. p. cm. – (Culture and civilization in the Middle East ; 23) Includes bibliographical references and index. 1. Fortification–Egypt–History–To 1500. 2. Military architecture–Egypt–History–To 1500. 3. Siege warfare–Egypt–History–To 1500. 4. Egypt–History, Military–To 1500. 5. Architecture, Ayyubid–Egypt. 6. Architecture, Mameluke–Egypt. I. Title. UG435.E3R36 2011 725'.18–dc22 2010007402 ISBN 0-203-84516-1 Master e-book ISBN

ISBN: 978-0-415-56925-5 (hbk) ISBN: 978-0-203-84516-5 (ebk)

Contents



List of figures List of maps List of tables Acknowledgements Notes on transliteration Abbreviations



Introduction

1 Ayyubid fortresses in the late twelfth and early thirteenth centuries

viii xiii xiv xv xvii xviii 1 4

2 Mongolian siege warfare and the defense of Mamluk fortresses

52

3 Laying the foundations: fortification work conducted during the reign of Baybars (1260–77), Qalāwūn (1280–90) and alAshraf Khalīl (1290–93)

81

4 Mamluk power on display: fortresses in the second half of the thirteenth century

125

5 Military architecture versus political and military organization

206



215 248 262

Notes Bibliography  Index

Figures

Unless otherwise noted photographs are by the author 1.1 Ayyubid fortresses from the late twelfth to the early thirteenth centuries drawn on the same scale (Mrs Miri Shmida, cartography laboratory, The Hebrew University of Jerusalem). Plan of Mount Tabor adapted from A. Battista and B. Bagatti, La Fortezza Saracena Del Monte Tabor (ah 609–15; ad 1212–18) Jerusalem, 1976, Tav. 33. Plan of ‘Ajlūn adapted from C. N. Johns, “Medieval ‘Ajlun,” QDAP, vol. 1, 1931. pl. 21. Plan of al-Ṣubayba adapted from Deschamps, Les Chateaux des Croisés et Terre Sainte II: La defense du royaume de Jérusalem, Paris, 1939, fig. 1. Plan of Qal’at Sadr adapted from J. Barthoux, “Description d’une forteresse de Saladin découverte au Sinai,” Syria 3, 1922: 49 1.2 Mount Tabor, plan and aerial photograph (1945). Plan adapted from Battista and Bagatti, Monte Tabor; Tav. 33). Aerial photograph, courtesy of the archives of the Geography department at the Hebrew University of Jerusalem 1.3 ʿAjlūn, masonry in towers built during the first phase (1184). Curtain wall between towers 1 and 2. Plan adapted from Johns, “ʿAjlun,” pl. 21 1.4 ʿAjlūn, curtain wall between towers 7 and 8 belonging to the second phase (1184–1214). Plan adapted from Johns, “ʿAjlun,” pl. 21 1.5 An angle along the northern wall at al-Ṣubayba strengthened by masonry of higher quality 1.6 Al-Ṣubayba, the glacis below the southeastern gate 1.7 Main gates in Ayyubid fortresses. Adapted from Battista and Bagatti, Mount Tabor, Tav. 33; Deschamps, Chateaux, fig. 1; Barthoux, “Forteresse de Saladin,” 49 1.8 Al-Ṣubayba, reconstruction of the Ayyubid gate-tower. Adapted from M. Hartal, The Al-Ṣubayba (Nimrod) Fortress, Towers 11 and 9, Jerusalem, 2001, fig. 7. Courtesy of the Israel Antiquities Authority 1.9 Postern at the northern wall of al-Ṣubayba

15

20 30 30 31 33 36 37 38

Figures   ix 1.10 Suggested stages in the development of the gate at ʿAjlūn. Adapted from Johns, “ʿAjlūn,” pl. 21 1.11 ʿAjlūn, machicolations above the main entrance belonging to the last building phase (1214–15) 1.12 Masonry along Ayyubid curtain walls and the Crusader curtain wall at Vadum Iacob  1.13 Core and stone facing of the curtain wall at Vadum Iacob 1.14 Chart showing the increase in width of Crusader curtain walls 1.15 Towers at Qalʿat Ṣadr. Adapted from Barthoux, “Forteresse de Saladin,” 49 1.16 ʿAjlūn, tower 13 – the solid wall. Adapted from Johns, “ʿAjlun,” pl. 21 1.17 ʿAjlūn, tower 13 looking from the northeast and the plan of its three floors. Plan adapted from C. Yovitchitch, “The Tower of Aybak in ‘Ajlun Castle: an example of the spread of an architectural concept in early thirteenth centaury Ayyubid fortifications,” in Military Architecture in Greater Syria, ed. H. Kennedy, Leiden and Boston, 2006, fig. 6. Courtesy of Dr Cyril Yovitchitch 1.18 ʿAjlūn, reconstruction of a funnel-shaped arrow slit. Illustration by S. Rotem 1.19 The age of spacious arrow slits. Illustration by S. Rotem 1.20 Al-Ṣubayba, solid towers belonging to the first Ayyubid phase (1228) 1.21 Al-Ṣubayba, archers’ galleries along the southwestern corner. General plan adapted from Deschamps, Chateaux, fig. 1. Plan of tower 9 after Hartal, Al-Ṣubayba, fig. 3 1.22 Mount Tabor, the tower at the southeast corner 2.1 Chinese siege machines in the twelfth to thirteenth centuries. Adapted from Weapons in Ancient China, ed. Y. Hong, New York and Beijing, 1992, figs. 361, 362 2.2 The counterweight trebuchet, and catapult stones from the siege of Acre, 1292. Jami‘u’t-Tawarikh, Tabriz, Iran 1306–14, in Edinburgh University library, Scotland. 20. Ms. in D. Nicolle, Arms and Armour of the Crusading Era 1050–1350: Islam, Eastern Europe and Asia, 2nd edn, London and Mechanicsburg, PA, 1999, fig. 626m  2.3 Earthen walls in northeast China and inner Mongolia (tenth to thirteenth centuries). Aerial photograph of Ning Cheng courtesy of Ling Hough 2.4 The giant crossbow. Adapted from Weapons in Ancient China, ed. Y. Hong, New York and Beijing, 1992, fig. 363 4.1 Mamluk fortresses from the thirteenth and early fourteenth centuries drawn on the same scale. Mrs Miri Shmida, cartography laboratory, The Hebrew University of Jerusalem 4.2 Al-Ṣubayba, the monumental inscription from the reign of Baybars. M. Hartal, The Al-Ṣubayba (Nimrod) Fortress, Towers 11

38 39 40 41 42 44 45

45 46 46 47 48 48 56

57 59 62 126

x  Figures and 9, Jerusalem, 2001, fig. 187. Courtesy of the photographic archives, Israel Antiquities Authority 4.3 Al-Ṣubayba, Mamluk repairs along the Ayyubid walls 4.4 Al-Ṣubayba, the distribution of towers during the last Ayyubid phase and later after the Mamluk reconstruction  4.5 Al-Ṣubayba, seam between the Mamluk tower and the Ayyubid wall  4.6 Al-Ṣubayba. Note the quality of masonry along the external walls of Ayyubid and Mamluk towers 4.7 Al-Ṣubayba, smooth ashlar blocks inside the new Mamluk tower 7 4.8 Al-Ṣubayba, inside view of tower 5 4.9 Al-Ṣubayba, Mamluk arrow slits with geometric design on the external frame (tower 7) 4.10 Al-Ṣubayba, Ayyubid archers’ gallery along the south  4.11 Al-Ṣubayba reconstruction of tower 11 (Bilik’s tower). M. Hartal, The Al-Ṣubayba (Nimrod) Fortress, Towers 11 and 9, Jerusalem, 2001, fig. 35. Courtesy of the Israel Antiquities Authority 4.12 Al-Ṣubayba, secret postern in tower 16 leading to the core of the Ayyubid “keep” and the grand Mamluk hall. Plan adapted from Deschamps, Chateaux, fig. 1 4.13 Al-Ṣubayba, giant stones along the northern face of tower 11 4.14 The fortress of Safad, the round tower and inner gate. Plan adapted from C. R. Conder and H. H. Kitchener, Survey of Western Palestine: Memories of the Topography, Orography, Hydrography and Archaeology, London, 1881, vol. 1, plan opposite p. 249 4.15 Safad, the remains of the Mamluk inner gate on the southwest 4.16 Safad, plan of round central tower and Mamluk inner gate. Adapted from H. Barabé and E. Damati, “Le Chāteau de Safed sources historiques, problématique et premiers resultants des recherches archéologique,” in Archéologie et culture, la fortification au temps des Croisades, eds. N. Faucherre et al., Rennes, 2004, fig. 2. Courtesy of Mr Barabé, the Israel Antiquities Authority) 4.17 Safad, reconstruction of the central tower. Note the cistern below. Illustration by S. Rotem 4.18 Safad, reconstruction to the entrance to the central tower. Note the latrines and fine masonry. Illustration by S. Rotem 4.19 Safad, central round tower. Note the construction and the dimension of the stone blocks 4.20 Qāqūn, the tower (left) and the village from the southeast (1918). Aerial photograph, Royal Air Force, UK 4.21 Qāqūn, the tower (2004). Aerial photograph, courtesy of David Silverman 4.22 Qāqūn, plan of first floor and suggested reconstruction of the tower. D. Pringle, The Red Tower (al Burj al Aḥmar) Settlement in the Plain of Sharon at the Time of the Crusaders and Mamluks a.d. 1099–1516, London, 1986, 66. Courtesy of Prof. Pringle Cardiff School of

136 139 140 140 141 141 142 143 143 144 144 145

148 151

151 152 153 154 157 158

Figures   xi History and Archaeology, Cardiff University 4.23 Qāqūn, masonry and construction. Note the columns inserted in the southwest corner 4.24 Karak. Plan adapted from W. Müller-Wiener, Castles of the Crusaders, trans. J. Brownjohn, London, 1966, 4 4.25 Karak, the Frankish entrance along the northeast corner  4.26 Karak, Mamluk western galleries with their barrel-shaped vaults 4.27 Karak, western entrance into the Mamluk galleries and the oculus above the inner staircase  4.28 Karak, the Mamluk keep. Note the remains of the Frankish towers destroyed during the Keep’s construction  4.29 Karak, Baybars’ panther adorning the inscription on the southern keep 4.30 Karak, Mamluk and Frankish masonry 4.31 Karak, the western wall of the fourteenth-century Mamluk palace 4.32 Al-Bīra, the fortress in the early twentieth century and 2004 Courtesy of the Ashmolean Museum of Art and Archaeology, The Creswell Archive 4.33 Al-Bīra, aerial photo (U.S. Air Force, 1968) and schematic plan of the fortress. Plan adapted from A. Cihat Kürkcüoğlu, Birecik Monografisi, Ankara, 1996. Aerial photo, Corona Images DS1105– 1009DA010 1968/11/04 4.34 Al-Bīra, the town curtain walls and the octagonal tower in the southern corner. Courtesy of the Ashmolean Museum of Art and Archaeology, The Creswell Archive 4.35 Al-Bīra, cliffs adorned with curtain walls. Note the exit to the river bank (1, 2). Courtesy of the Ashmolean Museum of Art and Archaeology, The Creswell Archive 4.36 Al-Bīra, pillars inserted in the lower half of the southeast tower. Courtesy of the Ashmolean Museum of Art and Archaeology, The Creswell Archive 4.37 Al-Bīra, arrow slits along the southern glacis. Courtesy of the Ashmolean Museum of Art and Archaeology, The Creswell Archive 4.38 Al-Bīra, the large tower on the southwest (F). Note (1) the entrance; (2) the inscription on the ceiling, covered in soot; and (3) the main hall. Courtesy of the Ashmolean Museum of Art and Archaeology, The Creswell Archive 4.39 Al-Bīra, the large built entrance in the east, possibly the main gate. Courtesy of the Ashmolean Museum of Art and Archaeology, The Creswell Archive 4.40 Al-Bīra, rock carved passage in the west  4.41 Al-Bīra, vaulted halls along the east and northeast. Courtesy of the Ashmolean Museum of Art and Archaeology, The Creswell Archive 4.42 Qalʿat al-Rūm. Plan adapted from T. E. Lawrence, Oriental Assembly, ed. A. W. Lawrence, London, 1939, pl. 15

158 159 164 165 168 169 170 171 171 172 175

176 177 178 180 180

181 182 183 184 185

xii  Figures 4.43 Qalʿat al-Rūm, before and after the flooding of the valley 4.44 Qalʿat al-Rūm, southern moat  4.45 Qalʿat al-Rūm, the zigzag wall on the east  4.46 Qalʿat al-Rūm, archers’ galleries and vaulted hall. Note the dense line of machicoulis 4.47 Qalʿat al-Rūm, Armenian and Mamluk arrow slits 4.48 Qalʿat al-Rūm, the eastern gate  4.49 Qalʿat al-Rūm, carved and built entrances along the multiple gate system  4.50 Tall Ḥamdūn, Mamluk tower and glacis along the western inner wall 4.51 Tall Ḥamdūn, view from the southwest  4.52 Tall Ḥamdūn, prow-shaped tower along the lower curtain wall and site plan. Note the Mamluk galleries (1) and the new Mamluk tower along the Armenian walls (n). Illustration of tower adapted from C. Foss and D. Winefield, Byzantine Fortifications: An Introduction, Pretoria, 1986, 30. Plan adapted from MüllerWiener, Castles, 24 4.53 Tall Ḥamdūn, Mamluk and Armenian arrow slits  4.54 Baghrās, core of the Templar fortress. Plan adapted from MüllerWiener, Castles, 24 4.55 Baghrās, lower and upper defences along the eastern side. Note the round tower in the front. W. H. Bartlett, 1838, in A. W. Lawrence, “The Castle of Baghras,” in The Cilician Kingdom of Armenia, ed. T. S. R. Boase, Edinburgh and London, 1978, pl. 8 4.56 Baghrās, the Mamluk tower (C). Note the masonry of the tower wall, and the loose construction of the curtain wall (lower right)

186 189 189 190 191 191 192 196 196

197 197 200

200 201

Maps

1.1 Ayyubid fortresses in the twelfth and thirteenth centuries. Adapted from H. Kennedy, ed., Historical Atlas of Islam, 2nd rev. edn, Leiden, 2002, 22–3 1.2 Qalʿat Ṣadr and the route off the Darb al-Ḥājj al-Miṣrī that crosses the Sinai Peninsula 1.3 Frankish fortresses in the vicinity of Mount Tabor  2.1 China and Mongolia. http://www..paulnoll.com/china/Dynasty/ history-maps-dynasty-2.html  2.2 Īlkhānid invasion routes into Syria 3.1 The four main mountain passes in Cilicia. Adapted from A. M. Eddé-Terrasse, ʿIzz al-Dīn Ibn Šaddād: Description de la Syrie du Nord, traduction annotée de al-A’lāq al-Ḥaṭīra fī Ḍikr Umarā’ alŠām wal-Ğazīra, Damascus, 1984, map 4 3.2 Fortresses in the vicinity of Aleppo taken by Baybars, according to Ibn Shaddād, but never rebuilt. Adapted from Eddé-Terrasse, Ibn Šaddād, map 4) 3.3 Ismāʿīlī fortresses in northwest Syria. Adapted from Chateaux des Assasins, in P. Deschamps, Les Chateaux de Croisés en Terre Sainte, vol. III, pt. 2: La Défense du comté de Tripoli et de la principat ‘Antioche, Paris, 1937) 3.4 Mamluk fortresses in relation to the Īlkhānid state  3.5 Fortresses mentioned in the treaties between the Mamluk sultanate and the Armenian kingdom. Adapted from Eddé-Terrasse, Ibn Šaddād, map 4 4.1 The middle course of the upper Euphrates. Adapted from EddéTerrasse, Ibn Šaddād, map 4

6 12 16 59 67

86 90

98 108 113 173

Tables

1.1 Mount Tabor: construction dates of different sections of the fortress 1.2 Widths of curtain walls in Ayyubid fortresses dating from the late twelfth century to the first decades of the thirteenth century 1.3 Widths of curtain walls in Crusader fortresses dating from the late twelfth century to the first decades of the thirteenth century 1.4 Tower measurements 2.1 Chinese siege machines in the service of the Mongol army 2.2 The chuang zi nu (large mounted crossbows) 2.3 The sieges of al-Raḥba and al-Bīra 3.1 Baybars’ conquests according to Ibn Shaddād  3.2 Fortresses conquered by the Mamluks  3.3 The results of Mamluk raids and treaties in Cilicia 4.1 The rank of the building teams according to inscriptions at al-Ṣubayba 4.2 Towers added and renewed in Mamluk fortresses 

28 42 42 49 56 63 72 92 105 117 137 204

Acknowledgements

This study is based on my dissertation, which examined the Ayyubid and Mamluk rural fortresses and the concept of defense in Greater Syria in the late twelfth and thirteenth centuries. It was supervised by Professor Reuven Amitai and Professor Ronnie Ellenblum from the Hebrew University of Jerusalem. I would like to thank them both for their valuable advice, guidance and help in funding several of the field trips that were the bases of this research. Part of the study was financed by the Israeli Science Foundation and a grant from the Faculty of Humanities at the Hebrew University of Jerusalem. The librarians at the following institutions, who assisted all along the way, deserve special praise and thanks: the Jewish National Library in Jerusalem; the Fine Arts Library at Harvard; Mt Scopus Humanities and Social Science Library at the Hebrew University; the Rockefeller Library in Jerusalem; the library of the Institute of Archaeology and the Aerial Photographic Archives in the Department of Geography at the Hebrew University of Jerusalem. Special thanks are due to my editor Mrs Mira Riech, Mrs Miri Shmida from the Cartographic Laboratory at the Hebrew University of Jerusalem who prepared several of the maps, and Noam Tamari from Studio Ze Tel Aviv who worked on the photographs and graphics that accompany this study. Professor Gideon Shelach from the Hebrew University of Jerusalem invited me to join his excavations in Northeast China in the summer of 2006. This was followed by a short trip to Inner Mongolia. It was a marvelous experience, an opportunity to glimpse the great steppe and to visit several medieval Chinese fortifications. Parts of this study were written during this visit to China. I owe much to him for offering this trip and sponsoring my travels during that summer. I would like to thank Professor Arthur Segal from the Department of Archaeology at Haifa University who helped and advised on any problem concerning construction methods. His careful explanations and illustrations always clarified matters. I am equally in debt to the following two professors from the Hebrew University of Jerusalem, Professor Michal Biran, for her translation of the Chinese source on gunpowder, and Professor Shaul Shaked, who translated the Persian source on siege warfare. Difficulties in Arabic translation were solved with the help of Professor Amitai, Mr Ofer Efrati, and my khushdāshiyya Shai Shir, Amir Mazor, and Yonatan

xvi  Acknowledgements Meroze. Military technology was often discussed with Shai Amisar, who patiently explained the chemistry of gunpowder and eventually built a model of the counterweight trebuchet to illustrate its problems and advantages. Yinon Engoltz spent a hard and long day measuring and weighing Mamluk catapult stones in Acre. Asad Daʿbūs, the park manager at Qalʿat Nimrod, showed me several secrets of the fortress that I would never have found on my own. Haim Barabé explained the architecture at his excavation in Safad, and described the finds that provide the dating. One of my favorite trips was made together with the Shlomo Rotem, who drew and reconstructed with great skill some of the architectural ruins and brought them back to life. I am most grateful to colleagues and friends for hearing out my ideas, and their many helpful comments. Any mistakes and faults in this study are entirely my own. My parents’ encouragement and support has been unstinting and I dedicate this book to them with gratitude and love.

Notes on transliteration

Arabic terminology is italicized and transliterated according to Mamlūk Studies Review. Common Arabic words such as amir and sultan are written without diacritical points. Most of the sources in Arabic are translated into English by the author unless otherwise noted. Place names are brought in the Arabic version as they appear in contemporary medieval sources. The Frankish names are mentioned in brackets, for example Ḥisn al-Akrād (Crac des Chevaliers). When the context deals explicitly with Frankish historical events, the name place appears in the Latin or the old French version. Well-known place names appear in their current usage in English (Aleppo, Cairo, Damascus, and so on). The Mongol and Turkish names are transcribed as if they were Arabic.

Abbreviations

ADAJ AISM BSOAS DI DOP EI2 ESI HJAS IEJ JESHO JNES JRAS JSS NEA PEQ PPTS QDAP RCEA SI

Annual of the Department of Antiquities of Jordan Arabische Inschiften aus Syrien Mesopotamien Bulletin of the School of Oriental and African Studies Der Islam Dumbarton Oaks Papers Encyclopedia of Islam, 2nd edn (Leiden and London, 1960–2002) Excavations and Surveys in Israel Harvard Journal of Asiatic Studies Israel Exploration Journal Journal of the Economic and Social History of the Orient Journal of Near Eastern Studies Journal of Royal Asiatic Studies Journal of Semitic Studies Near Eastern Archaeology Palestine Exploration Quarterly Palestine Pilgrims Text Society Quarterly of the Department of Antiquities in Palestine Répertoire chronologique d’épigraphie arabe, ed. E. Combe, J. Sauvaget and G. Wiet, 17 vols (Cairo, 1931–) Studia Islamica

Introduction

During the summer of 1998 I participated in a survey of several fortresses in southern Jordan. The signing of the Israeli-Jordanian peace treaty in the winter of 1994 allowed our small group to walk the grounds of sites previously studied from photos and plans.1 On our return to Jerusalem a heated discussion took place, in which the Mamluk sultan Baybars (r. 658/1260–679/1277) was declared “builder of the great Crusader fortresses.” This preliminary survey slowly revealed the scale of mamluk construction work in fortresses previously ruled by the Franks. Crusader fortresses in the Levant have long since become pilgrimage sites for scholars studying twelfth- and thirteenth-century history and military architecture of the Latin East. The magnitude of research and publication on the subject is overwhelming. Other than the large citadels of Damascus, Cairo and some of the smaller central Syrian cities, twelfth- and thirteenth-century Muslim rural and frontier strongholds have not been thoroughly studied. Ayyubid and Mamluk phases built upon Crusader fortresses are briefly mentioned, occasionally ignored and sometimes attributed to the Franks. In the past two decades Muslim military architecture in the Levant has attracted a growing number of scholars combining different methods of research such as surveys, excavations and the study of the Arabic primary sources. Thus, the events that led to the building of twelfth- and thirteenth-century Muslim fortifications and their architectural development are slowly being constituted.

Historical background During much of the twelfth century the Franks dominated the military scene in the Levant. The unification of Egypt and a large part of Syria under the rule of Salaḥ al-Dīn gradually changed the balance of power in the region. Though the Franks had lost much of their military strength, Salaḥ al-Dīn’s successors could not maintain a firm united political or military front against the Franks. This situation led to a new status quo, and for a period of 67 years (1193–1260) the Frankish states and the Ayyubid principalities were engaged in a struggle in which no side managed to gain a definite advantage over its adversary. It is important to note, however, that between 1292 and 1244 there were also relatively long periods in which both sides coexisted peacefully. The balance of power tilted in favor of the Franks only when a

2  Muslim Fortresses in the Levant new Crusade arrived and the local armies received substantial reinforcements (the fifth Crusade, 1217–21; the expedition of 1228 led by Fredrick II; the Crusade of Theobald of Champagne and Richard of Cornwall, 1239–41; and the Crusade of Louis IX, 1250–54). After the battle of La Forbie (1244) the Franks’ dependence on western aid became even greater. Other changes in the military balance of power occurred when Ayyubid-Frankish alliances were formed between various factions. These only increased the already existing tension among the Ayyubid rulers. It is against this background of the declining power of the Crusader states that the first Ayyubid fortresses were built. The fall of the Ayyubid dynasty and the rise of the Mamluks to power stretched over almost a decade (1250–60). The establishment of al-Muẓaffar Quṭuz (r. 1259– 60) as sultan coincided with the Mongol invasion of Syria led by Hülegü (1260). The first Mamluk–Mongol full-scale battle occurred in the Jizrael Valley, near the spring of ʿAyn Jalūt, on 3 September 1260. The Mongols suffered a humiliating defeat. Their forces in Syria and those who survived the battle fled back east and across the Euphrates. During the next 32 years (1260–91) the Mamluks slowly eradicated the Frankish settlement and fought to prevent the Mongol-Īkhānids from crossing the Euphrates and invading Syria. The Franks, who had been the main political and military entity in the region during the Ayyubid period, played a lesser part in the politics of the region between 1260 and the fall of Acre in 1291. Although the Mongol invasion had been repelled, once the Īkhānid state was established it became the Mamluks’ chief concern and remained the main threat to the Sultanate’s existence until 1323, the year the Mamluk-Īlkhānid peace treaty was signed.

The geographical and chronological framework Throughout this research emphasis is laid on rural and frontier fortresses; large urban citadels such as those of Cairo, Damascus and Aleppo are rarely mentioned or discussed at any length. The research is based on several field surveys that focused on an architectural analysis of four Ayyubid and eight Mamluk fortresses situated in what are today the modern states of Jordan, Israel, southern Turkey (the region of the central upper Euphrates and the Amanus Mountains) and Egypt (the Sinai Peninsula). This partly correlates to what was historically known as Palestine, Bilād al-Shām, al-Jazīra, the Kingdom of Lesser Armenia in Cilicia and Diyār Miṣr. The fortresses dealt with in this study were built or reconstructed from the early 1170s until 1291, covering the entire Ayyubid period but focusing on only the first three decades of Mamluk rule. Part of the historical discussion covers a slightly longer period that stretches into the fourteenth century, after the fall of the Crusader kingdom in 1291, the death of the last Īlkhān in 1335 and the last central nomadic invasion led by Temür (Tamerlane) in 1401. The aims of this research are threefold. The first is to study the architecture and development of Ayyubid fortifications at a time when the Franks were still a domi-

Introduction   3 nant power in the region, and the early Mamluk military architecture that developed after the establishment of the Mongol­-Īlkhānid state in 1260. The second is to examine the relation between strongholds and siege technology, often defined by military historians as the dialogue between military architecture and the development of siege warfare. Does the development of military architecture depend on the arrival of new, stronger and more advanced methods of siege warfare? Is military success in this period solely related to military technology? Developments and changes in Ayyubid and Frankish siege warfare have been systematically studied by several scholars. Therefore the research here focuses on methods of siege warfare brought from northeast China, and used by the Mongols in the Middle East. Because new siege technology would have affected the development of Mamluk fortifications, one of the questions carefully re-examined is the employment of gunpowder by the Mongols in the mid thirteenth century. A critical approach and a rather strict division are made between the Mongol armies until 1260 and those of the Īlkhānid state 1260–1335. A decade after the Mongols established themselves they had already adopted local siege warfare. The new technology they encountered, the counterweight trebuchet, was sent east to the court of Qubilai Khan (r. 1260–94) in China, where it was greatly admired. It seems that the Īlkhānid armies gradually changed and were different from those that arrived with Hülegü. They may have recruited local Muslim siege engineers as well as the Northern Chinese siege teams that had been the core of Mongol siege forces for several decades. This difference, however, is not reflected in the Arabic sources, where both the army that arrived with Hülegü and the later Īlkhānid army are referred to as Mongol (Mughul, Tatār). The third aim is to examine the influence of the structure of the army and methods of rule on the concept of defense and the development of fortifications. How if at all does the existence of a strong centralized military rule affect the development of military architecture as opposed to a loose federation or a weak monarchy? The following chapters examine Ayyubid and Mamluk rural and frontier fortresses within a wider historical, political and military context. They provide a detailed description of their architecture, their spatial distribution, and an overall view of their role, links, and ties with the many Crusader fortifications in this region. This research is part of a long-term study that will no doubt develop and change as new data from excavations, surveys and historical sources is revealed.

1 Ayyubid fortresses in the late twelfth and early thirteenth centuries

The notion that “the arrival of the Crusades at the end of the eleventh century revolutionized military architecture”1 presents a picture that is only partially accurate. True, there were relatively few Muslim fortifications in the Levant prior to the advent of the Crusaders, but they were far from inferior to what the newcomers were about to build. Urban Muslim fortifications in the Middle East were without doubt more sophisticated and had advanced far beyond the rural and frontier castle. Among the finest examples are the great city gates built in Cairo under the supervision of the Amir al-Juyūsh Badr al-Jamālī, during the years 480/1092–485/1097, on the eve of the first Crusade. The curtain walls that partly surrounded the city of Cairo and the three massive gates, Bāb al-Nāṣṣir, Bāb al-Futūḥ and Bāb Zuwayla, admittedly designed by three Armenians, were by no means inferior to the first fortresses constructed in the Kingdom of Jerusalem, and appear to have set very high standards.2 William of Tyre describes with great awe the mighty walls and numerous towers surrounding the Muslim cities of Tyre (1124) and Ascalon (1153).3 By the time the Ayyubid sultanate was established around 1171 the Muslims were well acquainted with the art of military architecture and were employing their own architects and craftsmen. Towards the end of the twelfth century Ayyubid fortresses matched those of the Franks in both scale and design. However, they evidently developed along different lines of thought, to some degree as a result of the way the Ayyubid armies and fortress garrisons were composed, and partly in reaction to the methods of sapping and assault used by the Franks. The following chapter will examine the characteristics of Ayyubid rural and frontier fortresses, emphasizing the main architectural elements and comparing them with Frankish fortresses constructed in the region at the same time. After presenting the architectural aspects, it may become possible to attempt an explanation of what led to some of the disparities between Ayyubid and Frankish fortifications. In addition this chapter, will, I hope, provide an adequate background to early Mamluk fortifications and will help in understanding some aspects of their evolution.

Ayyubid fortresses in the late twelfth and early thirteenth centuries  5

A short explanation as to why the following fortresses were chosen and the current state of research The boundaries of the region under discussion are the Sinai Peninsula and what are today the modern states of Jordan and Israel.4 The following chapter is based on field work conducted in four fortresses: Mount Tabor, also known in contemporary Muslim sources as Jabal al-Ṭūr;5 the first two building phases of al-Ṣubayba (Qalʿat Namrūd) at the foot of Mount Hermon; the fortress of ʿAjlūn (Qalʿat al-Rabaḍ) in the northern part of the modern state of Jordan; and Qalʿat Jundī, referred to in the Arabic sources as Qalʿat Ṣadr;6 built in the north western region of the Sinai Peninsula (Map 1.1). The choice fell upon these fortresses because of the clearly documented construction dates given in inscriptions found on the site and in the historical sources. Almost all fortresses are in a reasonably good state of preservation. All four have been either excavated or surveyed and the finds have been published. Johns conducted a thorough study of ʿAjlūn in the late 1920s.7 His detailed conclusions were partly revised by a French team, headed by Dangles which surveyed the site in 2002.8 The Franciscan monks Bagatti and Battista published the excavations and survey of the fortress on Mount Tabor.9 Deschamps surveyed al-Ṣubayba, and was the first to chart and draw a plan of the fortress, still used today by most scholars. This fortress, at first attributed to the Crusaders, was recognized as an Ayyubid construction only in 1989, after Ellenblum and Amitai carefully surveyed the historical sources and the inscriptions.10 Hartal thoroughly explored the Ayyubid towers on the western side of al-Ṣubayba in his excavation conducted in the winter of 1993, but did not cover the rest of the site.11 Barthoux, who published his finds in a short paper, first visited the fortress of Qalʿat Ṣadr in 1909.12 Wiet translated and analyzed the inscriptions from photographs taken by Barthoux, publishing them in 1922.13 The site was later visited and surveyed by Tamari while researching the Darb al-Ḥājj (“Pilgrims’ Road”) in the Sinai Peninsula.14 In the 1990s Jean-Michel Mouton and Sāmī Ṣāliḥ ʿAbd al-Mālik surveyed the fortress and published a series of articles.15 The importance of the fortresses varies considerably – while Mount Tabor and al-Ṣubayba represent frontier fortresses, ʿAjlūn is located in a rural area and Qalʿat Ṣadr is on an important pilgrimage route. The fact that the four fortresses were constructed over almost sixty years, allows us to study the gradual changes that occurred throughout this period. As is to be expected, grand urban Ayyubid citadels such as that in Cairo and major cities in central Syria have been thoroughly studied.16 Several monographs have been published on individual Ayyubid fortresses, and yet there is no comprehensive work examining the subject. Creswell was the first to research Muslim fortifications built prior to 1250.17 In his book on Crusader Castles, Kennedy devotes a short chapter to Muslim castles of the twelfth and thirteenth centuries, touching briefly on the subject of Ayyubid fortresses.18 The two volumes by Korn on Ayyubid architecture in Syria and Egypt do not deal explicitly with fortifications but decidedly contribute to the general subject.19 Hillenbrand includes a lengthy chapter on Muslim fortifications, in her book The Crusades: Islamic Perspectives, which contains valuable information and

6  Muslim Fortresses in the Levant

Map 1.1  Ayyubid fortresses in the twelfth and thirteenth centuries (after Kennedy, Historical Atlas of Islam, 22–3)

gives a swift survey of the most important sites.20 Her views, however, cannot be accepted without reservations. First, it is necessary to differentiate between Ayyubid and Mamluk fortifications and not to treat the two periods as one. The geopolitical and the military situation in the Levant had changed considerably; the centralization of the Mamluk sultanate and the composition of the army are clearly reflected in both the spatial distribution of Mamluk fortresses and the scale of their architecture. Second, Hillenbrand’s statement that “almost all Crusader castles were built to withstand siege; Muslim castles were not,”21 is rather surprising. Crusader and Mongol armies besieged Ayyubid fortresses, as did rivals within the Ayyubid

Ayyubid fortresses in the late twelfth and early thirteenth centuries  7 sultanate who laid siege to one another’s cities and fortresses. While some fortresses managed to resist assaults, others failed. Early Mamluk strongholds stood fast against the Mongol-Īlkhānid armys. Siege warfare was decidedly a matter carefully dealt with and an integral part of Ayyubid and Mamluk fortresses. It is important to note that two of the four strongholds, ʿAjlūn and al-Ṣubayba, were built on sites that did not have remains of earlier fortresses, whereas Qalʿat Ṣadr and Mount Tabor were partially fortified, although both of them are essentially Ayyubid fortifications.22 The sites of those fortresses, their size and plan, were decided according to the immediate threats and problems that troubled the region, the needs of the local population, or, in the case of Qalʿat Ṣadr, the safety of pilgrim and merchant caravans whose safety was the concern of the sultan.23 Another issue that will be emphasized is the ownership and building initiation of fortress construction during the Ayyubid period. As opposed to the Franks, some of whose early fortresses were built and owned by wealthy agricultural landlords who formed part of the Kingdom’s nobility,23 none of the above Ayyubid fortresses were built by families of high standing. Most of the construction was initiated by the Sultan or by one of the high ranking Ayyubid princes or amirs.

The role of fortresses in the Ayyubid period Several scholars have noted that the development of fortresses is a close and ongoing dialogue between siege warfare and the various methods of defense,24 both of which were constantly advancing during the twelfth and first half of the thirteenth centuries. Ayyubid military architecture is no exception to this rule. It developed and responded according to advances in the field of siege warfare made by their immediate enemies. At the same time, it seems that the whole Ayyubid approach and attitude to frontier and rural fortifications was different in many ways from that imported by the Crusaders from Europe, further developed by the Franks and later adapted by the Mamluks. The dependence of the Crusader kingdom on its castle garrisons, during this period and especially in the thirteenth century is well described by Marshall: The Latin states were able to defend their territories against external threat and civil disorder by means of the army and the strong point. The dependence of the kingdom’s troops on the fortified sites is striking … Most of the available soldiers, rather, were dispersed among a series of individual garrisons … The relationship of a garrison to its strongpoint thus provided the basis of the kingdom’s military strategy and a framework for much of the military history of the period.25 Fortifications were regarded by the Franks as essential to the existence and safety of their kingdom, and fulfilled both military and civil functions. Rural and frontier fortresses did not play the same role or receive the same attention or priority among the Ayyubid Sultans and political and military nobility. Their part in Ayyubid military array seems to have been less significant to the defense of the Sultanate in comparison

8  Muslim Fortresses in the Levant with the role assigned to fortifications by the Franks and later by the Mamluks. This may have stemmed partly from the lack of strong central government throughout most of the Ayyubid period, or the want of adequate funds to build, man, supply and run a large number of fortresses.26 The rather loose military organization of the Ayyubid forces, and their ability to summon field armies did not create a dependence on castle garrisons. On the whole they did not suffer from the shortage of manpower which became an acute and perpetual problem in the Crusader kingdom. Ayyubid armies recruited Türkmen, Kurds and small contingents of Bedouins.27 The fact that the core of the Ayyubid armies was composed of military slaves purchased and trained by the Sultan meant that the army (theoretically) could be enlarged according to the Sultan’s needs and depending on the balance of his treasury. The structure of the army and its composition may have had a strong influence on their relatively limited use of fortifications and the design of fortresses in the region under discussion. Ayyubid armies were light and mobile: based on mounted archers accompanied by trained infantry. The military supremacy of the Ayyubids during the reign of Ṣalāḥ al-Dīn enabled them to bring military confrontations to the threshold of their opponents’ castles or to the open field, rather than enclosing themselves in fortresses or using their fortresses as logistic centers or for retreating in the face of a stronger enemy. Although the idea of fortifications was not strange to them, they did not rely upon fortresses for the safety and defense of the Sultanate and did not see them as a necessity. Furthermore, rural and frontier fortresses in the Ayyubid sultanate do not appear to have served as official centers of administration as they often did in the Crusader kingdom. This function seems to have been reserved for the central cities. To a large extent they also diverged from the Mamluk sultanate perception of the role of fortresses. In rural districts such as Safad, Karak or the coastal plains in the case of Qāqūn, Mamluk fortresses became centers for the local population.28 Both the governor responsible for civil and bureaucratic affairs and the citadel governor who saw to the military concerns of the fortress and the region were nominated by the Mamluk Sultan in Cairo and were often chosen from among his personal Mamluks. During the Ayyubid period, in small towns both positions were occasionally held by one nā’ib (governor); in 583/1187 Mujāhid al-Dīn Qāymaz al-Zaynī, one of Ṣalāḥ al-Dīn’s most trusted amirs, was appointed to be the governor of both the town and the citadel of Tiberias.29 The Mamluk Sultans, however, maintained a strict and clearcut differentiation between civil and military tasks, in order to reduce the possibility of amirs turning a fortress into the center of a military uprising. One of the important and positive aspects of this Mamluk decree was that it enabled each nā’ib to pay full attention to his particular tasks, whether military or civil.

Changes in the dialogue between siege warfare and defense A combination of different factors dictated the geographical distribution, the number of fortresses the Ayyubids built, and the nature of Ayyubid military architecture: the balance of power between the Franks and the Ayyubids, the different methods of warfare, the structure of the two armies and their preferences as to the type of military

Ayyubid fortresses in the late twelfth and early thirteenth centuries  9 confrontation. Although these factors contributed to the development of Ayyubid fortresses, it was the gradual shift in siege warfare capabilities among the Ayyubids and the Crusaders that had the greatest influence on Ayyubid military architecture. Much importance has been attributed to the use of the counterweight trebuchet in defense of fortresses and its influence on Muslim military architecture in the late Ayyubid and early Mamluk periods. Some scholars tend to ascribe the developments and changes in Ayyubid and Mamluk fortresses almost solely to the increase in use of this siege machine. To avert the danger posed by the counterweight trebuchet al-ʿĀdil devised new fortifications that utilized gravity-powered artillery for defense. Optimal use of the new artillery required that the machines be mounted on the platforms of towers in order to prevent the enemy artillery from coming within effective range. … As a result, towers became larger in order to accommodate counterweight trebuchets … Walls were thickened in order to offer more resistance to bombardment. The counterweight trebuchet had rendered the passive system of defense obsolete.30 This statement will be examined in the following pages, as it does not quite fit the results of the research conducted on the four selected Ayyubid fortresses in this particular region. The advances in siege warfare, both in sapping techniques and in the construction of heavy siege machines,31 achieved by the Muslims since the early decades of the twelfth century, gave their armies a lead that thereafter the Franks never managed to match throughout their stay in the Levant.32 Their successes in siege warfare during the first Crusade belonged to the past. During the late 1160s the ability of the Frankish armies to carry out successful full-scale sieges was rather doubtful. The five sieges initiated by the Franks in the 1160s during their invasions of Egypt (Bilbeis in 1164 and 1168, Alexandria in 1167,33 Cairo in 116834 and Damietta in 116935) illustrate the beginning of this decline. With the exception of the second siege of Bilbeis that lasted a day, the siege of Alexandria lasted almost four months while that of Cairo stretched over two months. In both cases the Crusader army capitulated and retreated. Marshall devotes a lengthy discussion to the siege methods used by the Franks since the late twelfth century. His summary and conclusions show the relatively few sieges initiated by the Franks and their relatively few successes. The length of the siege is an important factor: whereas the Frankish sieges tended to last weeks, often months, and in a few cases well over a year; Ayyubid and later Mamluk sieges were short and tended to last a few weeks36 and sometimes only days. The difference in length was due mainly to the Franks’ endemic shortage of manpower.37 This was part of the reason, although even when the Christian forces outnumbered the Muslim, as was the case in the sieges of Alexandria (1167) and Mount Tabor (1217), their superior numbers did not give them the expected advantages. At Mount Tabor the siege ended after seventeen days, when the Christian army decided to retreat and return to Acre under cover of night.38 A year later the Franks laid siege to Damietta (1218–19). They eventually took the city, but the siege lasted

10  Muslim Fortresses in the Levant 18 months (Ṣafar 615–Shaʿbān 616/May 1218–November 1219) and ended only after the garrison and population ran out of food.39 In 1250 the Muslim garrison at Manṣūra and the Ayyubid armies camped outside the town walls witnessed one of the largest Christian armies ever put in the field, and yet the Crusaders could not organize logistics or control, and use their forces in a way that would lead to victory.40 The size of the fortress and the strength of its garrison did not make much difference. The Christian forces could not manage to take even relatively small fortresses such as Qalʿat Ṣadr (1177–8), or, almost a hundred years later, Qāqūn (1271) which was attacked by Prince Edward, later King Edward I of England. It appears that at the same time that the Franks reached their most formidable achievements in military architecture, during the late twelfth and early thirteenth centuries,41 their armies slowly began to decline in the field of siege warfare. This gradual change had a strong and long-term impact on the design and development of the Ayyubid fortresses as well as on later fortresses that were conquered and rebuilt by the Mamluks. A different way of looking at these gradual developments and shifts in military capabilities is to examine the evolutionary theories developed in order to explain the arms race between and within species. Most arms races are asymmetric, in contrast to a symmetrical race in which both sides constantly improve. In any arms race, whenever one side produces a new adaptation it may enjoy a period of grace before the other side comes up with a suitable reply.42 But adaptations are costly, as they often come at the expense of certain capabilities in other fields.43 This may partly explain the Frankish loss of capabilities in the field of siege warfare while they were concentrating on improving their military architecture. Arms races may end in three different ways: (1) they may simply continue; (2) both sides can reach an equilibrium; (3) one side may be driven to extinction.44 The third possibility fits that of our historical scene. Returning to the matter at hand, it has often been suggested that fortresses in the Levant developed in a linear way, always advancing and improving. But if we maintain that military architecture is an ongoing dialogue between two or more opponents, then the decline of the military capabilities of one side will bring about an almost immediate change in the architectural development of all sides involved in the dialogue. And so while the Franks were determined to invest a great deal of thought, time and funds in improving their fortifications throughout the Levant during the late twelfth and early thirteenth centuries, the Ayyubid rulers who built strongholds in regions that came into conflict with the Franks planned them with regard to the decline in siege warfare capabilities among the Franks.

Enemies of the Ayyubid sultanate and potential besiegers Throughout most of the Ayyubid period the wellbeing of the Sultanate was mainly threatened by the Franks and by feuds among members of the Ayyubid house. The Frankish armies belonging to the king, the high-ranking nobility and the Military Orders were reinforced only when a new Crusade successfully completed the journey to the Latin kingdom and on the few occasions when independent European rulers

Ayyubid fortresses in the late twelfth and early thirteenth centuries  11 responded to the frequent requests for military aid and sent soldiers to join the Frankish army on a permanent basis.45 Other than the Franks, internal feuds that at times developed into what could be described as full scale wars among various factions and members of the Ayyubid family plagued the Sultanate soon after Ṣalāḥ al-Dīn’s death (589/1193). The founding of al-Ṣubayba is a good example of a fortress built as a result of such internal family disputes. On a few rare occasions members of the Ayyubid family joined forces with the Franks against their own kin; such an alliance was formed between the German emperor Frederick II and al-Malik al-Kāmil, the Sultan of Egypt (d. 635/1237). It was this alliance that probably brought about the construction of al-Ṣubayba.46 Bedouin tribes that harassed and threatened local governors and the sedentary population were at times a problem serious enough to justify the construction of a fortress accommodating a large well-equipped garrison able to maintain law and order in the region. It is more than likely that the local tribe of the Banū ʿAwf was the main reason for the building of ʿAjlūn, and not the Frankish strongholds in southern Jordan as suggested by Johns.47 Ibn Shaddād al-Ḥalabī describes the tribal feuds of the Banū ʿAwf that had become a threat to the safety of the region and the local governors.48 According to al-Qalqashandī they simply ignored the local Ayyubid governors. It was only after the fortress was built that the tribe began to acknowledge Ayyubid rule.49 In order to prevent the Banū ʿAwf from interfering with the construction, the amir ʿIzz al-Dīn Usāma, who supervised the building, convinced the tribe that the fortress would protect them from the Franks.50 The idea that the fortress was built as an answer to the construction of Belvoir (Kawkab) is questionable.51 As a rule fortresses did not clash with one another and there are very few examples of two fortresses guarding a strategic site and belonging to two different opponents. Fortresses that were expected to guard river crossings or mountain passes were built directly on the pass or ford, like Baghrās which guarded the Belen mountain pass on the Amanus range or the Crusader fortress of Vadum Iacob that guarded an important crossing over the Jordan. ʿAjlūn may have been built with the idea that it would protect a ford over the Jordan, but it is about 17 km. from the river. While ʿAjlūn was being constructed (1184) it was no doubt considered a frontier fortress, although by the time the construction work was finished or shortly there after, the frontier lines had changed and moved considerably due to the Crusader defeat at the battle of Ḥaṭṭin. Belvoir was captured by Ṣalāḥ al-Dīn (584/1188) after a lengthy siege and probably partly restored.52 It was held by the Ayyubids until al-ʿĀdil seized it from ʿIzz al-Dīn Usāma, who up until then had governed also ʿAjlūn. Al-ʿĀdil decided later to raze Belvoir to the ground (609/1213).53 Following the peace treaty between Richard of Cornwall and al-Ṣāliḥ Ayyūb (1241), the ruler of Egypt, the fortress was given by Pope Alexander IV to the Order of the Hospitallers in 1255.54 Belvoir was probably taken by Baybars eight years later while the sultan was campaigning in the eastern Galilee. It was abandoned and never rebuilt by the Muslims.55 Qalʿat Ṣadr is located on a route branching off the Darb al-Ḥājj al-Miṣrī which crosses the Sinai Peninsula from east to west, starting at ʿAjrūd and ending at ʿAqaba (Map 1.2). The fortress probably existed well before Ṣalāḥ al-Dīn came to power, the initial construction perhaps dating to the beginning of the twelfth century or even earlier.

12  Muslim Fortresses in the Levant Tamari, who conducted detailed research on the Darb al-Ḥājj, suggests that “Qalʿat Ṣadr should be conceived from the outset as having functioned within the framework of the Darb al-Ḥājj rather than the context of military activity of the Ayyubid Sultans in the twelfth century.”56 Yet in spite of its small dimensions and old-fashioned military architecture,57 this statement of Tamari’s should be reconsidered, for Qalʿat Ṣadr is often mentioned as a place where the campaigning Muslim armies marching to and from Egypt during the late 1160s gathered, halted for a short rest or camped for a few days.58 Cairo lay within only two days march. The distance from Qalʿat Ṣadr to the Port of Ayla (modern ʿAqaba) was three days; however, according to Abū Shāma, water had to be carried from Ayla for the duration of the journey.59

Map 1.2  Qalʿat Ṣadr and the route off the Darb al-Ḥājj al-Miṣrī that crosses the Sinai Peninsula

Ayyubid fortresses in the late twelfth and early thirteenth centuries  13 The first Crusader assault on the fortress took place in 573/1177–8,60 the second in 579/1183 along the road leading to the fortress. An Ayyubid contingent fell upon the Franks before they actually reached Qalʿat Ṣadr, the whole force was killed and the garrison returned with goods the Franks had looted from the regions they had just passed through.61 Although the fortress is small and could hardly supply a large army on the march, it seems that the proximity to Cairo and to the port of Ayla turned Qalʿat Ṣadr into a convenient site to organize an army that was advancing towards Cairo. Ṣalāḥ al-Dīn was quick to acknowledge the importance of the site and kept it well stocked, as the desert land around it could not supply the garrison’s needs. In the spring of 1184 a caravan laden with livestock, weapons and goods was dispatched to the fortresses of Ṣadr and Ayla after a Frankish attack that had occurred earlier the same year. The caravan was protected by a contingent led by Qaysar, the governor of the East (al-Sharqiyya, a province in the eastern Delta).62 Even though the fortress rates as one of the smallest, the Franks were not able to take it by force. The rugged terrain and the protection provided by the steep topography accounted for much of its fortifications.63 Thus it appears that fortresses were built and designed with due consideration and after assessing the enemy’s strength, general abilities in the field of warfare and particular abilities in siege warfare. Fortresses such as Mount Tabor, which bordered Frankish territory, were built primarily against probable Frankish aggression. This was more than likely the case also at Qalʿat Ṣadr (although it was built on a much smaller scale), which was refortified by Ṣalāḥ al-Dīn after the Franks attacked it on two occasions. Even at times of family feuds, it is unlikely that the Ḥajj route would be disrupted by an Ayyubid force. Al-Ṣubayba does not have a clear position. It was probably built with the intention of resisting a possible joint force of Crusader, Frankish and Ayyubid armies, but in actual fact it did not suffer a severe assault until the arrival of the Mongols in 1260. ʿAjlūn, as we have seen, was built against a local menace in the form of the Banū ʿAwf, who were growing in strength. While the fortress is well built, it is relatively small in size. The fortress was assaulted by an Ayyubid force, but was never besieged or threatened by the Franks. The builder and governor ʿIzz al-Dīn Usāma,64 according to Abu ‘l-Fidā’, was thrown into jail in 609/1212 by the Sultan’s son, al-Muʿaẓẓam ʿĪsā;65 the reason being al-ʿĀdil’s suspicion that Usama was conspiring against him.66 This was the only occasion during the Ayyubid period that the fortress of ʿAjlūn was attacked. Usāma’s garrison tried to defend it but eventually surrendered the fortress to al-Muʿaẓẓam.67 The latter then presented it to one of his faithful amirs, ʿIzz al-Dīn Aybak, governor of Ṣarkhad, who enlarged and strengthened it. The last Ayyubid owner of ʿAjlūn was al-Nāṣir Yūsuf (d. 658/1260); no significant changes were made during his reign. As in the case of al-Ṣubayba, the fortress was eventually besieged by the Mongols (658/1260). Obeying the order of the Mongol commander Kitbughā, al-Nāṣir Yūsuf (who was now a Mongol captive) persuaded the governor to surrender the fortress.68 Once ʿAjlūn was delivered to the Mongols it was razed to the ground, according to the source, although the archaeological remains, as in many other sites, show no signs or evidence of massive destruction. One more point should be made here. There is a great difference between the characteristics of Ayyubid urban fortifications and those of rural and frontier fortresses,

14  Muslim Fortresses in the Levant the principal one being: the scale of construction. The existence of a large civil population dictated the nature and scale of military urban fortifications. Muslim cities in this period were well established as the seats of rulers and their armies. Urban fortifications and citadels developed along different lines to those built in the periphery. A brief look at political and military affairs in the region shows that the large cities in central Syria, such as Damascus and Aleppo, were besieged mainly by armies belonging to princes of various ranks in the Ayyubid family.69 Thus, the main threat to the Syrian cities during the last decade of the twelfth and beginning of the thirteenth century was not Frankish. This may have had an effect on the construction of the fortifications as compared with fortresses built to stand against Frankish armies.

The making of the Ayyubid fortress at Mount Tabor The fortress on Mount Tabor is one of the largest extant Ayyubid fortresses in the region (Figure 1.1). As well as the fairly rich contemporary historical sources, the archaeological remains are well preserved. The religious importance attributed to Mount Tabor by Christianity, the fact that it was the site of one of the first fortified monasteries built by the Franks in the early twelfth century, the size of the fortress built by the Ayyubids, and the circumstances under which it was constructed drew attention of both Latin and Muslim historians, who have left a more detailed picture of it than most other sites seem to offer. Between them, the historians give a clear outline of what led to the building of the fortress, the Frankish siege of 1217 and a few insights on the logistics of constructing a fortress and running a building operation on such a grand scale. I have therefore chosen Mount Tabor to open this section, although chronologically it is neither the earliest nor the most advanced example of Ayyubid military architecture. Mark Twain, who visited Mount Tabor in 1869, was obviously not over-impressed with the ruins. But he describes the scenery in great detail and was very taken with the view spread out below him. There is nothing about Tabor (except we concede that it was the scene of the Transfiguration), but some old gray ruins, stacked up there in all ages of the world from the days of stout Gideon and parties that flourished thirty centuries ago to the fresh yesterday of Crusading times. It has its Greek Convent, and the coffee there is good, but never a splinter of the true cross or bone of a hallowed saint to arrest the idle thoughts of worldlings and turn them into graver channels.70 Mount Tabor provides an excellent view over the lower Galilee and the northeastern parts of the Jezrael valley. It rises from the flat surrounding landscape to a height of 588m above sea level and dominates the region and the main roads that run from the Mediterranean coast to the Galilee and further north. The mountain slopes are covered with a dense green growth of Mediterranean forest, creating an illusion of a possible source of water or of high rainfall. In reality there is no natural spring at the summit and water was collected in cisterns that were built within the fortress walls.

Ayyubid fortresses in the late twelfth and early thirteenth centuries  15

Figure 1.1  Ayyubid fortresses from the late twelfth to the early thirteenth centuries drawn on the same scale

The strategic importance of Mount Tabor either escaped the Franks’ notice or they may have simply decided to rely on the smaller forts that dotted the area such as Safforie,71 Jezrael72 and La Fève,73 to guard the important roads that were crucial for communication between the coastal plain and the Galilee (Map 1.3). Tancred (c.1076–1112), who conquered the Galilee, built a new monastery run by the

16  Muslim Fortresses in the Levant Benedictines near the existing Greek monastery dedicated to Elijah. Later a third monastery was established and dedicated to Moses. Although Tancred did not provide the monasteries a garrison, the main church was well fortified, as attested by the Russian traveler Daniel who visited the site in 1106–7. According to his description a wide wall of dressed stone blocks with doors of iron surrounded the large church of the Transfiguration.74 William of Tyre, describing the Ayyubid attack of 1182 on the monasteries and village of Dabbūriya at the foot of the mountain, says that both the monks and the villagers who escaped from the Muslim army found refuge in the monastery that was protected by a wall and towers.75 Ibn Wāṣil says that there was a Crusader fortress on Mount Tabor and that the Muslims destroyed it after taking Acre (1187).76 If there was a Crusader fortress besides the monasteries it may well have been destroyed during Ṣalāḥ al-Dīn’s conquest (although it is very doubtful that he would have destroyed it so completely as to remove all evidence of its existence). Another possibility is that the construction of the new Ayyubid fortress in 1211 covered whatever remains may have survived. From the existing archaeological evidence visible to the naked eye, and without further excavations, it looks as if the Muslim chronicler Ibn Wāṣil regarded the fortified monasteries as a Frankish fortress. According to Abū Shāma, the Muslims took Mount Tabor soon after the battle of Ḥaṭṭīn.77 It was probably then that the monks abandoned the monasteries. About a year before Ṣalāḥ al-Dīn’s death (589/1193) it was decided that the Galilee would be given to his son al-Afḍal. Important towns and fortresses were distributed among a number of Ṣalāḥ al-Dīn’s own mamluks, while some were given as iqtāʿāt to new amirs. The loyalty of al-Afḍal’s governors in both towns and fortresses thus

Map 1.3  Frankish fortresses in the vicinity of Mount Tabor

Ayyubid fortresses in the late twelfth and early thirteenth centuries  17 derived from their loyalty to and respect for his father. This was the case at ʿAjlūn where Usāma, one of the most important amirs who served under Ṣalāḥ al-Dīn, continued to govern the fortress until falling into disfavor in 1212.78 This reduced al-Afḍal’s ability to dictate or make changes in local or regional policies.79 Al-ʿĀdil’s rise to power in 596/1200 changed matters throughout the region. As noted above, it was al-ʿĀdil’s son who got rid of Usāma, using the ultimate reason – treason against the Sultan.

The reasons behind the construction and the Franks’ response It is not quite clear what precisely led to the decision to build a fortress on Mount Tabor, as it seems that raiding expeditions were being conducted by both sides. In 601/1204–5 the Franks, perhaps a group of newcomers that had just arrived from Europe, landed in Acre. Obviously lacking a strong commander, they decided almost of their own accord to raid the Muslim territories. Ibn al-Athīr claims that their raiding expeditions reached as far as Trans-Jordan, ravaging and plundering, and killing many villagers. Al-ʿĀdil retaliated and set out the same year to raid Frankish territory. Later in that year an agreement was signed between the two sides. Al-ʿĀdil gave the Franks Sidon, Nazareth and a few other cities and returned to Egypt.80 In 606/1210 a Muslim contingent left Damascus and raided the Frankish villages along the coast. Once the soldiers were satisfied they went up to Mount Tabor. No further details are given as to why they decided to stop and rest at Mount Tabor or whether they attacked the monasteries.81 This indeed was the nature of the warfare between the Franks and the Ayyubids during the early thirteenth century: raid and counterraid with periods of truce and no major large-scale battles.82 Returning to the question above, the fortress may have been built in order to strengthen al-ʿĀdil’s rule over the Galilee and to prevent the Franks from using Nazareth as a jumping point for raiding Muslim territories further north.83 Ibn Wāṣil and Ibn al-Athīr both remark that al-ʿĀdil decided to destroy Kawkab (Belvoir) and to build a fortress at Mount Tabor.84 The impression one is left with is that he was determined to have a stronghold of his own to supervise the region. It seems we must remain with Benvenisti’s suggestion, based on Ibn al-Athīr’s description, that it was the Franks’ repeated raids (although the sources document only the one raid during 601/1204–5 from Acre inland to the Galilee) that led the Sultan to build a new fortress.85 The construction of a large-scale fortress in order to solve an immediate problem that troubled the region was not a revolutionary idea. Although the decision seemed a sound one at the time, taking into consideration the local state of affairs, seven years later (614/1217) it turned out to have been an expensive mistake, or, in the words of Ibn Wāṣil: “Its building was pointless.”86 Jacques de Vitry notes the building of the fortress in a few brief sentences and records the Franks’ reaction. His description gives a clear image of the humiliation and fear the new fortress created: Immediately after the aforesaid John was crowned and anointed king, the Saracens, to the disgrace and injury of Christendom, and more especially to the end of that they might straiten the city of Acre, fortified Mount Tabor against us.87

18  Muslim Fortresses in the Levant The fact that Mount Tabor was, and still is considered a sacred Christian site, where Christ performed the act of the Transfiguration, no doubt added to the anger and humiliation that was felt among the Franks. According to Jacques de Vitry the Franks renewed their truce with the Sultan and used this relatively quiet period to try and summon help from Rome.88 The construction of a large-scale Muslim fortress caused unrest beyond the immediate region and the news made its way to the papal court. In his Quia Maior, presented on 11 November 1215 at the Fourth Lateran Council, Pope Innocent III spoke of the Ayyubid fortress on Mount Tabor and referred to it as a threat to the Crusader kingdom.89 The arrival of the Fifth Crusade in September 1217 upset the regional balance of power, which for the duration of the Crusade tilted in favor of the Franks. Their first move was towards Mount Tabor, whether because the Pope had proclaimed the fortress as a threat, or simply because it had become a convenient close-range target for the large armies that were gathering at Acre. Two Muslim sources give a detailed description of the battle, but only one of them mentions the Franks bringing siege engines. Sibṭ Ibn al-Jawzī (d. 654/1256) describes a fierce battle between the Crusader armies and the defenders that took place outside the fortress gate but does not mention siege machines. The Christian army attacked twice. During the first assault, which began on the second day of Ramaḍān, they tried to breach the curtain wall with their lances (rimāḥ) hoping to dislodge the stones with the strength of the lance. This would in fact have been quite feasible, as the stones are not huge and the quality of construction is rather poor. A Muslim mounted force together with infantry stormed out, met the attackers at the foot of the wall, killed some of the Christian troops and drove the rest back to the lower part of the mountain. A day later the Christian army collected every available man and climbed to the fortress, carrying a large ladder. This was poised against the wall near the gate and face-to-face combat raged between the two sides as the Franks tried to scale the gate. Finally the Muslim defenders threw nafṭ.90 The ladder caught fire and a group of Christian soldiers, including high-ranking commanders, were killed.91 Thereafter the Christian armies decided to return to Acre and did so at dawn on the 6th day of Ramadan. The fact that the Crusader army had come to the siege without a full range of siege machines or sapper units, thinking they could take the fortress by using one scaling ladder and lances, shows they probably underestimated both the fortress and its garrison and indicates a certain lack of organization. Even if we accept Ibn al-Athīr’s (d. 630/1233) description of the siege, and assume that the Franks arrived with an assortment of siege machines, as he explicitly says, this does not add much to the honor or prestige of the Frankish siege units.92 After seventeen days (Sibṭ Ibn al-Jawzī claims the siege was shorter) of siege, with no threatened arrival of Ayyubid reinforcements, the Franks decided to retreat. Although the siege had failed, the Franks were defeated and the fortress had not suffered any damage, al-ʿĀdil and al-Muʿaẓẓam decided not only to evacuate the garrison from their newly built fortress but also to spend a further year demolishing it.93 According to Oliver of Paderborn (d. 1227) it was the construction of ʿAthlīt by the Franks that led to the Muslim decision to destroy Mount Tabor.94 Ibn Wāṣil points out that if the Franks had managed to conquer Mount Tabor, their control of

Ayyubid fortresses in the late twelfth and early thirteenth centuries  19 the fortress and ensuing raids on the lands of Islam would have cut off (qaṭa‘at) or interrupted the traffic along the main roads leading to and from Egypt.95 Was the whole idea of building such a large-scale fortress a rash act? Were the initial calculations misleading? It seems that, while sufficient funds were found and allocated by the Sultan to finance the building, the calculations of manpower needed to garrison and maintain such a fortress, and the accessibility of an Ayyubid army that could provide military help in times of siege, those aspects were to a certain degree ignored. Reinforcements were a necessity, very few fortresses of any kind or definition could defend themselves solely with their own garrisons, however large or well equipped. Those two issues, so carefully studied and examined fifty years later by the Mamluk Sultan Baybars when he had to decide whether to rebuild or destroy the fortresses he had conquered, were evidently not given enough thought in the case of the fortress of Mount Tabor. It is more than probable that if Egypt had not been put under direct threat the fortress at Mount Tabor would have been spared. But as the fifth Crusade aimed for Damietta and settled for a long siege (May to early December 1219) that became part of an extensive attack,96 the Ayyubid army was in need of all its available forces. A large fortress such as that at Mount Tabor was useless if the enemy was camped on the bank of the Nile Delta ready to head south for Cairo. In spite of the brief role of Mount Tabor during the Ayyubid period, the historical sources combined with the archaeological remains and the numerous inscriptions provide sufficient evidence of how the Ayyubids went about the building of a fortress.

The logistics of castle building Once the decision had been made, al-ʿĀdil was determined to build on a grand scale. The fortress at Mount Tabor is one of the largest ever built by the Ayyubids. It extends over the entire mountain summit (Figure 1.2). The current division of the fortress area between the Franciscan and Greek Orthodox monasteries makes it difficult to roam around freely and grasp the vast size of the compound. Al-ʿĀdil’s army was summoned to the site to safeguard the work from any wouldbe interruptions by the Franks. The soldiers and their amirs, however, were mainly employed in the various construction works that did not demand experience or professional skill. According to Ibn Wāṣil they transported stone and participated in the actual building.97 Every man on the site was put to work. Professional craftsmen (al-Ṣunnāʿ) such as masons and carpenters were brought from all across the country.98 According to Maqrīzī 500 builders were employed on the site, not including the professional craftsmen, the soldiers and the amirs.99 At ʿAjlūn labor for the building of the fortress came from the local tribe of the Banū ʿAwf. We have no further information as to where the skilled craftsmen came from, although according to John’s study the masonry is typical to the region and the masons may have been local.100 As for al-Ṣubayba and Qalʿat Ṣadr, there is no mention in the sources or the inscriptions as to who the labor force was or where it came from. It is not quite clear who supervised the work at Mount Tabor. Ibn Wāṣil writes in one place that al-ʿĀdil ordered his son al-Muʿaẓẓam to build the fortress,101 and

20  Muslim Fortresses in the Levant

Figure 1.2  Mount Tabor, plan and aerial photograph (1945)

in another he says that the Sultan himself was present until the construction of the fortress was completed.102 This is rather unusual, however, due to the sheer size of this project and its proximity to the Frankish frontier it is quite likely that the Sultan himself was present, if only to see the first stages of this grand project started. The inscriptions found on the site carry the name of al-Muʿaẓẓam. Al-ʿĀdil’s name appears only on the main foundation inscription dating to 609/1212–13. A  Mount Tabor, the foundation inscription

Ayyubid fortresses in the late twelfth and early thirteenth centuries  21 (1) Those who spend their substance in the cause of Allāh and follow not up their gifts (2) with reminders of their generosity or with injury, (3) for them their reward is with their Lord. On them shall be no fear nor shall they grieve.104 Ordered the construction of this (4) blessed [fortress] our lord, the great Sultan al-Malik al-‘Ādil (5) the fighter of the holy war, the victorious, Sayf al-Dunyā wal-Dīn, the sultan of Islam and the Muslims (6) Abū Bakr b. Ayyūb, friend of the Commander of the Faithful, upon his return from the East with the (7) victorious army, arriving outside al-Ṭūr after the truce had ended (8) The work began on Sunday, 5 (9) Dhu al-Ḥijja in the year 609/24 April 1212–13 (10) under the command of the amir Ḥusām al-Dīn Lu’lu’ (11) ibn ʿAbd Allāh al-Malikī al-Muʿaẓẓamī. The following four inscriptions, commemorating the building of al-makān (a particular but unspecified site) dated to 609/1212–13; the bāshūra (outer fortifications) built in 610/1213–14; al-manzil (house, accommodation, probably referring here to the living quarters), 611/1214–15; al-makhzan (the warehouse?) constructed in 611/1214–15, and al-badanah (the curtain wall) dated to 612/1215–16, reveal the order of building and provide further information on the men that supervised the construction. B  Mount Tabor, al-makān

(1) In the name of Allah the Merciful, the Compassionate. Ordered the construction of this blessed site (2) our lord the sultan al-Malik al-Muʿaẓẓam Sharaf al-Dunyā wal-Dīn (3) ʿĪsā b. al-Malik al-ʿĀdil, may Allah perpetuate his rule. [its construction was] under the command of (4) the most noble al-amir al-kabīr Ḥusām al-Dīn Lu’lu’ (5), the needy of Allah’s mercy. [This was] in the months of the year 610 (1213–14). C  Mount Tabor, the bāshūra

22  Muslim Fortresses in the Levant (1)In the name of Allah the Merciful the Compassionate (2) Those who spend their substance in the cause of God, and follow (3) not up their gifts with reminders of their generosity. And he ordered the construction of this blessed bāshūra (4) our lord the Sultan al-Muʿaẓẓam (5) ʿIsa b. al-Malik al-ʿAdil Abū Bakr b. Ayyūb. The work began on the 5 of Muharram in the year 610 (23 May 23 1213–14) (6), under the command of the servant needy [of Allah’s mercy] Lu’lu’ al-Muʿaẓẓamī. As the inscription was not found in situ the word bāshūra may be translated as any one of the following: outer fortifications, barbican, a fortified tower or a gate.107 D  Mount Tabor, al-manzil (residence)

(1) In the name of Allah the Merciful the Compassionate. Verily we have granted thee a manifest Victory. That God may forgive thee thy faults of the past and those to follow fulfill His favour to thee and guide thee on the straight way.109 (2) Ordered the construction of this blessed (3) residence, [under] the noblest amīr, the great isfāhsillār (general of a cavalry unit)110 Ḥusām al-Dīn Lu’lu’ in the reign of our lord, (4) al-Malik al-Muʿaẓẓam Sharaf al-Dunyā wal-Dīn in the year 611 (1214–15) the month … E  Mount Tabor, al-makhzan (possibly the storeroom?)

(1) In the name of Allāh, the Merciful, the Compassionate. (2) Ordered the building of this blessed al-makhzan (3) our lord the Sultan al-Malik al-Muʿaẓẓam Sharaf al-Dunyā (4) wal-Dīn in the months of the year 611(1214–15), under the command112 of your servant (5) the needy of Allāh’s mercy, may Allāh have mercy upon him Ḥusām al-Dīn Lu’lu’. F  Mount Tabor, al-badanah (the curtain wall)

Ayyubid fortresses in the late twelfth and early thirteenth centuries  23

113

(1) In the name of Allāh, the Merciful, the Compassionate. Among the things ordered to be built is this blessed curtain wall (2) our lord the sultan al-Malik al-Muʿaẓẓam (3) Sharaf al-Dunyā wal-Dīn ʿĪsā b. al-Malik al-ʿĀdil (4) Sayf al-Dunyā wal-Dīn Abū Bakr b. Ayyūb in Rabʿī (5) II in the year 612/1214–15 under the command of the servant (6) the needy of Allāh’s mercy, Ḥusām al-Dīn Lu’lu’ al-Mūʿaẓẓamī. The inscriptions bear the names of three more figures who supervised the scores of workers and craftsmen. The first is ʿIzz al-Din Aybak; since he in the course of his long career participated in the construction of quite a few substantial military and civil buildings in the region, he should be properly introduced. He began as a mamluk of the ruler of Damascus. Later he was appointed ustādār (head of the sultan’s household) by al-Muʿaẓẓam and was given Ṣalkhad. Among the buildings in whose construction he is known to have taken part are the fortresses at Azraq, and ʿAjlūn, madrasas in both Jerusalem and Damascus, and khāns.114 He was obviously an experienced builder and administrator.115 He did not remain for the whole duration of the construction at Mount Tabor; only one inscription bears his name and it is dated 609/1212, which can be regarded as a fairly early stage of the building. His name appears two years later, in 1214, on an inscription at ʿAjlūn.116 The word al-khādim appearing next to his name may be translated literally as servant, but it often has the meaning of eunuch.117 Ibn Khallikān says Aybak was a mamluk and so the word khādim should be translated simply as servant. However, throughout the Ayyubid period many important fortifications were built, supervised and governed by eunuchs whose loyalty to the Sultan was thought to be unquestionable. The walls of Cairo and its citadel as well as the fortifications of Acre were entrusted to Bahā’ al-Dīn Qarāqush al-Asadī, who was one of Saladin’s most trusted amirs and more than likely a eunuch.118 G  Mount Tabor, inscription commemorating the work of Aybak ʿIzz al-Din

119

(1) In the name of Allāh, the Merciful, the Compassionate. Ordered (2) the building of this blessed site the (3) amir ʿIzz al-Din Aybak, the servant (khādim) (4), al-Malikī al-Muʿaẓẓamī, in the year (5) 609/1212 … .

24  Muslim Fortresses in the Levant The other two men mentioned in the inscriptions are the amir Ḥusām al-Dīn Lu’lu’ al-Muʿaẓẓamī who was in charge of the construction works and/or the administration on the site between the years 609/1212–612/1215–16.120 His name appears in six inscriptions, below that of the Sultan and his son, followed by the rank of al-Isfāhsalār or Amir al-kabir.121 A tombstone found on the site carries a similar name, suggesting he found his death at Mount Tabor.122 It is very likely that Ḥusām al-Dīn Lu’lu’ was also a eunuch as the two parts of his name suggest: Lu’lu’ is considered to be a name with a feminine ring and was common among eunuchs. His nasab (“pedigree”) in the foundation inscription was Ibn ʿAbdallāh, a custom that became widespread mainly among eunuchs probably long before the Mamluk reign.123 H  Mount Tabor, the reservoir

124

(1) In the name of Allāh, the Merciful, the Compassionate. (2) Ordered the building of this blessed pool (3) Rukn ad-Dīn Mankūwirsh Jamdār Dawla, and master (?) the Sultan (4) al-Malik al-Muʿaẓẓam may Allāh perpetuate his rule. Rukn al-Dīn Mankūwirsh Jamdār Dawla, whose name appears on an inscription commemorating the building of the pool, probably a reservoir (no date is stated), beside that of al-Muʿaẓẓam, is the third person who supervised the building at Mount Tabor; nothing is known about him other than his rank. An amir Jamdar was the commander of a section of the Sultan’s royal guards.125 Although the title in the inscription does not state he was an amir, he may have served in this corps. If the Sultan himself actually participated in the supervision of the fortress, this was probably only at the very beginning and not for the whole period of building. Most of the work, which continued over three years, was supervised partly by his son but mostly by high-ranking amirs. At ʿAjlūn, the Sultan may never have set foot there while it was being built. ʿIzz al-Dīn Usāma was assigned to the work by Ṣalāḥ al-Dīn in 1184–5; once the fortress was built he remained there as governor.126 It was possibly Usāma who supervised the second phase that was carried out sometime between 1184–5 and 1214. Two more towers were added in the northeast, but neither has an inscription. The third and last phase dates from 611/1214–15 and was supervised by our former acquaintance, ʿIzz al-Dīn Aybak, who participated in the early building stages of the fortress at Mount Tabor. His name appears in the inscription on the southern tower at ʿAjlūn. The policy of nominating a high-ranking amir to supervise fortress construction could be carried out only at times when the Ayyubid Sultan held firmly the reins of both government and army and was in full command of the Sultanate. In periods when the Sultanate was ruled by a weak figure and its territories divided, each local prince made his own decisions as to where and how fortresses were to be built. The building

Ayyubid fortresses in the late twelfth and early thirteenth centuries  25 of al-Ṣubayba coincided with a period when the Ayyubid sultanate was plagued with feuds and quarrels between the Sultan, al-Kāmil, and the Damascene coalition led by his brother al-Muʿaẓẓam. Both the first and the second stages of the fortress were built by al-ʿAzīz ʿUthmān, one of al-ʿĀdil’s younger sons, who was the governor of Banias. But the initiative, the order and the funds probably came from al-Muʿaẓẓam who ruled from Damascus.127 One other name that appears on three inscriptions at al-Ṣubayba, dating to 627/1229–30, is that of Abū Bakr b. Naṣrallāh b. Abū Surāqa al-Ḥamdhāni al-ʿAzīzī. The inscription states that he supervised the building (wa-tawallā ʿimāratahu). From his name it appears he did not belong to the military class, but was a civilian who may have been qualified in some specific craft, although there is no mention of this. I  Al- Ṣubayba, Abū Bakr b. Naṣrallāh b. Abū Surāqa al-Hamdhānī al-ʿAzīzī

128

(1) In the name of Allāh, the Merciful, the Compassionate. Ordered the restoration of this defended frontier [fortress] (thaghr), the sinning, erring servant, (2) needy of Allāh’s mercy, ʿUthmān b. our lord, the great sultan, al-Malik al-‘Ādil, the scholar, (3) the doer of good deeds, the holy warrior, the fighter on the border, the raider, the martyr, Abū Bakr b. Ayyūb, may Allah cover him with his mercy. (4) The beginning [of the work] on this felicitous tower was in the month of Rabī‘ I, the year 627 (February–March 1230) Its construction was supervised by (5) the servant needy [of Allāh’s mercy] Abū Bakr b. Naṣrallāh b. Abū Surāqa (?) al Hamadhānī al-ʿAzīzī.129 The last Ayyubid building phase at al-Ṣubayba was carried out in 637/1239–40, while the amir Mubāriz al-Dīn Khutlukh al-ʿAzīzī was governor of the fortress and al-Malik al-Saʿīd Fakhr al-Dīn Ḥasan, son of al-ʿAzīz ʿUthmân, was the ruler of Banias (r. 630/1233–644/1247). Humphreys has suggested that a few years after the death of al-Muʿaẓẓam (624/1227) Banias had turned into a quasi-principality acting as a client state.130 Construction in this last phase was supervised by the great amir (amir kabīr) ʿAzīz al-Dawla Rayḥān, a eunuch who served al-Malik al-Nāṣir Dāwūd in 629/1231–2.131 He was put in charge of building both the reservoir and the archers’ galleries that were added in the southwest corner. The inscription, which bears the names of the ruler of Banias, the fortress governor and the eunuch is placed above the sabīl (drinking fountain) built outside the semi-covered pool.132 The information provided by the sources concerning the initial building of Qal’at Ṣadr is relatively scarce; construction dates are given in several of the inscriptions found on the site.

26  Muslim Fortresses in the Levant J  Qalʿat Ṣadr the site supervisor

133

(1) Ordered the restoration of this Al-Malik al-Nāṣir Ṣalāḥ al-Dunyā wa’l-Dīn, under the supervision of his brother al-Malik (2) al-ʿĀdil Sayf al-Dīn. Its construction was supervised by the amir Ṣārim al-Dīn Barghash al-ʿĀdilī (3) and its construction ended in Dhu al-qaʿda in the year 578/February1183.134 The amir who supervised the Ayyubid building phase was Ṣārim al-Dīn Barghash al-ʿAdilī.135 The name implies he was an officer belonging to Ṣalāḥ al-Dīn’s brother, al-ʿĀdil. Since he supervised the site it seems that part of the construction work was linked or initiated directly by al-ʿĀdil, who in later years became one of the most renowned builders among the Ayyubid sultans. The supervision of the construction was seldom entrusted to civilians. The military men who acted as supervisors were often moved from one site to another as attested by the inscriptions. Officers such as ʿIzz al-Dīn Aybak supervised work at both ʿAjlun and Mount Tabor, while Ṣārim al-Dīn Barghash’s name appears on inscriptions at the fortresses of Qalʿat Ṣadr and later at Karak.136 The supervisors of al-Ṣubayba and Mount Tabor were both of the rank of amīr kabīr, while ʿIzz al-Dīn Usāma who supervised the construction of ʿAjlūn was one of Ṣalāḥ al-Dīn’s chosen officers. Since an army was often present on a site to safeguard the construction, and occasionally joined the building force, a high-ranking officer had to be present. The logistics of running a large building site were no doubt different from those of a battlefield, but experience and efficiency in handling large numbers of men, seeing to their daily needs, and finding solutions for large and small problems must have served these officers well on such grand-scale building projects. They may have been involved in the initial stages of planning, lending their own experience to improve and modify the construction of a fortress as they thought fit. They had all probably participated in siege warfare at one point or another in their military careers.

Negotiating and compromising: topography and geology and their influence on the choice of building materials The topography and geology to a large extent dictated the location, much of the plan and the type of building material used in the construction of the fortress. The architect,137 or the engineer, present at the site negotiated and compromised between the natural contours of the summit, in Mount Tabor, or the spur, in the case of al-Ṣubayba and Qalʿat Ṣadr, and the positive and negative properties of the local stone. The limestone at Mount Tabor is porous and provides poor quality building material; on the other hand, it requires relatively less work to quarry. Cutting and dressing building blocks or hewing out a moat in the bedrock was easier thanks to the softness of the local stone. The topography of al-Ṣubayba is more dramatic. The almost vertical cliff

Ayyubid fortresses in the late twelfth and early thirteenth centuries  27 running down to the riverbed of Naḥal Govta along the northern side of the fortress is truly forbidding, while the southern side of the spur has a relatively moderate slope. Where possible the fortress plan makes the most of those natural defenses and the curtain wall runs almost exactly along the topographical contour lines. The qualities of the rock are very different to those at Mount Tabor. The hard dolomite is more difficult to carve, and cutting and dressing the stone blocks as well as hacking out a wide moat must have been a more time-consuming task. Qalʿat Ṣadr uses the topography in a similar fashion. It is located at the highest edge of the spur. The southern and eastern walls of the fortress rise directly from the edge of the steep cliffs of Wādī Ṣadr, so that there is virtually no access to the fortress from those directions. Because the northern slope provides no real natural defense, an artificial cliff was cut from the bedrock in order to give the fortress wall better protection. Out of all four fortresses ʿAjlūn’s natural terrain gives the least protection, and it is altogether quite an enigma as to why this site was chosen. The small hill that the fortress is built on commands a fine view but this would hardly suffice as an answer. In fact, during many hot summer days the view can hardly be seen through the heavy haze that settles over the Jordan Valley. Since the topography provides the fortress with no natural defenses it was surrounded by a deep moat cut into the stone.

The quarries and their location All four fortresses were built from the local limestone, the quarries being the site of the fortress itself. At ʿAjlūn and Mount Tabor a large amount of the stone was quarried directly from the moats, while at al-Ṣubayba and Qalʿat Ṣadr the quarry was established on the site in order to create a reasonably level courtyard within the fortress. Unlike most Crusader fortresses and fortified sites, where the stone was quarried at some distance from the site, the Ayyubid architects and engineers did not go out of their way even if the local stone was of relatively poor quality.138 They exploited the local material and made do with what they had in the immediate vicinity. It is hard to judge according to four case studies, but it seems as if, unlike the Franks, they preferred building directly on the bedrock and deliberately chose rock platforms rather than soil. To a certain extent the choice of location must have had an effect on the expense of the building. The cost of draught-animals, carts and men to manage the loading and transportation of stone from the quarry to the site was saved. Money and time were more often than not in short supply and cutting costs by placing the quarry on the site was a practical and simple idea. Although most of the local limestone was relatively soft and easy to work, fairly durable to the climate, plentiful and free of charge (as opposed to bricks) it did, however, have disadvantages: “[The] exposure [of limestone] to fire, however, it can not bear, but splits and cracks to pieces at once.”139 This note of warning sounded by the Roman architect Vitruvius (1st century bc) was exploited by both Christian and Muslim sapper units, in tunneling under towers and walls. The tunnel would be filled with wood and set on fire, in order to bring down the construction while the intense heat of the fire cracked the stone, undermining the supports of the overlying constructions.

28  Muslim Fortresses in the Levant

The order of building Before analyzing the separate elements of the fortresses, methods of construction and their general as well as particular role in the scheme of defense the sequence of construction should be considered. How was a plan carried out? What part of the fortress was built first? The curtain wall and towers or the inner defenses and living quarters? Or perhaps the water cisterns? Was there a rule followed by all fortress builders? Or was each site built in a different order? One of the few attempts to follow the precise order of building was made during the excavations of the Crusader fortress at Vadum Iacob. The excavation showed that the site had still been under construction and the fortress was never completed. This enabled the study of the order of building and the length of time required to construct certain sections of the fortress.140 A similar exercise can be carried out at the fortress of Mount Tabor by following the dates given on the inscriptions, which state what exactly was built and when the work began or when it was completed. This may help reconstruct not only the order of building but also the time schedule. The fact that the fortress was built by the Ayyubids during a relatively short period, and was only occupied once again by the Hospitallers in 1255, for a few years before Baybars conquered it (1263),141 allows us to follow the process with reasonable accuracy (Table 1.1). The construction of the fortress lasted almost four years. It began with the building of the bāshūra. This immediately poses a problem as the word bāshūra can mean any of the following: barbican, bastion or gate.142 The only aspect common to the three is that they are all part of the outer defense of the fortress. It seems, however, that as the inscription commemorating the building of the bāshūra was found in front of the gate tower it probably refers to the building of the main gate.143 The reservoir is dated the same year as the bāshūra. As there is no natural source of water at the summit it clearly received priority. Even if the reservoir constructed by the monasteries was still functioning, an additional reservoir was necessary. The manzil was built during the following year. The term probably refers to the living quarters. Their exact location, however, can no longer be determined. During the same year, the third year of building, the warehouse (makhzan) was constructed. This must have been fairly large as we are told that when the fortress was evacuated and destroyed the military supplies were distributed among ʿAjlūn, Jerusalem, Karak and Damascus.144 The most puzzling element in this account is the curtain wall, which was the last important Table 1.1  Mount Tabor: construction dates of sections of the fortress Dates

Individual buildings in the fortress

5th of Dhū al-Ḥijja in the year 609/1212–13 5th of MuḤarram in the year 610/1213–14 610/1213–14 611/1214–15 611/1214–15 Rabʿī II in the year 612/1215–16

Foundation date of the fortress Bāshūra The building of the pool/reservoir Living quarters (mazil) Storerooms (makhzan) The building of the curtain wall

Ayyubid fortresses in the late twelfth and early thirteenth centuries  29 building project to be mentioned in the inscriptions. This is somewhat difficult to explain, since a large percentage of the inner buildings in many fortresses “use” the inner face of the curtain wall as a prominent part of their structure. Vaulted galleries running along the inside of the curtain walls, towers and barracks are all constructed and supported by the curtain wall. From the defense point of view, one would assume that when undertaking such a project in a hostile country the army and the building force employed on the site would feel more secure if they surrounded themselves with a wall. Judging from the information provided by the inscriptions it appears that the inner part of the fortress was given priority. The curtain wall was apparently the final addition. This is very different from what we know of the Crusader fortress at Vadum Iacob (1178–9), where the curtain wall was built first, allowing the large vaulted halls to be supported by it. One way of explaining the curtain wall construction at Mount Tabor is by assuming that the year given (612/1215–16) represents the date of its completion, not commencement, although the inscription does not explicitly say so.

The anatomy of Ayyubid fortresses Turning to the individual elements that comprise a fortress, it is probably best to work from the outer parts of the defense to the center of the fortress. Types of masonry Determining the chronology or creating a typological chart according to masonry is a tricky matter. It demands a certain amount of caution and a survey of a much larger number of fortresses. As all four of our fortresses are dated precisely by inscriptions found on the site, the aim here is to provide a few clear descriptions of Ayyubid masonry types, and compare the findings to masonry styles employed by the Franks and later by the Mamluks. In the introduction to The Fortifications of Armenian Cilicia, Edwards gives this subject of masonry a great deal of thought. His ideas and advice are important and relevant to anyone working in the field. The four fortresses presented in this study all show different types of masonry, in other words, there is no uniformity. This could be due to a combination of factors. Different kinds of rock are carved in different methods. Certain regions may have had their own local style and traditions of masonry work; or, if restricted in time and funds, the architect may have decided to compromise and make do with coarsely cut and dressed stone.145 The rusticated style at ʿAjlūn was, according to Johns, common in the region after the Roman period.146 The first Ayyubid construction phase, dated 1184, used crudely cut stones. However, there is some variation. The outer face of the towers is of better quality stones than the inner face and the curtain walls (Figure 1.3). In the second phase, built between 1184 and 1214, the masonry is of a slightly higher standard. Both the curtain wall and the towers themselves are a mixture of ashlar and marginal drafted stone blocks of a fairly even size, cut and dressed with greater care and precision (Figure 1.4).

30  Muslim Fortresses in the Levant

Figure 1.3 ʿAjlūn, masonry in towers built during the first phase (1184). Curtain wall between towers 1 and 2

Figure 1.4 ʿAjlūn, curtain wall between towers 7 and 8 belonging to the second phase (1184–1214)

ʿAjlūn, Mount Tabor and al-Ṣubayba follow suit. The external masonry of the towers is of a relatively higher quality than that found on the internal face of the tower and the curtain walls. However, Ayyubid masonry appears to be of a rather poor quality when one compares it to the symmetrical building blocks and high

Ayyubid fortresses in the late twelfth and early thirteenth centuries  31 quality masonry used from the second half of the twelfth century in Crusader, and later in Mamluk fortifications. Crusader masonry is mostly defined by thin fine-combed diagonal lines, stones with marginal drafting and mason’s marks carved into the face of the stone.147 It is hard to give a similar definition of Ayyubid masonry. The tools used by the masons did not leave any particular mark; there is no clear repetitive direction of strokes, and no masons’ marks. Nevertheless, one can distinguish three different types of masonry. Roughly hewn stones were used mainly for the construction of curtain walls. Marginal drafted stones are used in towers and sections along the curtain wall where the wall makes a sharp angle; this simply strengthened the structure (Figure 1.5). Arrow slits, main gates and minor entrances were usually built of smooth ashlar, which enables more accuracy in construction. The stones are often tapered, in order to give the stone better anchorage in the core of the wall.148 On the exterior, the protruding boss of the marginal drafted stones may have helped to absorb the initial shock from stone projectiles hurled from siege machines.149 At the citadel of Arsūf small craters can be seen along the wall where Mamluk catapult stones hit the wall during the siege of 1265. The outer surface of this curtain wall, built of plain smooth ashlar was evidently not suited to absorb the shock of a strong direct attack.150 The size of the stones is of considerable importance. When stones weighed close to a ton or over a couple of tons, loading, transporting and maneuvering slowed the progress of work. However, there are some convincing reasons as to why large-scale stones were preferred. Large stones were more durable when battering rams or siege machines were used to breach the walls. They were also no doubt harder to force out by sappers aiming to undermine the curtain walls or towers.

Figure 1.5 An angle along the northern wall at al-Ṣubayba strengthened by masonry of higher quality

32  Muslim Fortresses in the Levant One of the most striking exceptions is in the size of the stones used to construct the quoin of the southern tower at ʿAjlūn. The corner stones of this tower measure two and a half meters in length.151 Such measurements are rare in both Crusader and Mamluk fortifications. The first line of defense: moats and glacis The fortresses’ first built line of defense, after the steep ravines and natural cliffs, was the moat or the glacis. In some fortresses a combination of both was employed in order to protect the curtain wall from any of the methods of assault the enemy might choose. The first and most frequent method employed by sappers was undermining walls or towers. Tunnels were dug under the foundations, stacked with wood and set on fire. Moats as well as glacis also made the use of scaling ladders difficult, as they could not be propped directly against the wall. At the siege of Mount Tabor (winter 1217) the only point where the Franks used a scaling ladder was the main gate. Maneuvering with wooden siege towers, a method the Franks often employed in the first Crusade and during the first decades of the twelfth century,142 was made almost impossible as the moat first had to be filled in order to bring the siege tower as close as possible to the fortress wall. The glacis was a fire swept zone surrounding the entire perimeter of the fortress or certain sections.153 The advantages of a moat, a glacis, or both are thus obvious. It seems, however, that the better alternative as far as the Ayyubids were concerned was to defend the curtain walls by means of a natural cliff or a man-made cliff, as this provided adequate protection, was fairly cheap, and consumed little or no time if the cliff was simply part of the natural terrain. In most cases moats were not a matter of choice; they were clearly a necessity if the fortress walls were weak and poorly built and if the natural terrain did not provide sufficient protection. Al-Ṣubayba is the only exception in this group. The short moat hacked out along the western stretch of the first stage (1228) was rendered useless once the fortress was enlarged and covered the whole length of the spur.154 While all four fortresses had moats, glacis are rather scarce. They were constructed along very short stretches, below particular sections that were obviously more vulnerable than others, and needed extra protection either because the natural bedrock was not sufficiently deep, making sapping easier, or because of the moderate terrain at the foot of the wall. An interesting improvisation can be seen at ʿAjlūn where the natural rock was simply hewn at an angle, thus creating a glacis from the bedrock. At al-Ṣubayba a glacis was laid out only below the gate-tower in the southeast. It is built of small, roughly cut stones. In certain parts they are tightly interwoven with the natural rock (Figure 1.6). It may have been a structural necessity, a wall support rather than a defense against enemy sapping. Contrary to Crusader work, it seems at times as if more effort and thought were invested by the Ayyubids in the construction of a moat than in that of the curtain wall. In all four fortresses in this study, the moats were hewn out of the rock as opposed to being dug out of the earth and supported by counterscarps. They are all

Ayyubid fortresses in the late twelfth and early thirteenth centuries  33

Figure 1.6  Al-Ṣubayba, the glacis below the southeastern gate

dry moats, which was more often than not the rule in the Middle East throughout history. At Qalʿat Ṣadr, the moat exists only where the angle of the slope makes it relatively easy to access the foot of the wall. The wall itself was built after a vertical man-made cliff was hewn from the natural rock. At certain points the actual foundations of the

34  Muslim Fortresses in the Levant curtain wall were laid as high as 11m above the moat. The width of the moat varies from 3 to 7.5m.155 Mount Tabor has the most impressive of all the moats in this particular group, being of great length, breadth and depth. The moat surrounds the fortress on three sides, leaving the northern stretch of curtain wall to be protected by a fairly steep slope. It is hewn from the bedrock but, unlike the precise and neat work at ʿAjlūn, the inner side of the moat is not squared off. Because of the porous and relatively soft nature of the rock, the inner side of the moat, where the rock shows signs of cracking and crumbling, was faced with stone building blocks in order to prevent further deterioration. ʿĀjlūn is the only fortress that is surrounded by a moat from all sides. This appears to have been the only way to safeguard the site. On the southern side the moat is built up and supported by a counterscarp. The width and depth are truly forbidding: at certain points the depth reaches 8m and the width 10–15m. As the fortress was enlarged the earlier sections of the moat were built on and turned into cellars and storerooms.156 Considering the size of the moats surveyed here, it is quite obvious that filling the moat or tunneling underneath it was not an easy task. Filling in a moat of such dimensions as at Mount Tabor, ʿAjlūn and even Qalʿat Ṣadr would delay most enemy armies for a considerable length of time. The only attempt made by an army to fill a moat of a Muslim fortress occurred much later, during one of the early sieges conducted by the Mongol-Īlkhānid army against the fortress of al-Bīra (located on the upper Euphrates), in the winter of 1264–65.157 Gates The main gate is one of the few parts of the outer defenses whose structure exceeds the bounds of military architecture and partly enters the realm where aesthetics, symmetry and design play a role similar to that of monumental public buildings in urban locations. As well as the practical side of the matter at hand the defense of the entrance no doubt also had a psychological effect. There would often be a large inscription commemorating the ruler who had constructed the fortress. Occasionally this was accompanied by his royal emblem. The façade was clearly meant to inspire any passing travelers, neighboring inhabitants, local troops and the enemy, with respect, awe and fear. Yet gates had to be practical, allowing the passage of loaded carts, beasts of burden, men and their mounts as well as provide defense. With regard to defense, gates mark both the most fortified and most vulnerable part of many fortresses; being a break in the curtain wall and situated at ground level they were often the obvious focal point of an assault. As the doors were made of wood, the gate’s thickness, height and width were necessarily restricted. Being of inflammable material and lacking the strength of solid stone, wooden gates had to be concealed from the enemy’s siege machines, fire bolts and battering rams. The vulnerability of the material led to a number of creative ideas. On the whole, however, the Ayyubids followed ideas that were well established, and did not design new models of gate complexes.

Ayyubid fortresses in the late twelfth and early thirteenth centuries  35 A number of smaller entrances were built in accordance with the size and particular surroundings of each fortress. Some of the posterns, essentially designed as possible escape routes in an emergency, or as exit points for sorties, are located along the strongest and best protected section of the wall; some seem to have had no road or path that told of their existence. The most common structure found among the main gates is that of the “bent entrance,” which made the passage into the fortress more defensible. It was meant to slow down and delay the enemy, giving the defenders a chance to halt a charge. The bent gate appears in various forms and has no strict dimensions. Nevertheless it is interesting to follow its development in this group of fortresses and note the advantages and weaknesses of each form (Figure 1.7). The gate at Qalʿat Ṣadr was located on the northeastern stretch of the curtain wall. At the foot of one of the round towers is a bluff doorway to a non-extant passage.158 The main gate is therefore the only entrance into the fortress. This almost seems to be the rule in small fortresses, and can be seen at both ʿAjlūn and the first phase built at al-Ṣubayba. Although this gate is not built on a grand scale it requires friend or foe to take three sharp turns, each built on a different plane. An inscription commemorating Ṣalāḥ al-Dīn’s construction of the gate and the two adjacent towers in 583/1187 was placed above the entrance, decorated with a round shield crossed by a sword. Directly below the shield is a six-pointed star.159 Although the construction might not be of the strongest or of the highest quality, the design and layout made it a fairly complicated arrangement. In times of peace, but even more so in times of attack, it is hard to imagine a horse led through such a complex,160 or a group of enemy soldiers bursting their way in with ease. The entrance design of round solid towers flanking the gate can be found in Umayyad fortresses, khans and palaces built along the lines of the Roman and Byzantine castrum (military camp).161 The combination of tower and “bent entrance” Since al-Ṣubayba almost doubled in size during the second building phase in 1230, two new gates were required to serve the new western quarters. The first gate tower at al-Ṣubayba was constructed on the western side (Figure 1.8). It followed the plan of the bent entrance, but was encased in a tower and built on a grand scale with exceptionally thick walls (2.5–3.7m). The entrance was protected by a portcullis, and the doors were probably coated with sheets of iron to shield them from fire.162 The gate tower was well protected from siege machines by having the outer entrance face south, making it almost impossible to station siege machines directly opposite it. The main defense of the tower was provided by archers posted on the roof behind merlons. Following what was by 1230 a well-established tradition, the gate was adorned with an inscription commemorating the work of al-ʿAziz ʿUthmān.163 This tower served as the main entrance also during the Mamluk period but underwent further reconstruction that bears witness to some of the finest achievements of Mamluk builders. In addition to the three large gates, al-Ṣubayba had one small exit in the north, a short and narrow vaulted corridor running along the width of the curtain wall and

36  Muslim Fortresses in the Levant

Figure 1.7  Main gates in Ayyubid fortresses

leading to the steep slope of Naḥal Govta. It may have been used to send or receive a messenger if the fort was under siege, or as an exit for an infantry sortie (Figure 1.9). Whoever planned the fortress at Mount Tabor was extremely cautious or simply very economical with the use of gates. Only three gates were built in a fortress whose circumference measures 1530m. The main gate at the center of the eastern side was probably encased in a tower.164 It is difficult to reconstruct this structure since even the foundations can barely be seen and the new gate, built by the Franciscans, demolished the little that had survived the effects of weather and time. Bagatti and Battista reconstructed a simple straight entrance, although it seems they had little real evidence to support this idea. There may have been a two-point turn, which seems more common in both Frankish and Muslim gates built in the thirteenth century.

Ayyubid fortresses in the late twelfth and early thirteenth centuries  37

Figure 1.8  Al-Ṣubayba, reconstruction of the Ayyubid gate-tower

The main gate that awaits the visitor to ʿAjlūn is most impressive. As you walk through the long, beautifully lit corridors, up staircases and under arched entrances, the complexity and length are striking. This site, however, is somewhat misleading. The maze of inner walls makes it difficult to follow or distinguish the different stages of construction that the main gate underwent during the three Ayyubid building phases.155 The first gate built at ʿAjlūn was a simple straight entrance with no complicated sharp turns or angles (Figure 1.10a). It was placed on the west and was protected by flanking fire from the towers at its sides. During the second building phase (1184– 1215) it became a two-point turn system (Figure 1.10b). In the third and last phase

38  Muslim Fortresses in the Levant

Figure 1.9  Postern at the northern wall of al-Ṣubayba

Figure 1.10  Suggested stages in the development of the gate at ʿAjlūn

(1214–15) the main entrance was relocated to the corner tower in the east (Figure 1.10c). A corridor with a staircase led to a wide, arched gate with a portcullis. The external archway was decorated with a crude pattern of pigeons. The gate was given further protection by adding machicoulis allowing archers a vertical view down to the foot of the wall and moat. This is the only Ayyubid fortress of this group where the machicoulis can still be seen (Figure 1.11). They resemble a closed hanging balcony projecting from the wall resting on stone corbels. From them archers could fire in a direct vertical line or drop molten lead, boiling oil, rocks and missiles.166 They are small and cramped and were probably difficult to operate from. Other than the main gate, the fortress did not have any posterns. As suggested above, this was possibly due to the relatively small size and compact plan, although an emergency exit would appear a necessity even in small fortresses. The bent gateway was a simple construction. Although there are variations in size and form, the Ayyubids used this design and improvised upon it as they saw fit. The gate tower with the bent entrance as it appears in al-Ṣubayba can be found in

Ayyubid fortresses in the late twelfth and early thirteenth centuries  39

Figure 1.11  ʿAjlūn, machicolations above the main entrance belonging to the last building phase (1214–15)

Frankish fortresses of both the twelfth and thirteenth centuries;167 it continued to be used throughout the Mamluk period with relatively few innovations. Curtain walls The construction of all four fortresses on solid bedrock had two important advantages. The first was the almost absolute protection given to the walls and towers against enemy sappers. The second advantage was that there was no need to dig a foundation trench for the wall. The first stone course in al-Ṣubayba and Mount Tabor was laid directly on the bedrock surface. This method of construction can still be seen today in Druze villages near al-Ṣubayba. The basic structure of the curtain walls in Frankish and Ayyubid fortresses is similar. In both, the main part is constructed of a thick and wide core, traditionally consisting of stones neither dressed nor hewn. What holds the core together and gives solidity to the whole structure is a very generous amount of mortar. The core is faced with courses of stone blocks (Figure 1.12). The variations on this relatively straightforward building technique are surprisingly quite interesting. The core plays an extremely important role in the construction. If it is poorly composed the sapper’s work is made much easier. Once the stones of the outer wall are removed, breaking through the core would take little time and effort. In both Mount Tabor and al-Ṣubayba the core was made of rocks and mortar with no apparent method and order. This technique, or rather lack of one, may have caused flaws within the core, considerably weakening it.

40  Muslim Fortresses in the Levant

Figure 1.12  Masonry along Ayyubid curtain walls and the Crusader curtain wall at Vadum Iacob

At ʿAjlūn the masonry along the curtain wall is of highest quality. It may well be that the small size of the fortress allowed the architect to spend a substantially higher percentage of his funds on finer quality masonry and better building techniques. The Franks, on the other hand, devoted a great deal of attention, time, labor, funds and thought to the building of their curtain walls. By the time the fortress of Vadum Iacob was being built (1178–9) the construction of walls had developed almost into a fine art.168 If there was a manual explaining what material to choose and the method of curtain wall building it was probably carefully studied and followed by the Vadum Iacob supervisor of the works. The masonry work is of extremely high quality although the local limestone is poor and tends to crumble and crack quite easily. The stones are large and of fairly even size. The high protruding bosses provide a fair protection from catapult stones. In addition, the stone courses of the outer wall are wider and larger than those of the inner wall. This provides an extra defense against bombardment from siege machines. The core itself is neatly constructed of similar sized stones. Although the stones are not dressed they were carefully selected, their shape and size being roughly uniform (Figure 1.13). They are laid in neat courses, using an extremely generous quantity of mortar. It is important to note that the core at Vadum Iacob is built out of hard basalt, which was probably the only way to compensate for the soft and porous qualities of the local limestone that was used to face the wall.

Ayyubid fortresses in the late twelfth and early thirteenth centuries  41

Figure 1.13  Core and stone facing of the curtain wall at Vadum Iacob

The wall at Vadum Iacob is 4.3m thick, which has no parallel in the four Ayyubid fortresses presented in this study. One explanation to this great width could be the fact that very few towers were planned along the circumference, thus the main protection was the width.159 The width at Vadum Iacob can perhaps also be seen as corresponding to fear, lack of confidence in your capabilities to defend and protect your fortress and acknowledgment of your enemy’s superiority in the field of siege warfare. The width of the Ayyubid curtain wall at Mount Tabor is 2.9m. In al-Ṣubayba it measures 2.3m along the northern section of the fortress, which is built above the almost vertical cliff of Naḥal Govta; the southern wall is slightly wider and measures 2.7m. The additional 40cm is probably due to the relatively mild slope on that side that makes it much easier for an enemy to approach and attack. Qalʿat Ṣadr illustrates a similar idea. The curtain wall’s thickness changes according to the topography. Where the cliffs provide sufficient protection the walls measure only 1.5m, while along the northern side where the gentle slope provides no defense the curtain wall measures 2.8m, almost double the thickness around the rest of the fortress.170 This change in width can be seen also at ʿAjlūn. In this case the topography provides no protection whatsoever and one would expect a uniformly wide wall surrounding the whole fortress. The curtain wall built during the first phase measures 2.2m, but the northeastern section built during the second phase, measures approximately 3m.

42  Muslim Fortresses in the Levant The differences in width, quality of masonry, inner core and size of stones between the curtain walls of Crusader fortresses built in the second half of the twelfth century, and Ayyubid fortresses of the same period are remarkable (Tables 1.2, 1.3; Figure 1.14). The construction of the curtain wall is an important issue, for it appears that Table 1.2  Width of curtain walls in Ayyubid fortresses dating from the late twelfth century to the first decades of the thirteenth century Ayyubid fortresses & dates of construction

Width of Ayyubid curtain walls

Qalʿat Ṣadr (1173) ʿAjlūn first phase (1184) ʿAjlūn second phase (1214) Mount Tabor (1217) Al-Ṣubayba (1128–1230)

1.5–2.8 2.2 3 2.9 2.3–2.7

Table 1.3  Widths of curtain walls in Crusader fortresses dating from the late twelfth century to the first decades of the thirteenth century1 Crusader fortresses and dates of construction

Width of Crusader curtain walls

Bethgibelin (1136) Karak (1142) Belvoir (1168) Bethgibelin (1168) Vadum Iacob (1178–9) Chateau Pelerin (1218) Caesarea (1251–2)

2.1 2.5 4 4.2 4.3 8 5

1 A lengthy discussion is devoted to the growth and improvement of Crusader curtain walls in Ellenblum, R. Crusader Castles and Modern Histories (Cambridge, 2007), chap. 11. The measurements of the curtain walls at Bethgibelin are quoted from Shapira, Y. “The tower in the Crusader Kingdom of Jerusalem,” M.A. thesis, The Hebrew University of Jerusalem (2000),13–14.

Figure 1.14  Chart showing the increase in width of Crusader curtain walls

Ayyubid fortresses in the late twelfth and early thirteenth centuries  43 the Mamluks did not see any reason to improve the methods and quality of curtain walls in the fortresses they inherited from the Ayyubids, and more often than not the curtain walls were left just as they were found. This was not for lack of skill, as it is quite easy to distinguish the fine masonry and careful construction of the immense Mamluk towers at al-Ṣubayba that stand in great contrast to the rather poor Ayyubid curtain walls. Towers, arrow slits and fighting galleries Towers have often been defined by form and attributed to a particular tradition of building in a certain region or period.171 There have been several attempts to follow the development of towers through the twelfth and thirteenth centuries, making comparisons between those built by the Muslim rulers and those built by Crusaders and Armenians.162 The results were seldom conclusive. My initial research supposition was that the strength of an Ayyubid fortress relied to a great extent on its towers and the fighting capabilities of its garrison archers. Although it is best to draw conclusions at the end, I present this assumption at the head of the discussion since my first field surveys showed the towers in this group to be constructed with greater care and attention than most of the curtain walls. Qalʿat Ṣadr is the earliest fortress of the four. It has two distinct types of towers, belonging to two different building phases. The smaller rectangular towers (measuring: 4 × 3m, 4 × 4m and 5 × 5m), were probably constructed when the fortress was first built, in the early 1170s. According to Barthoux, who surveyed and charted the fortress, the towers were used mainly for accommodation.173 The round towers guarding the main gate (measuring 3m in diameter), the L-shaped tower in the northwest and the two large round solid towers (5m and 7m in diameter), were all built by Ṣalāḥ al-Dīn in 582/1186, as attested by the inscriptions found at the site (Figure 1.15).174 This is the only site in the group where solid towers are to be found, although it would seem that this building technique gives towers extra protection against mining and any type of projectiles. Solid round towers are fairly rare in rural and frontier fortifications of the Ayyubid period in this region. However, towers of similar size and design can be seen at the fortress of Habonim north of Caesarea (known in the Arabic sources as Kafr Lab and in the Latin charters as Cafarlet),175 probably first built by the Umayyad Caliph Hishām (r. 742–3) and the fortresses of Mīnat al-Qalʿa (Ashdod Yam) dated to the Umayyad period.176 Other examples of solid turrets can be found at Ascalon, where the towers along the city walls, originally constructed during the Byzantine period, were rebuilt by the Umayyads and later repaired by the Fatimids.177 On the whole, the towers of Qalʿat Ṣadr are fairly modest in size. The number of soldiers they could each have accommodated during an assault was small (fewer than 10), and as the rooftop was the only fighting level it seems very doubtful that the towers were ever used for the operation of siege machines.178 As for archers’ galleries, remains of arrow slits can be seen only along one very short section of the curtain wall. Along the wider parts of the wall there is a chemin de ronde, connecting the towers, allowing men to move round the fortress walls quickly.179

44  Muslim Fortresses in the Levant

Figure 1.15  Towers at Qalʿat Ṣadr

The towers built at ʿAjlūn, al-Ṣubayba and Mount Tabor developed in a fairly similar way, with no surprises or striking innovations. In spite of the many advantages of the round tower, it seems to have gone out of favor. This “mode,” which can hardly be explained, changed again and round towers made a return in the first decades of the Mamluk period. In the interim the Ayyubids found rectangular towers more than adequate for their needs. Frankish fortifications followed a similar pattern. Round towers are rare among fortresses in the Kingdom of Jerusalem during the twelfth century.180 They reappear in the thirteenth century, the best example being Crac des Chevaliers. There are bound to be exceptions, one of them being the D-shaped tower at Montfort Castle, built by the Teutonic Knights.181 In general it appears that L-shaped towers are more of an architectural coercion. They were added to castles that were being refortified, and seldom appear in the initial plan.182 The L-shaped tower situated along the southern curtain wall of ʿAjlūn clearly illustrates the advantages of this particular shape. Built in the reign of the last Ayyubid governor, ʿIzz al-Dīn Aybak, (612/1214–15), it hugs the corner of the previous building phase. Thus one could avoid the complications of breaking the curtain wall in order to insert a new tower. The most striking aspect of this tower was the method of construction. The walls, which are 3m thick, are built only of ashlar, unlike the smaller towers of the previous stage that were constructed in the traditional way – a core of rubble mixed mortar faced on both sides with hewn stones (Figures 1.16 and 1.17).

Ayyubid fortresses in the late twelfth and early thirteenth centuries  45

Figure 1.16  ʿAjlūn, tower 13 – the solid wall

Figure 1.17  ʿAjlūn, tower 13, looking from the northeast, and the plan of its three floors

Although closely spaced, the towers at ʿAjlūn barely project from the curtain wall; this reduced the defenders’ view and extended the ‘dead ground’ at the foot of the wall. The flaw was evidently noticed and corrected during the last building phase. The arrow slits in the towers of the first building phase (1184) at ʿAjlūn are surprisingly small and extremely awkward to maneuver in, let alone fire a bow from. The design resembles the shape of a funnel; they are low and narrow allowing enough room for only one archer (Figure 1.18). Its earliest appearance in the region is in the Crusader fortress of Belvoir (1168). It is difficult to trace the origin of this design. Similar funnel-shaped arrow slits were built at White Castle in South Wales, probably

46  Muslim Fortresses in the Levant

Figure 1.18  ʿAjlūn, reconstruction of a funnel-shaped arrow slit

Figure 1.19  The age of spacious arrow slits

during the years 1184–88; at Falaise, in Normandy, where they date to the late twelfth and early thirteenth century;183 and later in fortresses built by Edward I (r. 1272– 1307) in North Wales at Caernarvon and Beaumaris.184 An experiment conducted by Renn and Jones at White Castle shows that the archer drew his bow, aimed and shot while he stood at the threshold of the arrow loop.185 The funnel-shaped arrow slit had become obsolete by the time the Ayyubids decided to enlarge the fortress of ʿAjlūn in 1214–15. It was replaced by an arrow loop with a wider angle of fire and a wide spacious chamber large enough to accommodate two archers that could operate simultaneously (Figure 1.19), assuming that while one was shooting the other could organize his supply of arrows and rest

Ayyubid fortresses in the late twelfth and early thirteenth centuries  47

Figure 1.20  Al-Ṣubayba, solid towers belonging to the first Ayyubid phase (1228)

before it was his turn to move into the front, thus the rate of fire could continue without interruption. Every stage at ʿAjlūn represents an important advance and improvement in the tower structure. The height increased and the number of floors grew. The towers projected further and were armed with a greater number of arrow slits. Arrow slits were better planned, more spacious and provided a wider angle of fire. The first Ayyubid building phase at al-Ṣubayba in 1228,186 is somewhat similar to the first fortress built at ʿAjlūn.187 It was relatively small, with five solid-based, towers tightly fitted into a neat rectangular frame around a central hall (Figure 1.20). The second phase (1230) presents a line of important developments. While the fortress of ʿAjlūn was to a certain degree confined to the great moat that surrounded it, at al-Ṣubayba the terrain presented fewer limitations. During the second building phase at al-Ṣubayba the fortress almost doubled in size, making the most of the terrain and stretching along the whole length of the spur. Not only the scale of the fortress changed, but the design and layout were completely different from the initial symmetrical fort perched on the highest point of the spur. The rectangular fortress that represented the first stage may now be described as a donjon, even though it was not designed as one. The number of towers grew significantly, from five to twelve; three were built on the west and eight on the south. Towers were built only where the natural terrain did not provide sufficient protection. Along the south there were now eight towers. The northern curtain wall, which was well protected by the vertical cliff of Naḥal Govta, is completely bare of towers. The towers were connected to one another by galleries, narrow vaulted corridors that run along the inner side of the curtain wall and are pierced with arrow slits (Figure 1.21).188 The fortress towers had three levels of fire, while the galleries along the curtain walls probably had two levels of fire, the lower manned by archers and roof tops by teams operating siege machines. In contrast to Qalʿat Ṣadr, ʿAjlūn and al-Ṣubayba, Mount Tabor had one phase of construction. Though it was built only seventeen years earlier than al-Ṣubayba the two fortresses have very few features in common. Its twelve towers are small and sparsely scattered.189 This is a relatively small number in view of the extremely impressive circumference of the curtain wall.

48  Muslim Fortresses in the Levant

Figure 1.21  Al-Ṣubayba, archers’ galleries along the southwestern corner

The only tower that still stands at full height is at the southeast corner (Figure 1.22). It has one floor, an open central hall with six spacious arrow-loops stemming off it. The only evidence of the existence of galleries is in the plan drawn by the Franciscans. Paul Chevedden, who had conducted an extensive research on siege machines and fortifications during the Ayyubid period, reached the conclusion that “al-ʿĀdil drastically changed the construction of defensive circuits and the building of towers, which, became larger, higher, thicker and more closely spaced.”190 This change, says Chevedden, occurred due to the introduction of the large counterweight trebuchet into the defense of fortresses.191 Most of the towers described above are very modest in size. However, each fortress has one tower that is significantly larger than the rest. They were all capable of carrying a manjanīq, although it was probably not a large counterweight trebuchet. Chevedden’s conclusions were drawn mainly from the survey of towers in urban fortresses such as Damascus, Aleppo and Bosra. Although towers in rural and frontier fortresses grew and doubled in dimensions their average size was still smaller than that of the massive urban towers. In addition, they were sparsely placed along the

Figure 1.22  Mount Tabor, the tower at the southeast corner

Ayyubid fortresses in the late twelfth and early thirteenth centuries  49 Table 1.4  Tower measurements Name of fortress

Size in meters

Qalʿat Ṣadr 1170 (?) Qal’at Ṣadr 1187 ʿAjlūn 1184 ʿAjlūn 1214–15 Mount Tabor 1211

4 × 3, 4 × 4, 5 × 5 5, 7 (diameters of round tower) 8 × 8, 9 × 9, 11 × 11, 15 × 15 22 × 24 (L-shaped tower) 5 × 3.75, 8.25 × 7.5, 11.25 × 11.25, 12.25 × 15, 12.5 × 17.5 10 × 10, 19 × 12.5 14 × 15, 20 × 8, 18 × 22

Al-Ṣubayba 1228 Al-Ṣubayba 1230

curtain walls, as can be seen in both Mount Tabor and al-Ṣubayba where the distances between the towers run from 75 to 150m. To summarize the observations derived from archaeological surveys and measurements taken at four Ayyubid fortresses requires a great deal of caution. The four fortresses do not give a clear-cut answer as to one particular direction of development towards which Ayyubid fortresses were moving. It is almost impossible to isolate particular elements, dimensions or ideas that are shared equally by all four fortresses. It is important to remember that the Ayyubid period of construction was relatively short when compared to that of the Crusaders, which went on for almost two centuries. However, some conclusions may be reached and a few interesting ideas suggested concerning the general Ayyubid strategy, which differed greatly from that of the Franks and the Mamluks. The choice of location and surroundings is one of the few elements shared by all four fortresses. More often than not, the ground on which the fortress was built was chosen mainly because of the topography that made the most of the natural defenses provided by the contour lines of the site. The width of the curtain walls was dictated to a large degree by the terrain, thus saving time and expense in construction. ʿAjlūn is an exception: its moat was intended to compensate for the lack of steep ravines and jagged cliffs. This approach was not unique to Ayyubid fortifications and many Crusader fortresses as well were built on mountainous sites chosen because the topography provided an excellent defense in addition to the man-made fortifications. However, the Ayyubids exploited the topography to a much greater extent than the Franks. Towers might be completely omitted along stretches of wall that were protected by the natural terrain. Curtain walls that ran along the edges of cliffs and ravines were reduced to a crudely constructed wall that one could almost describe as the bare necessity. The plans of the fortresses differ from one another to a great extent. There was no general idea or pattern that prevailed or was preferred above others. The only closely matching plans can be discerned in the first stages at ʿAjlūn and al-Ṣubayba. However, when those fortresses were enlarged, they each adopted very different courses. One of the strongest and most dominant features of Ayyubid fortresses was the tower. And yet the design and development of towers was similar in twelfth century Muslim and Frankish military architecture. Until the second half of the twelfth

50  Muslim Fortresses in the Levant century, both Crusaders and Ayyubids built relatively small round or square towers, some of which were solid with one or two fighting floors. Their number on the whole was modest and they often barely projected from the curtain walls. The best Frankish examples can be seen at Bethgibelin (1136)182 and Karak (1142), and the inner walls at Belvoir (1168); the clearest Ayyubid examples are from Qalʿat Ṣadr (1186) and Mount Tabor (1211–17). This pattern took a sharp turn in the early decades of the thirteenth century. Almost overnight the Franks doubled the scale and size of their towers. Towers of the concentric Frankish fortresses at Chastel Pelerin, Crac de Chevaliers and Arsuf have no match in Ayyubid rural and frontier military architecture. Those towers suited the new dimensions of the Frankish concentric fortress. In most cases Ayyubid towers were built with greater care and a higher quality of masonry than that displayed in the fortress walls. Keeps, almost a trademark of the Crusader fortifications, were a European feature brought to the east by the Franks. They were not common in Ayyubid fortresses, and seem to have been built almost as a second thought as the result of a later decision to enlarge the fortress. This can be seen quite clearly in the second phase at al-Ṣubayba and the final stage at ʿAjlūn. One of the most striking differences between Crusader and Ayyubid fortresses is in the quality and size of their curtain walls. While the Franks sharply upgraded their curtain walls during the second half of the twelfth century, the Ayyubids kept on using the same methods and building on the same scale to which they were accustomed. Crusader curtain walls, as shown in Table 1.2, constantly grew in width, and quality of the masonry was improved and maintained. The Ayyubids, on the other hand, built walls that were of a relatively lower quality of construction. The strength of their fortresses depended to a great extent on the towers and the fighting capabilities of the garrison archers. This fits well with the general Muslim military concept in which archers played a dominant role. If the skill and strength of the garrison archers was as great as that displayed by Ayyubid mounted archers, we may conclude that their efficacy was not impaired by the modest size of the towers and the relatively long distances between them. The arrival of the concentric fortress marked the climax in the development of Crusader fortresses. Belvoir represents the first of this type. This architectural plan was meant to provide the ultimate defense. This idea was taken and further developed in the early thirteenth century; the most impressive examples are the massive fortress of Crac de Chevaliers and the double curtain walls at Chastel Pelerin. The Ayyubids did not adopt or imitate the Franks’ latest innovations and achievements. The fortress at Mount Tabor, built in the midst, both geographically and chronologically, of these architectural changes, bears little resemblance to those being built in the Crusader kingdom. This was, however, most definitely not for lack of technical knowledge, funds or skilled craftsmen. The walls at both Mount Tabor and al-Ṣubayba display the Ayyubid feeling of superiority in the field of siege warfare and their strong awareness of the decline in this specific field among the Frankish armies. If we look at the larger picture discernible in this study of Ayyubid rural and frontier fortresses, it appears that those built in remote areas that did not border Frankish territory were much smaller, as in the case of Qalʿat Ṣadr and ʿAjlūn. Mount Tabor and the final Ayyubid stage at al-Ṣubayba are significantly larger. The initial reason

Ayyubid fortresses in the late twelfth and early thirteenth centuries  51 for the construction of those two fortresses was directly connected to the military and political conflict with the bordering Crusader kingdom. The need for a large garrison in order to reduce Crusader raids into Ayyubid territories, and the short distance from Acre, were probably the main reason for al-ʿĀdil’s decision to build on such a grand scale. The same argument applies to al-Ṣubayba, which was probably built in anticipation of a large force comprising both Crusader and Egyptian-Ayyubid armies. It was no doubt obvious to the Ayyubid coalition in Damascus that a large garrison was needed in order to defend the main road to their capital and if not to defeat it at least delay any force that aimed to attack the city. This concept changed in the early Mamluk period. The most important frontier fortresses such as al-Bīra and al-Raḥba, which faced the Sultanate’s most formidable enemies, the Mongol-Īlkhānids, were of a fairly modest size. The fortifications that were rebuilt by the first Mamluk Sultan, Baybars (r. 1260– 77), followed some of the initial architectural lines established by the Ayyubids. Like the Ayyubids, they relied on their archers and accordingly increased the size of the towers and the number of floors, thus enabling a devastating line of fire from along the curtain walls and the towers. This will be shown in greater detail in the following chapters. But, as well as improving and strengthening their fortresses the Mamluks made certain that each of their frontier fortresses was supported by a relief force that in times of siege could arrive within a few days or a couple of weeks. The main fault of the Ayyubid system was not in the way the fortresses were constructed or designed but rather in their inability to organize a reliable system of reinforcements capable of relieving a fortress under siege.

2 Mongolian siege warfare and the defense of Mamluk fortresses

The evolution of military architecture is to a large extent linked to the development of siege warfare. The emergence of new ideas and technologies within the field of siege warfare was often followed by changes in the method of fortification and the defense of strongholds. This sequence could be reversed: innovations in the field of military architecture encouraged the development of siege machines and sapping. In some cases a substantial period would elapse before a suitable architectural solution was found, since the conversion and upgrading of an “old” fortress to fit new methods of siege warfare was a complex and often expensive task. The Mongolian army that prompted the strengthening and building of fortresses in the early decades of the Mamluk period did not resemble any of the contemporary Middle Eastern armies. Neither in size nor in structure could it be compared with the Ayyubid or the Frankish forces that fought in the Levant. Thus, the Mamluk fortifications built by Baybars and his successors should be studied in the light of Mongolian technology and methods of siege warfare. The aim of this chapter is to provide an account of Mongolian siege warfare on the eve of Hülegü’s campaign to the west (1253). In addition, it will follow and examine developments in the field of siege warfare within the Īlkhānid state. Did the Mongols possess knowledge or technology that was not known amongst the Frankish, Ayyubid, Armenian and Mamluk armies? The power of the Mongol army lay in its vast numbers of mounted archers, their swift and skilled maneuvers on the battlefield and their impressive ability to cover great distances in a way that very few contemporary armies were capable of matching.1 However, the military skills needed to conduct a successful siege were not found in the Mongol army that was gathered on the steppe by Chinggis Khan. Planning and building dams, ramps and siege-machines required engineers, carpenters and trained soldiers.2 Siege warfare demanded expertise and knowledge that were practically foreign to the nature of the Mongol army. The following excerpt, taken from the Secret History of the Mongols refers to Ögedei (r. 1229–41). The Great Khan is enraged with his son, Güyük for the disrespect he had shown towards Batu during a feast in celebration of a successful campaign.3 It clearly displays Mongol opinion concerning the difficulties involved in siege warfare and the inferior status attributed to those who fought in sieges.

Mongolian siege warfare and the defense of Mamluk fortresses  53 We [Ögedei] shall place him [Güyük] in the vanguard: We shall make him climb the town walls Which are as high as mountains Until the nails of his ten fingers are worn away; We shall place him in the garrison army We shall make him climb the town walls Which are made of hard-pounded earth Until the nails of his five fingers are ground down.4 Throughout the Mongol Empire’s existence and during the various stages of its army’s development, open battlefields were always favored over engaging an enemy fighting from strongholds and fortified cities. Embarking on a long siege could delay the army for a considerable length of time and disrupt a campaign that was carefully planned according to a strict schedule.5 Siege warfare forced the army to halt, to set up a camp that could serve them if necessary for an extended period and to procure sufficient pasture and water for thousands of horses and the large herds that were an integral part of Mongolian armies.6 Chinggis Khan had quickly learnt and acknowledged the shortcomings of the Mongolian army and it was he who initiated the recruitment of foreign artisans. This became a well-established policy among his successors. Craftsmen essential to the army’s campaigns were recruited from the lands that the Mongols conquered.7 The status, religion or origin of these people was of no significance to the Mongols, and their recruitment was determined only by their skill, knowledge and experience. The need for siege units first arose when Chinggis Khan moved across the steppe into northern China (1211–1234) and faced the walled cities of the Chin Empire.8 From this time onwards siege units were to be regularly recruited among the conquered states of northern China, and were to serve the Mongol army until the late thirteenth century. In 1211 a Chinese officer by the name of Chang Pa-tu supervised the Mongol army’s siege machines. After him came Hsueh T’a-la-hai who commanded both the siege machines and the navy.9 When Chinggis Khan invaded Central Asia he was said to have had with him ten thousand soldiers who could work the siege machine batteries.10 The recruitment of siege units among the Northern Chinese continued during the reign of the great Khan Möngke (r. 1253–9) who conscripted blacksmiths, carpenters and gunpowdermakers.11 Hülegü followed suit and in 1253 when the campaign to the west was being organized “he sent messengers to Cathay (North China) to bring a thousand households of Cathaian catapult men, naphtha throwers and crossbowmen.”12 During Qubilai Khan’s reign siege experts were still being recruited among the Northern Chinese and teams were brought from the region of the Huai River. Only in 1272 was this trend interrupted, when Qubilai turned to his nephew the Īlkhān Abagha and asked him to send two Muslim engineers who specialized in building siege engines.13 Three years later he established a Muslim siege unit that included soldiers and artisans specializing in siege weapons of the type used in the Eastern Mediterranean.14 From the year that Chinggis Khan began to besiege the cities of Northern China and up to the time Hülegü set out on his campaign, the Mongols were constantly

54  Muslim Fortresses in the Levant engaged in warfare. During those forty years the Mongol army besieged tens of strongholds and cities possessing larger, stronger and more advanced fortifications than those found in the Levant. Complex siege operations were carried out, some demanding the building of dams, the diversion of rivers and the raising of large earth ramparts. Apart from the necessary engineering knowledge, the scale of these operations required hundreds if not thousands of laborers.15 Prisoners taken from the local populations around the besieged city were often the source of manpower.16 One of the most detailed accounts of Mongolian methods of siege warfare is given by John of Plano Carpini, an emissary of Pope Innocent IV, who traveled in 1245–7 to the court of the Great Khan. They [The Mongols] reduce fortresses in the following manner. If the position of the fortress allows it, they surround it, sometime even fencing it round so that no one can enter or leave. They make a strong attack with engines and arrows and they do not leave off fighting by day or night, so that those inside the fortress get no sleep; the Tartars however have some rest, for they divide up their forces and they take it in turns to fight so that they do not get too tired. If they cannot capture it in this way they throw Greek fire … If they are still unsuccessful and the city or fort has a river, they dam it or alter its course and submerge the fortress if possible. Should they not be able to do this, they undermine the city and armed men enter it from underground; once inside, some of them start fires to burn the fortress while the rest fight the inhabitants. If they are not able to conquer it even in this way, they establish a fort or fortification of their own facing the city, so as not to suffer any injury from the missiles of the enemy; and they stay for a long time over against the city, unless by chance it has outside help from an army which fights against the Tartars and removes them by force. While they are pitched before the fortification they speak enticing words to the inhabitants making them many promises to induce them to surrender into their hands.17 Hülegü’s campaign to the west was the last of its kind under the united Mongol Empire.18 Although the source says that Hülegü recruited a thousand soldiers from northern China who specialized in siege warfare, it is not clear whether the Īlkhānid state established by the Mongols in 1260 maintained this custom and continued to recruit Northern Chinese siege experts. The exact size and composition of Hülegü’s army is not known, but it is worth looking at some of the figures and conclusions suggested by scholars who have studied Hülegü’s journey. None of the armies that joined Hülegü had their own siege units; it seems they were mainly made up of mounted archers. According to Martinez the armies that marched to the west under Hülegü’s command were similar to those that fought under Chinggis Khan.19 The size of Hülegü’s army has been estimated by a number of historians. According to Allsen the core of the army numbered seventy five thousand.20 A large Mongolian body, similar in size and composition, which had been in Asia Minor, Afghanistan and the west of the Caucausus, joined Hülegü before he entered the Middle East.21 Thus Hülegü had under his command roughly a hundred and eighty thousand soldiers. Both Smith and Allsen suggest that the army was considerably larger, eventually

Mongolian siege warfare and the defense of Mamluk fortresses  55 numbering three hundred thousand men, of whom only half were Mongolian. Smith estimated that from fifteen to seventeen tümens (a tümen roughly equals ten thousand men) were composed of both Mongolian and Turkish (Turikc) soldiers; to those he added contingents belonging to local vassals who were fully subordinated to the Mongols. The number of tümens composed of non-Mongols was, in Smith’s estimation, thirteen.22 We have no further information about possible siege units, other than the teams recruited by Hülegü on the eve of his campaign.23 By the time Hülegü set out for the West, Mongol siege warfare had come to rely and depend heavily if not completely on the Northern Chinese contingents. In the four decades that elapsed between the death of Chinggis Khan and the reign of Möngke, the Mongolian Khans faithfully observed this policy of recruiting specialists in this particular field, and did not train their own men in the art of siege warfare. Seeing that Hülegü besieged cities and fortresses in the Middle East by employing Chinese units and technology, a survey of the siege methods in use in northern China until the mid thirteenth century is in order.

Chinese siege machines in the thirteenth century The most common feature of Chinese siege machines during this period is the fact that they were propelled by manual power. As well as the type and design of the machines, their strength and efficiency depended on the size of the team that pulled the ropes. Their greatest drawback was the large number of people required to work them. While even simple siege machines needed forty men, the larger ones demanded a force of two hundred and fifty men. This obviously made the teams an easy target for the enemy. Chinese siege machines can be divided into three main categories: catapults that fired round clay balls;24 siege machines that hurled inflammable materials; and large and powerful crossbows. The Pao: the Chinese catapult The word Pao (Phao), means literally to hurl or to throw. The Pao was mounted on a base that resembled a pyramid or a light cart, which made maneuvering it during a siege a slightly easier task. The type of wood used for the central arm was of great importance; in most cases it was either oak or black birch, soaked in water for at least six months. The length of the central beam varied between 5.58 and 7.75m. One end had a leather sling that held the catapult-stone and the other had ropes to pull on. A more advanced type of catapult machine had several beams and could hurl five or seven stones simultaneously (Figure 2.1).25 This latter type of siege machine does not appear to have been known among the Franks or the Mamluks. In addition, there were also siege machines where the beam was mounted on an axis, enabling the machine to be aimed in different directions without having to move the entire structure. Most siege machines were built on those basic lines. The main difference between the various types was in the length of the central beam and the number of beams. Those two components dictated the number of men needed to operate the

56  Muslim Fortresses in the Levant

Figure 2.1  Chinese siege machines in the twelfth to thirteenth centuries

machine and the weight of the stones that could be hurled. Table 2.1 lists the various types of siege machines and the number of men needed to operate them, the distance each machine could fire and the weight of the stone or clay projectiles. A number of historical narratives concerning the Mongol invasion into the Middle East provide information that does not comply with the figures given in the above table. However, the figures proposed provide a reliable basis for estimating the force of Chinese siege machines. An interesting aspect of Chinese siege machines is the use of clay balls as projectiles. This was recommended by Ch’en Kuei (ca. ad 1130), author of the twelfth-century war manual Shou-ch’eng lu.26 The counterweight siege machine first appeared in China during Qubilai’s reign (1260–94). We have no written evidence of the use of counterweight siege machines by the Īlkhānids other than an illustration in the manuscript of the Jāmī‘ al-tawārīkh (dated 1306–14) at the Edinburgh University Library.27 The most convincing evidence of the penetration of Islamic siege technology from Syria to the Īlkhānid state and further into the Chinese Empire is dated 1272. During Qubilai’s wars against the Song state he had sought the advice and help of his nephew, the Īlkhān Abagha, Table 2.1  Chinese siege machines in the service of the Mongol army Number of men on Weight of stones the crew in kg.

Range in meters

Type of siege machine and number of beams

  40   50   70 100   70–80   90–100

  86.75   86.75   86.75 104   86.75   86.75

single-beam catapult five-beam catapult (whirlwind) single beam (crouching tiger) two-beam catapult five-beam catapult seven-beam catapult

   1.2    1.8    7.2   15   94.2 150

Source: This table is based on Weapons in Ancient China, ed. Y. Hong (New York and Beijing, 1992), 266. Hong’s data are taken from Wu Jing Zong Yao (‘The collection of the most important military techniques’) written by Zeng Gonglian at the order of the northern Sung dynasty (r. 960–1126).

Mongolian siege warfare and the defense of Mamluk fortresses  57

Figure 2.2  The counterweight trebuchet and catapult stones from the siege of Acre, 1292

asking for two Muslim siege engineers to be sent to his court.28 This request indicates that siege technology in the Eastern Mediterranean was more advanced than that of the Īlkhānid state and of the Chinese siege units employed by Qubilai. We may also conclude that the Īlkhānid army had begun to use the counterweight trebuchet soon after the state was established. Two Muslim engineers who specialized in building siege machines were brought to Qubilai’s court in 1272. ʿAlā’ al-Dīn was from Mayyāfarīqin and Ismāʿīl from Aleppo. At the court they built what is known in the Mamluk sources as “the large Frankish manjanīq”; in China it received the title “the Muslim phao.” A description of this siege machine is given in the Yuan-Shi. Its main advantages were that it did not require a large team; the projectiles were considerably heavier and could cover a greater distance. The weight of the stones could reach 90kg and more.29 According to the source, when they hit the ground they made craters that measured 2.17m in diameter. The sound tore through the sky and the earth and the damage was considerable.30 This appears to be the first Chinese description, if somewhat exaggerated, of the counterweight trebuchet (Figure 2.2). Inflammable materials In addition to stone and clay projectiles, vessels containing inflammable materials were a favorite type of ammunition in Chinese siege warfare.31 Certain siege machines had names such as “Hao Pao,” i.e. a fire-hurling siege machine. Pots with wicks of flax or cotton were used, containing a combination of sulphur, saltpeter (potassium nitrate), aconite,32 oil, resin, ground charcoal and wax. This recipe in fact contains all the basic ingredients for making gunpowder, but the emphasis in the recipe given above was on producing toxic fumes; it was called “tu-yao yen-ch’iu,” literally a ball of smoke and fire.33 The frequent use of inflammable materials in Chinese siege warfare was most probably due to the fact that a large percentage of both public and private buildings were constructed of wood and bamboo. Even houses made of mud bricks had a wooden frame. In addition to clay roof tiles, thatched roofs could be found in both northern

58  Muslim Fortresses in the Levant and southern China. The materials used for thatching included straw from wheat or rice, canes and various types of wild grass.34 It is no wonder then that the main dread of the besieged was of fires inside the city. Pots containing inflammable materials were hurled into a densely populated city where many structures were of wood.35 Occasionally, even the city towers along the curtain walls were a combination of bricks and wood;36 this can be seen in illustrations of the city Shao Hsing dating to the thirteenth and fourteenth centuries. A few examples of wooden gate towers date to the fourteenth and sixteenth centuries.37 In some cases a high wooden palisade was erected parallel to the stone or brick curtain wall.38 The immense destruction caused to the city of Bukhara when it was besieged by Chinggis Khan illustrates the effectiveness of inflammable materials in siege warfare. He [Chinggis-Khan] now gave orders for all quarters of the town to be set on fire; and since the houses were built entirely of wood, within several days the greater part of the town had been consumed, with the exception of the Friday mosque and some of the palaces, which were built with baked bricks … mangonels were erected, bows bent and stones and arrows discharged … For days they fought in this manner.39 In contrast to the widespread custom in large parts of China and some parts of Central Asia of building with wood and cane, in the Eastern Mediterranean public, private and military buildings were constructed mainly of stone.40 This was both plentiful and relatively cheap when compared to the price of wood. Although inflammable materials were used in siege warfare by the Ayyubids, the Franks and the Mamluks, their effect was considerably less disastrous. At this point of the discussion we should note that the most common material for the construction of curtain walls in China and Central Asia was rammed earth sometimes faced with stone and/or sun-dried mud bricks (Map 2.1 and Figure 2.3).41 In contrast to Middle Eastern and European strongholds that were built on prominent sites (hills, mountains, cliffs and the like) “no such preference for an elevated, easily defensible site, however, seems to have nourished city site selection in China.”42 According to Di Cosmo, wall building was a military concept and a technology that originated on the Central Plain. Walls asserted the state’s political and military control over certain areas. “Like roads, walls provide the logistic infrastructure to facilitate communications and transportation, vital elements for armies employed in the occupation or invasion of foreign territory.”43 This function attributed to walls in China differs from their function in the Eastern Mediterranean. Chinese walls were not constructed only for military purposes. They were often located on a river bank and their first and at times most important task was to protect the population from floods. In certain areas in the northwest the walls acted as screens against sand storms.44 Forbidding earthen ramparts almost became the trademark of Chinese fortifications, although certain sections such as the gates and towers were constructed of bricks or stone. Although they were effective their main fault was the large area of “dead space” below the wall. The straight lines and relatively few towers did not provide the defenders with sufficient flanking fire.45

Mongolian siege warfare and the defense of Mamluk fortresses  59

Figure 2.3  Earthern walls in northeast China and inner Mongolia (tenth to thirteenth centuries) Map 2.1  China and Mongolia [bottom left]

Was gunpowder brought over to the Middle East by the Mongols? One of the most controversial and intriguing issues debated among a wide range of scholars is the arrival and development of gunpowder. The differences of opinion concerning the use and efficacy of gunpowder during the thirteenth century call for a short survey. The majority opinion is that gunpowder was used by Chinese armies in the mid thirteenth century and even earlier. This group includes the following scholars: Goodrich, Chia-Shêng, Allsen, Khan, Martin, Chase, DeVries and Needham. Needham has conducted the most extensive archaeological and historical research on the subject of the development of gunpowder and firearms in China,46 arguing that by the mid thirteenth century the Chinese could destroy city gates and walls using gunpowder.47 He cites the four sieges in which gunpowder was used: Kayfeng (1232), Merv, Samarqand,48 and probably also during the siege of Baghdad (1258).49 Martin, who carried out a thorough study of the Mongol army, holds a

60  Muslim Fortresses in the Levant similar opinion.50 Khan, who surveyed the sources that cover the Mongol invasion of the Delhi sultanate, found sufficient evidence that gunpowder was used by the Mongol forces.51 Allsen also believes that gunpowder began to be utilized around the same time, but presents his opinion with some caution. Among the one thousand households Hülegü brought with him on his campaign to the west, he thinks there were men who knew how to produce gunpowder. Allsen says that at the battle near the Amu Darya (1220) the Mongol army fired rockets launched with the help of gunpowder.52 Like Needham, Allsen claims that during the siege of Baghdad the Mongols used some form of gunpowder. Many scholars rely to a great extent on the studies published by Needham. Franke and May are currently among the few who clearly state that even at the end of the Yuan Dynasty (1368) gunpowder was not a weapon that could determine the outcome of a battle. According to May “its use remains speculations.”53 Franke’s conclusions are based on a survey of the archaeological finds and the historical sources, all of which show that the number of casualties as a result of the use of gunpowder was very low. Ayalon demonstrated the problems of terminology in the Mamluk sources. His conclusion was that there is no definitive evidence that firearms were being used during the thirteenth and the first half of the fourteenth century, and that gunpowder was no more than an incendiary.54 Returning to the question of whether the Northern Chinese contingents in Hülegü’s army employed gunpowder technology in siege warfare requires an examination of the three sources that give an account of his army on the eve of his journey west. ••

••

••

The earliest source was written by ʿAlā’ al-Dīn ‘Aṭā Malik Juwaynī, a Persian official who served the Mongols (d. 681/1283). The Ta’rīkh-i jahān-gushā (‘History of the world-conqueror’) is an encyclopedic history of Chinggis Khan and his times that documents the events till 1257. The second source was written by the Persian historian Rashīd al-Dīn Faḍl Allāh Abū ‘l-Khayr (d. 718/1318). The Jāmi‘ al-tawārīkh (‘Complete Collection of Histories’) is the first universal history and was compiled ca. 1310 (an earlier version was completed in 1306–7).55 It records the history of the Īlkhānid state established by Hülegü soon after the fall of Baghdad and the execution of the Abbasid Caliph (1258). The latest source is the Yuanshi, a Chinese official history of the dynasty founded by the Mongols, compiled ca. 1370.

Neither Juwaynī nor Rashīd al-Dīn, both of whom describe Hülegü’s campaign, mention the use of gunpowder by the siege units of the Mongol army. It is perhaps possible that they were not familiar with this new weapon and described it by the terminology often used for the various incendiary weapons that were known among Middle Eastern armies. However, this does not seem to be the case. Both works contain only a short paragraph on the whole subject of the Northern Chinese siege units and neither notes anything unusual or new concerning the arsenal of Hülegü’s army. Thus it seems more than likely that Mongol methods and weapons did not appear strange or unknown in the field of siege warfare.

Mongolian siege warfare and the defense of Mamluk fortresses  61 A close examination of the Persian and Chinese texts is important to fully understand the matter at hand. Juwaynī says that among the siege units there were soldiers specializing in hurling nafṭ. The Persian term he uses is nafṭ andāzān.56 Rashīd al-Dīn mentions these very same units and uses the same terminology to describe the men who hurl nafṭ. In addition, he writes that among those siege units were teams he terms the charkh andāz. The Persian word charkh literally means “a round object,” while andāz means “to throw” or hurl.57 Rashīd al-Dīn is in fact referring to the teams operating the siege machines that hurled round catapult stones. The Chinese source is more complex than the two Persian ones.58 The Yuanshi describes the recruitment of craftsmen/artisans for the siege warfare units in the armies of Chinggis Khan and Ögedei.59 In 1252 Hülegü’s siege units were summoned from the same source of manpower. The term used for a particular craftsman in Chinese is a combination of two words. The first describes the raw material utilized by a specific craft. The second is jiang which can be best translated as smith. The list in the Yuanshi includes blacksmiths who work with iron, smiths who work with wood, i.e. carpenters, and smiths who work with gold, i.e. goldsmiths. The last raw material mentioned in the list is huo – fire. In contrast to the first three combinations which use the word jiang, huo appears on its own. It is therefore difficult to translate and reach a definite conclusion concerning its exact meaning in this context. The term hou jiang does not appear at all in the Hanyu Dacidian (the Chinese 12-volume dictionary) or in the Ciyuan (a dictionary orientated towards classical Chinese). The term hou jiang appears once more in the Yuanshi and there too comes directly after a mention of goldsmiths. Further, the combination hou jiang never seems to appear in any of the other earlier or later official Chinese histories such as the Mingshi, Jinshi, or Songshi. It is therefore difficult to conclude that the Mongols used gunpowder in siege warfare or to assume that they had men in their siege units who knew how to produce gunpowder. The fact that the Persian sources are silent and do not mention a dramatic debut of a new technology brought by the Mongols, as well as the fact that they simply stick to terms that were well-known in descriptions of siege warfare in Persian texts, further supports the view that gunpowder did not arrive in the Middle East with Hülegü’s Mongol army. Chuang zi nu: the giant crossbow The chuang zi nu was one of the most dominant siege weapons in the Mongolian army during this period. Its construction and operation were complicated and among the Chinese experts recruited by Hülegü were men skilled in the assembly and operation of those giant bows.60 The structure of this siege machine was similar to that of the crossbow carried by infantry archers. The word chuang zi nu means literally “a giant crossbow shaped like a bed.” It was made by using one, two or three giant crossbows connected to each other and mounted on a cart so that the machine could be maneuvered with ease (Figure 2.4). The crossbows were constructed from wood and bone, giving them flexibility and strength. The cord was drawn with a winch and the arrows were similar in size to spears and more than one arrow could be fired each time.61

62  Muslim Fortresses in the Levant

Figure 2.4  The giant crossbow

A bow of this type was used during Hülegü’s siege of the Assassins’ fortress at Maymundiz.62 Apart from arrows the size of spears, it fired burning torches that according to Juwaynī caused many casualties.63 Many parallels can be drawn between the various types of siege machines used in the Middle East and in Northern China. There are however a number of important differences, the main one being that all siege machines in China were propelled by manpower (see Table 2.2). The Middle East had clearly made an advance, and besides the manually operated siege machines, it was using the counterweight trebuchet, which was more powerful and capable of firing heavier stone projectiles to a greater distance. It also required a considerably smaller team. The ropes pulled by men were replaced by a set of pulleys and a wooden box filled with stones that acted as a weight.64 The disadvantages of this new siege technology were that its size and weight meant it could not be maneuvered with ease, not even along short distances, and it had to be built and assembled by a professional team of carpenters and engineers.65 The Chinese on the other hand had the multiple armed siege machines that were not known to the armies of the Middle East even after their encounter with the Mongol army. The size and firepower of those siege machines must have been impressive if one takes into account that each machine could fire between five and seven stones at a time. There is also a great difference in the type of projectiles used, since clay catapult balls were unheard of amongst Franks, Ayyubids and Mamluks. Taking into account that a team of forty was needed for the operation of the smallest Chinese siege machine, this may explain the great number of a thousand skilled men that Hülegü recruited from Northern China for his siege units. The use of incendiaries was common to both Middle Eastern and Northern Chinese armies, though it seems that they were more effective in Chinese and Central Asian warfare since wood, cane, and thatched roofs, all highly inflammable, were widespread building materials. The clay pots known as nafṭ have been found in a

Mongolian siege warfare and the defense of Mamluk fortresses  63 Table 2.2  The chuang zi nu (large mounted crossbows) Model

Number of men per team

Distance

Two bows Three bows Three bows1

  4–7 20–30 70

243–260m 347–520m 347–520m

Source: Y. Hong, ed., Weapons in Ancient China (New York and Beijing, 1992), 263. The data are taken from the book Wu Jing Zong Yao (‘The collection of the most important military techniques’), written by the order of the government of the northern Sung. 1 The difference is probably in the type and size of the arrow fired.

number of archaeological sites on land and at sea though they were more successful in naval warfare where wooden ships and sails could easily catch fire.66 Apart from the differences in siege machines employed by the armies, there was a vast difference in the way the Northern Chinese siege units in the Mongol army utilized the topography and natural surroundings of the area round a besieged city. This is one of the most impressive characteristics of siege warfare under the Mongol commanders. Rivers were diverted, dams built and mud ramparts or brick walls were constructed in order to seal the city and cut it off from its surroundings. These large earthen works show an outstanding degree of thought and initiative on the part of well-trained engineering teams and skilled craftsmen, capable of planning and operating such largescale projects. The Frankish, Ayyubid and Mamluk siege teams dug sapping tunnels and worked a variety of different siege machines, but they never attempted to change the course of rivers and/or the area near the besieged city. The main reason was lack of human resources: without a huge labor force such ideas could not be carried out. There can be no doubt that Eastern Mediterranean siege machines were more advanced. The counterweight trebuchet could hurl stones that weighed 95kg and more to a distance of 300m. On the other hand, the strength of the siege units employed by the Mongols apparently did not depend on superior siege technology or on changing the topography round the besieged city or fortress, but lay rather in the ability to recruit and position mass numbers of siege machines and teams that could run the entire siege operation day and night without halting. The siege of Nishapur (1221) is a good example. The force sent to besiege the city numbered 10,000 men under the command of Toghachar, a son-in-law of Chinggis Khan. The garrison was assisted by the entire population. Juvaini mentions mangonels, naphtha and crossbows used by the defenders. The city’s defense held for six months. It was only when Tolui arrived in April of 1221 with a significantly larger force that Nishapur was taken. The walls were breached within three days.67 Juwaynī describes the effort Tolui’s teams put into gathering stones for the siege machines: “although Nishapur is in a stony region they loaded stones at a distance of several stages and brought them with them. These they piled up in heaps like a harvest, and not a tenth part of them were used.”68 D’Ohsson and Martin describe the siege machines at Nishapur, giving exact numbers. The Mongols placed 3,000 giant crossbows, 300 stone-hurling siege machines, 700 machines that hurled pots of nafṭ, 4,000 ladders and 2,500 piles of stones brought from the nearby mountains.69

64  Muslim Fortresses in the Levant The assault continued day and night. Assuming this source is reliable, even if the figures are exaggerated they give an idea of the immense strength of the siege contingents that fought in the service of the Mongol armies. The source of manpower available to the Īlkhānid state differed from that of the Mongol empire.70 It is rather doubtful that Hülegü’s heirs continued to recruit men from Northern China for their siege contingents. This raises a number of questions concerning the composition, number and origin of the men in the Īlkhānid siege units. The chroniclers provide very little information. On the strength of previous Mongol policy it is more than likely that Muslim engineers were recruited and local siege technology was used along with that brought over from Northern China. But any assumption is bound to remain so for lack of firm evidence. However, it is possible to overcome this obstacle by analyzing the sieges conducted under Hülegü’s command and comparing them to those carried out in later years by his successors. This may enable us to draw some conclusions concerning the development and changes of siege warfare that occurred in the Īlkhānid army.

Sieges conducted by Hülegü while campaigning to the west and after establishing the Īlkhānid state (1256–65) A survey of the sieges conducted during this period displays a set “Mongol method” composed of a number of stages repeated with very slight variations. Before beginning a siege, Hülegü would send an emissary with a clear message to the governor or the ruler of the city or the fortress; the message conveyed a simple and straightforward offer − surrender and avoid a military confrontation.71 The next stage was determined according to the answer. But in fact the only reply that would prevent a siege was surrender, complete submission and acknowledgment of Mongol sovereignty. Once it was decided to proceed and besiege the site, the first days were devoted to felling trees, building siege machines, collecting catapult stones, surveying the besieged site, searching for its faults and weaknesses. In addition points to position the siege machines were looked for, so they would gain every advantage of height or a well-protected area that would ensure the teams’ safety. Before the siege of Maymundiz began, Hülegü surveyed the area and examined the fortress gates.72 He then ordered his best men, chosen from his private guard, to carry logs for the construction of siege machines to the nearby summits that surrounded the fortress. In some sieges the Mongols dug a moat or built a wall or an earth rampart around the besieged city at quite an early stage. During the entire period of organization the army was exposed to the enemy’s attacks and was in a highly vulnerable position. In most cases the besieged force had prepared itself well in advance, and at this early stage it still had plentiful supplies of food, water and weapons; in addition, the morale of the population and garrison was still usually fairly high. Assaults were launched in this phase using both siege machines and small contingents of mounted men who raided the Mongol camp and quickly returned to the safety of the walled city. An example of such behavior can be seen during the siege of Mayyāfarqīn (656/1258–657/1259). At first, the morale at Mayyāfarqīn was high; a number of raids were carried out on the Mongol siege camp, which caused severe losses. But when food supplies became short

Mongolian siege warfare and the defense of Mamluk fortresses  65 and famine began to spread within the city the raids ceased.73 The siege of Gird Kūh (spring 651/1253), carried out by the Mongol vanguard commanded by Ketbugha, began by digging a deep moat and building a high wall round the siege camp. This provided sufficient protection from enemy raids.74 Meanwhile Ketbugha organized his men round his camp in a pattern known as järgä,75 a formation borrowed from the hunting methods used on the plains, but to no avail: the Mongolian forces suffered severe losses, and the fortress remained steadfast. The army eventually retreated and Gird Kūh was taken only eighteen years later during the reign of Abagha. When a siege camp was eventually organized both sides employed mainly siege machines, and teams were worked night and day. In the siege of Baghdad an iron ram was used, but this was a rather rare instance. One of the most informative narratives of the siege of Baghdad is given by Waṣṣāf, who in one short sentence describes everything that flew through the sky between the two camps: catapult stones, arrows, spears and javelins.76 It seems more than likely that the mounted archers, who were the most dominant force in the Mongol army, dismounted and fired from the ground. This would have increased the number of fighters considerably and provided both a stronger attack and better fire cover for the teams operating the siege machines. On a few occasions the Mongols tried to surround and seal off the city or the fort in order to prevent the population from escaping and food and supplies from entering. The best example of construction work carried out by the Mongols was probably that used at the siege of Baghdad. After a few weeks of siege the Mongols decided on a change of tactics. This entailed a great investment of time and work, since it included the building of towers that rose above the city walls; once they were completed siege machines were mounted on them, thus gaining the advantage of height. The towers were built of mud bricks, collected from the local town industry, which was positioned outside the city walls. In addition, the water tunnels leading into the city were blocked so that no one could escape.77 Another example of this tactic of surrounding a besieged city can be seen in the siege of Aleppo in December 657/1259. At the outset of the siege, a large earthen rampart known as a chappar was built around the city. While it was being built teams of sappers tried to undermine the foundations of the city wall.78 When a siege lasted longer than expected, and a city did not show any signs of surrendering, its fortifications remaining solid, the siege was often raised and the army left. In a few cases the Mongols returned and renewed the siege at a later stage. This was done in Arbela (Irbil). The attack commanded by the Mongolian general Uruqtu Noyan took the city, but the citadel remained strong and its garrison refused to surrender. The siege commenced at the end of the winter; the Mongols abandoned the siege and returned in the summer, when the unbearable heat eventually brought about the surrender of the garrison.79 The fate of a city or a fortress garrison depended on whether the city yielded promptly or resisted and the duration of its resistance. Cities that surrendered were usually spared. Large-scale massacres were carried out in places that refused to submit. In a number of cases the massacres were followed by orders to rebuild the city. After the surrender of Baghdad Hülegü appointed Arab and Persian officials to oversee the reconstruction and administration of the city. In many cases, however, the sacked

66  Muslim Fortresses in the Levant cities or strongholds were left in ruins. In some the destruction was so severe, that recovery was long and slow.80 The impression one is left with is that the majority of the siege teams remained with the core of the Mongolian army under Hülegü’s command, and that they were reserved for cities and strongholds that were strategically important or of great symbolic significance, such as the Assassin strongholds, or Baghdad, the seat of the Caliph,81 or the fortress of al-Bīra that sat on a ford on the upper part of the Euphrates. Cities and strongholds of secondary importance were besieged by forces recruited from the local population and a small body of siege experts who were probably sent from the main Mongol force. After Hülegü’s son Yoshmut encountered difficulties during the siege of Mayyā­ farqīn he was assisted by the two sons of Badr al-Dīn Lu’lu’, the ruler of Mosul. An experienced siege engineer accompanied them. The town eventually fell once its food supplies had been consumed and famine struck the population and garrison.82 The same combination of local forces and a Mongolian contingent besieged Arbela and Mārdīn. Those examples leave little doubt that shortly after the Mongols’ arrival in the Middle East they were exposed to siege machines such as the counterweight trebuchet that was in use throughout the region. Hülegü may have recruited local siege experts, as Mongol rulers had done since the reign of Chinggis Khan, but none of the sources explicitly say so. The only information we have is that of the recruitment of a siege expert from among Badr al-Dīn Lu’lu’s men who fought alongside Hülegü’s force in the siege of Mayyāfarqīn. In contrast to battles waged in the open field the element of surprise was completely absent from the Mongolian sieges. All the various stages were well prepared in advance. Urban strongholds were occasionally restored by the Mongols. Local commanders were selected to govern places that were strategically important.83 But on the whole Mongol fighting methods did not depend or rely on fortresses or fortified cities.84 Unlike the Mamluks, we do not know of fortresses built or reconstructed by the Mongols in Syria or along the Euphrates. Their policy was, moreover, very different from that of the Armenians and the Franks, who based their rule and defense on fortresses and fortified towns. This is best described by Smail: “In the twelfth century, as well as the thirteenth, the walled towns were the ultimate expression of Latin authority in the region.”85 The idea of a “central place,” a fortress or a fortified city that serves as an administrative, financial, religious centre and also provides protection for the local population,86 was not foreign to the Mongols, but it did not fit their perception of the defense of their own territories.

The Īlkhānid sieges of the fortresses of al-Bīra and al-Raḥba The sieges conducted after the establishment of the Īlkhānid state were mainly carried out against the two frontier fortresses on the Euphrates, al-Bīra and al-Raḥba (Map 2.2). The obvious Īlkhānid targets and the constant Mamluk state of alert left little room for surprise attacks. During Hülegü’s reign al-Bīra was besieged twice. While

Map 2.2  Īlkhānid invasion routes into Syria

68  Muslim Fortresses in the Levant the first attempt was certainly a success the second was a great failure. This raises some questions as to the siege units and the conduct of the whole affair; the more so since the area and the immediate surroundings of the fortress must have been well known by then to the Īlkhānid army. Al-Bīra is situated on a natural ford along the northern section of the upper Euphrates,87 approximately 60 km west of the modern city of Urfa (medieval Edessa). The Mongols considered it as the gate to Syria, and for that same reason it became a key stronghold in the Mamluk defense. The first Mongol siege of al-Bīra was carried out in Dhu al-Hijja 657/December 1259. The garrison fought as best as it could but the Mongol army outnumbered it and the fortress fell within two weeks.88 Three years later (659/1262) it was recaptured by al-Barlī (a Mamluk of the late Ayyubid prince al-ʿAziz Muḥammad). Baybars rebuilt al-Bīra and granted it special status because of its strategic importance. Although the fortress was well-built, consistently maintained and the garrison reinforced, its ability to withstand the enemy depended on the ability of the Mamluk field army to arrive within days or a couple of weeks at the most.89 Al-Bīra was besieged for the second time approximately a month before Hülegü’s death. By the winter of 663/1264–5 the military balance in the region had changed considerably. The Mongol defeat at ʿAyn Jālūt did not change Mamluk perceptions of the Mongols, but it no doubt broke the Mongol military image of an invincible army. The stability achieved within the Sultanate during the early 1260s allowed Baybars to organize the Mamluk frontier and prepare it for the next Mongol move. The strengthened Mamluk garrison and the well-organized relief forces from the Syrian cities considerably reduced the Īlkhānids’ chances of recapturing the fortress. Between December 1264 and January 1265 the Īlkhānid army erected seventeen siege machines round the fortress of al-Bīra. A section of the moat was filled with wood in order to bring the siege machines closer to the wall.90 The Mamluk garrison was quick to act and the Mongol operation was foiled by the wood being set on fire. The Īlkhānids’ second attempt to scale the walls failed and in mid February they abandoned the site after hearing that a Mamluk relief force was on its way from Cairo. Baybars ordered repairs to be taken care of by his own troops.91 The fortress was restored, weapon and food supplies were renewed and the garrison reinforced; thus the fortress was made ready for the next Īlkhānid assault. The methods of siege warfare conducted under Hülegü’s command and those of his successor seem much the same. The main difference between the two sieges, separated by five years, was in the size of the army. During the first siege Hülegü arrived with the greater part of the Mongolian army. It seems likely that in December 1264 the force that besieged al-Bīra was substantially smaller. The Īlkhānid force that besieged al-Bīra in 1264–5 under Durbai’s command had only one tümen – 10,000 men, at the most. An Armenian force was supposed to join them, but when they heard that the Īlkhānids were retreating they decided to turn back,92and thus never participated in the battle at all.93 The difference in the size of the armies is of considerable importance especially since both fortress and garrison had been strengthened and reinforcements could reach al-Bīra given a relatively short warning.

Mongolian siege warfare and the defense of Mamluk fortresses  69 The third and fourth sieges of the fortresses occurred simultaneously in the winter of 671/1272 during Abagha’s reign (1265–82). While half the force marched towards al-Raḥba the other besieged al-Bīra. The Mamluks followed suit. The Syrian force led by Fakhr al-Dīn al-Ḥimṣī and Alā’ al-Dīn al-Hājj was sent to al-Bīra while Baybars hurried to al-Raḥba. The Īlkhānid army that camped in front of al-Bīra was composed of 3,000 Mongols and approximately 3,000 Seljuks from Rum. Siege machines of different types were positioned at the foot of the fortress. According to Waṣṣāf Baybars had promised the garrison at al-Bīra that reinforcement would arrive within a week. Nevertheless he added that if they were delayed the besieged may surrender and abandon the fortress.94 There was no need for this alternative plan, since the Īlkhānid army was defeated after engaging the Mamluk relief force in close combat, leaving behind its siege machines. Although we lack information concerning the number and type of siege machines we know the strength of the Īlkhānid force. The size of the army had dropped significantly, once again, and it numbered 6,000 men of whom only half were Mongolian. The siege of al-Raḥba, conducted simultaneously with that of al-Bīra, was carried out by an Īlkhānid army of a similar size. With the arrival of Baybars the siege was abandoned, and the Īlkhānids retreated to the eastern bank of the river where they built a fortified camp. This turned out to be a grave mistake, mainly due to the superior numbers of the Mamluk army. Baybars was determined to attack this Īlkhānid camp. The Mamluks began to cross the river; the Īlkhānids quickly discovered them and began to attack while the Mamluks were still crossing. If it was not for the numerical superiority of the Mamluk army, Baybars might have decided to retreat. The Īlkhānid army was defeated, suffering heavy losses. The numbers of the Īlkhānid army, at this siege, have been estimated at three to six thousand.95 All four attempts to take al-Raḥba and al-Bīra with a relatively small force failed. This may have taught the Īlkhānids an important lesson, since future sieges were conducted with a considerably larger army. The fifth attempt to take al-Bīra dates to the winter of 674/1275. The Īlkhānid army led by Abtai numbered 30,000 men; half are said to have been Seljuks and soldiers recruited from the nearby cities of what is today modern Iraq. The siege units must have been well manned and organized, as we are informed that 70 siege engines were built. Yūnīnī mentions only 23 Frankish siege machines, while Ibn al-Furāt states there were only 11. Although the army was considerably larger, the number of siege machines possibly higher than ever before, and the Mamluk army was delayed, the Īlkhānid force lifted the siege within nine days. The garrison at al-Bīra stood fast and managed to hold their ground without reinforcement.96 The fierce cold and signs of treason in the Seljuk force are among the main reasons given for the Īlkhānid failure. The siege machines were set on fire and the army retreated.97 After the fifth Īlkhānid rout a noticeable change occurred in their strategy. From this point on invasion routes bypassed the two Mamluk fortresses on the Euphrates. In addition, once the Īlkhānid forces entered Syria they maneuvered around the strongholds and fortified cities of Aleppo, Hama and Homs, making great efforts to avoid siege warfare. The routes they chose speak for themselves and are a quiet recognition of the limitations and flaws of their siege units. Whether because their siege technology and methods ceased to advance or because they were less experienced and

70  Muslim Fortresses in the Levant numerically fewer than Hülegü’s siege contingents, there was a clear attempt to move the fighting to the battlefield and lure the Mamluks into close combat. The bypassing routes The first attempt to take roads that bypassed al-Raḥba and al-Bīra was made towards the end of Abagha’s reign. In Jumda II 679/October 1280 the Īlkhānid force invaded Syria from the north (see Map 2.2), and took ʿAyn-Tāb, Darbassāk and Baghrās from the Mamluks.98 The two latter fortresses guard the Belen mountain pass that commands the main road crossing Amanus range and running down to Syria. Meanwhile a second part of the Īlkhānid army moved from the east. A third section joined them and they all gathered at Aleppo. The local population and the Mamluk garrison at Aleppo fled. This may allow us to assume that the Mamluk fortresses mentioned above were also abandoned a short time before the arrival of the Īlkhānid army. The flight of the Mamluk garrisons and the city population cleared the way for the Īlkhānid army and if it did encounter any resistance this was nominal. The retreating Mamluk forces halted at Damascus, where they reorganized their lines and returned to fight the Īlkhānid army pursuing it back across the river and further north to ʿAyn-Tāb.99 A year later (680/1281) Abagha decided to invade Syria, using almost the same route. He sent a vast army (estimated between forty and eighty thousand men) under the command of Mengü-Temür.100 The invasion ended with the crushing defeat of the Īlkhānid army at the battle of Homs.101 Ghazan handled the Mamluk front in a way similar to that of his predecessor, and spent much of 1299 to 1303 in trying to conquer Syria. He invaded Syria three times within a space of four years, and but for his death there would have been a fourth attempt. In his renewed efforts he chose a new crossing point for his army. In 699/1299–1300 the Īlkhānid forces crossed the Euphrates near Qalʿat Jaʿbar, and from there marched for five days to reach Aleppo.102 The army maneuvered around the city, avoiding a siege, and continued south, bypassing Hama and luring the Mamluk army into an open battle field near Wādī Khaznadār, where it was defeated.103 The city of Hama surrendered immediately after the battle and the Īlkhānid army marched south into Damascus. Within a few weeks the entire army left with Ghazan, a small contingent remaining behind in Damascus. The only siege conducted during this campaign was that against the citadel of Damascus, which refused to capitulate. The Īlkhānid army set up a battery of siege machines before Bāb al-Barīd, as well as in the famous Damascus mosque. The area was sealed off so that no one would be able to sabotage the siege engines. Although an Īlkhānid garrison remained in the city it was busy looting. The Mamluks took advantage of the chaos and sent out small units from the citadel to attack the Īlkhānid force. A few weeks later the Īlkhānids retreated, as it became increasingly difficult to fight the citadel garrison, continue the siege and hold on to the city.104 A year later, in 700/1300–1, Ghazan invaded Syria for the second time. The Īlkhānid army marched along the same route and reached Sarmīn, where the campaign ended due to severe rains. Thus the second invasion came to a halt.105

Mongolian siege warfare and the defense of Mamluk fortresses  71 For his third campaign (702/1303) Ghazan chose a very different course. The army, which numbered one tümen, crossed the Euphrates at al-Ḥilla east of Karbalā. It is easy to justify this choice of a ford as it was not defended by the Mamluks and was far from any of their possible sources of intelligence. Though there was no danger of the army suffering from lack of water, since it would march along the river, the time and distance this journey consumed was considerable. Once the army had crossed, Ghazan turned north and marched to al-Raḥba which surrendered within a few days. According to Rashīd al-Dīn, Ghazan sent an envoy, headed by Rashīd al-Dīn himself, to conduct the negotiations with the local population, which eventually decided to surrender.106 Thus there was no need for a siege. Ghazan’s next step is somewhat surprising, for he decided to end the campaign at this point. It is difficult to explain this move. Rashīd al-Dīn says that the hot weather and the danger of seasonable flooding caused Ghazan to abandon the original plan. The Armenian historian, Hetʿum, gives a different account. Hetʿum claims that Qaidu (ruler of Central Asia, a member of the house of Ögedei, d. 1301) threatened to invade the eastern frontier of the Īlkhānid state,107 which caused Ghazan a quick change of plan. The Īlkhānid force that moved parallel to that of Ghazan entered Syria from the north, reached Homs, and continued to Damascus where it was defeated by the Mamluk army at Marj al-Suffar in Ramaḍān 702/April 1303.108 All the invasions described above were made with the idea of confronting the Mamluks on an open battle field. There is no evidence in the sources that siege units joined the army or that the Īlkhānid force was equipped with siege engines. Ghazan was determined not to engage in siege warfare. He by-passed the large Syrian cities and tried to avoid conflict with them. The last Īlkhānid attempt to invade Syria took place in 712/1312–13 during the reign of Öljeitü. The army under the command of the Īlkhān left the new capital of Sulṭāniyya, turned towards Mosul, crossed the Euphrates at Qirqsiya (ancient Carchemish) and marched south to al-Raḥba.109 Abū ’I-Fidā’ gives different details. According to him the Īlkhānid army passed through Sīs, meaning that they came from the north. It is possible that the army moved in two heads as it had done in the past.110 Öljeitü, influenced by Qara-Sunqur and Aqqush al-Afram (who defected to the Īlkhānid side shortly before the campaign), was convinced that al-Raḥba would surrender without a siege. This notion was soon proved mistaken. The town put up a fierce defense, and though the Īlkhānid army set siege machines and hurled nafṭ its losses were severe. Īlkhānid supplies were far from adequate and after three weeks a shortage of food was beginning to be felt.111 Rashīd al-Dīn, who was present, advised the Īlkhān to answer the request of the besieged and provide them with an amān (guarantee of safety). The whole event seems like an effort to avoid humiliation. It is more than likely that the attempt to persuade the Mamluks to surrender was made to save face and to enable the Īlkhānid army to retreat without it appearing to have suffered a severe defeat. Approximately a month after the siege had begun, Ramadan 712/January 1313, the Īlkhānid force crossed the Euphrates and headed back east.112

72  Muslim Fortresses in the Levant Table 2.3 summarizes the attempts to conquer the Mamluk fortresses on the Euphrates.The most striking point that emerges from this table is the decrease in the size of the army in comparison with the force under Hülegü before the establishment of the Īlkhānid state. At first it seems that the cause of these failures was the small size of the forces deployed, yet even when the numbers were considerably increased the Īlkhānid army could not subdue those fortresses. The claim that the Īlkhānid army was occupied with matters on the frontiers with Chaghatai and the Golden Horde does not fully explain the matter. The failure was clearly not only due to the size of the army. Although we do not have all the details, and historical accounts are lacking, it seems that part of the reason was the deterioration of the Īlkhānid siege units. None of the large-scale engineering works that were carried out during the campaigns of Chinggis Khan or on Hülegü’s journey to the Middle East were attempted by the Īlkhānid army. It is important to note that, although both fortresses were modest in size, their location made it impossible to flood or mine their walls and towers. Even if an original engineering idea had been implemented, such projects demanded a great deal of labor, hundreds if not thousands of workers (in addition to the fighting men), as well as advanced engineering knowledge and suitable teams of artisans. The number of men allocated to the campaigns was relatively small and hardly sufficient for carrying out large-scale projects that would have included shifting quantities of earth. Even when the army was bigger there is no sign of any new initiative in the conduct of siege, no new ideas or techniques. The fact that the Mamluk relief force was able to arrive within two weeks prevented the Īlkhānids from carrying out large-scale engineering projects since they often required weeks and sometimes even months of preparation. Even the simplest method of siege warfare – surrounding the site and starving the garrison and the local population until they surrendered – was not possible due to the short time that passed until the arrival of the Mamluk reinforcement. Table 2.3  The sieges of al-Raḥba and al-Bīra Fortress

Year

Size of army

Number of siege machines

Results

1  Al-Bīra

1259

Up to 70,000

Unknown

Fortress taken within two weeks

2  Al-Bīra

1264–5

1 tümen =10,000

17

The siege failed

3  Al-Bīra

1272

6,000; half were Seljuks from Rum

Unknown

The siege failed

4  Al-Raḥba

1272

3,000–5,000

Unknown

The siege failed

5  Al-Bīra

1275

30,000; only half were Mongol

70

The siege failed

6  Al-Raḥba

1303

1 tümen = 10,000

Unknown

The Mamluk force surrendered within days

7  Al-Raḥba

1312–13 Unknown

Unknown

The siege was raised after a month.

Mongolian siege warfare and the defense of Mamluk fortresses  73 This starvation method had been successful in the siege of al-Raḥba in 496/1103, when Shams al-Mulūk Duqāq, ruler of Damascus, and Ẓahīr al-Dīn Tughtigin Atabeg (who was married to Duqāq’s mother) prevented food supplies from entering the besieged city. The population surrendered after a long struggle that lasted six months until March of that year.113 In 500/1107 al-Raḥba was besieged by Jāwalī, a general of the Seljuk sultan Muḥammad. The siege lasted eighty seven days. The city walls were breached and once the city had fallen the citadel was taken within five days.114 Although these sieges took place at the beginning of the twelfth century they still give a clear idea of how long one would need in order to starve out a city. If the city was well stocked, supplies could last as long as six months. The Īlkhānids’ ambition to conquer the Mamluk sultanate was genuine. It was driven by the Mongol imperial ideology: the belief that they were given the mandate by the heavens to conquer and rule the world. This ideology was maintained by the Īlkhānids.115 The efforts invested in these military campaigns were by no means negligible. Nevertheless, as shown above, there were two main difficulties that prevented them from conquering Syria, shortage of time and their relatively small forces. It is important to note that the Mamluks were not conceived as a threat to Īlkhānid existence. Apart from a number of raids launched by the Mamluks beyond the Euphrates, in all the phases of the conflict they never sought to conquer Īlkhānid territory.116 It thus seems the Īlkhānids did not fear the Mamluks; they never stationed troops along the Euphrates and there is no information of fortresses manned by their army. The shortage of time was a problem the Īlkhānid army could not overcome. The changes in strategy that occurred during the reign of Abagha in 680/1281 were the result of the decision to avoid siege warfare, which required both time and large forces. The new strategy emphasized the use of mounted archers, who had always been the core of Mongol military strength. The Mamluk strategy along the Euphrates was carefully organized and clearly more intricate than any other contemporary border or frontier. They depended on their fortresses along the Euphrates and invested considerable time and funds in organizing their manpower. If the fortress garrisons together with the larger Mamluk relief forces failed to sound the alert and hold back the Īlkhānid army, the villages and cities of northern Syria would suffer severely.

The defense of Mamluk fortresses Bearing in mind that the Mongols arrived with no new siege technology, any improvements or changes in Mamluk military architecture were the outcome of the composition, the size, or the particular methods of defense used by the Mamluk garrisons. The following pages will examine Mamluk methods of defense from within the fortress. Compared to siege warfare, where there appears to be more of an opportunity to exercise and develop innovative ideas, the possibilities of defense are rather limited. A careful preparation for a lengthy siege was crucial. The garrison and the entire town or village population gathered and stocked food, water and weapons. The area round the fortification was stripped bare. Water sources were sealed or contaminated

74  Muslim Fortresses in the Levant and pasture that could support the besieging enemy camp was set on fire. Buildings that might provide shelter were removed so that the area below the walls was left exposed. When Ghazan was besieging the citadel of Damascus (Rajab-Jumādā II 699/ February–March 1300), the amir commanding the Mamluk garrison of the citadel, Sayf al-Dīn Arjawāsh, ordered all the buildings below its walls to be burnt down.117 Great emphasis is laid on the importance of having your city fortifications well-prepared even in times of peace, when one least expects the enemy to attack. Although it dates to the late fourteenth century and is considered a second-rate war manual, the Tafrīj al-kurūb fī tadbīr al-ḥurūb by ʿUmar Ibn Ibrāhīm al-Awasī al-Anṣārī (d. 811/1408) provides important information concerning the general policy sultans should follow as regards the fortification of their capital cities and fortresses throughout the Sultanate.118 It gives detailed instructions on the duties of fortress commanders in times of war and in times of peace, as well as the preparations to be carried out prior to a siege and while the siege is going on.119 Some of the instructions are rather obvious and at first sight appear slightly banal. But the very fact that they were written as a set of rules within a few years of Temür’s (Tamerlane) invasion120 implies that during Barqūq’s reign (792/1390–801/1399) old rules and regulations concerning the upkeep and running of fortresses were no longer being followed. The author, ʿUmar Ibn Ibrāhīm, was by no account an active member of the Mamluk army. Scanlon describes the manual in the following words: “The work reeks of the library and of the court, rather than of the camp and the battlefield. Its tone is that of the observer rather than a participant.”121 And yet the information was gathered from high-ranking Mamluk amirs who had served in Syria. There seems to be little consensus among contemporary Muslim chroniclers. While Maqrīzī does not think much of him or of his work, Ibn Taghrī-Birdī says he was a respected member of the court.122 ʿ Umar Ibn Ibrāhīm describes how the commander of a fortress should keep it in a constant state of alert (mutayaqqiẓ, ).123 The fortress must be strengthened, and each man should be assigned to a particular position within the defense. The commander must be certain he has a sufficient number of men and weapons, including siege machines. But above all he must see to it that the water and food supplies are well stocked. In addition, knowing the composition of the enemy’s army is essential in order to prepare accordingly.124 The composition and size of Mamluk garrisons Fortress garrisons were constituted on similar lines to that of the Mamluk field army. They included infantry archers who probably used both crossbows and composite bows, teams that operated siege machines, and mounted men and infantry who could carry out raids and skirmishes against the enemy camped outside the walls. The main difference was in the hierarchical importance and size of each group. While mounted archers dominated on the battlefield, the importance of infantry archers and siege machine teams was evidently greater in a fortress garrison. The number of men a

Mongolian siege warfare and the defense of Mamluk fortresses  75 fortress could accommodate ran from a few tens (Qāqūn) to as many as two thousand (Safad). Large urban citadels could maintain additional forces within the city quarters. But even in large cities the balance of power tilted in favor of the besieging army. Neither side could usually withstand a lengthy siege. While the besiegers had the constant anxiety of dwindling supplies and intercepted retreat routes, the garrison’s fears were much the same, for they knew their supplies would be exhausted and their manpower would lessen due to battle injury, exhaustion and death. Mamluk strategy was based on a fairly simple scheme – the ability of the garrisons along the frontier to withstand a short siege until the arrival of Mamluk reinforcement. The next stage was to engage the Īlkhānid force on two fronts: that of the garrison and that of the relieving forces. At this point it will be useful to briefly survey the Mamluk garrisons along the Euphrates, which, notwithstanding their modest size, played a crucial part in the defense of the frontier. Garrisons were recruited and composed in several different ways. The method of recruitment is of special interest because of the structure and hierarchy of the Mamluk army. The vast majority of the sultan’s Mamluks (who had a superior social and economic status) rejected the idea of serving in remote posts along the frontier far from the center of government in Cairo. Stationing a Mamluk in Syria was rather like a form of severe punishment, similar to exile. A large percentage of Mamluks serving in Syrian citadels and fortress garrisons consisted of Mamluks owned by local amirs and “free men.” Their military and social rank was lower than that of the sultan’s Mamluks. In reality much of the responsibility for defense of the north frontier fell on the Syrian contingents and garrisons, since many of the battles and sieges took place on Syrian soil. In consequence, a high proportion of the Syrian contingents was made up of infantry and siege experts.125 Since the infantry would have slowed down the army on the march from Egypt, there was an advantage in having infantry contingents based in various Syrian cities.126 Fortress and citadel commanders (nuwwāb) were appointed directly by the sultan and were subject to him alone. The sultan’s involvement did not end with choosing his nuwwāb and in many cases he participated in selecting the garrison’s soldiers. If the conquered fortress had a well-trained garrison that could be trusted to transfer its loyalties to the new Mamluk sultan, it was maintained. Occasionally it was strengthened. A good example is the organization of Karak by Baybars after it surrendered in 661/1263. The Ayyubid garrison was kept in place and paid from the sultan’s treasury. Their amirs received a rise in pay. In addition, Baybars sent a number of his own Mamluk archers from the Baḥriyya to join the garrison on a permanent basis. A similar procedure took place in the fortress of Kahybar, which was reinforced with Mamluk soldiers from the Baḥriyya. Ayalon raised some question concerning the composition and quality of the Baḥriyya who were chosen to serve in the Syrian fortresses during the reign of Baybars and that of his successors. According to Ayalon by the early 1260s their rank had thinned and many of the regiment’s men were fairly old.127 After the siege of Safad (1266) Baybars appointed a commander of the fortress, a second-in-command and an arsenal supervisor, who immediately saw to the collecting of arrows and siege machines left over in the Mamluk camp. Mamluk

76  Muslim Fortresses in the Levant soldiers were brought over from Damascus to strengthen the Safad garrison. Maintaining the garrison cost 80,000 dirham per month. The land in the vicinity of Safad, once owned by the Franks, was transferred to the amirs but there is no further information concerning the daily expenses of the fortress, i.e. if they were paid through local taxation or by the sultan.128 If there was no core from which a garrison could be built, a new garrison was established. At the end of the siege of Marqab (684/1285) Qalāwūn constructed an entirely new garrison that included 1,000 infantry archers and troops (jarjya, ), 400 craftsmen, 150 amirs, infantry soldiers who were probably armed with javelins, swords and mace heads and 550 Mamluks from the Saliḥiyya, the Manṣuriyya and the Baḥriyya. The entire battery of siege machines was moved from the Mamluk siege camp to the fortress, together with all the pots of naphtha and wooden logs remaining from the siege. Garrison salaries, the costs of rebuilding the fortress and daily expenses were paid out of taxes collected from the local population.129 Along the coastal plain, in addition to the small garrison at Qāqūn, Türkmen tribes were settled in the vicinity by Baybars who entrusted the region’s safety and defense to them.130 In more remote places the majority of the garrison was recruited from the local population,131 occasionally reinforced by a small number of Mamluk soldiers from the sultan’s army or from the Mamluk contingents belonging to one of the local Syrian amirs. The Mamluk incursion into Cilicia during the summer of 697/1298 ended with the conquest of three Armenian fortresses: Tall Ḥamdūn, Marʿash and Nujayma. The recruitment of garrisons for all three was entrusted to the governor of Damascus. Soldiers of all ranks and specialties were recruited by order of the governor, who organized the garrisons and saw that they were properly armed and taken care of.132 During especially tense periods garrisons were temporarily reinforced, as was done by al-Nāṣir Muḥammad when he went on pilgrimage to Mecca during the winter of 713/1313.133 The garrison’s size was not dictated by the strategic importance of the site. This peculiar situation probably resulted from the fact that most of the fortresses were not built by the Mamluks themselves and there was not always a correlation between the initial size of the fortress, the military or political changes that occurred in the region and the needs of the Sultanate. Under Frankish rule Marqab served as an administrative center for a fairly large regional population.134 When it was taken by Qalāwūn, he composed a garrison that numbered 2,100 men, although the fortress was not located on or near the frontier and may have even lost its importance as an administrative center. We have no information about the size or composition of the fortress garrisons at al-Raḥba and al-Bīra, which were often besieged by Īlkhānid armies. In comparison to Safad or Ḥiṣn al-Akrād the two Euphrates fortresses were fairly modest. Although the fortresses were well built and planned, with several levels, underground storage and vaults, it is doubtful whether they could accommodate more than a few hundred men each. It is possible that the local population was counted upon, and was expected to take an active part in the defense; though none of the sources explicitly says so. The only mention of the role of townspeople in the defense of al-Raḥba is during the siege

Mongolian siege warfare and the defense of Mamluk fortresses  77 of 1313. Once the besieging army retreated, the locals went out and collected all the siege machines left in the Īlkhānid camp.135 The finest example we have of local participation in the defense of a city comes from Damascus. In 699/1300, after the Īlkhānid army had left, the guarding of the city walls was entrusted to the townsmen.136 Once the city had recovered, the Mamluk governor of the citadel decided to carry out important reforms in order to improve the local population’s fighting skills. Targets were placed in the madrasas and mosques, and men were ordered to practice archery. Shopkeepers were instructed to hang their weapons inside their shops. In addition, each of the city markets was assigned an amir who organized the local civilian population.137 Long-range weapons: bows and siege machines and their role in   fortress defenses The garrison’s advantages were mainly its knowledge of and familiarity with the surrounding area, the protection provided by the fortifications, and the fact that it was usually positioned on a higher plane than the enemy. That the besieged have many advantages over the besiegers was pointed out by the Prussian general Clausewitz (1780–1831): The attacked possesses advantages far in excess of the attacker. He enjoys political sympathies and the moral advantages, which are derived from defending his own territory. He also enjoys the advantages of fighting on terrain more familiar to him than to the enemy, and his operations are conducted on his own lines. Often, he has the advantages of fighting from prepared positions. The conclusion is that in the world of practical fact the defensive is the stronger form of war. In addition it is a stronger form in a philosophical sense.138 Although Mamluk garrisons often raided Īlkhānid siege camps, the fortresses’ defense was hardly based on mounted men, and the outcome of a siege was seldom determined by mounted archers. Both the besieging camp and the fortress garrison based their tactics on long-range weapons. To a large extent these weapons dictated the type of soldiers and the size of the garrison. Most of the burden inside the fortress fell on the teams that operated the siege machines, and on the archers. Since the number of siege machines one can station in a fortress is fairly limited, the majority of the garrison would have been composed of archers. Types of bows The crossbow’s weight and somewhat slow and cumbersome drawing mechanism required the archer to find shelter during the long minutes he was occupied in fitting a bolt and drawing the bow. It was because of such qualities that the crossbow was to a certain extent more suitable to fortress defense than fighting in the open battlefield. In addition, training crossbowmen was not as time-consuming as training archers using composite bows. Composite bows were the dominant weapon of the Mamluk

78  Muslim Fortresses in the Levant army and were no doubt used also by fortress garrisons. Their main advantage was speed of fire. A professional, well-trained archer could fire 12 arrows a minute,149 as opposed to a crossbowman who could fire only 2.140 The range of both types is 100 to 250m.141 Their force, length of range and accuracy made the archers the most important element of Mamluk fortress garrisons. This had a great influence on how the Mamluks rebuilt the fortresses they had taken. The Kitāb ghunyat aṭ-ṭullāb fī maʿrifat ramy an-nushshāb (‘Essential Archery for Beginners’) is a Mamluk manual dated to c. 1368 that is dedicated solely to archery. It is attributed to Ṭaybughā al-Ashrafī al-Baklamishī al-Yūnānī, of whom we know very little. The signature on the front page shows that the book stood in the personal library of the Mamluk amir in charge of the arsenal. This manual deals mainly with types of bows and arrowheads as well as with exercises for mounted archers. It does, however, give some instructions and advice to infantry archers. This is summed in a paragraph of four lines in which he explains how to aim from a position at the head of a curtain wall, at enemy soldiers along the foot of the wall. This paragraph concludes with the author’s directions that the archer (i.e. a garrison archer) should be experienced and that this exercise requires a great deal of training.142 Of all weapons in this period, the influence of the bow on military architecture appears to be the most marked. The number and position of towers, their size and shape, the number of floors in each tower, as well as the number, design and location of arrow slits, were all determined according to the advantages and disadvantages of the bow, and strove to provide the best possible conditions for archers.143 Siege machines The detailed description of the defense of Mayyāfāriqīn during the Mongol siege (656–7/1258–9) illustrates the use of siege machines from within the fortress. There was someone there [in the city] manning a catapult who was extremely accurate and had killed many with his projectiles. The commanders were helpless to do anything about him. Badruddin Lu’lu’ also had an extremely good catapult engineer he summoned. He set up a catapult directly opposite the catapult in the city, and the two shot stones from the pan at the same time. The stones collided in midair and shattered. The people on both sides were astonished by the mastery of the two catapult engineers. In the end the catapult outside was burned.144 Siege machines can be studied according to both the historical sources and the archaeological evidence that is slowly being unearthed. One of the difficulties at this stage is identifying the exact type of siege machine. The area and space needed in order to station and operate the contraption is rather large. One should take into account the base, the length of the central beam, the catapult stones piled nearby and a team of men. The size of the team depended on the type of machine. If it was a counterweight siege engine the number of men needed to operate it was at least four; if it was manually worked, i.e. by a team pulling on ropes, it would need tens of men.

Mongolian siege warfare and the defense of Mamluk fortresses  79 The number of siege machines that a fortress could maintain can be determined according to the size of the towers or the flat platforms that could be used for positioning the siege machines. According to Chevedden a great change in the dimensions of towers occurred during the early twelfth century. The diameter of towers grew from 5m to 30m. Chevedden’s study focused on the citadel of Damascus, and proposed that it was al-ʿĀdil (596–615/1200–18) who initiated this change in the defense concept. He raised siege machines to the tower tops and integrated them in the citadel’s defenses. The change allowed the besieged to ward off the enemy and placed attackers and defenders on even grounds. Chevedden’s main argument is that the major change in the size of towers coincided with the arrival of the counterweight trebuchet. This revolutionized not only siege methods but also methods of defense from within the fortress, and had a considerable effect on the construction of fortifications.145 When you examine this idea and test it on citadels and small frontier fortifications there seem to be very few cases that follow the pattern of inserting grand-scale towers along the curtain walls. At al-Raḥba, for instance, the two largest towers are the southwest and northwest corner towers measuring 8.5 × 8m and 8 × 6m. If siege engines were located on the tower tops they would have had to be of a fairly modest size. At Ḥiṣn al-Akrād there are more than nine towers that could have been used for stationing siege machines, some no doubt could have supported counterweight trebuchets. It is important to bear in mind that the changes made by al-ʿĀdil came after the severe earthquake of 1202 that destroyed many fortresses and central Syrian citadels. The large towers may well have been constructed to withstand the immense force of severe earthquakes.146 Mongol siege warfare and Mamluk methods of defense The maintenance of a constant state of alert in the garrisons, and of the Mamluk armies in Syria and the sultan’s armies in Cairo, and also the ability to maneuver large contingents in a relatively short space of time can be achieved only by a centralized regime. The sultan’s efforts to repair the damage after every Īlkhānid siege, see to the supplies and reinforce the garrison are all well documented. This ability to respond and repeat these steps after every siege, without faltering, was one of the most remarkable aspects of the Mamluk military structure. Although the majority of the MongolĪlkhānid sieges failed, and their army was defeated at the battles of ʿAyn Jālūt and Homs, the Mongols and later the Īlkhānid armies managed to enjoy and maintain the image of a state with a superior military force. The fear the Mongols instilled in the Mamluks kept the Sultanate in a state of alert for almost six decades. The Mamluk defense system established by Baybars remained much the same even after the Īlkhānid army changed its strategy and avoided siege warfare whenever possible. The last Īlkhānid siege of al-Raḥba (1312) set off the same chain of reactions – a message was sent to the inner Syrian cities, which were quick to organize reinforcements together with the sultan’s army that arrived from Cairo under the command of al-Nāṣir Muḥammad, just as had been done in earlier decades. The dependence of the Euphrates fortresses on reinforcements seems at times like a perilous gamble. Any

80  Muslim Fortresses in the Levant delay would have ended with their fall, as neither fortress had a large garrison. The Mamluks, however, did not modify their concept of defense and the reinforcements seldom failed to arrive. As far as siege technology is concerned, the knowledge possessed by Hülegü’s siege units was no greater than that generally found at the time in the Middle East. The siege machines used by both Franks and Mamluks were equal to those of the Mongols. The acquisition of Muslim siege machines by Qubilai via Abagha implies that the Īlkhānids were using counterweight siege machines by 1271. However, this did not give them any advantage and it seems that none of the siege methods and machines that the Īlkhānids used against the Mamluk fortresses could surprise Mamluk defenders or undermine the Mamluk defenses.

3 Laying the foundations Fortification work conducted during the reign of Baybars (1260–77), Qalāwūn (1280–90) and Al-Ashraf Khalīl (1290–93)

The uniting of Egypt and Syria under a centralized government in Cairo became an urgent issue under the Mongol threat of invasion and the remaining enclaves of the Crusader kingdom and principalities in the Levant. In addition, absorbing the Ayyubid entities and creating a geographical and political continuity from Cairo to Aleppo was necessary for the establishment of the new regime. Mamluk fortifications did not arise out of the desire to expand Mamluk rule. What led to this policy was the need to protect and guard the Sultanate’s territory. The decision was undertaken in the face of the overwhelming numerical superiority of the Mongols in the early 1260s,1 and the fact that the Mamluks were operating on three different fronts: Mongolian, Frankish and Armenian. In light of the political and military situation, building and maintaining a defense line was by and large an easier and more practical procedure than defending the frontier by capturing territories east of the Euphrates. The Mamluk army did not have the means,2 training or logistic framework that enabled the Mongols to hold vast territories, fight in distant lands far from the center of government and maintain themselves for long periods. Although permanent Mamluk armies existed in Syria, the core of the army was stationed in Cairo. The Mongol invasion commanded by Hülegü began with the destruction of the Ismāʿīlī centers south of the Caspian Sea. It was followed by the sack of Baghdad (1258), the execution of the Caliph and the conquest of the Jazīra (1259). It concluded when the armies crossed the Euphrates into Syria (1260) and advanced into Trans-Jordan and Palestine. This long campaign left behind a trail of destruction, and destabilized the entire region. While the Ayyubids decided to play a passive role, the Armenian king was quick to turn his kingdom into a client state in the early 1240s. The Mamluks refused to bow their heads and surrender. The victory at the battle of ʿAyn-Jālūt did not diminish the threat of Mongol power or strength in the eyes of the Mamluks; it probably gave them a better understanding of Mongol fighting abilities and the nature of the new enemy that was there to stay. The political instability within the newly built Sultanate did not prevent Baybars from constructing a complex defense that continued to serve the Mamluk sultanate long after his death. As well as fortresses along the frontier a great deal of thought

82  Muslim Fortresses in the Levant and funds were invested in fortresses that protected the lines of the barīd,3 trade and pilgrim routes. Mamluk attacks on Armenian territory, the remains of the Crusader kingdom and principalities during the second half of the thirteenth century were almost always concluded with the conquest of a number of strongholds. If it was decided to keep them, reconstruction work began almost immediately. In addition to the reconstruction of fortifications, the field army was strengthened and carefully administrated.4

Flaws and weaknesses along the Sultanate’s borders and frontiers 5 The term “border” is somewhat problematic when used in surveying the geographical regions that the Sultanate ruled during the period under discussion. In comparison with modern Middle Eastern borders, carefully drawn, mapped, fenced, mined and marked with signs written clearly in three languages, the borders of the Mamluk sultanate in the second half of the thirteenth century and the beginning of the fourteenth were vague. The location of fortresses only partly illustrates contemporary border lines, and one should be cautious in drawing conclusions since many fortresses acquired by the Mamluks, for example those in Cilicia, were in the heart of Armenian territory. Tall Ḥamdūn, for instance, was for a long time a Mamluk enclave in a foreign and hostile territory (Map 2.2). This fortress was seldom directly involved in the defense of the Mamluk frontier and its main task was to keep a watchful eye on movements within the Armenian kingdom. A large number of Mamluk fortresses are not located along a definite frontier. In fact, one can draw a clear border only along one section in the east, on the upper part of the Euphrates. The upper Euphrates marked a natural border and was treated as such by both the Īlkhānids and the Mamluks, each of whom saw it as the definite boundary of their territory. While the Mamluks seldom crossed the river and did not wish to collect taxes from the population east of it, the Īlkhānids clearly knew that any attempt to cross the river would be seen as an act of aggression and a declaration of war. Thus the Mamluk fortifications along the Euphrates enable us to draw the only linear border in the Sultanate. Al-Bīra, al-Raḥba and Qalʿat al-Rūm are three out of five fortresses along the Euphrates that were rebuilt and garrisoned by the Mamluks before 1292. The border between the Armenian kingdom and the Mamluk sultanate can not be seen as a line drawn between fortresses; though from Abū’l-Fidā’s account it appears that the Jayḥān River was perceived by both sides (for lack of Armenian choice in the matter) as a border. During the first four decades of the fourteenth century the name of the Jayḥān River comes up several times as a marker between Mamluk and Armenian territories. The agreement signed between the Armenian king Constantine and the Mamluk sultan Lachin in 697/1298 states: The army demanded that the River Jayḥān should be the boundary between the Muslims and the Armenians, and that he should surrender all the fortresses and territory south of the River Jayḥān. Constantine agreed to this and surrendered to the Muslims all the territory south of the River Jayḥān including Hamūs,

Laying the foundations  83 Tall Ḥamdūn, Kuwayrā, al-Naqīr, Hajar Shaghlān, Sarfadkār and Marʿash, all of which are as strong fortresses as one could wish.6 The fortresses south of the Jayḥān often changed hands between Armenians and Mamluks. The chronicler Badr al-Dīn al-ʿAynī quotes Ibn al-Kathīr in his work ʿIqd al-jumān fī ta’rīkh ahl al-zamān, which describes the Mamluk raid on Sīs in 703/1305. After a long siege of the fortress of Tall Ḥamdūn, held at the time by the Armenians, the Armenian king Hetʿum II appealed to the sultan al-Nāṣir Muḥammad to bring the raids that were causing havoc and destruction in his country to an end. The agreement states that the area between the Jayḥān River and Aleppo would belong to the Muslims while the land to the north and the east of the river would belong to the Armenians.7 In 737/1339, after the Mamluks had besieged the port Ayās, the two sides agreed that the Mamluks would raise the siege if the Armenians handed over all the fortresses east of the Jayḥān. The list included eight fortresses: Kāsīsh, al-Naqbar, Kuwayrā, al-Hārūnyah, Sarfadkār, Ayās, Bānās and Bakhīmh.8 While some of these appear in previous treaties others are mentioned here for the first time. In contrast to the borders between the Mamluk sultanate, the Īlkhānid state and the Armenian kingdom, there was no river or mountain range to mark a border or frontier between the Sultanate and the Crusader kingdom and principalities. The borders of these two entities can not be defined. 9 The reconstruction of fortresses along the Euphrates and south of the Jayḥān was directly linked to the arrival of the Mongols, the establishment of the Īlkhānid state and the Īlkhānid-Armenian alliance. Most Mamluk fortresses, however, are distributed in central Syria and Trans-Jordan. Explaining the Mamluk investment in refortifying fortresses in these regions can only be done by examining each fortress separately, taking into account regional conditions, possible threats, the needs of the central government, and the local population. The decisions made by Baybars and the sultans who succeeded him do not always relate directly to the defense of the frontiers or to threats from the Mongols, Armenians or Franks. This probably can best be seen when studying fortresses in Trans-Jordan. While some served as local administrative centers before the Mamluk period, others were rebuilt in order to serve the logistics of the Mamluk communication lines – guarding and maintaining the barīd routes. An important group of fortresses served as stages along the ḥājj, and were charged with the safety of pilgrim and merchant caravans. The influence of climate and topography on borders and the spatial distribution of fortresses The following section gives a description of the topography and climate along the Sultanate’s borders, and particularly of regions that could be defined as natural barriers which armies would avoid confronting. The topography was carefully considered when the Mongols and later the Īlkhānids planned their invasion routes into Syria. These routes determined Mamluk priorities concerning the location of their fortresses.

84  Muslim Fortresses in the Levant The eastern frontier The Mamluks’ longest frontier ran parallel to the middle course of the upper Euphrates. The river can not be perceived as a natural barrier. The Mongols could and did cross it at various points. Although they preferred shallow ford crossings, there are descriptions of how they crossed rivers holding on to the tails of their ponies, with all their equipment and weapons.10 Geographically this section is a coherent unit. Both sides of the river can be defined as a semi-arid hilly region;11 the annual rainfall average is between 100 and 200 mm and the river is in flood during the spring months of April and May.12 Crossing the Syrian Desert (Bādiyat al-Shām) that stretches along southern Syria was a route few armies attempted due to meager pasture and few reliable water sources. The annual rainfall in this region is 100 mm. All the fortresses along this stretch of the river had already been built by the end of the Ayyubid period.13 It was impossible, practically speaking, to build defenses along the entire length; thus the only section where the Mamluks rebuilt and maintained strongholds was the stretch closest to the northern and central Syrian cities. The Mamluks never possessed a fortress along the fords south of al-Raḥba. Other than the location of the Syrian cities, what determined the position of the Euphrates fortresses from the outset was the topographical nature of the river and the character of the river flow. Where the river flows in a steep valley, the current is fast and the banks difficult to climb. All the fortresses were thus built beside comfortable crossing points, fords where the water is shallow, the current relatively slow and the river not too wide or boggy. The three most important fords are located near al-Bīra, Qalʿat Jaʿbar and al-Raḥba.14 This section of the river runs parallel to the two major northern Syrian cities of Aleppo and Hama. The distance between Aleppo and al-Bīra is approximately 140 km.15 Defending ����������������������������������������������� this stretch was easier as reinforcements could be called upon from those cities. However, if the Īlkhānid armies were not stopped on the Euphrates the central Syrian cities were in immediate danger and faced a numerically superior army. A good illustration is the conquest of Damascus by Ghazan in December 1299. Shams al-Dīn Abū ʿAbd Allāh, an eye-witness, claims the Īlkhānid army numbered 100,000 men, three times the size of the Mamluk force inside the city.16 This enabled Ghazan to gain control over the city almost without a struggle. While the Syrian littoral underwent several political changes due to the conquest and destruction of Frankish cities and fortresses, the eastern frontier remained unchanged for the next sixty years, until the fall of the Īlkhānid state (1335). The fortresses along the Euphrates still retained their importance and continued to serve the Sultanate during the fourteenth and fifteenth centuries following the invasion of Temür (Tamerlane, 1370–1405) and the rise of local rival entities that gradually settled in Īlkhānid territories. Although Mamluk investment in both fortresses and garrisons was considerable, the eastern frontier could not be sealed and the fords could not be completely barred against the Īlkhānid armies. Of the six invasions that took place between 1280 and 1312 the Īlkhānids managed to cross the river and march towards central Syria on four occasions: the second invasion by Abagha (1281); the first and second invasions

Laying the foundations  85 by Ghazan (1299–1300, 1301); and during Ghazan’s third invasion when his auxiliary forces crossed the river and reached Damascus (1303). The fact that the river was crossed on several occasions demonstrates the faults and flaws of the Mamluk defenses. Garrisons could carry out raids in the vicinity of the fortress, but they could not confront a full-scale army. Their main tasks were to alert the Mamluk army, to warn and inform the sultan of the coming of Īlkhānid forces and to try and delay them until the arrival of the Mamluk reinforcement from the central Syrian cities and/or Egypt.17 As long as the Mamluk army was well organized and capable of sending relief forces, this array of strongholds functioned and held the eastern frontier secure. The northern neighbor The northern frontier of the Sultanate bordered the Armenian kingdom of Cilicia, a vassal and a close ally of the Īlkhānid state.18 The regions northeast of Ayn Tāb are semi-arid, with hills rising as high as 400–600m. The land gradually descends as it nears the banks of the Euphrates. The western region is dominated by the Amanus mountain range, averaging 1800m in height. Steep ravines carve the mountain slopes,19 and large areas are covered with dense Mediterranean forest. The area has a high rainfall averaging 1,000–2,000 mm a year. The Amanus acted as a barrier between the Mamluk sultanate and the Armenian kingdom. The four passes along it, however, are all accessible to both infantry and cavalry (Map 3.1).20 Reckoning from the south, the Belen pass is guarded by two fortresses, Baghrās and Darbassāk. The second pass, narrow and 12 miles long leading from Iskandarūn to Payas (Bāyās), is known in the Mamluk sources as Bāb al-Malik. The third is the Quwara pass along the northwestern slope, known today in modern Turkish as Demir Kapi. Once across, one could continue on the Cilician plain to Adhana, al-Massīsa and Ayās. The last pass is Darband Arslān (Amanus Gate), a narrow valley stretching eastwards across the range.21 Like the Quwara pass it leads into the Cilician plains. On the western side of this valley stands one of the best preserved fortresses – Tall Ḥamdūn. The fortresses guarding the four passes were often attacked by the Mamluk armies. Although the Sultanate wished to add them to the array already in its possession they were often won and lost, and this cycle continued during much of this period. At the same time the Armenians fought to acquire and control the passes and fortresses that were the key to the defense and economy of their kingdom. Any army that wished to march from Cilicia to northern Syria had to cross one of these mountain passes. They were a military asset that all entities in the region wished to hold. The Mamluk city closest to the Amanus range was Aleppo. It was both a target for the joint Īlkhānid-Armenian armies and a point of departure for the Mamluk armies that raided the Armenian territories. In addition to their strategic importance one must note the economic value of this region. The short distances between the northern Syrian cities and the trade routes that passed through Cilicia down to the Mediterranean ports turned the mountain passes along the Amanus into a military arena. However, throughout much of the second half of the thirteenth century and the early fourteenth no one managed to dominate the region. ʿ

86  Muslim Fortresses in the Levant

Map 3.1 The four main mountain passes in Cilicia

The Crusader enclaves The three Crusader entities, the Kingdom of Jerusalem, the County of Tripoli, and the Principality of Antioch, formed enclaves that broke the geographical and political continuity along the Syrian littoral within the Sultanate. The attempts of the Īlkhānids to form an alliance with the Pope and with European monarchs in order to receive their support in fighting the Mamluks constituted a threat that hovered over the Sultanate and could not be ignored. In addition, the Principality of Antioch and the other Frankish settlements in Syria occasionally cooperated with the Īlkhānids and spied on their behalf.22 By the early 1270s the Crusader kingdom had lost large parts of their territories along the coast. The aggressive policy of Baybars was carefully followed by his successor. Except for Chastel Pelerin and Acre, most of the fortresses and towns along the southern coast and further inland had capitulated by 1265. Antioch fell in 1268, Qalāwūn sacked Tripoli in 1289 and with it fell many of the northern coastal towns and strongholds, among them Marqab. Acre, the last important city, was besieged in 1291 and fell to al-Ashraf Khalīl. The destruction of the Crusader enclaves eventually gave the Sultanate political and geographical continuity from Cairo to Aleppo. These changes had a direct influence on the distribution of fortresses in the Levant; some of the strongholds conquered and rebuilt in the early decades gradually lost their military importance due to geopolitical changes throughout the region.23

Laying the foundations  87

To build or not to build: the military, political and economic priorities that dictated which fortresses should be reconstructed Throughout the second half of the thirteenth century Mamluk military activity barely ceased. The Mamluk forces in Syria and the sultan’s army in Cairo were constantly employed in siege warfare, raids and full-scale battles. Though an efficient and mobile force, the army could not fight on several fronts simultaneously. Each of the political entities concerned – the Franks, Īlkhānids, Armenians and Ismāʿīlīs – had its own interests as well as its own army. Both the size of each entity’s army and methods of warfare were quite different. The Franks, at this stage, fought mainly from within their fortresses. They carried out raids in their vicinity but rarely initiated open-field battles. This defense policy which relied on strongholds and fortified cities was carefully followed until the end of the Crusader period.24 In contrast to the Mamluk ability to organize efficient relief forces, throughout most of the second half of the thirteenth century each Frankish stronghold had to stand and defend itself, knowing that relieving armies were rarely available. Although they were well versed in the art of siege warfare, Frankish forces could not threaten Mamluk strongholds without considerable assistance from an outside military ally. One of the few Frankish assaults on a Mamluk fort was led by Prince Edward of England, who tried to conquer Qāqūn in 1271. Even when a considerable reinforcement had arrived from Europe, little effort was made to join forces with the Īlkhānid army or the Armenians,25 and even modest ambitions such as capturing Qāqūn (by far one of the smallest Mamluk strongholds) were unsuccessful. The English prince finally returned home without achieving any of his aims. The Īlkhānid army, on the other hand, was highly mobile in comparison to the Franks, the Armenians and the Mamluks. It did not suffer from a lack of manpower and could summon a large force. Mounted archers composed the core of the army and fighting in the open field was clearly preferred to any other form of combat. The Franks were familiar with this form of warfare though they did not adopt it in the East. As far as fortresses were concerned, the Īlkhānids did not follow the practice of establishing strongholds. Camps were set up in the field and packed and folded when the army was ready to move. There are no contemporary accounts attesting to Īlkhānids building or restoring fortifications along the Euphrates and there appears to have been no large permanent Mongol presence along the eastern frontier.26 The third army that often clashed with Mamluk forces throughout this period was the Armenian, which combined various elements and styles of warfare acquired from the Byzantine Empire, Central Asia and Western Europe. It was composed of mounted archers who rode beside heavily armoured knights.27 The kingdom was guarded and secured by fortresses built along mountain passes, central junctions and the Mediterranean ports.28 According to Molin the Armenians knew that their military forces were not sufficient to encounter the Mamluks on the open battlefield. They therefore chose strongholds in mountain areas to lure the Mamluk forces into rough, steep and forested terrain that would be difficult for a large mounted force to negotiate.29

88  Muslim Fortresses in the Levant After 1264, the Armenians never initiated raids or crossed into Mamluk territory of their own accord, though they never avoided or refused any of the Īlkhānids’ invitations to join forces. Among all these armies the Ismāʿīlīs were exceptional. This small group established itself in a chain of fortresses along the southern Nasseriyya Mountains. Their methods of warfare and their military activities were based on a small group of trained assassins who were sent out to murder carefully chosen political, religious and military leaders.30 Although this type of warfare did not cause mass destruction of villages, cities, crops and the like, it could easily destabilize the entire region.31 The only army that had both the knowledge and the real capabilities of engaging in siege warfare and could threaten Mamluk fortifications was that of the Īlkhānids. While the Armenians supported and often joined Īlkhānid campaigns they could not mount a full scale siege with their own forces. The Ismāʿīlīs did not possess a field army that could employ siege warfare methods. And, although the Franks had both the knowledge and the experience, they did not have sufficient manpower to carry out either long or short siege campaigns. Unlimited variety By the mid thirteenth century numerous fortresses were scattered throughout the region under Mamluk rule. The majority were built in the twelfth and early thirteenth centuries by the Ayyubids, the Franks, the Armenians, the Ismāʿīlīs and numerous local rulers. Points of particular military or economic significance, sites that overlooked central roads and junctions as well as strategically important mountain passages and fords – most such sites were fortified long before the Mamluks came to power. Some fortresses represented the finest and most advanced military architecture of the time. The nature of these strongholds, the quality and standard of construction depended on the importance rulers ascribed to the site. One should bear in mind, however, that the importance of fortresses changed in accordance with political developments, which often brought about significant modifications in the layout of frontiers and borders. The first impression one gets from a preliminary survey of Baybars’ military activity is that the Mamluks’ chief aim during their first decades in power was to destroy as many fortresses as they could, thus reducing their adversaries’ hold on the region. It seems that what guided Baybars was the knowledge that a considerable amount of time and money would be required to rebuild a fortress whose towers and walls were badly damaged, food, grain and weapons supplies burnt or confiscated and whose garrison was put to death or taken into captivity. More than any other political entity, the Franks, consistently lacking human resources, found it increasingly difficult to man and replace garrisons in fortresses they had lost. Even where the archaeological remains show that the fortress was not completely destroyed, the damage to the infrastructure was enough to prevent the Franks from returning to it – they simply did not have the forces and funds to re-take and repair their strongholds. This lack of a central organized regime and of unity

Laying the foundations  89 amongst their forces hampered any attempt to regain lost territory. Thus, fortresses taken by the Mamluks were never recovered by the Franks. In contrast, the Armenians in Cilicia managed to recover their fortresses, via diplomatic negotiations or by besieging and re-taking them in battle. They quickly rebuilt, manned and supplied them. In one case where funds were in short supply they appealed to the Pope. The fortifications at the port of Ayās, in need of reconstruction after a severe Mamluk assault, were restored using papal funds. The last Ismāʿīlī strongholds were captured by Baybars by 1273, though it is not quite clear whether the Mamluks ever used any of them for their own purposes. Ibn Baṭṭūta, who passed through the district in 1326, says that the Ismāʿīlīs were allowed to hold and maintain five of their fortresses and that Mamluk sultans occasionally still employed Ismāʿīlī assassins.32 To build or not to build “One part [of the Muslim armies] uproots Frankish fortresses, and destroys [their] castles, while another part rebuilds what the Mongols destroyed in the east …”33 Mamluk means, however, were fairly restricted; because the core of their army was permanently quartered in Cairo, the number of fortresses the Sultanate could afford to rebuild, arm and man, was fairly limited. It seems Baybars carefully calculated the number of strongholds he could efficiently maintain, and was often forced to relinquish some fortresses for lack of sufficient forces to garrison them or provide them with reinforcement in times of siege. Many fortresses were simply deserted and never rebuilt during his reign. To get an idea of the size of garrisons in the larger fortresses one can use the composition of the Frankish garrison at Safad. The entire force numbered 1,700 men. In times of war it rose to 2,200. There were 50 knights, 30 sergeants, 50 turcopols and 300 crossbowmen. The total number of fighting men was only 430. An additional 820 men were responsible for the daily running of the fortress, and we are told that there were also a further 400 slaves.34 If Karak and Ḥiṣn al-Akrād were of similar size then they must have been garrisoned at the expense of the Mamluk army. The capacity of the sultan to allocate men from his own army or from the Mamluk armies stationed in Syria, for the sake of constituting garrisons was fairly limited. Ibn Shaddād gives a list of twenty-two strongholds in the district of Aleppo that were destroyed and abandoned. It is more than likely that most of them were ruined during the first Mongol invasion. His account does not provide details as to who built or owned them. They may have belonged to local noble families that were not directly connected or affiliated to the central regime. The fortresses are all referred to as Ḥiṣn ( ), which does not offer further information about their size or structure since there seems to have been no differentiation in the terminology. Both Ḥiṣn and Qal ʿa mean stronghold; neither refers to a specific type or size of fortification.35 The following is the list as it appears in Ibn Shaddād’s account: Sunayāb, Salʿaān, Sarbak, Tall Ramāl, Bāsūtṭā, Bātrakh, Shīḥ al Ḥadīd, Karmīth, Marāsīyā, ʿAnāqīb, Hāb, Basarfūth, Ārīnyā, Īnnab, Tall-Kashfahān, Zardanā, Āzramān, ʿAim, Salqin,

90  Muslim Fortresses in the Levant

Map 3.2 Fortresses in the vicinity of Aleppo taken by Baybars, according to Ibn Shaddād, but never rebuilt.

Tall-ʿAmār, Tall-Khālid and Ārmanāz (Map 3.2). Only half of these fortresses have been identified. The closest to Aleppo are 40–50 km away, while the most distant are approximately 80 km from the city. At the end of the paragraph Ibn Shaddād adds that there are many more fortresses apart from those he has mentions in his list.36 The balance between the garrisons and the Mamluk field army as well as the dependence of each on one other should be taken into account when one tries to follow the decisions made by Baybars as to which fortresses should be left in ruins and abandoned and which fortresses should be held and rebuilt. Al-Bīra and al-Raḥba were chosen by Baybars to guard the Euphrates frontier. It seems that preference was given to sites whose maintenance did not require an enormous outlay and thus they did not become a burden on the local population or the sultan’s treasury. A further two fortresses were added to the defense of the Euphrates frontier in later years, when the Mamluk army had grown considerably in size and the geopolitical situation had changed. Qalʿat al-Rūm was taken and rebuilt after Acre had fallen in 1291; it was of outstanding size and probably had a garrison similar in numbers to that at Safad, approximately 1,700 men. Qalʿat Jaʿbar was rebuilt by al-Nāṣīr Muḥmmad in 1333–4, a decade after the peace treaty was signed between the Īlkhānids and the Mamluks. It is difficult to explain the immediate threat that brought about the establishment of these fortresses. This large-scale

Laying the foundations  91 military construction program was implemented during times of peace, when the Mongols were occupied elsewhere and the Mamluks were not busy on other fronts.37 Although most Mamluk fortification work started almost immediately after a fortress was conquered it sometimes took several years for projects to be completed. Most fortresses could not endure a siege of more than a few weeks. In 1259 al-Bīra was conquered within two weeks by a section of the Mongol army headed by Hülegü. In the winter of 1312–13 the garrison at al-Raḥba managed to withstand siege for a whole month.38 The Mamluk army played a key role in the support of the Euphrates fortresses, for resisting an Īlkhānid siege force could only be done by a full-size and well-trained field army. On a number of occasions the Īlkhānid army left the scene before or while the Mamluk forces were making their way north. Without the support of the Mamluk field army the strength and existence of the frontier strongholds was questionable. An interesting point to note is the fact that reinforcements were seldom sent from any of the neighboring strongholds. They always arrived from the central Syrian cities and/or directly from Cairo. It seems the number of men per garrison was carefully calculated and that even in times of emergency the garrisons could not afford to send reinforcements to a neighboring site. The decisions made by Baybars may seem at times to be random. Although it is possible to rationalize most of his moves, one should keep in mind that unlike many of the Frankish, Ayyubid, Armenian and Ismāʿīlī strongholds that were built on virgin soil, Baybars avoided building new fortresses. It seems that to some extent his decisions were influenced by the distribution of existing fortifications. This policy was partly impelled by a feeling of urgency, and possibly by the need to reduce outlay. The field army enjoyed a higher priority, and a large percentage of the sultan’s funds were invested in buying Mamluks and arming his forces. Above all, it was difficult to justify the building of new fortresses in view of the fact that the region already had numerous strongholds of every type and size. There are three extant lists that summarize the conquests of Baybars. The first is that of Ibn ʿAbd al-Ẓāhir, who served as Baybar’s private secretary (kātib al-sirr). His short account lists the fortresses in Syria and Trans-Jordan that were destroyed by the Mongols during their first invasion. Once Baybars became sultan he immediately saw to the reconstruction of all the ruined strongholds; moats were cleaned and fortresses strengthened or enlarged. Soldiers, funds and military supplies were sent to each site. The list includes Damascus, al-Ṣalt, ʿAjlūn, Ṣarkhad, al-Ṣubayba, Shayzar and Shumaymish.39 Many other sites rebuilt by Baybars are not mentioned and it seems that the list follows the main route of destruction left by the Mongols. Although none of the sites listed were close to the Frankish or the Mongol frontier, they were all rebuilt in the sultan’s reign on the supposition that if the Īlkhānids invaded they would take a similar route. The most complete list is that provided by Ibn Shaddād, who divided the sites and fortresses captured by Baybars into six categories: (1) sites conquered from the Franks; (2) sites in which revenues were shared between the Franks and the Mamluks; (3) sites taken from the Armenians; (4) places taken from Muslim rulers; (5) Ismāʿīlī strongholds; (6) sites taken from the Mongols. The list is not arranged in a strict

92  Muslim Fortresses in the Levant chronological or geographical order. Although most of the important sites are noted some, such as al-Ṣubayba are missing, and a number of Armenian strongholds in Cilicia are ignored. Interesting additions are sites where Baybars had reached an agreement with the Franks according to which certain fortresses were left in their ownership and the revenues shared.40 Ibn al-Furāt gives a similar account, basing himself on Ibn Shaddād; however, he did not copy the work word for word and omitted important sites such as Karak, Shawbak and the Ismāʿīlī strongholds.41 Since none of these lists gives a full account that might reveal the sultan’s priorities, the only way to gauge the preferences of Baybars and his successors is by conducting a chronological survey of his conquests and the reconstruction work that followed. The first fortresses to be rebuilt by Baybars were along the frontier with the Īlkhānid state, which was clearly given precedence over other regions in the Sultanate. Table 3.1  Baybars’ conquests according to Ibn Shaddād Sites conquered Sites where by Baybars from revenues were the Franks shared between Mamluks and Franks

Sites recovered from the ruler of Sīs

Sites previously owned by Muslim rulers

Fortresses of the Ismāʿīlīs

Sites taken from the Mongols

Al-Marqab Caesarea (Margat) Arsūf Bāliynās Safad Anṭarṭūs Tiberias (Tartosa) Jaffa Al-Shaqīf (Beaufort) Antioch Baghrās (Gaston) Al-Quṣair (Cursat) Ḥiṣn al-Akrād (Crac des Chevaliers) Al-Qurayn (Monfort) Ḥiṣn al-ʿAkkār (Gibelacar) Ṣāfītha (Chastle Blanc) Maraqīya (Maraclea) Ḥalbā (Aleb)

Darbasāk Darkūsh (Dargos) Talmīsh Raʿban Marzubān

Damascus Baalbek ‘Ajlūn Buṣra Ṣalkhad Al-Ṣalt Homs Tadmor Ṣahyūn Blātuns Burzya Al-Raḥba Shawbak Karak

Kahf Qadmūs Manīqa ʿUlayqa Khawābī Rṣāfa Masyāf Qūliy’a

Al-Bīra Aleppo and its region Shayzar

Source: ʿIzz al-Dīn Muḥammad b. ʿAlī ibn Shaddād al-Ḥalabī, Ta’rīkh al-malik al-ẓāhir, ed. A. Ḥutait (Wiesbaden, 1983), 322–3.

Laying the foundations  93 The political changes, and the alternating periods of prosperity and decline within the Īlkhānid state, did not reduce or modify the Mongolian threat in Mamluk eyes. Throughout the next sixty years the fortresses and garrisons along the Euphrates were maintained at the highest level. Baybars gradually worked his way from the periphery to the center of the Sultanate. The first fortresses to be rebuilt were al-Bīra (1262)42 and al-Raḥba (1264), situated on the bank of the Euphrates. The Īlkhānids crossed at the same Euphrates fords and seldom used different routes. Most invasions began by crossing at al-Bīra and heading south towards Aleppo (Map 2.2). Topography, the structure of the river, and the pasture and water sources before and after reaching the crossing did not leave much room for maneuvering.43 At the beginning of 1260 Hülegü turned to Syria after conquering the northern Jazīra. The following description of his crossing explains what guided the Mamluks in choosing certain points along the river to fortify. And the King of Kings commanded, and bridges were tied over the Euphrates at Melitene, and Kalaʿat al-Rhomatia and at Birah, and at Circesium (Osrhoene),44 and troops crossed over into Syria together with the King of Kings himself … and troops, fighting men were posted against the fortress of Birah and against the fortress Kal’at Najm and against that of Kal’at Jaʿbar and Calonicus and Balish and the people that were in them were killed.45 Hülegü crossed near al-Bīra, divided his army and besieged al-Bīra, Qalʿat Najm and Qalʿat Jaʿbar. The three fortresses fell within a matter of weeks and Aleppo was conquered shortly afterwards. In December 1299 Ghazan, unlike his predecessors, decided to cross south of Qalʿat Jaʿbar, five days march to Aleppo. A year later (in January 1301) he returned and crossed at the same place.46 In Ghazan’s third invasion the army crossed at al-Ḥilla (east of Karbalā), and marched north along the bank of the river up to al-Raḥba.47 The last Īlkhānid invasion into Syria took place in 1312, during Öljeitü’s reign. The army left Mosul and crossed the Euphrates at Qirkasiya,48 some 60 km north of al-Raḥba. The need to rebuild the two fortresses was a direct response to continual attempts of the Īlkhānid army to cross the fords at the nearest point to the northern Syrian cities. It is therefore surprising that the fortifications of Aleppo were not restored at the same time as those of al-Raḥba and al-Bīra, particularly since the city was the closest to the border and was exposed to assaults from both north and east. Aleppo’s fortifications were rebuilt shortly before the Mongol invasion of 658/1259. The account of the restoration reveals the expenses involved and the effect this had on the fortification plan. In 622/1223 the Atabeg of Aleppo, Shihābā al-Dīn Tughrīl, took it upon himself to build a 500 dhirāʿ section of the city wall (i.e 340m; one dhirāʿ = 68cm) and to restore a number of towers that were in ruins.49 He summoned craftsmen and asked for their advice. They suggested that he move the moat to the lower part of the slope and raise the height of the walls. When the Atabeg calculated the costs and realized that this project would take a great deal of time to complete, he decided to dismiss the advisors and ignore their advice. Olive and mulberry tree trunks were used to rebuild

94  Muslim Fortresses in the Levant the foundations of the wall,50 and thus the grand project came to an end. According to Ibn Shaddād it was at this section that the Mongols breached the walls in 1260: “When the Mongols attacked [Aleppo] the sappers could undermine the wall only at this section (where the wall was supported by tree trunks of olive and mulberry).”51 Ibn Shaddād, who was born and grew up in the city, managed to escape, and returned to Aleppo after it was re-taken by the Mamluks. His knowledge of the city makes him a reliable source. His account of Hülegü’s invasion of 1260 tells of the destruction of the urban fortifications, the city’s public buildings, and entire neighborhoods. Aleppo was struck once again during Abagha’s invasion in 1280. The urban fortifications remained in ruins for 32 years, until Qalāwūn began rebuilding the citadel. The work was only completed in 1290–1 during the reign of his son, al-Ashraf Khalīl. Abū’l-Fidā’, who describes the invasion, does not elaborate on this issue and one cannot obtain any information concerning the scale of building.52 The Īlkhānid invasions of 1300 and 1301 may have brought further destruction to the city fortifications. It was, however, only during Barqūq’s rule (1382–99) that the fortifications were renewed. It is not difficult to explain why Baybars ignored the need to refortify Aleppo. The earlier example of the Atabeg’s rejection of expert advice because of the heavy costs involved was probably a factor in Baybars’ decision. It seems likely that raising local taxes to help fund part of the outlay was not feasible, since the economy of the region had been badly damaged by the Mongols. Villages were deserted and a large part of the urban and farming population had fled. In addition there was a sharp decline in commerce along the trade routes between Syria and Iraq.53 If the sultan had decided to go ahead and restore the damage the greater part of the expense would have had to come out of his own treasury. The fortifications along the eastern frontier, however, were of the utmost importance as long as the Īlkhānid state existed and they were given priority by Baybars and his successors. In 1292 the sultan al-Ashraf Khalīl conquered Qalʿat al-Rūm, refortified the site and established a new garrison.54 Two years before the official fall of the Īlkhānid state in 736/1335–6 the sultan al-Nāṣir Muḥammed ordered the governor of Damascus to rebuild the fortifications of Qalʿat Jaʿbar, strengthen its garrison and nominate a nā’ib.55 By the time Abū Saʿīd, the last Īlkhān, died (1335) there were four Mamluk fortresses along a stretch of 375 km of the upper Euphrates – al-Raḥba, Qalʿat Jaʿbar, al-Bīra and Qalʿat al-Rūm. The conquest of the inland fortresses At the same time as the conquest of al-Bīra and al-Raḥba the Mamluks turned their attention to fortresses located on or near the main routes running between Syria and Cairo. Chronologically, al-Ṣubayba does not fit this general theme. It was one of the first places that fell to the Mamluks. In the aftermath of Ayn Jālūt (1260) the Ayyubid prince al-Azīz Haṣan b. ʿUthmān, lord of al-Ṣubayba, was executed and the site passed into the hands of the sultan. The fortress was manned and restoration began, lasting several years. An inscription found on one of the towers, states that it was completed in 1275.56 The importance of al-Ṣubayba derives from its location close to Damascus,57 providing protection against enemies advancing from the west, invasions heading towards the northern littoral or aiming to use the coastal road going south. ʿ

ʿ

Laying the foundations  95 The fact that a fortress had been taken by the Mongols was a sufficient reason for Baybars to fortify and garrison the site. It appears that most of the work carried out during the first decade of his reign was done out of fear that the Īlkhānid army would return and retrace the footsteps of the Mongol invasion of 1260. The fortresses along the main Trans-Jordanian highway going from north to south were rebuilt for similar reasons. The arrival of the Mongol army headed by Ketbugha on the banks of the Jordan,58 prior to the battle of ʿAyn Jālūt, emphasized how crucial it was to secure the main roads running through the Sultanate. In addition to the fear of losing Karak, Ṣarkhad, Bosra, ʿAjlūn and al-Ṣalt once again, the fortresses in TransJordan were a means of establishing Mamluk authority. They continued to serve as local and regional administration centers, providing security for the local population and for pilgrim and merchant caravans traveling to and from the holy cities of Mecca and Medina. Due to the existence of Frankish enclaves along the Mediterranean coast, the route across Trans-Jordan became the main line of communication between Cairo and Damascus. Karak passed to the Mamluks in 1263. New sections were built while others were renovated and the garrison strengthened.59 Shawbak was taken soon after. Its governor swore allegiance to the sultan and the Christian population took an oath on the New Testament.60 ʿAjlūn probably fell during the same year. Ibn ʿAbd al-Ẓāhir includes it among the fortresses destroyed by the Mongols and restored by Baybars when he rose to power.61 There is no detailed information as to how and under what circumstances the fortress was transferred to the Mamluks. During the same year (1263) Baybars ordered the destruction of the fortress at Mount Tabor. The fortress was built by al-ʿĀdil and al-Muʿaẓẓam in 1211,62 and counted as one of the largest in the region. In 1255 Mount Tabor reverted to the Franks. On receiving permission from the Pope the Order of the Hospitallers rebuilt and garrisoned the fortress.63 Baybars’ decision to destroy it probably stemmed from problems similar to those encountered by the Ayyubids. The need for an exceptionally large garrison and the high cost of maintaining the fortress may have been beyond the means of the Sultanate at the time. Mount Tabor was not an administrative center and though it commanded inland routes these were obviously not of great importance to the Mamluks. According to Burchard of Mount Sion, who wrote in 1283, Mount Tabor served as the Mamluk sultan’s hunting grounds.64 In 1264, a number of Syrian nuwwāb were sent to restore Shaqīf Tīrūn (Cave de Tyron), located 20 km northeast of Sidon. The fortress had been captured and destroyed by the Mongols and remained in ruins until Baybars ordered it to be rebuilt, armed and its storehouses filled with supplies.65 It is a fairly small fortress and its garrison would have been limited in number. It seems likely that Baybars decided to rebuild the site because it overlooked the coast. Its status was thus similar to that of Qāqūn. After the restoration of the Euphrates and Trans-Jordan fortresses owned by the Ayyubid princes and amirs, Baybars turned to the Frankish strongholds. A different and more complex set of considerations came into play when contemplating the conquest and restoration of the large inland Frankish strongholds, and the policy differed greatly from that followed by the Mamluks when dealing with the Frankish coastal fortifications.

96  Muslim Fortresses in the Levant The remaining Crusader enclaves along the coast did not have sufficient military power to face the Mamluks in battle or to threaten the new Sultanate. The possibility, however remote, that a new Crusade would arrive by sea to re-conquer Jerusalem and Egypt, was the main cause for the aggressive policy conducted by the Mamluks. The Ayyubid and Mamluk lack of a strong navy brought about the scorched earth policy that was carried out methodically against the Crusader strongholds along the coast. This policy had been well established by the end of Ṣalāḥ al-Dīn’s reign. The Mamluks did not regard the sea or the navy as important as their land forces, which enjoyed greater attention, funding and respect.66 The Mamluk navy did not raid Frankish strongholds and could not prevent the Franks from landing on the Sultanate’s coast.67 The navy was of modest size and it anchored only in Egyptian ports. Thus it seems that Baybars had no option but to follow Ṣalāḥ al-Dīn’s policy.68 In the years 1265–6 Caesarea, Arsūf, Haifa and Jaffa fell into Mamluk hands one after the other.69 A different way of guarding and supervising the coastal region without investing in a large navy, was by settling the area with Kurdish, Mongolian and Turkish populations loyal to the sultan.70 The conquest of Qāqūn can be linked to this policy of Baybars. Although Qāqūn is not located directly on the coast it kept a watchful eye on the Frankish movements along the shoreline and served as an administrative center for this new population which had recently been settled by the sultan. Safad and Hunin were conquered in the same year (1265),71 neither of which is located on the Sultanate’s frontier. Their Mamluk garrisons were never put to the test; they were never besieged or attacked and were seldom called upon to assist in large-scale campaigns. Their main role until the fall of the Crusader kingdom was to help establish the Mamluk regime and safeguard the local population. While Qāqūn may have served the new Muslim population that was settled in the region, Safad was officially turned into a niyāba – a large regional administrative center. In addition to refortifying the citadel Baybars built markets, a caravanserai and mosques.72 According to Johns, ʿAjlūn underwent a similar development.73 Although they were refortified and garrisoned, the military role of these fortresses was considerably reduced even before the Frankish kingdom fell. In 1268 Baybars conquered Qalʿat Shaqīf (Beaufort). The women and children were sent to Tyre, while the men were divided among the Mamluk armies that participated in the siege. The new section of the fortress that had been added by the Franks was destroyed. Once the siege ended and Baybars decided to hold on to the site and refortify it, the amir Ṣārīm al-Dīn Qāymāz al-Kāfará was appointed as its nā’ib. Twenty-six siege machines were positioned by the Mamluks around the fortress and parts of it may well have needed to be rebuilt. Baybars appointed the amir Sayf al-Dīn Balāban al-Zaynī, who was charged solely with the responsibility for the reconstruction work. The garrison that was set up in the newly acquired fortress was composed of mounted men (ajnād, ) and infantry (rajjāla, ).74 During the same year Baybars conquered Antioch, capital of the Crusader principality.75 Antioch’s citadel was given to Bīlīk al-Khaznadār and Badr al-Dīn Baysarī al-Shamsi, who were put in charge of organizing and distributing the booty. Once this was completed Baybars returned and burnt down both the citadel and the city. Antioch’s grim fate prompted nearby strongholds to ask for an amān (guarantee of safety).76

Laying the foundations  97 The amir Bīlīk al-Ashrafī granted them an amān and received the fortresses.77 In order to ensure their loyalty and neutralize their military strength, Bīlīk al-Ashrafī decided to imprison all the men.78 In the spring of 1268 three Frankish fortresses along the southern border of the principality of Antioch were taken by the Mamluks: Dayr Kūsh (Dargous), Dubbīn and Talmīs (Cavea Balmīs). Ibn al-Furāt describes those fortresses in the following manner: “These strongholds were a blockage in the throat and choking in the breast (of Islam).”79 The garrisons in all three fortresses surrendered, and they passed into Mamluk hands in a good state of preservation. All three are located along the northern section of the Orontes, close to the river bank. They had been taken by Hülegü in the winter of 1260 when he conquered Aleppo and were returned to Bohemond VI, ruler of Antioch, the original owner. Ibn Shaddād gives only a short account of those conquests. He concludes by noting that each of the three fortresses had a small adjacent suburb and a mosque and that the sultan appointed a governor and a judge.80 A comprehensive survey conducted by Major in the northern region of the Orontes has shown the exact nature of the strongholds. The term shaqīf indicates that they were in fact cave fortresses.81 Al-Quṣayr (Cursat), a Frankish fortress that belonged to the Catholic Patriarch of Antioch was taken by Baybars only in 1275.82 The sources do not indicate whether the Mamluks restored this site. The fortresses of the Military Orders were constructed and run almost like independent islands; in many ways they were self-sufficient. Garrisons at Safad, Marqab and Chastel Pelerin (ʿAthlīth) numbered over a thousand men. The Military Orders’ fortresses represent some of the finest contemporary military architecture, a great deal of thought and money having been invested in them, so that they could withstand the advances made by the Ayyubid armies in siege warfare.83 They show a fine understanding of defense: siege machines could be mounted on their towers,84 and large numbers of archers could be stationed along the towers and walls.85 Nevertheless, the fortresses fell one by one to the advancing Mamluk forces. While some were reduced in the course of a successful siege, others fell thanks to clever trickery. Qalʿat Shaqīf and Baghrās were conquered two years later in 1268.86 La Roche de Roussel (Çalan) and Port Bonnel (Ḥiṣn Rūsūs) were taken without battle; their garrisons simply abandoned the sites. In 1271 the Mamluks conquered the Templar fortress Chastle Blanc (Ṣafītha), Crac des Chevaliers (Ḥisn al-Akrād) 87and Gibelacar (Ḥiṣn al-ʿAkkār).88 The latter two belonged to the Hospitallers. Within six years (1265–71) the Mamluks had succeeded in destroying the entire infrastructure of the Templars in the East.89 During the summer of 1271 the Teutonic Order lost Montfort (Qalʿat al-Qurayn). Ibn ʿAbd al-Ẓāhir describes this affair in a few brief sentences, but he gives the Mamluks’ reason for taking this stronghold: and this Ḥiṣn al-Qurayn … was among the fortresses that would harm Safad … and after the conquest of Ḥiṣn al-Akrādand ʿAkkār he [i.e. Baybars] saw that he cannot leave its conquest to his successor.90 Large and forbidding fortresses fell within a few weeks. The lack of reinforcements was a fault the Franks could not overcome. Though the fortresses stored sufficient

98  Muslim Fortresses in the Levant quantities of water, arms and food, their fate was sealed if there was no military force available to threaten the besieging enemy camp. In addition to the decline in Frankish military abilities during the thirteenth century,91 one must consider the remarkable efficiency displayed by the Mamluk siege armies. They used every known method in the art of siege warfare: tunnels were dug under towers and curtain walls, pots containing nafṭ were thrown,92 ladders and ropes were used to scale fortifications and every type of siege machine was employed.93 In the early 1270s Baybars set out to eradicate the eleven Ismāʿīlī fortresses in the Nasseriyya Mountains (Map 3.3). The existence of an extreme Muslim sect whose loyalty to the Mamluks was rather doubtful was hardly conducive to the establishment of the Mamluk regime in Syria. In the spring of 1270 Masyāf was conquered and a Mamluk garrison established. In 1271–2 al-Khawābī, al-Ruṣāfa and ʿUlayqa

Map 3.3  Ismāʿīlī fortresses in northwest Syria

Laying the foundations  99 were taken; the latter was given a Mamluk garrison to safeguard it. In 1273 Qadmūs and Manīqa surrendered to the sultan. Al-Qadmūs, Manīqa, ʿUlayqa, Masyāf and al-Kahf were probably returned to the Ismāʿīlīs by 1326. Ibn Baṭṭūṭa, who passed through the region in that year, reports that the fortresses mentioned above can be entered only by members of the Ismāʿīlī sect.94 Al-ʿUmarī (d. 749/1349) mentions al-Kahf, al-Qadmūs, al-Khawābī, al-Ruṣāfa and Masyāf as stops along the barīd route.95 Thus they must have been integrated in the Mamluk array of fortifications. The conquest of the Ismāʿīlī strongholds illustrates the use of fortresses to strengthen Mamluk rule and subdue hostile populations. Very little field work has been done in this region and it is difficult to assess what, if at all, was the scale of Mamluk re-building in those fortresses. Almost parallel to the campaign in the Nasseriyya Mountains, Baybars took Ṣahyūn and Baalbek. Restoration work began almost immediately on the latter. The large inland fortresses were to assist in establishing Mamluk rule, to continue to serve as administrative centers, to see to law and order in the region, and to provide security for the local population and those who traveled along the Sultanate’s roads. Some of them stored military supplies such as siege machines and surplus grain that were transported to different parts of the Sultanate when required. In contrast to the policy implemented inland, along the Euphrates Baybars had chosen to restore only two fortresses. Throughout his reign the sultan was conscious of the need to preserve a fine balance between the strength and size of the garrisons and the size of his field armies in the Syrian cities and Cairo. The key seems to have been to hold fewer fortresses and invest in a strong mobile field army.

The restoration of fortresses during the reign of Qalāwūn Qalāwūn added a number of important strongholds; none, however, were located along the frontier. It seems that, rather than initiate the conquest and restoration of new fortresses, he devoted much of his time to finishing work that Baybars had begun. The fortresses of Safad,96 Baalbek and Ḥiṣn al-Akrād were completed during his reign.97 The rise of Qalāwūn to power did not bring any changes in Mamluk policy towards the Īlkhānid state. Towards the end of his first year in power, news of Īlkhānid preparations for a new invasion into Syria arrived in Cairo. In October 1280 the Īlkhānid army conquered Baghrās, Darbassāk and ʿAyn Tāb, all located on or near the main route leading to Aleppo and all owned at the time by the Mamluks, according to a treaty signed during Baybars’ reign. The Mamluk garrisons abandoned the fortresses and fled as the Īlkhānid army arrived. Aleppo too was deserted; the Īlkhānid army took advantage of the situation and looted the city.98 Getting reinforcements to the Mamluk fortresses in Cilicia became one of the main problems during Qalāwūn’s reign and in later decades. In contrast to the fortifications along the Euphrates that could receive help within days or a couple of weeks at the most, the Mamluk garrisons in Cilicia were often left to fend for themselves. Abū’lFidā’ describes one of the occasions on which the Mamluk garrisons abandoned the fortresses in Armenian territory during Ghazan’s invasion in 699/1299–1300.

100  Muslim Fortresses in the Levant When Ghazan came with the Mongol hosts to Syria, the Armenians had designs on the territories which the Muslims had conquered from them. The Muslims were unable to hold them, so the garrison-troops and foot soldiers there left them and evacuated them. The Armenians occupied them, regaining Hamūs, Tall Ḥamdūn, Kūbar, Sarfandkār, al-Naqīr. Of all these citadels, only Hajar Shaghlān was left to the Muslims. The Armenians occupied the other territories to the south of the river Jayhān.99 The problem of the defense of the Cilician frontier was only solved when the Armenian kingdom fell in the last quarter of the fourteenth century and its entire territory was occupied by the Mamluk sultanate.100 Although it was not the first time the Īlkhānid army managed to penetrate Syria in depth, the panicked desertion of Aleppo by its inhabitants and garrison may have emphasized the need to further invest in inland strongholds and strengthen the urban fortifications. Restoration of Aleppo’s citadel began in Qalāwūn’s reign.101 The strengthening of the Damascus fortifications had begun during Baybars’ reign, but most of the work was completed by Qalāwūn.102 Qalāwūn’s policy towards the Franks in the early years of his reign was significantly more moderate than that of Baybars. The sultan was determined to use all the diplomatic means available in order to keep the peace with the Frankish enclaves. The reason for this policy, so different from that of his predecessor, was probably the growing tension along the Mamluk–Īlkhānid border (news of Abagha’s coming invasion reached the Mamluk court towards the end of 1280).103 Qalāwūn knew that his army was not large enough and could not fight on two fronts at the same time. Thus, between the years 1282 and 1286 the sultan signed treaties with the Frankish authority in Acre, the commanders of the Templars in Acre, Chastel Pelerin and Tortosa and also the ruler of Tyre, Beirut and Sidon.104 This period of restrained policy came to an end at the beginning of 1285, when Qalāwūn besieged and took the Hospitallers’ fortress of Marqab (Margat) located in the southern region of the principality of Antioch. It was the first large Crusader fortress the Mamluks conquered and maintained on the coast. In deciding to keep and rebuild Marqab Qalāwūn strayed from the destructive policy conducted by Baybars along the southern coastal plain. The decision to garrison Marqab was probably the result of a number of developments in the region. There was a need to assert law and order amongst the Druze, the Maronites, the Mutawālī and the Nizārīs, who would not submit to Mamluk rule.105 Another reason for the conquest of Marqab was its location near the merchant route that runs parallel to the coast and links Cilicia and Syria. The commercial activity of the Genoese and the growth in the volume of trade justified the Mamluk need to control and supervise mercantile development along this route and direct part of the profits into the Sultanate’s treasury. The economic value of the northern coast may have grown considerably after the destruction of ports and towns along the Palestinian coast in the 1260s. According to Northrup, the reason behind the conquest of Tripoli in later years was that a large percentage of the naval traffic sailing parallel to the north Syrian coast dealt in slaves purchased for the sultan’s army in Cairo.106 The fear that the Franks or for that matter any one at all would

Laying the foundations  101 interrupt the trade along this route may have caused Qalāwūn to increase security by holding the small harbors in this section of the coast.107 The location of Marqab and Ṣahyūn, between the northern harbors of al-Suwaydiyya (St Simon) and Latakia that served the merchants of Aleppo and the port of Tortosa (south of Marqab) that served Homs, enabled the Mamluks to keep a watchful eye on all three harbors. From Abū’l-Fidā’s description it appears that defensive aspects were taken into account. The decision to keep the fortress was reached even before it was captured. Once his mind was made up Qalāwūn tried to conclude the siege as soon as possible and in a peaceful manner in order to reduce the damage to the structure of the fortress from the Mamluk siege machines: When the mines had got the better of the castle-walls, its people asked for terms. The sultan agreed out of desire to keep it in being, for if he had taken it by the sword and destroyed it, its restoration would have been troublesome.108 Qalāwūn’s debate concerning the fate of Marqab – to restore or destroy, as described by Ibn ʿAbd al-Ẓāhir – gives a rare insight into the manner in which decisions of this type were taken. The final decision was made by the sultan alone, but it was preceded by a discussion with the amirs of the highest rank. Some differences of opinion were voiced but the sultan supported the idea that the fortress should be kept, and the damage repaired. Qalāwūn declared that the importance of Marqab lay in its ability to provide support and defense for the fortresses in the close vicinity.109 The taxes collected from the local population financed the restoration of the fortress, its daily running and the garrison’s pay. A large garrison was established, numbering a thousand men from two military groups: the archers (uqjyya, ) and the troops (jarjya, ), who were commanded by amirs from the aṣḥāb (followers). In addition there were Mamluks recruited from the Baḥriyya, the Sāliḥiyya and the Manṣūriyya, four hundred craftsmen and one hundred and fifty soldiers.110 Once Marqab was restored, Qalāwūn turned to the fortress of Maraqiyya. Maraqiyya was built on a small rocky island 50m from the coast.111 The island fortifications were renewed by Bartholomew de Ravendel and funded by the Prince of Tripoli. A successful siege required the support of a naval force. The Mamluks, however, could not assemble a fleet for this campaign. Maraqiyya was eventually taken after Qalāwūn sent a threatening letter to the Prince of Tripoli, who, according to the sultan, was responsible for the Frankish renewal of the fortress. The sultan’s terms were clear and simple. The Prince of Tripoli and Bartholomew, lord of Maraqiyya, were to demolish the fortress. If the prince did not comply, the sultan, at the head of his armies would conquer parts of his principality. The Franks quickly came to terms with the sultan’s decree. A group of Mamluks and two high ranking amirs were sent to supervise the destruction of the fortress.112 A large percentage of the profits made by the Frankish principalities and the Military Orders that maintained fortresses close to the shoreline was due to their role as mediators and providers of various harbor services to merchants. The conquest of Marqab and Qalāwūn’s military campaign along this short stretch of coast line that included the harbors of al-Suwaydiyya, Latakia, Tortosa and Jubail stemmed from the economic importance and the potential profits that were to be made. The

102  Muslim Fortresses in the Levant harbors mentioned above prospered under Ayyubid rule and were mostly used to export Syrian goods.113 This commercial activity largely depended on peace in the region and the safety of the inland trade routes. Marqab did not protect a frontier. Its Mamluk garrison was mainly responsible for seeing to the local sects who refused to submit, watching over the coast, safeguarding the harbors, merchants, and roads leading from the mainland to the sea. In treaties signed by both Baybars and Qalāwūn (prior to the conquest of Marqab) the Mamluk sultans received a high percentage of the taxes collected in the ports. Baybars ����������������������������������������������� negotiated with the Order of the HospiAfter the conquest of Ḥiṣn al-Akrād,������������������������������������������������ tallers at Marqab. According to the treaty of 669/1272 the Hospitallers were obliged to give the sultan half of their profits from taxes collected at the ports on goods exported and imported, and any other commodity that the Order taxed.114 A treaty signed by Qalāwūn and Bohemond VI of Tripoli determined that taxes collected in the port of Latakia would be divided between the two.115 Although all this provided a fine income, with the conquest of Marqab the Mamluk sultan could reap the entire revenue without having to share it. Qalāwūn was no doubt confident that commerce would continue to prosper. The fact that the Franks no longer ruled the region would not stop Italian and other merchants from trading. In 1288 Qalāwūn promised to safeguard the Genoese merchants who chose to trade in Syria.116 Four years after the fall of Marqab the sultan besieged Tripoli (688/1289). The small Cypriot fleet that came to the help of the besieged was unable to save the town. Tripoli fell within a month. The sultan ordered its destruction and a new city was built near the riverbank. With the conquest of Tripoli two more harbor towns – Beirut and Jubail – and the fortresses in their close vicinity were delivered to the Mamluks.117 The news concerning the conquest of Tripoli reached the Armenian king within days, as he could not allow himself to remain indifferent, he quickly dispatched a messenger to the sultan. Ibn al-Furāt tells us that the messenger “asked for the sultan’s mercy and tried to appease his mind.”118 The sultan demanded the two Armenian fortresses Bahasnā and Marʿash, northwest of al-Bīra. The messenger returned to his king with the sultan’s demands. The Armenian king firmly refused, with profound apologies (iʿtidhārāt, ); instead he offered the sultan a large sum of money and numerous presents.119 The sultan’s demand for the two fortresses and the king’s determination to keep them were to surface frequently in future negotiations during Qalāwūn’s reign. One of the interesting results of Qalāwūn’s campaign, which began with the conquest of Marqab and ended with the fall of Tripoli, was the temporary or partial evacuation of Ḥiṣn al-Akrād. According to Maqrīzī, the garrison under the command of Sayf al-Dīn Balabān al-Ṭabbākhī and the vanguard (probably arriving shortly before Tripoli was besieged) left the fortress early in the summer of 1289 for Tripoli, which had just been taken. Tripoli was given to al-Ṭabbākhī and the town received the status of an administrative regional centre- a mamlaka (or niyaba). When al-Ṭabbākhī left, he took with him a hundred and fifty soldiers, fifteen amirs of the rank of ten and ten other amirs.120 Maqrīzī does not give any further information concerning Ḥiṣn al-Akrād. According to Irwin, shortly after the taking of Tripoli Ḥiṣn al-Akrād ceased to be the base for Mamluk military operations in the area and most of their activity shifted to

Laying the foundations  103 Marqab.121 It seems that after the fall of the principalities of Antioch and Tripoli there was no real need to hold two large fortresses in the region. A large garrison was therefore established in Marqab and the number of men in Ḥiṣn al-Akrād reduced. Marqab’s proximity to the coast and the ports probably gave it preference over Ḥiṣn al-Akrād. Qalāwūn’s conquests in northwest Syria ended with the fall of Sahyūn. The fortress was first taken by Baybars. It was later restored by the rebellious Mamluk amir Sunqur al-Ashqar who turned it into the center of a small semi-independent principality. In 685/1287 the sultan took action against Sahyūn, and the fortress fell within a few weeks. Qalāwūn decided to maintain it and appointed a new nā’ib.122 Qalāwūn’s reign was shorter than that of his predecessor. Even if we take this into account, the number of fortresses he conquered and the scale of building during his reign were significantly less. Most of the work was an attempt to complete projects started by Baybars. The infrastructure laid down by Baybars answered most of the military needs of the Sultanate, so that it seems there was less pressure on Qalāwūn to conquer and restore additional fortresses. Like his predecessor, he carefully weighed the abilities of his military forces against the number of fortresses he could maintain, and balanced his military strategy against sound economic calculations.

Contributions made by al-Ashraf Khalīl The three short years of al-Ashraf Khalīl’s rule were most impressive as far as military achievements and accomplishments in the field of military architecture went. The conquest of Acre was followed by the surrender of all Frankish coastal strongholds. A few months later the sultan marched his army northeast to the Euphrates and besieged the large Armenian fortress of Qalʿat al-Rūm (Hromgla). Like his father, al-Ashraf Khalīl was obliged to continue work that he had not initiated. The fortifications of Aleppo were completed in 690/1290–1.123 Damascus was strengthened once again,124 and the fortress at Baalbek was further enlarged.125 Following the treaty with the Armenian king in 692/1293 the fortress at Bahasnā was given to the Mamluks. The sultan appointed the amir Sayf al-Dīn Tughan al-Manṣūri as its nā’ib and ordered him to strengthen the fortress.126 One of this sultan’s most impressive military moves was the conquest and restoration of Qalʿat al-Rūm (691/1292).127 A two-line inscription at the citadel of Homs commemorates this event.128 The reasons for the decision to take and rebuild this fortress are not evident. Stewart who thoroughly examines the sources concerning the conquest of Qalʿat al-Rūm, writes: “It is not clear if there was any specific reason why al-Ashraf chose this year to attack Hromgla, other than the fact that he wanted to and was able to.”129 It is difficult to find a fortress conquered in order to satisfy a sultan’s whim. The logistics, costs and difficulties of the campaign and restoration of a fortress the size of Qalʿat al-Rūm were no small matter. The organization of such a campaign and the decision to maintain the stronghold required clear and sound reasoning: political, economic and military factors, one or all of these together had to be considered. Fortresses of no immediate use to the Mamluks were left in ruins, neutralized by killing the garrison, expelling the soldiers and able-bodied men or taking them

104  Muslim Fortresses in the Levant hostage. Food supplies were used or destroyed and arms confiscated. When we examine the decisions taken by both Baybars and Qalāwūn it becomes the more obvious that their rebuilding and garrisoning of sites was carried out after considerable thought. Qalʿat al-Rūm was no different. Qalʿat al-Rūm is situated 40km up the Euphrates, north of al-Bīra, on a ford that had been used throughout the Hellenistic and Roman periods. Prior to its conquest by the Mamluks it was the seat of the Armenian Catholicos. The existence of a large fortress on a shallow ford close to the Īlkhānid-Mamluk border only a few miles from one of the most important Mamluk strongholds (al-Bīra) was almost certainly perceived as a threat. It is interesting to note that Qalʿat al-Rūm is not mentioned in any of the treaties signed between the Mamluks and the Armenians during the rule of Baybars or Qalāwūn. A failed, first attempt to take the fortress had been made in 1279 during the brief reign of Berke Khan (1277–9).130 The news of al-Ashraf Khalīl’s siege soon reached the court of the Īlkhān. His quick reaction indicates just how important the fortress was to the Īlkhānids. Geikhatu (the new Īlkhān who assumed power in the summer of 1291) acted in great speed, sending two contingents to reinforce the fortress garrison, while his coronation ceremonies were still going on. Qalʿat al-Rūm fell before the Īlkhānid reinforcements arrived.131 Once the fortress was taken the sultan changed its name to Qalʿat al-Muslimīn. ʿAlam al-Dīn Sanjar al-Shujā‘ī was put in charge of the reconstruction and according to Abū’l-Fidā’ he “built and refortified it very completely.”132 The timing of this conquest is not difficult to explain. The sultan was exploiting the military momentum gained after the victory over the Franks in Acre and the new political situation created with the final destruction of the Crusader kingdom. A letter addressed to the High Council (al-majlis al-sāmi), and to the honorable Qāḍī gives an account of the siege and describes Qalʿat al-Rūm as the gate to the lands east of the Euphrates.133 The letter was distributed throughout the Sultanate, announcing the sultan’s victory and future plans. It ends with a statement that indicates al-Ashraf Khalīl’s ambitions. 134

And this is what will be after this conquest, if God wills, I shall conquer the East, Rūm and Iraq, and rule the lands from the east where the sun rises to the west where it sets. The location of Qalʿat al-Rūm made it an important acquisition. Above all, this Euphrates crossing was a link between east and west. A network of roads along and across the river had existed for centuries and had served empires and kingdoms from the earliest times.135 The decision indicates a shift in Mamluk policy; until then the Sultanate had seldom sent its armies across the Euphrates, except for a number of raids. Qalʿat al-Rūm was apparently to be the key to future conquests in the east. Al-Ashraf Khalīl was threatening the lands in the heart of the Īlkhānid state, hoping to reclaim Baghdad, and bring it back under Muslim rule.136

Laying the foundations  105 Table 3.2  Fortresses conquered by the Mamluks (see Map 3.4) Fortresses conquered by the Mamluks

Owners prior to the The fate of the fortress Mamluk conquest under Mamluk rule

al-Ṣubayba 1260 al-Bīra 1262 ʿAjlūn 1263 (?)

Ayyubids Ayyubids Ayyubids

Rebuilt Rebuilt Rebuilt

Karak 1263 Shawbak 1263

Ayyubids Ayyubids

Rebuilt Rebuilt

Al-Raḥba 1264 Baalbek 1271 al-Ṣalt 1263 (?) Ṣarkhad (1270–1)1 Shayzar (1260)2 Qalʿat Jaʿbar ʿAyn Tāb Buṣra Mt. Tabor 1263 Belvoir (1263) Shaqīf Tīrūn 1264 Caesarea 1265 Haifa 1265 Arsūf 1265 Qāqūn 1265 Hunin 1265 Safad 1266 Beaufort 1268 Jaffa 1268 Antioch 1268 Baghrās 1268 Dayr Kūsh 1268 Dubbīn 1268 Talmīs 1268 ʿAkkar 1271 Ṣafītha 1271 Montfort 1271 Crac des Chevaliers 1271 Ṣahyūn 1272 al-Quṣayr 1275 Marqab 1285 Tripoli 1289 Maraqiyya 1289 Acre 1291 Tall Ḥamdūn 1266

Ayyubids Ayyubids Ayyubids Ayyubids Ayyubids Ayyubids ? Ayyubids Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks Franks

Rebuilt Rebuilt Rebuilt Rebuilt Rebuilt Rebuilt Unknown Rebuilt Destroyed

Franks Franks Franks Franks Franks Franks Armenian

Rebuilt Unknown Rebuilt Rebuilt Destroyed Destroyed Rebuilt

ʿAmūdin 1966 al-Tīna 1966

Rebuilt Destroyed Destroyed Destroyed Rebuilt Rebuilt Rebuilt Rebuilt Destroyed Destroyed Rebuilt Rebuilt Rebuilt Rebuilt Unknown Destroyed Destroyed Rebuilt

Remarks

The sources do not mentioned exact year of conquest or restoration. Destroyed by al-Ashraf in 1294

Restored in 1333–4

Kawkab

Q. Shaqīf

Chastel Blanc Q. al-Qurayn Ḥisn al-Akrād

Rebuilt on a new site Rebuilt in the last decade of the 13th c. Adamodana Al-Tinat, Canamella

106  Muslim Fortresses in the Levant Table 3.2 ctd Fortresses conquered by the Mamluks

Owners prior to the The fate of the fortress Mamluk conquest under Mamluk rule

Remarks

Darbassāk 1268 Raban 1268 Marzabān 1268 Shīh al-Ḥadīd 1268 Bahasnā 1268 Anavarza 1268 Kaynūk (Göynük) 1273 Tall Bashir 1285 Gargar 1285 Kakhtā 1285 Mar‘ash 1289 Qalʿat al-Rūm 1292 Hāmīs 1298 Kuwayrā 1298 Nghir 1298 Hajar Shaghlan 1298 Sarfandkar 1298 Tumlu Masyāf 1270 Ruṣāfa 1271

Armenian Armenian Armenian Armenian Armenian Armenian Armenian

Rebuilt Rebuilt Rebuilt Rebuilt Rebuilt Rebuilt Destroyed

Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty

Armenian Armenian Armenian Armenian Armenian Armenian Armenian Armenian Armenian Armenian Armenian Ismāʿīlī Ismāʿīlī

Unknown Rebuilt Rebuilt Rebuilt Rebuilt Rebuilt Rebuilt Rebuilt Rebuilt Rebuilt Unknown Rebuilt Destroyed

Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty Handed over in a treaty

ʿUlayqa 1271 Khawābī 1273 Manīqa 1273 Kahf 1273 Qadmous 1273 Quliyʿa 1973 Sarmīn Jabal Sumaq Jabal al-Baḥrā

Ismāʿīlī Ismāʿīlī Ismāʿīlī Ismāʿīlī Ismāʿīlī Ismāʿīlī Ismāʿīlī Ismāʿīlī Ismāʿīlī

Rebuilt Destroyed Destroyed Destroyed Destroyed Destroyed Destroyed Destroyed Destroyed

Returned to the Ismāʿīlīs according to Ibn Baṭṭuṭa

Notes 1 Meinecke, M., “Ṣalkhad,” EI2 8: 995. The date of reconstruction is according to inscriptions found on the site. 2 Mouton, J. M., “Shayzar,” EI2 9: 441. Dated according to inscriptions found on the site.

Laying the foundations  107

Map 3.4  Mamluk fortresses in relation to the Īlkhānid state

108  Muslim Fortresses in the Levant The following account represents one of the most unusual examples of al-Ashraf Khalīl’s policy concerning fortifications. The fortress of Shawbak, taken by Baybars in 1263, was destroyed by al-Ashraf Khalīl in 692/1293–4. The only reason the sources give for this is the growing strength of the local Bedouin, who slowly gained control of the site, settled inside the fortress and probably disrupted security in the region. Rather than send a Mamluk contingent from Cairo to suppress the Bedouin, al-Ashraf Khalīl decided to destroy Shawbak. ʿIzz al-Dīn Aybak al-Afram al-Ṣaliḥī was ordered to carry out the sultan’s plans. Al-Afram disagreed and voiced his opinion before al-Ashraf Khalīl: Oh prince how can you destroy such a fortress, it is a well built site where people dwell, and is amongst the strongest Muslim strongholds.137 The sultan did not change his mind and punished al-Afram severely for what he thought was bad advice and a somewhat arrogant stand. Al-Afram was arrested and his vast property confiscated.138 Al-Aynī, who describes this incident, criticizes the sultan’s decision and says that the destruction of Shawbak showed poor judgment.139 Table 3.2 sums the conquests and work of the first three Mamluk sultans. ʿ

The Cilician question: Mamluk fortresses within the boundaries of the Armenian kingdom The Armenian kingdom was perceived as the northern “entrance gate” to the Mamluk sultanate. Furthermore, parts of Cilicia were still occupied by the Ayyubids in the mid thirteenth century. The area between al-Bīra and ʿAyn-Tāb was officially part of al-Nāṣir Yūsuf ’s kingdom.140 Between 660/1262 and 662/1264 five raids were carried out into northern Syria. Apart from the fourth in which the army of King Het‘um was reinforced by contingents of Īlkhānid soldiers and Frankish knights from Antioch, the campaigns were initiated and organized by the Armenians. None of the campaigns achieved lasting political or territorial results.141 The Armenian invasions had a long and lasting impact on the relations with the Mamluk sultanate. During most of the second half of the thirteenth century, from as early as 1266, the Mamluks tried to establish a firm hold in Cilicia by restoring and strengthening the fortresses they conquered. Apart from Mongol-Armenian attempts to invade Syria from Cilicia, a number of plans drawn up in Europe for the launching of a new Crusade assigned Cilicia an important role.

Plans for a new Crusade, and the prospect of a Mongol, Armenian and European alliance against the Mamluks A series of new grand campaigns was being planned in western Europe during the second half of the thirteenth and the first decades of the fourteenth century. The aims of this new Crusade were to conquer Egypt and the territories the Franks had lost in the East. After the fall of Acre (1291) a number of proposals were written, some of which were presented to the Pope.

Laying the foundations  109 In a number of such proposals, written by different authors, Cilicia was to serve as a base for organizing the armies before entering Syria. The Mamluks must have been aware of this new European initiative. Thus, part of the need to demonstrate constant military presence in Cilicia probably arose from the fear of an alliance of European, Armenian and Īlkhānid forces. This idea is conveyed in the accounts written by Het‘um and by Fidenzio of Padua.142 In reality, though, the Armenian kingdom had lost much of its strength due to internal strife over the throne; and the will and ability of European rulers to organize a new Crusade is questionable.143 During Philip IV’s reign (1285–1314) a program for a new Crusade written by a certain Philip of Taranto was presented to the Pope. This Crusade was to conquer Byzantium, cross the Armenian kingdom and march to the Holy Land. Those plans were forgotten soon after the death of the Pope and of the king of France in 1314.144 One of the figures who supported the idea of a new Crusade to the East and actively tried to promote and organize a campaign was Marino Sanudo, a Venetian merchant and chronicler (1270–1343). Much of his life was devoted to planning a Crusade that was to revive the Kingdom of Jerusalem, though emphasis lay on the conquest of Egypt. He prepared a written account titled “Liber secretorum Fidelium Crucis super Terrae Sanctae Recuperatione et Conservatione” that was read and accepted by Pope John XXII in 1321. Marino Sanudo had traveled widely in the Eastern Mediterranean and was acquainted with the geography, climate, culture and population of the region. His valuable knowledge made him an important figure whose opinions and advice were discussed in the courts of popes, monarchs and the nobility. His book was presented to popes Clement V and John XXII, Edward II of England and to two French kings, Charles IV and Philip VI.145 The economic importance of the Armenian kingdom was much emphasized in this composition. According to Sanudo, the Mamluks could only be weakened and eventually defeated by an attack on the Sultanate’s revenues. A trade embargo should be imposed, the Egyptian coast besieged and the land routes running from central Asia, India and Persia to the Cilician harbors opened to trade. This land route was to replace the sea routes that went from India to Aden and to Alexandria via the Red Sea. 146 The ocean trade routes that were controlled by Egypt from as early as the tenth century were still an important source of revenue in the early Mamluk period.147 An important issue raised and stressed by Sanudo was the need to damage and severely limit the Mamluk slave trade which provided the core of the Sultanate’s army. 148 Sanudo’s advice eventually influenced some of the decisions made by European rulers. In the late thirteenth and early fourteenth centuries a number of initiatives were made to send European forces to strengthen the Armenian kingdom. While most of those enterprises never passed the final planning stage, some actually set out but failed to reach their destination. In the late thirteenth century Pope Boniface VIII (1294–1303) asked the Armenian king to try and maintain his position until the European force reached Asia Minor; it is not quite clear if a force was ever sent.149 During the reign of Pope Clement V (1305–14) a Genoese force seems to have been preparing to leave for the Armenian kingdom, though apparently it never set sail. Assisted by the Order of the

110  Muslim Fortresses in the Levant Hospitallers, the Pope managed to summon a force of 5,000 men that was to spend five years in the Armenian kingdom. This force was to be the vanguard of the Crusade headed by Philip VI of France. The contingent left for Asia Minor but its voyage ended in Rhodes.150 During the years 1300–4 the Order of the Hospitallers had sent reinforcements to the Armenian kingdom, but it could not maintain this support for long and it soon came to a halt. In 1306 the Armenian king Leon VI sent a delegation to the papal court demanding a new Crusade, in order to rescue his kingdom that had been deserted and betrayed by the Īlkhānid state. A similar request was sent in the following year by King Oshin (1307–20).151 The most impressive plan of all was proposed by the Īlkhān Öljeitü in 1307. The Īlkhānid delegation which came to Poitiers asked the Pope to support a Crusader army that would march to Cilicia. The Īlkhān committed himself to send 100,000 mounted men, 200,000 horses and 200,000 rations of grain to meet the Crusader army when it reached the Armenian kingdom.152 Returning to our Italian advocate of the Crusades, Marino Sanudo not only roamed through the courts of Europe voicing his ideas on the economic importance of Cilicia, but even raised funds to assist the Armenians in warding off Mamluk raids and full-scale campaigns.153 In 1323, he proposed sending a French vanguard of 10 ships carrying 1,300 knights and their mounts. The French king Charles IV feared this force would not suffice, and claimed that the Armenians could not be trusted. He insisted that the number of men be doubled. This of course raised the cost of the expedition. The size of the force supposed to arrive by sea encumbered the entire plan, which was eventually dropped mainly because of the great expense involved.154 Although the ideas of Marino Sanudo were not implemented, Pope John XXII clearly understood the importance of maintaining the Christian presence in Asia Minor. He aimed to restore Hospitaller property,155and promised and convinced the Armenians that the Order would contribute to the defense and safety of their kingdom. A slightly different plan for a large Crusader force was drawn up during the reign of Philip VI (r. 1328–50), but it never materialized due to the growing political friction between the European rulers.156 The Armenian historian Hetʿum (Hayton), who in 1306 was living close to the papal court at Poitiers, was the author of yet another book concerning the revival of the Crusades titled: ‘La flor des estoires la terre d’Orient’. It was written in answer to a direct request from the Pope. The book contains Hetʿum’s own ideas as to the route this new Crusade should follow.157 It is not surprising that he suggests attacking Syria, and that Cilicia should be used as a meeting and organization point before heading into Syria. The Armenian kingdom, he wrote, had suitable harbors, numerous horses and was rich enough to provision the armies and their mounts with foodstuffs and fodder.158 His ideas seem to have been ignored. The Armenians had probably lost much of their credibility due to the schism between the two churches (Rome and the Armenian church) and the political chaos within the kingdom. Another proponent of a new Crusade to the East was Otto of Grandson, who suggested sending a European force to the Port of Ayās where it would be met by the Mongol and Armenian armies.159 Since he himself had commanded a force that

Laying the foundations  111 defended Acre in 1291 and was perceived by many as a fierce and brave fighter, his opinions aroused some interest in the various European courts. Throughout this period European presence in Cilicia was negligible. Reinforcements sent from the west amounted to small contingents of a few thousand men at the most. It seems, however, that the rumors of preparations for a new Crusade, the few scant European attempts (that mostly failed) to assist the Armenians, and the idea of creating a joint force with the Īlkhānids were enough to alert the Mamluks.160 It is against this background that we can understand their efforts to try and establish a Mamluk base and permanent footing in Cilicia. Thus, they emphasized their presence by occupying Armenian strongholds, maintaining Mamluk garrisons, raiding the kingdom and leaving behind trails of destruction. The Mamluk idea behind maintaining a permanent force in Cilicia was to create a secure and controlled zone that would protect the northern Syrian towns: a chain of fortresses that would raise the alarm and hold back the enemy until the arrival of Mamluk reinforcements. Although the Armenians and Īlkhānids seldom invaded Syria from the north, the Mamluks remained alert. The fortresses rebuilt by the Mamluks in Cilicia varied in plan and method of construction as well as in building materials. Some of the strongholds were first established by the Byzantines; most were built by the Armenians and a few by the Military Orders. Because of the instability along the Mamluk-Armenian frontier, fortresses often changed hands. The lack of continuity and the fact that they were often held for short periods, makes it difficult to establish their exact lineage. Most of the strongholds within the Armenian kingdom that passed into Mamluk hands are mentioned in a series of treaties signed between the two sides during the second half of the thirteenth century and the early decades of the fourteenth. The treaties seem to have left the Armenians little room for negotiations. The military balance tilted in favor of the Mamluks and it changed only when the Armenians received assistance from the Īlkhānids. The Īlkhānid state, however, did not always rush to support, defend or fight beside its client kingdoms. In 664/1266, when al-Malik al-Manṣūr, the Ayyubid lord of Hama, set out with a Mamluk army towards Cilicia,161 the Armenian king Hetʿum tried to recruit help from the Īlkhānid army. The Īlkhān refused due to problems on his eastern border.162 In 1271 matters evolved in quite a different manner. The Mamluk campaign was canceled or postponed for fear that Abagha would send a reinforcement of 20,000 soldiers to the Armenian kingdom.163 Although the Īlkhānid state carried considerable weight and could easily change the balance of power in the region, it was the Mamluks who dictated and decided the main clauses in many of those treaties. Whatever the two sides ultimately agreed upon, there was always a clause that required the Armenians to surrender a number of fortresses to the Mamluks. In addition there was always a clearly phrased clause that forbade the Armenians to restore existing fortresses or build new ones. A condition of this type was inserted in the treaty signed in 684/1285.164 That this clause was included was due to the remarkable capacity the Armenians had demonstrated in refortifying fortresses abandoned and damaged by the Mamluk garrisons. In certain cases the Armenians initiated the building of new fortresses in order to improve the safety of central trade routes.165

112  Muslim Fortresses in the Levant The survey below of Mamluk raids into the Armenian kingdom and the treaties that followed defines the fortresses the Mamluks tried to gain control over. By following the military and political maneuvers involved one may be able to better understand the problems, the advantages and disadvantages of fortifications in this particular region. Mamluk presence in Cilicia: raids followed by treaties The first Mamluk assault on the Armenian kingdom (664/1266) was probably a direct response to a sequence of raids carried out by the Armenians into northern Syria in 1262–4. Bar Hebraeus gives a very different explanation for this first raid. According to his account a Mamluk delegation left for the Armenian court in 1266, with the following offer: the king would surrender and pay tribute to the Sultanate. In return Baybars would open the routes from Syria to Cilicia and make safe the transport of food supplies to the kingdom.166 From the choice of words in Bar Hebraeus’ account it appears that the matter was of a sensitive nature. King Hetʿum I would have given the sultan a positive answer if he had not feared Abagha’s reaction. The Mamluk raid is described by both Smpad and Bar Hebraeus in a similar manner. The capital, Sīs, was sacked, the population massacred,167 and the fortresses along the frontier – Adana, Ayās, Mesis and Ṭarsūs, Tall Ḥamdūn, Karanjil, Ḥamūṣ, al-ʿAmūdayn (which belonged to the Teutonic Order) and Sarfadkār – left in ruins (Map 3.5).168 In 666/1268 King Hetʿum surrendered several fortresses to Baybars. In return the sultan released the king’s son who had been taken hostage two years earlier.169 The fortresses Baybars received were Darbassāk, Marzabān, Shīḥ al-Ḥadīd, Raban, Bahasnā and al-Zarb.170 The king changed his mind at the last moment and insisted on retaining Bahasnā. The sultan, however, did not relent. Once Hetʿum was informed of the sacking of Antioch, he agreed to the initial terms of the treaty. As a guarantee, a hostage was held in the sultan’s court until all the Armenian fortresses were safely transferred to the Mamluks.171 The second Mamluk raid took place in 671/1273. The army left Aleppo, and following Baybars’ orders took the Armenian fortress of Kaynūk (Göynük), situated on the bank of the Aq Su River, northeast of Marʿash. The Mamluks did not garrison the site. The direct reason for the raid was an attack by the inhabitants of Kaynūk on Mamluk merchants and agents.172 In 673/1274–5 the Mamluks again invaded Cilicia, because the Armenians had neglected to pay the tribute determined in the treaty signed in 1268. Furthermore, their attempt to restore a number of fortresses and build a new fortress at the foot of the Taurus mountains was a clear breach of the treaty.173 There is an almost tangible element of revenge, and one of the raid’s aims seems to have been to make sure the Armenians were aware of their position, to punish them for being such loyal allies of the Īlkhānids and eager participants in each Īlkhānid invasion into Syria. A chain of destruction and ruin was left throughout the country and a large amount of booty was collected. The Mamluks did not take further advantage of the kingdom’s weakness and did not demand any of its strongholds as they had done on previous occasions.

Laying the foundations  113

Map 3.5  Fortresses mentioned in the treaties between the Mamluk sultanate and the Armenian kingdom

Bar Hebraeus gives this raid a long and detailed account. To begin with, he says, the Armenians managed to drive back the Mamluks with an army that numbered 500 cavalry and foot soldiers. The Mamluk army reorganized and returned to attack Ayās. Many of the inhabitants found refuge in the small island close to the town. From Ayās, the Mamluk army turned to Mopsuestia (Mesis). The population was massacred and the town burnt down. The Mamluks then marched to Ṭarsūs, Cyricus and Sīs. The population of Sīs found refuge in its citadel though it was eventually set on fire. The Mamluk army was joined by a Türkmen force numbering 10,000 men, and laid waste to large parts of the country.174 A treaty between Qalāwūn and Leon III (r. 1269–89) was signed in 684/1285, for a period of ten years. Many of the fortresses handed over to Baybars are listed again. This evidently means that before Qalāwūn’s rise to power in 1280 the Mamluks had lost many of the Armenian fortresses they had gained through military and diplomatic maneuvers. The Armenians’ contribution to the destruction of Aleppo and their participation in the battle of Homs (October 1281), which ended in Īlkhānid defeat, forced Leon III to reach an agreement with the Mamluks. The treaty contains a list of fortresses surrendered to the Mamluks shortly after or while negotiations were being held. The following strongholds are noted: Baghrās, Darbassāk, Tall Bāshir, Gargar and al-Kakhtā. One of the clauses repeats an earlier

114  Muslim Fortresses in the Levant treaty forbidding the Armenians to restore and build fortresses on their lands.175 Within four years (1289) a new agreement was signed. It was initiated by the Armenian king immediately after the Mamluk conquest of Tripoli. The sultan agreed, and Marʿash was handed to the Mamluks. This fortress, apparently, was soon back under Armenian rule, since it is mentioned a few years later in an agreement signed in 1293. While al-Ashraf Khalīl was preparing the conquest of Qalʿat al-Rūm plans were also being made for the capture of Sīs. By then the Armenians had suffered several devastating raids and the king hastily sent a large and honorable delegation carrying precious gifts to the court in Cairo (692/1293). It was decided that the Mamluks would be given three sites: Tall-Ḥamdūn, Marʿash and Bahasnā. Once again the Armenians tried to bargain with the sultan over Bahasnā. Al-Ashraf Khalīl refused to compromise. A nā’ib was appointed and a garrison sent to man the fortress. It is difficult to explain this recurring procedure of the Armenians, who were determined to try and hold Bahasnā. Although the sources do not provide a detailed description of the fortress and its surroundings, the nature of the negotiations between the sultan and the Armenian delegation implies that both sides attributed great importance to this site. Al-Ashraf Khalīl was considerably more determined than his father, who had decided to yield to the Armenians’ request after receiving their marvelous gifts (treaty of 1289). The gifts did not appease Al-Ashraf. The sultan followed the departing delegation and warned the Armenian herald that if the fortress was not handed back he would not hesitate to raid the kingdom. Al-ʿAynī gives an interesting historical account that may partially explain why the Mamluks wished to command this site. The land round Bahasnā, he writes, had numerous rich estates that were tilled and sown. In the past the estates had belonged to the Ayyubid ruler of Aleppo, al-Malik al-Nāṣir. After the Mongol invasion (1260), the Armenian king asked Hülegü to grant him Bahasnā. From al-ʿAynī’s point of view, Bahasnā and the surrounding area belonged to the Mamluks due to the fact that thirty years earlier this had been a district of Aleppo.176 In the late 1290s the Mamluks took advantage of the internal chaos that struck the Armenian kingdom and Lachin invaded Armenia with a large army made up of Mamluks from Egypt and troops from Safad, Tripoli, Damascus, Hama and Homs.177 Their aim was to take Sīs. In April 1298 Tall Ḥamdūn was besieged and on 23 June the herald announced its fall to the people of Damascus. Marʿash, Nujayma,178 and Hāmīs (Çardak Hamus),179 probably fell shortly afterwards. According to al-Yūnīnī, the sultan’s governor issued an order to recruit soldiers from every level of society for the garrisons of Tall Ḥamdūn, Marʿash and Nujayma. He adds that the garrisons were well equipped. In August 697/1298 an Armenian delegation representing King Hetʿum II went to Cairo to ask for fair terms in negotiations for a truce.180 If there were any results al-Yūnīnī does not mention them. Abū’l-Fidā’ describes the change of rulers and the accession of Constantine to the Armenian throne. The new king was quick to make a political declaration. He promised to be a loyal subject of the Mamluk sultan, and to act as governor on behalf of the sultan in Cilicia. The Jayḥān River was to be the border between the two entities; lands and fortresses south of the river were turned over to the Mamluks. The following fortresses were given to the sultan, according to Abū’l-Fidā: Hāmīs, Tall Ḥamdūn, Kuwayrā, Hajar Shaghlan, Sarfandkar, Marʿash

Laying the foundations  115 and al-Naqīr. Abū’l-Fidā’ says that Lachin garrisoned them all; he criticizes the sultan and says this was a mistaken move. Abū’l-Fidā’ lived long enough to witness the return of all eight fortresses to the Armenians after Ghazan’s invasion.181 His account shows that the main problem was not conquering the Cilician strongholds, but manning and maintaining the garrisons. Intensive Mamluk activity continued through al-Nāṣir Muḥammad’s second reign (1299–1309). In the summer of 1302 the Armenian capital was raided by a Mamluk army from Cairo assisted by contingents from Hama. The entire region was looted, crops were burnt and a large number of local inhabitants were taken into slavery.182 No attempt was made to conquer adjacent towns or fortresses. This campaign had a disastrous ending. Qashtamur, a Mamluk of Qara Sunqur’s, was appointed as the commander of the force. He is described as being “weak-witted, a poor organizer and fuddled with wine.”183 The Mamluk soldiers were quick to recognize his weaknesses; they refused to obey orders and the commanders lost control over them. Further, the army was not well informed and the enemy’s strength was under-estimated. The governor of Sīs called upon the Īlkhānids and Franks for urgent help. A large Īlkhānid force and a group of Frankish knights arrived and joined the Armenians. The Mamluks were caught by surprise and the Aleppo contingent retreated. A great number of men were killed and a significant number of Mamluks were taken into captivity.184 After this the Armenians enjoyed a relatively long period of peace that lasted for just over a decade; it ended in the spring of 715/1315 when Mamluk forces decided to raid Malaṭya. This was the northernmost point the Mamluks had reached since they began campaigning in Cilicia. Abū’l-Fidā gives a long and detailed account of the events that led the Mamluks to start this campaign. He writes that marriages between Christians and Muslims were fairly common in Malaṭya, which led to a growing support for the Mongols amongst the Muslim population. The Muslims apparently began to inform the Mongols about Mamluk raids leaving from Bahasnā, Qalʿat al-Rūm, Gargar, and Kakhtā, as these passed near Malaṭya. The local population attacked the Mamluk forces; some soldiers were killed while others were taken captive. The sultan sent a Mamluk force from Cairo to Damascus under the command of the amir Sayf al-Dīn Baktamur Abūbakrī. It was joined by contingents from all the central Syrian cities. The army commanded by the governor of Damascus, Sayf al-Dīn Tankiz al-Nāṣirī, met the Cairo contingent at Aleppo and reached Malaṭya within nine days. The city was surrounded by Mamluk forces. At this stage a large and honorable party set out to negotiate on behalf of Malaṭya’s governor. The southern gate was left open and the Mamluk forces charged in and began to loot the city. Muslim ownership was not respected and Muslim inhabitants were made captive, though later it was decided to release them despite the fact that the Muslim community had shown loyalty to the Mongols. The city was burnt down and the army did its best to undermine the fortifications.185 The Mamluks were above all determined to warn the Muslim population against future cooperation with the Mongols and the Armenians. They did not actually wish to annex Malaṭya or turn it into a frontier stronghold to keep watch on the region.

116  Muslim Fortresses in the Levant In the spring of 720/1320 the Mamluks raided Sīs again on the sultan’s orders. A Mamluk force left Cairo and met the Syrian forces near Harim in north eastern Syria. The Jayḥān River had risen and it claimed the lives of a large number of soldiers who drowned while crossing. The highest rate of casualties was suffered by the Türkmen troops who came from the Syrian littoral. The army burnt crops and collected herds of cattle which were driven back to Syria. Abū’l-Fidā writes that the city was besieged but does not add any further information (he himself did not take part in the raid). He does not explain what prompted this raid. The Mamluk forces plundered Cilicia for six whole weeks.186 As with the previous raid no effort was made to occupy Sīs in order to establish a garrison and hold the city for any length of time. The Armenian king, Oshin, died in the same year. The destruction and ruin left throughout his kingdom by the Mamluk raids no doubt hastened his end: “He witnessed the burning of his lands, the devastation of its places, the killing of his subjects and the driving of their beasts. So his sufferings were redoubled and he perished in Jumada 1 of this year.”187 The first quarter of the fourteenth century ended with a large scale raid on the port of Ayās. The conduct of the siege (spring 722/1322) suggests that the Mamluk army was thoroughly informed and arrived prepared with a team of sappers, siege machine and engineers recruited well in advance. The city was defended from both land and sea.188 While the citadel fell without too much difficulty, the conquest of the seabound fortress required large-scale earthworks. Two causeways were built, and two large siege machines were rolled up opposite the fortress walls. The siege ended after the Armenian garrison set fire to the fortress and escaped. The Mamluk army saw to the destruction of the citadel and left.189 The citadel was rebuilt a year later with help from Pope John XXII.190 The information gathered from the sources does not always match that of the archaeological survey. According to Edwards, four fortresses were rebuilt by the Mamluks during the thirteenth and fourteenth centuries: Anavarza (al-Zarb), Sīs, Tall-Ḥamdūn and Tumlu.191 It seems more than likely that some of the fortresses turned over to the Mamluks did not require any restoration work. Those that did may well have been rebuilt by local Armenian workers, so that it would be difficult to identify the additions. In Anavarza and Tumlu, I could not find any remains of Mamluk construction. If the Mamluks did any rebuilding in either of these fortresses they must have used local methods and/or Armenian workers. In contrast, Tall-Ḥamdūn, only a short distance from Anavarza and Tumlu (40–50 km), has a distinct and unmistakable Mamluk style similar to those seen in Mamluk fortresses in Trans-Jordan (Karak and ʿAjlūn) Syria (al-Ṣubayba) and Qalʿat al-Rūm on the upper Euphrates.192 Mamluk interests in Cilicia were neither vague nor concealed. The need to defend northern Syria led to an uncompromising policy towards the Armenian kingdom in Cilicia. The necessity to hold fortresses and Mamluk garrisons in Cilicia, and maintain a permanent presence there resulted from the fear that Cilicia could be used by the Īlkhāns, the Crusaders or the Armenians as a jumping-off point to reconquer Syria. The economic aspects are an important factor in this discussion. The opening of an alternative land route connecting the Eastern Mediterranean with central Asia

Laying the foundations  117 Table 3.3  The results of Mamluk raids and treaties in Cilicia.

 1

The results of Mamluk raids in the Armenian kingdom

Fortresses ceded to the Mamluks following the signing of a treaty.

1266

1268

 4

Destruction of Sīs, Adana, Mesis, Ayās and Ṭarsūs Tall Ḥamdūn, Karanjil, Ḥamūṣ, al-ʿAmūdayn, Sarfadkār. Baghrās is taken and retained in Mamluk hands 1273 Kaynūk taken but not restored 1274–5 Destruction throughout the kingdom of Armenia and collection of a great quantity of booty. Attack on the port of Ayās 1285 No raids

 5

1289

 6

1292

 7

1298

 2  3

 8  9 10 11 12

1302 1305 1316 1320 1322

No raids. Following the Mamluk conquest of Tripoli the Armenian king initiates a treaty. Qalʿat al-Rūm taken, restored and garrisoned Tall Ḥamdūn is conquered by the Mamluks. It is mentioned in a treaty together with several other fortresses. Sīs raided Raid on Sīs fails. Ayās attacked. Malatya taken. Sīs raided Sīs taken Ayās is besieged.

Darbassāk, Marzabān, Raban, Shīh al-Ḥadīd, alZarb, Bahasnā

1285

Darbassāk, Baghrās, Gargar, Kakhtā, Tall Bashir 1289 Bahasnā negotiated but remains in Armenian hands. Marʿash ceded. 1292–3 Darbassāk, Baghrās 1298

Tall Ḥamdūn, Marʿash, Kuwayrā, al-Naqīr, Hajar Shaghlān, Sarfadkār, Hāmīs.   Marʿash, Hāmīs and Nujayma are garrisoned by the Mamluks

and partially or completely replacing the sea route between India and Alexandria via Aden and the Red Sea to Alexandria would have had a devastating effect on the Sultanate’s trade. The fortifications the Sultanate sought to control were vital components in protecting the trade routes that crossed the region. The conquest of Baghdad (1258) and the fall of the principality of Antioch (1268) caused the trade routes between the Persian Gulf and the Eastern Mediterranean to gradually move northwards. Thus, at the beginning of the fourteenth century Tabrīz became a thriving commercial city. A large percentage of the merchandise in its markets arrived from the Persian Gulf. Part of those goods traveled westwards from Tabrīz to the Cilician port of Ayās. The route between Tabrīz and Ayās is given a detailed description by Francesco Balducci Pegolotti, a well-educated entrepreneur from Florence employed by the Florentine banking house of the Bardi. The details of his travels are found in his book La Pratica della Mercatura. The trade route starting from Tabrīz went west through the following cities: Khoy (Xoy), Arzanjān (Erzincan), Sivas, Kaysariye and Geben (Gabon).193 It then turned south and ran parallel to the Jayḥān River and down to Ayās. The port of Ayās was

118  Muslim Fortresses in the Levant one of the few connecting Asia and the Eastern Mediterranean. It gradually became a prosperous station from which eastern goods were exported to the west and European products traveled east (linen was one of the main European products exported to the east). The Genoese and Venetian merchants had warehouses in the nearby town of Mesis just north of Ayās.194 The Tabrīz–Ayās route has a number of advantages: the merchants using it received substantial benefits from both Armenian and Īlkhānid rulers. Merchants passing through the Armenian kingdom were given generous rights. Well-garrisoned strongholds secured the caravans traveling between Persia and the region of the Black Sea and the ports of Cilicia.195 Safety along the trade routes was the main condition for the existence of trade in the region. A large portion of the Armenian kingdom’s wealth derived from taxes collected from merchants crossing its territories and using the services provided at its ports. Part of these revenues was reinvested in maintaining fortresses and building new ones.196 The destruction of Frankish ports along the coast of the Levant in the second half of the thirteenth century may well have contributed to the rise of Armenian trade. The Armenian king was quick to grant Genoese and Venetian merchants trading rights in Ayās. Most of the goods in Egyptian and Levant harbors and markets could be found in Persia and brought to the Cilician ports. The arrival of the Italian merchants boosted the trade and the economy of this Armenian port.197 This change did not escape Mamluk eyes. The commercial activity along the Cilician land route and in the ports probably diverted profits that would otherwise have made their way into the sultan’s treasury. To this one should add the trade in slaves that passed through the Armenian kingdom en route to Egypt. Although only a small percentage of the slave trade moved through Ayās, it seems plausible that the Mamluks wished to keep a close watch on this valuable commodity. The Īlkhānids made their own contribution by reducing the taxes collected along the trade routes that passed through their lands. According to Ashtor, the taxes in the Īlkhānid state ran at about 4–5 percent in comparison to the 10–20 percent collected by the Mamluks from merchants trading in Egypt. The Italian merchants were highly respected at the court of the Īlkhān; some even served as diplomatic emissaries when the Īlkhān corresponded with the papal court or any of the European monarchs.198 Although most European traders ignored papal bans on commercial activity between Europe and the Mamluk sultanate, these restrictions may have led to a search for new markets and alternative routes. The volume of trade between Persia and Europe grew steadily during the last quarter of the thirteenth century and the first three decades of the fourteenth. Ashtor adds that the Īlkhāns may have hoped commercial ties would help build political relations and possibly a military alliance with Europe.199 The trade routes described above were dominated by Genoese, Venetians, Pisans and Sienese. Genoese and Venetian consulates opened in Tabrīz and Sivas; a Genoese colony was established in Trebizond and some merchants founded communities in cities between Tabrīz and Ayās. The latter became a thriving port where Italian, Egyptian, Syrian and other merchants met. Tabrīz, on the other hand, became the starting point for European traders heading east to China and India.200

Laying the foundations  119 This route however, had several disadvantages and the revival of trade was shortlived. Though taxes were reduced, the great distances and the time caravans had to spend on the road turned transportation by land into an expensive undertaking. With the death of Abū Saʿīd (736/1335), the last Īlkhān, political unrest struck the region.201 The Jalayirid dynasty that rose to power could not restore peace and quiet and European merchants gradually left the markets and ports and turned elsewhere.202 The last section of this route, from Kaysariye south to the port of Ayās, is important in order to complete this discussion. The numerous Mamluk raids, aimed at disrupting both stability and economic affairs within the Armenian kingdom, were partly concentrated against the strongholds and towns along this very last section. Ayās, Sīs and Mesis, where merchandise was stored, were plundered and burnt down several times. Since the kingdom relied heavily on foreign merchants, the damage caused to the economic infrastructure must have been considerable. Towards the end of the thirteenth century as the Armenian kingdom slowly declined the Mamluks’ aims changed; rather than raiding and retreating they seem to have wanted to conquer and hold the port of Ayās as well as the trade routes that crossed Armenian territory. Assaults on Ayās began during Baybar’s reign. The port was sacked in 1266 and 1275. During the 1280s it was raided twice by the Türkmen.203 In 1305 and 1322 the port was attacked again. It was finally captured in 1337 and remained under Mamluk rule. The reasons for trying to maintain a Mamluk presence, whether permanent (fortresses and garrisons) or temporary (raids), are fairly complex. The most evident is the Mamluks’ need to defend their northern border. No less important was their wish to punish the Armenians for their loyalty, cooperation and participation in the Mongol invasions into Syria. However, it appears that the economic motives carried some weight. The series of raids and treaties show a political and military pattern that the Mamluk sultanate found hard to break. The Armenians were just as persistent in fighting to retain their fortresses. Their attempts to regain control and rebuild their fortresses after each Mamluk raid are striking. The fortresses were vital to both Armenian trade and to the security of the northern Mamluk borders. The Mamluk concept of a defense line along the northern border was similar to that built along the Euphrates frontier – a model that proved its strength against the Mongols for over sixty years. Thus, the Cilician fortresses were relied on to establish Mamluk military presence in the region, and sound the alarm if the enemy approached. In reality, however, both the Armenians and the Mamluks found it difficult and at times almost impossible to hold and maintain the fortresses for any length of time. The Armenians appear to have suffered from a severe shortage of manpower and could not withstand Mamluk raids and full-scale attacks without calling on considerable reinforcements. The only occasion on which they gained the upper hand and defeated a Mamluk army was when Īlkhānid and Frankish forces joined them (summer of 1305).204 The fortresses that were of strategic importance to the Armenians became a permanent target throughout the Mamluk period; they were

120  Muslim Fortresses in the Levant plundered and burnt down time after time. Nevertheless, as soon as the opportunity offered itself the Armenians did their utmost to restore their strongholds. Strangely enough, it seems that the Mamluks, in possession of a stronger military force, faced almost the same difficulties. Holding a large number of fortresses in a hostile territory was a difficult task for a state that had the core of its army stationed permanently in Cairo. Although most of the Mamluk raids were successful few, of their victories were translated into long-term political or strategic gains. In fact it was only when the Armenian kingdom fell (1375) that the Mamluks fully ruled Cilicia. By that time, however, their concept of frontier defense had undergone some significant changes. During the fourteenth century the northern frontier defenses of the Sultanate were entrusted to Dulkadir and Ramadan – the two buffer states that arose in this region. The Mamluks themselves remained in command of the large cities.205 It is doubtful whether Baybars and Qalāwūn would have supported this new strategy, but the region had changed considerably and the thirteenth-century military concept of defense that relied solely on Mamluk forces was no longer suitable.

In search of conclusions The defense system along the eastern borders would probably not have been built if the Mongols had not invaded the Middle East. The establishment of the Īlkhānid state east of the Euphrates meant that an almost constant threat hovered over the Sultanate. It is difficult to assert that this policy of strengthening fortresses, started by Baybars, would have been carried out only in order to help establish Mamluk rule in Syria, Palestine and Trans-Jordan. The restoration of fortresses along the frontier and inland, the presence of Mamluk garrisons and armies in the fortresses and in the large Syrian cities, the sultans’ frequent journeys through Syria, the construction of an intelligence network, and the barīd that connected the frontier with the main urban centers and the capital most certainly helped establish the new Mamluk regime throughout the region. The spatial distribution of fortresses changed, when frontiers and borders shifted due to the rise or fall of political entities in the region. As the Frankish territory gradually receded during Qalāwūn’s reign, a number of fortresses restored by Baybars slowly lost their military importance. This trend continued for a short time after the fall of the Frankish kingdom. Thus Qāqūn, Safad, al-Ṣubayba and Ḥiṣn al-Akrād were no longer of great importance to the Sultanate’s defense. The number of fortresses the Sultanate could efficiently maintain was restricted by military and economic considerations. The lack of a large navy meant that almost all the Frankish fortresses along the coast had to be destroyed. This, as Ayalon observed, was the only way to guarantee that the Franks would not reconquer the coastal towns. One of the most interesting aspects of the Mamluk fortifications is the influence of the organization of Mamluk government and military society on the spatial distribution of fortifications. Though the Mamluk army grew considerably between the days of Baybars and the end of the reign of al-Ashraf Khalīl, the army remained stationed in Cairo and the large Syrian cities. The number of Mamluks the sultan could allocate

Laying the foundations  121 to garrisons outside the urban citadels was rather limited. Even when amirs were given large agricultural estates and villages as an iqṭāʿ, all of them chose to remain close to the center of government, in Cairo or Damascus.206 Thus no fortresses or fortified manor houses were built in the rural areas by high-ranking Mamluk officers. In contrast, both the Frankish and the Ayyubid landed aristocracy lived on their estates and were responsible for the foundation of fortresses and strongholds in rich agricultural areas. In later decades some of those fortified sites were handed over to the Military Orders and significantly enlarged. The Ayyubid nobility built considerably less, though it seems the subject has not yet been thoroughly researched. Another important factor determining the map of Mamluk fortresses and the Mamluk scale of construction is the centralization of the Mamluk regime. In comparison with Ayyubid rule, the Mamluk system gave the sultan substantial military and political power. He was an autocratic ruler.207 The Ayyubid dynasty, even at its height, was never unified under one central ruler for any length of time. The territories belonging to the Ayyubids were ruled by various members of the family and by a number of high-ranking amirs. Their sultanate more often resembled a family federation and, except for some military restrictions, each ruler within the federation had full autonomy. The Ayyubid rulers in Asia Minor often carried the title of sultan.208 The internal dissension among the Ayyubids can be seen in the spatial distribution of their fortresses. A fair number of fortresses constructed by various princes simply ignored the military, economic and political policies of the Ayyubid sultan in Cairo.209 Most of those strongholds answered immediate local interests and were intended to solve regional problems, such as tribal threats and internal family conflicts among Ayyubid princes. In contrast to the Mamluks, who during their first three decades had to face enemies on three different fronts, the only foreign enemies the Ayyubids faced were the Crusaders. The case of ʿIzz al-Dīn Aybak al-Muʿaẓẓamī, a Mamluk of al-Muʿaẓẓam ʿIsa, provides further evidence of the liberty Ayyubid rulers had concerning the construction of fortifications. In 1214 ʿIzz al-Dīn Aybak received Ṣalkhad and its surrounding region as an iqṭāʿ from the Ayyubid sultan, and governed for the next forty years building mosques, irrigation systems, khans and fortifications.210 The change in the physical location and structure of royal and aristocratic quarters had an important influence on the development of Muslim military architecture. Starting from the eleventh century, fortresses served as the official living quarters of the ruling elite and their military entourage.211 This trend continued into the twelfth century, during which the Ayyubids constructed large citadels. Ṣalāḥ al-Dīn initiated the building of the Cairo citadel, which remained the center of government throughout the Mamluk period and long afterwards. During the twelfth and thirteenth centuries citadels were built in Damascus, Aleppo and Homs and in their vicinity. In contrast to the city fortifications, the spatial distribution in the rural areas lacked an overall plan. This is not surprising since there was no one authority that commanded, planned and organized construction throughout the Ayyubid sultanate. The four fortresses built on the Euphrates fords clearly show this lack of central governmental planning. Each stronghold was ruled by a different Ayyubid prince and reflected his own local interests and the regional family disputes. During the early

122  Muslim Fortresses in the Levant Ayyubid period both Qalʿat Jaʿbar and Qalʿat Najm were ruled from Aleppo. The latter was considered a key fortress in the defense of the territories that belonged to the city.212 Qalʿat Jaʿbar later changed hands and during al-ʿĀdil’s reign (d. 615/1217) it was annexed to the Jazīra. The entire area was known as the eastern territories and was in fact an independent Ayyubid province governed by al-ʿĀdil from his capital in the Ḥarrān.213 According to Maqrīzī, al-ʿĀdil built and enlarged the fortress,214 and kept his treasury in it.215 Al-Raḥba was given by Nūr al-Dīn as an iqṭāʿ to Asad al-Dīn Shirkūh (d. 564/1166), Ṣalāḥ al-Dīn’s uncle, and the fortress remained in the family’s hands until the Mongol invasion. The history of al-Bīra is more complicated. From 1099 to 1150 the fortress belonged to the Crusader county of Edessa. It was briefly ruled by the Byzantines and eventually fell to the Urtuqid governor of Mārdīn.216 By the mid thirteenth century it was part of the kingdom of al-Nāṣir Yūsuf (d. 658/1260), before it was besieged and taken by Hülegü in 1260.217 The fortresses in the Crusader kingdom and principalities were built by a similar array of rulers. Some of the independent political entities were often stronger and wealthier than the king, the count or the prince. Most of the large impressive fortresses that date to the first half of the thirteenth century (Chastel Pelerin, Crac des Chevaliers, Safad, Marqab and others) were built under the auspices of the Military Orders. The latter were the only bodies that could afford to build, garrison and maintain fortresses on such a scale.218 The Armenian fortresses are the only example we have that resembles the system of the Mamluk sultanate. The building of fortresses was initiated and supervised by the king and probably funded by the crown. Since a high percentage of the kingdom’s revenues came from providing services to merchants and collecting taxes on imported and exported goods, most Armenian fortresses were built in order to safeguard the merchant caravans traveling along the main routes.219 The fortresses rebuilt by Baybars, Qalāwūn and al-Ashraf Khalīl best reflect the differences between Mamluk governance and that of the Ayyubids and the Franks. Whether conquered, rebuilt or even destroyed, most of the strongholds served the political, economic and military interests of the Mamluk sultanate itself, and not those of the aristocracy, religious institution, high-ranking amirs or wealthy and influential traders. The decision as to which fortresses should be maintained was made solely by the sultan, who occasionally consulted his most prominent amirs. When amirs doubted or disagreed with the sultan’s final decision, his response was fierce and uncompromising as was the case when al-Ashraf Khalīl ordered the destruction of Shawbak. Fortress governors were chosen only from the sultan’s amirs and were appointed directly by him. When a governor died or was moved to a different post the sultan himself appointed a new one in his place. Once Qalāwūn had established himself in power he replaced all the fortress governors in Syria who had been chosen by Baybars, appointing governors from among his own amirs in their stead.220 All construction work concerning restoration and strengthening of fortifications, whether in urban or rural areas, required the sultan’s approval. High-ranking governors such as Shams al-Dīn Qarā Sunqur al-Jukandār al-Manṣūrī, the nā’ib of Aleppo, had to receive Qalāwūn’s permission before building the mosque and the citadel.221

Laying the foundations  123 Due to this highly centralized rule, the total number of fortresses was less in comparison to the number of fortresses held by the Ayyubids,222 and the Franks.223 Ibn Shaddad gives a list of 22 fortresses in the vicinity of Aleppo that were taken but never rebuilt. As noted above, the reasons for the drop in the number of fortresses were linked directly to the size of the Mamluk army, its preference to be stationed in urban centers and possibly the financial constraints of the Sultanate’s treasury. However, the main reason appears to have been the geopolitical changes and the unification of Ayyubid and Crusader territories under one political entity. This reduced conflicts and hostilities throughout Syria, Palestine and Trans-Jordan, and the number of fortresses needed for defense purposes was accordingly smaller. It is almost impossible to propose a theory or a historical framework that would offer a comprehensive explanation for the spatial distribution of the Mamluk fortresses. There will always be a number of sites that do not fit the assumed framework, and any attempt to bind all the fortresses together and impose one over-riding idea on them, will never fully or accurately represent the historical scene. The fortresses along the Euphrates, bordering the Īlkhānid state, were restored by the Mamluks in order to defend the fords and raise the alarm against Mongol invasions. Even after this stretch of the river was fortified and two additional large strongholds were taken and restored (Qalʿat Jaʿbar and Qalʿat al Rūm) the effectiveness of the fortifications and safety of the Syrian population largely depended on the ability of the Euphrates fortresses to alert the Mamluk contingents in Syria and Egypt and to confine the fighting to the area east of the river. The Mamluk priorities were fairly clear. The Armenian frontier did not enjoy the same status as the Īlkhānid. The Sultanate’s limited ability to provide relief forces restricted the number of strongholds and garrisons it could maintain in Cilicia. The region of Trans-Jordan cannot be defined as a military front and it rarely demanded the Mamluk army’s attention. The main reason for Baybars’ restoration of the Trans-Jordan fortresses was the considerable interest they attracted during the first Mongol invasion. Though very few were severely damaged, the key fortresses were besieged and taken by the Mongols, who maintained the Ayyubid governors providing they were loyal to the new regime. Thus it became essential to break up the Ayyubid principalities and bring their territories under the authority of the Mamluk sultan in Cairo. The appointment of new fortress governors certainly guaranteed the establishment of the Mamluks in the region. It is most important to remember that the main transport and communication routes between the two Mamluk capitals, serving the sultan’s couriers, pilgrims on Hajj and merchants, ran through the entire length of Trans-Jordan, thus adding to the importance of the Trans-Jordan strongholds. Other than Shawbak, which was destroyed during al-Ashraf Khalīl’s reign, all the main fortresses were restored; each played a different role and there was a definite hierarchy within this group. Karak topped the list it became an administrative center and a place of refuge for the sultan and his family during stormy political periods in Cairo. Mamluk policy towards the Franks remained essentially the same throughout the reign of the three sultans. Qalāwūn for a short time appears to have followed a moderate policy.

124  Muslim Fortresses in the Levant As important as the military aspects may be, they were not always the main consideration that led to the final decisions concerning the fortresses that should be maintained. Economic and social aspects and permutations within those two fields require careful consideration. Those aspects should be examined both on regional and more local levels. Although Qalāwūn had taken Marqab and destroyed Maraqiyya, his approach in northern Syria was restrained. Some ports and a number of fortresses were rebuilt and garrisoned. Their defensive properties were of relatively little importance and it seems commercial activity and the profits that were to be made were the main reasons for Qalāwūn’s policy. The fortune of most of the fortresses was decided by the end of Baybars’ reign. Qalāwūn and al-Ashraf Khalīl did not significantly change the map of Mamluk fortifications. A few fortresses were added and construction work continued on sites that Baybars had not had time to complete. The status and importance of certain strongholds changed as the Frankish states were gradually reduced in size. This trend became more apparent once the Crusader kingdom fell and a peace treaty was signed with the Īlkhānid state (1323).

4

Mamluk power on display Fortresses in the second half of the thirteenth century

By the end of Baybars’ reign fortresses were well integrated in the Mamluk military array. If there was ever a short period of trial and error it left little trace in early Mamluk fortifications. While the Mamluks evidently followed some existing patterns and models that had long been used and improved by the Franks and Ayyubids, it is yet to be considered whether they brought new ideas into the field of fortifications, which had by then almost reached the peak of its development.1 As with the Ayyubid period, the few works dedicated to the subject of Mamluk military architecture are mainly monographs giving a detailed description and analysis of a specific site. The Mamluk fortifications of Cairo and the central Syrian cities such as Damascus and Aleppo have been explored and studied in greater detail. Deschamps carried out an extensive survey of Karak, al-Ṣubayba and Crac des Chevaliers (Ḥiṣn al-Akrād), but his emphasis was on the Crusader architecture of each site.2 Pringle surveyed and excavated towers on the Sharon plain that have a Mamluk building phase, but his research centered on the Crusader period. Hartal’s excavations at al-Ṣubayba concentrate on Bīlīk’s tower; where Mamluk construction methods are extensively described. Some new information has emerged in the past decade concerning the Mamluk architecture at the site of Ḥiṣbān in Jordan.3 During the late 1980s small excavations were conducted at Karak and at Shawbak, giving precise dating to some of the construction work carried out during the fourteenth century.4 Edwards has composed a short description of Mamluk construction work in Cilicia; his research, however, is dedicated to Armenian fortresses.5 Because the Mamluks mainly added to and built upon existing fortresses their work has been consistently overlooked or else documented as a phase within Crusader, Ayyubid or Armenian fortresses. It has rarely been studied as an independent subject. Eight fortresses are discussed in this chapter (Map 3.4; Figure 4.1).6 Starting from the Mamluk inland fortifications, Al-Ṣubayba is of particular importance, since the Mamluk construction is a direct continuation of the Ayyubid work. It is the only rural fortress in this study that was built by the Ayyubid rulers and was further restored and enlarged by the Mamluks, with no Frankish intervention.

126  Muslim Fortresses in the Levant

Figure 4.1  Mamluk fortresses from the thirteenth and early fourteenth centuries drawn on the same scale

Mamluk power on display  127 Safad, located in the Upper Galilee has only been partly excavated, and a considerable amount of work is still required in order to fully understand the scale of restorations conducted by the Mamluks. The historical sources, on the other hand, are remarkable and give a detailed account that few sites offer. Qāqūn is one of the few examples of the smaller fortresses restored by the Mamluks; it is also the only Frankish stronghold on the Sharon plain that was not destroyed by Baybars. The Mamluk fortifications in Trans-Jordan are represented by Karak, parts of which are securely dated by inscriptions. The fortress was extensively built and almost doubled its size in the early years of Baybars’s reign. Qalʿat al-Rūm and al-Bīra represent the border fortifications which secured the Euphrates fords. From the Euphrates I move west into Cilicia, choosing Baghrās and Tall-Ḥamdūn, which guarded the passes over the Amanus mountain range. As well as endeavoring to represent different geographical regions within the Mamluk sultanate, the eight fortresses are all well dated, either by historical sources or inscriptions found in situ and sometimes by both.

The language of construction work, time schedules and officials The historical sources Very few periods in history were as fortunate as that of Julius Caesar and Augustus in having an architect and military engineer who wrote about the buildings of his time. Vitruvius,7 in his ten volumes On Architecture (written in the second half of the first century bc), gives a detailed survey of methods of construction, the various materials available, problems concerning constructions and other valuable information. His books cover public, military and private buildings. No matching Mamluk source has yet surfaced. Although the Mamluk historical sources are numerous and rich, none deal explicitly with architecture. Several of the contemporary chroniclers were highly respected scholars, but they all appear to have lacked the knowledge and understanding of architecture and construction required in order to give a clear and detailed description of the buildings standing before them.8 Strongholds and city fortifications in the Mamluk sultanate are frequently listed in contemporary works, but the descriptions are often poor and brief, usually coming after a relatively long account of the siege in which the fortress was taken. In some cases sweeping summary accounts are given concerning a large number of strongholds and the restoration works carried out by the sultans. Ibn ʿAbd al-Ẓāhir’s list of ten fortresses and citadels provides only very general information: “their moats were cleaned and their curtain walls widened.”9 No further details are given regarding the buildings, and one is left with a slight suspicion that the account may be somewhat unreliable. This feeling is strengthened when Ibn ʿAbd al-Ẓāhir’s information is examined in the light of the archaeological remains, as at least one of the fortresses in this report did not have a moat and three of the sites did not have their curtain walls rebuilt or widened.10 In describing the organization of Karak by Baybars, Ibn ʿAbd al-Ẓāhir states that the fortress was carefully inspected from the outside and the inside and that the Sultan ordered

128  Muslim Fortresses in the Levant the reconstruction of what was necessary.11 But that is as far as the report goes; no further details are given as to what was damaged or which weak sections needed to be strengthened. The siege and reconstruction of Safad is mentioned by several chroniclers, it will be used here to show the main problems arising from the different sources. Ibn ʿAbd al-Ẓāhir’s (d. 692/1292) writing reeks of court politics. His excessive admiration for Baybars, arising from his appointment as kātib al-sirr and his position as official biographer of the sultan, often surfaces. Much of his account of the reconstruction of Safad concerns the Sultan’s fine leadership.12 It appears that Ibn ʿAbd al-Ẓāhir received his information from someone who participated at the siege and was present in the reconstruction. Ibn al-Furāt (d. 807/1405) relies heavily on Ibn ʿAbd al-Ẓāhir and adds little to our knowledge of fortress construction. Fortifications were obviously of no interest to him or he simply did not see it as his task to report on the subject. His account concerning the reconstruction of Safad is poor in comparison to the detailed description of Baybars’ siege and the new Mamluk garrison that was placed in the fortress. The only buildings he describes are the two mosques that were erected by the sultan, one in the fortress itself and the other in the suburb (rabaḍ).13 The most informative of the thirteenth-century Mamluk sources for this subject is the Ta’rīkh al-malik al-ẓāhir written by Ibn Shaddād (d. 684/1285).14 This gives a much more detailed account than any other contemporary source. True, the descriptions are often brief, but they appear to be accurate and reliable. Professor Rabbat maintains that Ibn Shaddād’s descriptions are factual and no political motives of any kind are perceptible in this author’s work.15 Occasionally he gives a more comprehensive and thorough picture, as in the case of Safad.16 Ibn Shaddād may have visited the site after the work was completed. It is difficult, however, to follow or distinguish how or why certain fortresses receive a full and long account, while others no less important are barely acknowledged. Immediately after the lengthy report on the rebuilding of Safad, Ibn Shaddād goes on to al-Ṣubayba where his report is completed in five short lines. Maqrīzī (d. 845/1442) is not a contemporary source and he often relies heavily on Ibn al-Furāt and Ibn ʿAbd al-Ẓāhir.17 Nevertheless, his descriptions occasionally present important and interesting details that are not mentioned elsewhere. When strongholds are described, it is often done from the perspective or view of a person standing in the besieging camp, who observes the outer walls, towers and gates. The buildings inside the fortress walls are rarely given any consideration. Thus the main sources for structures such as water cisterns, storage houses and the like are inscriptions, archaeological excavations and surveys. Mosques are possibly the only buildings within the fortresses that are mentioned both by chroniclers who paid some attention to architecture and by those who almost ignored it and concentrated solely on historical events. The construction of mosques and the conversion of churches into mosques are almost always spoken of as an important part of the general restoration work carried out by the Mamluk sultans.18 Again, architectural details of the mosques are usually omitted and the reference is more of a political statement concerning the bringing of yet another fortress under the rule of Islam.

Mamluk power on display  129 Architectural terminology found in the sources The vocabulary used to describe fortresses in the chronicles is generally very simple and straightforward. The following section concerns the basic architectural terminology and some of the problems that may arise from the texts due to the several meanings that some terms carry. The most common words for fortress are qal ʿa or ḥiṣn. Other words for fortress which I have only seldom come across are maʿqil ( ),19 and ṣarḥ ( ).20 On a few occasions fortresses on the frontier are distinguished as such by the term thughūr (singular thaghr, ).21 Occasionally the landscape, steep ravines, mountains and riverbanks are given some consideration if they are regarded as an important part of the fortress defenses. However, the most common opening in many accounts is “this was a strong fortress” (qalʿa ḥaṣīna, ),22 or “this is one of the most impregnable fortresses” (hādh al-ḥiṣn min amnaʿ ḥuṣūn, ).23 An early fourteenth-century chronicle repeatedly describes fortified sites as “a fine fortress” (qalʿat laṭīfa, ).24 There are seldom more elaborate introductions to a fortress. There seems to have been a structured frame that few chroniclers strayed from; once the geographical location of the fortress was established the chronicler usually worked his way through the various parts of the fortifications. The bāshūra ( ) is a key word in many descriptions. There is no real equivalent in English, and there appear to have been several meanings. The word refers to a whole section of the outer fortifications, which may include any or all of the following structures: barbican, bastion, a gate or a fortified tower. In some cases the outer suburb that often grew up beside the fortress walls is also referred to as the bāshūra.25 The bāshūra features in most siege accounts, as it was the area that suffered most and received the full blow of the Mamluk attack. As a result it was often the first section of the fortress to be rebuilt and the works involved were generally quite substantial. Moats, whether wet or dry, appear as al-khandaq ( ). The depth and width are sometimes given if the moat was truly impressive or forbidding.26 In Ayyubid inscriptions the word most frequently used for describing curtain walls is al-badanah ( ), literally to encircle.27 The term sūr ( ), which has the exact same meaning, seems to have been more common both in Mamluk inscriptions and among chroniclers.28 Main gates are often included as part of the bāshūra; every so often they are given a full description and are then referred to simply as bāb ( ). Ibn Shaddad uses the Persian word darkāt or dargāh ( ) to describe the gate at al-Ṣubayba, but this term is hardly ever used.29 While curtain walls seldom receive a detailed description, towers play a major role in many accounts and are mentioned by almost every chronicler. Occasionally the number of towers built and reconstructed by the Mamluks is given and if they are of exceptional features such as height or outstanding construction (e.g. the size of the stones) precise details are cited. The word for tower is burj ( ); so far I have not encountered a specific word that denotes “keep” or “donjon”. Keeps are referred to simply as larger towers. Strangely enough, arrow slits are never described or even mentioned, possibly because they were such an obvious part of the fortress that they did not seem of any importance to

130  Muslim Fortresses in the Levant contemporary historians. As opposed to arrow slits machicolations, ṭalāqāt ( ), are noted, though not very frequently.30 Secret passages, abwāb sirr ( ), are rarely taken into account, probably due to the fact that this is the kind of detail that would only be known to those who lived in the fortress or paid a lengthy visit to it.31 Glacis, musaffaḥa ( ), are also seldom mentioned; the word safḥa literally means the foot or base of a mountain.32 The general descriptions and plain accounts may indicate that many chroniclers received their information from secondhand sources and never actually walked the ground of the fortresses they describe. The vocabulary and especially the variety of verbs related to Mamluk building activities reveal some important and interesting facts. Unlike their predecessors, the Mamluks were adamant about extracting the most from the already existing fortresses in the region, and displayed no real will to construct new strongholds. This approach is clearly conveyed to the reader through the chroniclers’ vocabulary and choice of words. A good example is perhaps the inventory of building activities conducted by Baybars, given by al-Ṣafadī (696/1297–764/1363). Al-Ṣafadī constantly switches between five different verbs: (1) jaddada ( ), to renew; (2) aṣlaḥa (  ), to better, correct, or to repair; (3) ʿammara (  ), to build, to reconstruct, to restore or to repair; (4) ansha’a ( ), to build, to construct, to set up or erect (from scratch); (5) banā ( ), to build, to construct. Other verbs occurring in the sources though not often used are (6) shayyada ( ), to build, to construct, to erect or set up;33 (7) ramma ( ), to repair or to renovate;34 (8) zāda ( ), to add;35 and (9) aḥkama ( ), which can mean to strengthen or to build in a thorough or an exact manner.36 Al-Ṣafadī’s inventory includes religious, civil and military buildings. The verb most commonly used in speaking of fortresses and fortifications is jaddada, meaning to renew. When the verb ʿammara, which can mean to build or to reconstruct, is used, each case should be cross-checked separately with other sources of information (archaeological remains, inscriptions and other chronicles). Banā which simply means to build, occasionally comes after the author has pointed out that only particular sections were built, as in the following example:

… and he built [only] what had been ruined in the fortress of ʿAyn Tāb.37 Because this policy of building on already well-established strongholds was maintained throughout the thirteenth and early fourteenth centuries, the scale of building even in the largest of the Mamluk fortresses can hardly be compared to the construction building of a new stronghold from the foundations. Thus the scale of building was considerably smaller than that of the Franks and the Ayyubids, and it is therefore surprising that Mamluk construction work often stretched over a relatively long period.

Mamluk power on display  131 The importance of looking at the time schedules Dates provided in both chronicles and inscriptions offer a unique opportunity to draw up fairly accurate work schedules. Mamluk fortresses were seldom built in a hurry. Practically speaking, they were already fully built. Once secured and garrisoned, construction work seems, in some cases, to have stretched over a considerable number of years. Ibn ʿAbd al-Ẓāhir’s list mentions Damascus, al-Ṣalt, ʿAjlūn, Ṣarkhad, al-Ṣubayba, Shayzar, Shumaymis, Bosra, Baalbek and Homs; he states quite clearly that those fortresses were rebuilt and their moats cleaned. He does not give a complete account of the destruction caused by the Mongols or a systematic description of the repairs called for in each of those sites. The degree of damage can often be assessed according the scale of reconstruction carried out by the Mamluks. At ʿAjlūn the work undertaken by Baybars was minimal; the fortress was remarkably well preserved and whatever destruction the Mongols may have inflicted, the repairs did not necessitate a large investment. The last building phase dates to the Ayyubid period, and there is little evidence of Mamluk construction.38 At al-Ṣubayba the Mamluks invested a great deal of thought and work. The final size of the fortress was, however, established during the Ayyubid period, almost three decades before the taking of the fortress by Baybars. Yet the renovations at al-Ṣubayba went on for years. True, the scale of Mamluk construction here was altogether more substantial than at other sites; several large towers were added, a gate enlarged and sections of the walls repaired, as will be described later in greater detail. Still, the time it took can hardly be justified. The fortress passed into the hands of the Mamluks in 1260; according to an inscription found on one of the new Mamluk towers the reconstruction was only completed in 1275.39 In contrast, fortresses along the Euphrates, bordering the Īlkhānid territory, were rebuilt within a short space of time. It would be wrong to assume that the Mamluks built at leisure, but in comparison with the work schedules known from Frankish and Ayyubid sites they most definitely did not build in great haste. The entire construction at Mount Tabor, the largest Ayyubid fortress in the region, lasted approximately four years, from 609/1212–13 to 612/1215–16. The building of the curtain wall of the Frankish fortress at Vadum Iacob, measuring approximately 410m in circumference and 4.3m in width took eleven months (October 1178–August 1179). It may have not been altogether finished by the time the castle was destroyed, but it functioned and was no doubt very close to completion, as can be seen from the surviving archaeological remains. According to Oliver of Paderborn, who describes the building of the massive towers and double curtain wall of ʿAtlīth, work began in February 1218 and was finished in mid April of the same year.40 The rebuilding of the first phase of the fortress at Safad by the Templars began in December 1240 and was completed within two and a half years, in the middle of 1243.41 Both Safad and ʿAtlīth rank amongst the largest fortresses in the Crusader kingdom. Such time schedules clearly suggest that the Mamluk reconstruction at al-Ṣubayba was by no means done in a hurry, for it could not possibly have taken fifteen years to restore the curtain wall, enlarge three towers and add four new ones. Building fortifications at a relatively slow pace may also suggest that the Mamluks were not at all alarmed or worried by

132  Muslim Fortresses in the Levant the state of the fortifications they acquired, and that the condition of the fortresses taken by Baybars was on the whole far better than contemporary chroniclers would perhaps like us to believe. Thus, the description by Ibn Shaddād,42 as well as that by Ibn ʿAbd al-Ẓāhir, of the chain of destruction left by the Mongols should be carefully examined and taken with a grain of salt. To destroy a stone fortress one needs a considerable amount of time, manpower and heavy siege machines. Al-ʿĀdil and his son al-Muʿaẓẓam spent an entire year demolishing their own fortress at Mount Tabor.43 Qalāwūn sent siege machines to assist in the destruction of Maraqiya (Maraclea).44 The fortress structures no doubt suffered damage under the wrath of the Mongol army in 1260, but it seems it was easier to kill the Muslim garrisons, destroy food and weapon supplies and burn the crops in the surrounding countryside. This rendered the defenses ineffective for a considerable length of time. Perhaps the emphasis in Ibn ʿAbd al-Ẓāhir’s narrative on the Mongol destruction and the restoration of strongholds by Baybars should be laid on the latter part of the sentence: “and he carried to them (i.e. to the fortresses) supplies (military supplies, ) and sent to them Mamluks and soldiers;” rather than on the first part that refers to the fortifications: “the fortresses were rebuilt during his reign and he cleaned their moats and widened (strengthened) their walls.”45 Was there a government office that was responsible for the construction of fortresses? In reading the descriptions referring to the conquest and rebuilding of fortresses, it becomes clear that the initial and final decision whether to rebuild a fortress or leave it in ruins was made by the sultan himself. There are few exceptions to this rule. Once the decision was taken, the sultan often left the site and nominated an amir to supervise the reconstruction work. In some cases this amir became the nā’ib of the qalʿa. But on a few occasions he was there only to oversee the building. One of the most interesting figures to play a significant part in the rebuilding and organization of fortresses was ʿIzz al-Dīn Aybak al-Afram (d.1295), who was appointed amir jāndār by Baybars.46 He continued to hold this post throughout the reign of Qalāwūn and most of that of al-Ashraf Khalīl, who eventually put an end to al-Afram’s long career.47 During the Ayyubid and early Mamluk periods this post of amir jāndār was considered one of the most important offices in the Sultanate. It was given only to an amir of a hundred or an amir ṭablkhāna (an amir of forty). Under the Ayyubids the officer who held this post was not involved in building projects, military or civilian. He officially commanded the doorkeepers and esquires; he was the chief of the royal guards and was in charge of the zaradkhāna (arsenal and prison). He was also one of the inkwell carriers.48 According to Ayalon the post declined in importance during the Mamluk period.49 Irwin presents a different picture, according to which the official duty of the amir jāndār was to guard the Sultan’s tent. But “the formal description set down in chancery manuals gives a misleading impression of the real function and importance of the emir’s responsibilities. Aybak al-Afram’s responsibility for guarding the sultan was purely nominal. Until his death in 1295, he acted as master of public works and as senior military engineer.” 50

Mamluk power on display  133 An office of this kind was foreign to the Ayyubids and probably also to the Franks. Crusader fortifications must have been designed and built under the supervision of various persons, for the simple reason that fortresses in the Levant were owned by several different bodies (the Military Orders, the king and the nobility). The situation in the Ayyubid principalities was no doubt very similar; although certain amirs were constantly employed by the sultan to carry out public works, and some of them supervised the building of more than one fortress.51 In the early decades of the Mamluk sultanate the office developed and changed considerably. The amir jāndār may have been one of the high-ranking officials within the Mamluk court specially entrusted to oversee building projects; this would include fortresses that the Sultan had decided to restore. He served as an adviser on military architecture and was often sent out to supervise work on the site. Aybak al-Afram was also responsible for the transportation and operation of siege machines. Irwin presents him as a military engineer though the meaning of the word muhandis and the status of the holders of such titles is rather obscure. Beyond the field of military architecture, he also supervised large irrigation projects. Aybak al-Afram participated in the rebuilding of several fortresses. When the amir Ṣārim al-Dīn al-Kāfirī was appointed by Baybars as the commander of Ḥiṣn al-Akrād (Crac des Chevaliers), Aybak al-Afram and ʿIzz al-Dīn Aybak al-Shaykh were put in charge of the restoration work.52 The same pair saw to the reconstruction of Qalʿat al-Qurayn (Montfort).53 During the siege of Arsūf Aybak al-Afram cut the logs and supervised the building and operation of the siege machines. When the siege ended successfully he was highly commended by the sultan for his diligence.54 After the siege of Caesarea he received a robe of honor for his work and effort in the operation of the siege machines.55 During the siege of Safad he was sent to help bring the siege machines from Jacob’s Ford.56 When al-Ashraf Khalīl was preparing to besiege Acre, Aybak al-Afram was sent to Damascus to build siege machines and organize their transfer to Acre.57 As a senior adviser Aybak al-Afram must have felt sufficiently experienced and confident to disagree with al-Ashraf Khalīl, who wished to demolish Shawbak (1293). What led to the sultan’s decision was the strength of the local Bedouin tribe, which gained control of the fortress. The destruction was to be executed by Aybak al-Afram. Aybak al-Afram thought the fortress was an important stronghold and should be kept and maintained. This was to be his last act as amir jāndār. The sultan was angered by his objection and had Aybak al-Afram severely punished; he was thrown into jail and his property was confiscated.58 He was released during the early reign of al-Nāṣir Muḥammad, after the murder of al-Ashraf Khalīl, and reestablished his status and position. He died two years later, in 1295. Aybak al-Afram regained the respect of the grandees of the court, as we are told that the sultan himself and many high-ranking officials attended his funeral.59 Although fortifications were still being constructed in the fourteenth century most of the important strongholds were conquered and rebuilt during the first three decades of the Mamluk sultanate. Thus once the Crusader kingdom fell and the Īlkhānid state had signed a peace treaty with the Mamluks, the post of amir jāndār may have lost much of its significance.

134  Muslim Fortresses in the Levant

Inland fortresses One of the finest examples of Mamluk fortifications is that at al-Ṣubayba. The fortress is situated on a long narrow spur at the foot of Mount Hermon, six km west of the Druze town of Majdal Shams. Although most of the fortress was built by the previous Ayyubid regime, the first fifteen years of Mamluk construction, beginning in 1260, represent the state of the art in Mamluk military architecture.60 The literary sources as well as the inscriptions and archaeological remains give a clear picture and absolute dating of the work conducted on the site. This in many ways provides a key to Mamluk military architecture during its formative decades. Al-Ṣubayba, a short history The execution of the last Ayyubid ruler of the fortress, al-Saʿīd Ḥasan b. al-ʿAzīz ʿUthmān, was ordered by Qutuz (r. 657/1259–658/1260), due to his collaboration with the Mongol commander Kitbugha. This had led to the capitulation of the Muslim garrison at al-Ṣubayba in the spring of 1260. He also made the grave error of joining the Mongol force at the battle of ʿAyn Jālūt (3 September 1260).61 The Mongols, who took al-Ṣubayba, had no intentions of retaining the fortress. They pillaged and destroyed as much of the site as they could. However, it seems that Ibn Shaddād exaggerated when he described the extent of the Mongol destruction.62 His account consists of one short sentence: “the Mongols ruined (akhrabahā, ) al-Ṣubayba and they did not leave any remains [of the fortress].”63 Al-Ṣubayba was not alone in the treatment it received at the hands of the Mongol forces. But, unlike the Franks, who lacked manpower and funds, and were not able to restore the damage wrought to their coastal fortresses, the Mamluks dealt with the destruction left by the Mongols within a relatively short time. The military importance of the fortress underwent considerable mutations in the political and military shifts that took place after 1228–30, when it was first built. The Ayyubid ruler of Damascus erected the fortress in haste, apparently expecting his capital to be attacked by a Crusader-Frankish force that joined the army of the Ayyubid sultan in Cairo. By the end of 1260 the entire political and military balance had changed. The Mamluks had just defeated the Mongols at the battle of ʿAyn Jālūt, the border had moved further east, and al-Ṣubayba was no longer a frontier fortress as it had been during most of the Ayyubid period. Nevertheless the fortress attracted a great deal of attention, mainly due to its owner’s high rank in the Mamluk sultan’s court. An exception to the rule: exclusive ownership and its influence on the quality of construction The figure who played the most important role in the restoration of al-Ṣubayba was the Mamluk amir Badr al-Dīn Bīlīk al-Khaznadār al-Ẓāhirī. Much has been written about this remarkable personage by contemporary chroniclers and modern historians; the following is a summary of a few of his acts that concern the fortress of al-Ṣubayba.

Mamluk power on display  135 In 1260, when Baybars ascended the throne, he nominated Bīlīk al-Khaznadār as his nā’ib al-sulṭana (viceroy),64 the highest rank in the Mamluk sultanate after the sultan himself. At the time Bīlīk al-Khaznadār was only eighteen years old.65 Al-Ṣubayba was given to him together with Banias and the surrounding area. Several chronicles mention this but the terminology used differs and does not allow us to reach a clear answer as to whether the land became his private possession (milk, freehold appanage) or an iqṭāʿ, that cannot be passed on to an heir.66 Al-Ṣubayba appears to present the only case where the sultan did not directly nominate the governor (nā’ib al-qalʿa) of the fortress. The amir Baktūt al-Khaznadārī was chosen by Bīlīk from among his own Mamluks to govern the fortress. This procedure was not repeated; under Qalāwūn the nā’ib of al-Ṣubayba was appointed by the sultan, as was the custom in the rest of the realm.67 Again in contrast to the sultan’s usual method, the restoration of the fortress was initiated by Bīlīk.68 The sultan did not overrule this decision, perhaps because of the great confidence he had in this amir. Although Bīlīk probably owned the property, it is doubtful how much time, if at all, he actually spent on the site, since he held the post of nā’ib al-sultana until his death (16 Rabīʿ I 676/August 1277).69 The fact that the fortress belonged to the second in command to the sultan had considerable influence on the rank of the amirs who commanded it. In addition, it seems that the building force was carefully chosen and possessed higher professional abilities than those employed on other sites. Expense was obviously not spared and thus the construction work at al-Ṣubayba was of an exceptionally high quality. The monumental inscription (Figure 4.2) found on the site lists the officers in charge of the building and gives the precise names of craftsmen who took part in the reconstruction of the fortress. Baybars and Bīlīk are listed at the beginning, with their titles, but probably only the last five men mentioned in the inscription (see diagram below) were actually present on the site and participated in the construction work. Not surprisingly in the Mamluk period, those of military rank received greater honor than the skilled craftsmen. Thus the nā’ib of the fortress, Baktūt, comes third and another Mamluk officer, the amir Sanjar, is directly after him. The professional civilian craftsmen are presented after the military men. The first civilian is a certain ʿAbd al-Raḥmān, who carries the titles of ustādh and muhandis. The exact profession of the muhandis in the Mamluk period is rather vague. According to Mayer the term used at the time to denote an architect was shādd al-‘amā’ir or bannā’ or muhandis. The mi’mâr (stonemason) was probably the lowest rank among the professional craftsmen.70 Doris Behrens-Abouseif concluded that the term muhandis refers to land surveyors, engineers and perhaps contractors.71 She proposed a definition of the social class of each group: while the building supervisors belonged to the Mamluk ruling class, the skilled craftsmen often remained anonymous and came from the lower social classes. Rabbat suggests that the term muhandis is probably closest to the definition of architect: “it is the only term that indicates the professional craftsmen with a wide range of technical aptitudes and theoretical knowledge that we associate today with a designer-engineer. Basically, a muhandis was a surveyor with a primary training in geometry and perhaps hydrography …”72 In this particular case of al-Ṣubayba, the architect is not only acknowledged but is given an honorable place in the large inscription that adorned the fortress walls.

Figure 4.2  Al-Ṣubayba, the monumental inscription from the reign of Baybars

Mamluk power on display  137 The fact that the term ustādh (master) precedes that of muhandis is significant: “this title (i.e ustādh) could not have meant more than ‘qualified builder’ or ‘foreman’. But inscriptions prove that its application was not haphazard. Certain architects are called masters from a given date onward, thus proving they were not entitled to it before.”73 As to the origins of ʿAbd al-Raḥmān (al-ustādh, al-muhandis) and ʿAbd al-Wahhāb (al-miʿmār) no clue is given, nothing is known of them or their previous works. They may well have been brought from Egypt or Syria or from beyond the Sultanate’s borders: “architecture has always been an international craft and medieval architects moved to where there was most to be built and where patronage could be expected.”74 It is interesting to compare the respective ranks of the teams that dealt with the building of fortresses in the Ayyubid and Mamluk periods. The Mamluk supervisors were evidently of a higher rank than the team that oversaw the last Ayyubid building phase (1239–40). The Ayyubid construction work was supervised by a eunuch who was a great amir (al-amīr al-kabīr). His name appears before that of the governor of the fortress. Compared to the Mamluk nā’ib, his responsibilities were fewer and less demanding. The Mamluk nā’ib, Baktūt, not only commanded the fortress, but also administrated the lands that belonged to his master and commanded Bīlīk’s cavalry unit which was permanently stationed in Damascus.75 Baktūt too was given the title of great amir. Humphreys has a lengthy discussion on the title “al-amīr al-kabīr” in which he concludes that while the Ayyubids used the title very loosely, during the Mamluk period it was generally equivalent to the rank of amir of a hundred horsemen (amīr mi’a).76 The Mamluks appointed officers of higher rank simply because al-Ṣubayba belonged to the nā’ib al-sulṭana and the fortress was accordingly staffed with amirs from his own body of Mamluks, who belonged to a higher class.

Table 4.1  The rank of the building teams according to the inscriptions at al-Ṣubayba Ayyubid

Mamluk

Al-Saʿīd Ḥasan b. al-ʿAzīz ʿUthmān (prince) ↓ ʿAzīz al-Dawla Rayḥān al-ʿAzīzī (eunuch, amīr kabīr) ↓ Mubāriz al-Dīn (Mamluk amir [?], governor)

Baybars (sultan) ↓ Bīlīk (na’ib al-sultana) ↓ Baktuk (Mamluk amīr kabīr, governor of the fortress) ↓ Sanjar (an amir) ↓ ʿAbd al-Raḥman (al-ustādh, al-muhandis) ↓ ʿAbd al-Wahhāb (al-miʿmār) ↓ Yusuf (inscription writer)

Note: The Ayyubid order follows an inscription of 1239–40; the Mamluk order goes according to the inscription of 1275

138  Muslim Fortresses in the Levant As for al-Ṣubayba’s defensive importance, the role of the fortress was of greater significance for the protection of Damascus under the Ayyubids than for the defense of the Mamluk sultanate. Ibn Shaddād presents a short report, and is rather sparing of words in describing the reconstruction carried out when the site passed into Bīlīk’s responsibility. He begins with a brief account in which he uses the verb “to renew” (jaddada) in referring to the fortress in general. When he goes into detail he switches to the verb “to build” (ʿammara). Ibn Shaddād’s account ends with the description of a minaret that was added to the already existing mosque, a residence built for the nā’ib al-sulṭana and a bridge that led to the fortress.77 When Ibn Shaddād’s account is correlated with the archeological remains it appears to be quite accurate, in contrast to his report of the destruction carried out by the Mongols, which as noted above is somewhat exaggerated. Though the construction stretched over several years, it was completed during the reign of Baybars. No later buildings are recorded in the Mamluk sources. The return of the round towers The most dominant structures at al-Ṣubayba, probably the trademark of Mamluk military architecture, are the round towers built along the southern Ayyubid curtain wall. While towers of this type are rarely seen in Ayyubid rural and frontier fortresses in southern Syria and Palestine, they were introduced on a lavish scale in the early Mamluk period and feature in almost every fortress the Mamluks rebuilt. They were known in late twelfth and early thirteenth century Crusader fortresses, although they are not very common.78 The two main advantages of round towers are they are less prone to damage from siege machines, as the surface hit by projectiles is smaller, 79 and the defenders stationed in the tower have a wider angle of vision. The other striking change manifested during this Mamluk building phase is the increase in the number of towers. A new round Mamluk tower was inserted between every two rectangular Ayyubid towers. This no doubt improved the bowmen’s efficacy and further reduced the dead ground below the walls. By the end of the Mamluk reconstruction phase the number of towers along the southern curtain wall was nine; as opposed to five in the final Ayyubid phase. One of the most interesting features in this site is the fact that the Mamluks who demonstrated such high ability in the construction of towers were willing to make do with the poor structure of the Ayyubid curtain walls. The walls were repaired, but their structure and width did not change (Figure 4.3). The Mamluks decided to counteract the weaknesses and faults of the curtain walls by almost doubling the number of towers. This approach had in fact been initiated by the Ayyubids, who, as was shown in the previous chapter, laid greater emphasis on the design and construction of their towers than on their curtain walls. The Mamluks improved and further developed this concept. Thus the dense line of towers provided the strongest part of the defense, while the curtain walls became an almost negligible factor. The flanking fire from the series of towers would have been extremely powerful if every archer took his position.

Mamluk power on display  139

Figure 4.3  Al-Ṣubayba, Mamluk repairs along the Ayyubid walls

All four round towers share the same basic structure and features. The Mamluk towers were inserted into the Ayyubid curtain wall, four sections being broken to make space for them. Once completed, the new tower was carefully reconnected or “stitched” to the Ayyubid wall (Figures 4.4, 4.5). The diameter of the round towers runs between 10 and 18m; their base is completely solid. The walls were built in a similar fashion to the curtain walls. The stones are marginal drafted leaving a large projecting boss, which would enable the wall to withstand the impact of stone projectiles. Each course lies neatly on the one below (Figures 4.6, 4.7), in contrast to the external walls of the Ayyubid towers where the only well-dressed stones are to be found in the corners. All the southern round towers have only one floor. The flat rooftops provided ample space for siege machines and archers who could operate from machicoulis that hang over the rim. The teams working the siege machines had their own external staircase that led to the roof (Figure 4.5, right). The movement of soldiers in the tower and on its roof was planned so that they would not obstruct one another. The existence of only one floor is somewhat surprising as the solid structure of the towers suggests they could have carried more than one floor. It seems that due to the increased number of towers along the walls, the Mamluks decided not to raise their new round towers to more than one floor. While the towers are perfectly round from the outside, the inner halls are either hexagonal or pentagonal. The ceiling is a high dome with a large pillar to support it in the center (Figure 4.8). Each rib of the hall has a large and spacious arrow slit from which two archers could shoot at any time. The structure of the arrow slits is similar in design to those of the previous Ayyubid phases, the only difference being that they are larger and provide a clear view of the foot of the wall. An interesting feature of the arrow slits is their external stone facing.

140  Muslim Fortresses in the Levant

Figure 4.4 Al-Ṣubayba, the distribution of towers during the last Ayyubid phase and later after the Mamluk reconstruction

Figure 4.5 Al-Ṣubayba, seam between the Mamluk tower and the Ayyubid wall

Mamluk power on display  141

Figure 4.6 Al-Ṣubayba. Note the quality of masonry along the external walls of Ayyubid (left) and Mamluk (right) towers

Figure 4.7 Al-Ṣubayba, smooth ashlar blocks inside the new Mamluk tower 7

142  Muslim Fortresses in the Levant

Figure 4.8 Al-Ṣubayba, inside view of tower 5

Their outer frame has a distinct boss, shaped in a variety of geometric designs (Figure 4.9). The raised boss added further protection from siege artillery.80 The towers were connected to one another via an archers’ gallery that ran along the curtain walls. This consisted of narrow vaulted passages that were pierced with arrow slits (Figure 4.10). The western side of the fortress was not ignored, but the general approach was quite different. As the three Ayyubid towers along this section were close to one another there was no space and no need to add new towers. Thus the Mamluks enlarged only two towers by encasing them in a substantially bigger structure. The largest and most complex of those new towers is “Bīlīk’s tower,” as Hartal named it, which had three floors and was turned into a fortified residence (21.4 × 20.75m) (Figure 4.11). The cistern dug at its base and a shaft built along the entire height enabled each floor to draw water. A secret passage was constructed, on a grand scale. It consisted of a wide vaulted staircase measuring 27m in length that led out on to the northern slope. A smaller, but no less impressive passage was constructed along the northeast from the large round tower to the grand Mamluk hall (Figure 4.12). The grand and lavish scale of Bīlīk’s tower and the fine quality of the masonry indicate that as well as having an important role in the defenses it may have been used as the quarters of the nā’ib of the fortress and his high-ranking amirs. It seems that by the mid thirteenth century it was difficult to find new and original ideas in the field of military architecture, especially since there were no innovations in siege warfare that might have acted as triggers. Nevertheless, one of the most impressive features of Bīlīk’s tower is the outstanding size of the stones, the largest of which weighs 37 tons (Figure 4.13).81 This tower has no parallel among fortresses of the earlier generations. The only evidence we have of Ayyubid work carried out with stones of enormous dimensions comes from Ibn Shadādd’s description of the citadel of Aleppo.82 Although it was not a new idea it was not a common feature of military

Mamluk power on display  143

Figure 4.9 Al-Ṣubayba, Mamluk arrow slits with geometric design on the external frame (tower 7)

Figure 4.10 Al-Ṣubayba, Ayyubid archers’ gallery along the south

144  Muslim Fortresses in the Levant

Figure 4.11 Al-Ṣubayba reconstruction of tower 11 (Bilik’s tower)

Figure 4.12 Al-Ṣubayba, secret postern in tower 16 leading to the core of the Ayyubid “keep” and the grand Mamluk hall

Mamluk power on display  145

Figure 4.13 Al-Ṣubayba, giant stones along the northern face of tower 11

architecture. Perhaps it was an experiment carried out by an innovative architect, the purpose being to build the ultimate tower that no siege machine, mining team or earthquake could bring down. Huge building stones appear also in Safad, which had a remarkable Mamluk keep built in the center of the fortress (see below). Latrines: small but important improvements The additions planned by the Mamluks made garrison life more comfortable. Latrines were built of stone and set in a hidden corner within the towers so that some form of privacy was provided. The latrine was connected to a channel, flushed with water and drained into a separate cistern.83 Although the construction was simple it meant the building needed a substantial amount of water for flushing the system clean, as well as for drinking and washing. It appears latrines were regarded by the Mamluks as a basic necessity rather than an item of luxury. Water cisterns Although the region of Mount Hermon has a fairly high amount of rainfall the long summer months are hot and dry. The supply of water in the fortress depends on accumulation of run-off water during the winter months. There are nineteen water

146  Muslim Fortresses in the Levant reservoirs at al-Ṣubayba. The majority seem to have been built by the Ayyubids. The Mamluks added a few, possibly because the garrison was enlarged and the water supply had to be increased. It also may indicate a certain improvement of the standard of living within the fortress. It seems more than likely that the towers and galleries were used for housing the garrison; they were spacious and had many of the basic requirements. Although much of the fortress is well preserved we know nothing of its inner buildings such as the whereabouts of the armoury, stables, storage, kitchens, dining quarters, smithy and mosque. The sources are silent and much of the central grounds of al-Ṣubayba will have to be excavated before any of those questions can be answered. Safad The medieval fortress of Safad is located in the upper Galilee, in the center of the modern town that carries the same name. The grounds of the fortress are covered to a large extent by a pine grove that serves as a local park; its outskirts have long since been paved and partly built over. Only very few sections of the structure can be seen today. Excavations have been conducted on the site since the early 1950s and in recent years by the Israeli Antiquities Authority. Although important evidence has been uncovered, the area excavated is still very small.84 The sources, on the other hand, are detailed and a fairly clear outline of the site can be drawn if they are carefully followed. The five main sources Medieval Safad was a small village. It was only after the Crusaders decided to build a fortress there that the site began to attract the attention of Muslim chroniclers.85 Five of the six main sources, including a French author, are written by contemporaries who lived in the second half of the thirteenth and in the early fourteenth century, when the fortress was being built by the Franks and rebuilt by the Mamluks. The anonymous French writer gives a comprehensive account in his work titled “De constructione castri Saphet,” where he describes the building of the thirteenth-century Crusader castle, emphasizing the first phase of the work (1240–3).86 This source has been referred to and analyzed by most scholars who have discussed the construction of Frankish Safad at any length; therefore I shall refer to it only briefly. It is the only account that gives a clear idea of the fortress as the Mamluks saw it in 1266. From 1266 onwards the additional buildings and repairs carried out by the Mamluks are presented by four Muslim authors. Ibn Shaddād and Ibn ʿAbd al-Ẓāhir both give long and elaborate accounts of the work conducted by Baybars immediately after the siege. Each emphasizes different parts of the fortress. Of all the Muslim historians who wrote about Safad in this period, al-Dimashqī (d. 727/1327) who died there,87 and al-Ṣafadī (696/1297–764/1363), who was born there and later returned for a short period,88 are probably the only ones who saw the fortress with their own eyes. Their accounts are similar, short and rather colorful. Al-Ṣafadī adds an important fact to the chronology of the Mamluk building and states that the large central tower was completed during the reign of Qalāwūn.89 Al-Dimashqī’s attention was caught by the fascinating mechanism

Mamluk power on display  147 of the well with its chain of buckets that drew water from the cistern beneath the round central tower. His precise account leaves little doubt that al-Dimashqī had seen and studied the fortress carefully.90 It was the stronghold’s impressive size rather than its military importance that attracted the interest of most of the Muslim contemporary chroniclers. None of them mentions its military significance. It is difficult to explain the Mamluk decision to rebuild this fortress. The reconstruction of Safad had little to do with the larger concept of the Sultanate’s defense. According to Tracy, the larger the population the greater the chances that a city will be fortified.91 It seems the main reasons for the reconstruction were probably the social and economic status of the town, the need to establish Mamluk authority in the area and provide security for the town’s population and for villages in the vicinity. The hinterland of Safad and its agricultural prosperity are all vividly described by the anonymous French chronicler. The wealth and the revenue that could be collected from the local population were no doubt taken into account when it was decided to invest in rebuilding and garrisoning the site. A short history of Safad under Crusader rule One cannot ignore the 164 years during which the Franks slowly constructed the fortress at Safad, which in due course became one of the most formidable Frankish fortresses in the region. It was probably first built by the Crusaders in 1102–3.92 It later became the property of the crown, but the exact date is unknown. In 1168 King Amalric decided to hand it over to the order of the Templars.93 After the battle of Ḥaṭṭīn (1187), Safad was taken by Ṣalāḥ al-Dīn and remained under Ayyubid rule until 1240.94 Ibn Shaddād and Ibn al-Furāt give a list of its Ayyubid governors, while Ibn al-Furāt declares that no construction work was done during the Ayyubid period.95 Ibn al-Athīr lists Safad as one of the fortresses reinforced by the Ayyubids in 588/1192.96 No archaeological evidence of Ayyubid construction has been found to date. It is possible that none exists, as the fortress suffered two severe disasters within a short space of time. The first was a powerful earthquake that struck the region in 1202.97 The second was due to al-Muʿaẓẓam ʿIsā, who in 1219 demolished the fortress at Safad, as he had done at Mount Tabor, since he feared the return of the Franks.98 Following the treaty signed between Richard of Cornwall and al-Malik al-Ṣāliḥ Ismāʿīl of Damascus the Templars returned to Safad.99 At this point the fortress was either completely in ruins or in very poor condition. Thus began the last and most significant Frankish building phase at Safad. Within two and a half years the Templars built the largest concentric fortress in the kingdom (Figure 4.14). Two formidable curtain walls were constructed. The inner wall reached the height of 44m and was 3.5m wide; the external wall measured half the height. The moat between the two walls was 15m deep and 13m wide. Seven large round towers were built along the external wall, each rising to the height of 48m and measuring 22m in diameter. The total circumference of the fortress walls was 850m.100 It is not surprising that after six weeks Baybars turned to trickery in order to bring an end to the siege.101 According to Ibn-Shaddād and Baybars al-Manṣūrī, the defenders

148  Muslim Fortresses in the Levant decided to surrender, and asked for an amān; the sultan granted this but did not keep his word, and the Templar garrison was executed in cold blood to the last man.102 The Mamluk building force and the work on the bāshūra An important part of Ibn ʿAbd al-Ẓāhir’s account of the repairs at Safad is devoted to the organization and division of the work among the Mamluk army. His description gives some details of the sultan’s participation in the reconstruction work together with his soldiers and amirs. This may be a way to glorify the sultan, although it seems that Baybars indeed differed from many of those that followed him in that he trained, fought and worked with his men. Ibn ʿAbd al-Ẓāhir begins with the clearing and widening of the moat, in which the entire army participated as well as the sultan and his entourage. The reconstruction of the towers was supervised by the amir Sayf al-Dīn al-Zaynī and the work was divided amongst the Baḥriyya and the Ḥalqa.103 Thus the main working force was the army itself, as no mention is made of slaves, captives or local villagers. Ibn Shaddād describes the sultan’s departure from Cairo on his way to start the restorations at Safad; in addition to the army, the amirs and his personal entourage a small group of skilled craftsmen joined the sultan.104 The damage caused to the bāshūra during the weeks of the siege must have been considerable. Siege machines had been set up on every side of the fortress and sappers

Figure 4.14 The fortress of Safad, the round tower and inner gate

Mamluk power on display  149 worked day and night. According to Baybars al-Manṣūrī, tunnels were dug under the curtain walls, the towers, the fortress foundations and at the corners.105 Although the fortress was surrounded, the Mamluk force focused on the bāshūra and this area suffered the most. Ibn Shaddād says that “the fortress had a bāshūra that was built by the Franks and when our lord the Sultan ruled it was destroyed to its foundations, and it was built anew from hirqilī ( ) masonry.”106 Al-Ṣafadī’s description minimizes the damage: “He renewed the bāshūra and rebuilt it from hirqilī stones and [built] in the same manner the towers and the curtain walls (badnāt).”107 At all events it is clear from both accounts that large parts of the external circuit walls and towers needed repair. The term hirqilī comes from Heraclius, the Byzantine emperor, or Hercules, the mythological hero.108 The use of the word in this particular context may mean either the re-use of Byzantine stones (possibly from a Byzantine ruin in the near vicinity) or stones of great strength, characteristics attributed to Hercules.109 The term appears to be quite rare; and is reserved for the description of fortifications. It is used once again by Ibn Shaddād in a description of the Aleppo fortifications built by the Ayyubid prince al-Malik al-Ẓāhīr Ghāzī in 611/1213: He tore down the bāshūra that used to be in it and poured earth on the tall and rebuilt it with hirqilī stones.110 In this particular context, Ibn Shaddād used the term hirqilī to convey the strength and size of the stones used by the Mamluks. The idea of using large stones may have been inspired by the Mamluks’ contemplation of the results of their own siege. Strengthening the outer fortifications by using larger stones would help endure the impact of heavy siege engines, and make sapping and removing the stone blocks an all but impossible undertaking. The walls were given further protection by a glacis (musaffaḥa).111 Ibn Shaddād notes that the towers had machicoulis. The shape or number of towers built or repaired is not mentioned in any of the sources. Ibn-ʿAbd al-Ẓāhir does not contribute much to our knowledge of the structure of the towers along the external curtain wall. Nevertheless, like Ṣafadī, he says that those towers were merely repaired rather than built anew from the foundations. No section of the moat or of the towers described by the sources has been excavated, and only fragments can be seen protruding from the earth along the northern side of the site. We thus have no choice but to rely on the literary descriptions. Taking into account the already massive structure of the seven towers built by the Templars (22m in diameter and 48m in height) it seems that the Mamluks only restored the outer circuit of towers and did not add new ones or widen them as they did in al-Ṣubayba. The fact that the walls at Safad were almost twice as thick as those at al-Ṣubayba may have been the reason that the Mamluks made do with a relatively small number of towers along the outer wall.

150  Muslim Fortresses in the Levant The main gate and the secret passage Ibn Shaddād describes the Mamluk gate complex. He begins with the wooden drawbridge stretched across the outer moat and operated by a simple pedal (dawāsa, ). Once across, one came to a T-shaped junction with a gate on the right and another on the left. This structure has no close parallel among Ayyubid or Frankish fortresses, which followed the traditional and conservative structure of bent gates and gate-towers.112 A large Mamluk gate complex was uncovered in the southwest corner during the excavations conducted in 1995 (Figures 4.15, 4.16). This is not the gate described by Ibn Shaddād, but it was built along similar lines. It served as an entrance to the inner bailey. Certain sections of the previous Frankish structure were incorporated in this new gate. The quality of both masonry and construction is rather poor. The local Safad limestone is soft and porous and of a lower quality than that further north in the vicinity of al-Ṣubayba. This, however, is not an adequate reason for the generally low standard of the work. The dating of the gate to the Mamluk period relies on the design of the arrow slits along the passage and the fact that Mamluk coins were found in the foundation levels. It is possible that the building material was simply recycled and taken from the Frankish structures that were demolished during the siege. Since the gate was not located on the external curtain wall it was not expected to withstand heavy attack, thus the quality of the work was of no great importance. Ibn ʿAbd al-Ẓāhir is the only Arabic source that describes the building of the secret passage by the Mamluks (Maqrīzī quotes him almost word for word).113 Secret passages existed prior to the Mamluk conquest of the fortress and they are briefly mentioned by the anonymous French chronicler as “underground tunnels between the outer wall and the [inner] ditch.”114 Pringle suggested that the secret passage attributed by the chroniclers to Baybars was originally built by the Templars.115 As there is no archaeological evidence all we have in this case is the written word. Hartal’s thorough examination of the secret passage at al-Ṣubayba shows the Mamluks were sufficiently aware of the importance of such passages, and equally capable of building them whether long or short, above or below ground.116 A free-standing central tower With the help of the archaeological remains and of the sources it is possible to get a reasonably accurate picture of what this free-standing tower of outstanding height looked like (Figure 4.17). It suggests some new ideas in the field of military architecture that may be ascribed to the Mamluks, although once introduced at Safad they were not repeated elsewhere. The great round tower is mentioned by Ibn Shaddād, Dimashqī and Ṣafadī. Though today it is largely in ruins, this immense structure that probably served as a keep, still dominates the summit. One of its most unusual features is its complete isolation; the tower is a free-standing structure. It was not incorporated in the inner

Mamluk power on display  151

Figure 4.15 Safad, the remains of the Mamluk inner gate on the southwest

Figure 4.16 Safad, plan of round central tower and Mamluk inner gate

152  Muslim Fortresses in the Levant curtain walls nor did it form part of a chain of towers. The walls that can still be seen surrounding it date to the Ottoman period. The idea of a huge free-standing tower within a fortress is rare in both Frankish and Muslim fortifications. The best Frankish examples are Chastel Blanc (Safita) and Gibelet (Jubayl) dating to the twelfth and early thirteenth centuries. Al-Safadī writes: “He built in the fortress a huge tower that was very tall and rose to the height of eighty dirāʿ, and its building was completed in the reign of al-Manṣur [Qalāwūn].”117 The diameter, according to al-Dimashqī, was 40.6m.118 The sources are not unanimous concerning the height, giving it as between 46 and 96.6m. Ibn Shaddād estimates it was one hundred dirāʿ high (58m), and says that from the top one could see whoever walked along the moat or around the fortress.119 The Mamluks did not take into account the tectonic activity of the region; the tower partly survived into the late Ottoman period, though a series of earthquakes in 1759 and again in 1837 submitted this massive construction to pressures it may not have been designed to withstand. What remains today is the main entrance and the cistern carved out of the bedrock below it. The entrance is encased in a solid stone structure that served as the tower’s base (Figure 4.17). The rest of the tower can be reconstructed from the information provided by the sources. Dimashqī’s account describes two winding passages each wide enough to enable five horses to climb to the top. Ṣafadī replaces the five horses with two camels.120 This path was designed in the shape of a snail shell (ḥilazūn, ), a wide spiral ramp or staircase encased in the walls and leading to the flat rooftop (Figure 4.18). The latrines (manāfiʿ, ), which are mentioned by al-Dimashqī, can still be seen at the entrance. There was no shortage of storage place and the water in the large cistern could supply the whole garrison and the community of the nearby area.121 The bell-shaped cistern held approximately 800 cubic meters of water. This tower is the only building attesting to Mamluk construction carried out with stone blocks of outstanding size, the kind that Ibn Shaddād and Ṣafadī described as

Figure 4.17 Safad, reconstruction of the central tower. Note the cistern below

Mamluk power on display  153

Figure 4.18 Safad, reconstruction of the entrance to the central tower. Note the latrines and fine masonry

hirqilī stones (Figure 4.19). The soft limestone of the external wall may have had a few advantages; Duffy suggests that in general an extremely hard stone is considered to be a poor material since it would easily break under the impact of projectiles.122 The only features shared with Bīlīk’s tower at al-Ṣubayba are the large building blocks and the cistern dug out below the tower. Other than that, the tower at Safad remains unique both in its structure and dimensions. A water supply that allowed bath-houses Bath-houses feature in the accounts of both Ibn Shaddād and Dimashqī. Archaeological remains have yet to be found. From Ibn Shaddād’s description of the water system and the evidence of cisterns, it seems that the fortress at Safad relied on rainwater collected inside the fortress grounds. In addition water was brought from birkat (pool) burj al-yatīm to a square tower near the gate of the fortress. This pool also supplied the bath-house.123 Dimashqī provides a detailed description of the mechanism of the water-wheel that was built above a well inside the fortress. The well (3.5m in diameter and 64m deep)124 may have been fed by an underground aquifer as well as rain water collected from the roofs and channeled to the well. The mosques within the fortress Baybars returned to Safad after attending to various affairs in the region. Once he arrived, says Ibn ʿAbd al-Ẓāhir, “He went up to the fortress to pray in the tower where he used to sit. There he saw a large statue (ṣanam) of St. George. He ordered

154  Muslim Fortresses in the Levant

Figure 4.19 Safad, central round tower. Note the construction and the dimensions of the stone blocks

to tear it out (qalaʿa), break it, purify the place and build a miḥrāb instead of it.”125 It is difficult to conclude from this account that it occurred in the large round tower,126 since neither a tower nor a chapel is mentioned by the anonymous French author. Other than the hall that was converted into a mosque, Ibn Shaddād says that the church within the fortress was turned into a mosque.127 The act of building mosques became an important symbol of Muslim victory. Even in chronicles that only briefly note the reconstruction work, there is often a word or a short sentence referring to the building of a mosque or the conversion of a church into a mosque by the sultan. The elusive inscription Our knowledge of the inscription commemorating Baybars’s conquest and reconstruction of Safad comes from Ibn ʿAbd al-Ẓāhir and Maqrīzī.128 It was, they say, inserted in the curtain wall, but it has yet to be found: We wrote in the Psalms (Zabūr) after the message (given to Moses) My servants the righteous, shall inherit the earth.129 Truly it is the party of Allāh that will achieve felicity.130 And he ordered this stronghold and its fortification to be renewed and the completion and improvement of its building, after he liberated it from the captivity of the cursed Franks and he restored it to the authority of the Muslim rulers and transferred it from the possession of the Templars to the believers and he returned it to the faith as it was in the beginning. And there he dealt the infidels defeat and sorrow, he exerted great efforts, he replaced unbelief with belief, the sound of bells with the sound of the call to prayer and the Gospel with the Qur’ān. And he took it upon himself to carry earth and stones on his head from the moat together with his entourage. The sultan al-Malik al-Ẓāhir

Mamluk power on display  155 Abū al-Fatḥ, Baybars. And may [God] grant a portion of his [Baybars’ heavenly] remuneration to any Muslim ruler who takes possession of this fortress and any mujāhid who makes his dwelling in it; and let him [God] not deprive him of mercy, secretly and openly and the [supreme] reward is preserved for the pious until the day of judgment.131 The sultan is the only personage mentioned by name in the inscription. His soldiers are described as mujāhidūn but none are further identified. The inscription communicates contemporary political propaganda to the local population that has just been occupied by a new regime. Its main intention is to reassure the people that the conquest was a righteous act. It emphasizes the legitimacy given to the sultan by God himself, who had ordered the campaign and the restoration of the fortress. Furthermore, it promises that those who participated in the siege will be rewarded. An inscription of this type was another sign of sovereignty, rather like the minting of a new coin.132 An important fact revealed here is that the construction project at Safad was partly financed by the sultan’s own treasury. This was not always the case; funds for restoration and the garrison’s upkeep were occasionally collected from the nearby region, as at Marqab.133 Karak’s garrison was paid by the sultan.134 The rich agricultural land around al-Ṣubayba, which belonged to Bīlīk, probably provided part of the resources for the building of the fortress. Qāqūn The fortress of Qāqūn is situated in the eastern perimeters of the Sharon, 1 km north of the modern settlement of Gan Yoshiyya. Although it was the only site restored by Baybars in the region of the Sharon, none of the Mamluk sources provide a detailed architectural description of its fortifications. Ibn ʿAbd al-Ẓāhir says it became a strong fortress, but his emphasis is on the town’s public buildings, the reservoir, the newly-built mosque and the khan.135 The town of Qāqūn, which became a wilāya (an administrative center of a district) in the reign of al-Nāṣir Muḥmmad b. Qalāwūn,136 was by then a well-established stage in the barīd between Tīra and Faḥma (a section of the route that ran from al-Gaza to Jenin and Damascus), and in addition it maintained a dovecote.137 Qalqashandī (d. 821/1418) describes Qāqūn as a pleasant town with a bath-house, and a mosque. It had no surrounding walls, but it had a fine (laṭīfa) fortress.138 Earlier Frankish sources are of little help. The fortress Caco (Qāqūn) is briefly mentioned by Fulcher of Chartres and William of Tyre in 1123, but no description of the site is given.139 The village passed into the hands of the lord of Caesarea in the last quarter of the twelfth century. It was taken by Baybars in 1265 together with Caesarea and Arsuf. Qāqūn is mentioned again during Baybars’s reign in the Mamluk chronicles,140 and in the Gestes des Chiprois, which recounts King Hugh of Cyprus and Jerusalem and Prince Edward of England’s attempt to conquer the site in 1271. Whereas the Muslim accounts refer only to the battle, the Templar of Tyre, a contemporary Frankish source adds a short descriptive phrase:

156  Muslim Fortresses in the Levant … and they besieged some Saracens within a tower at Qaqun. It was very strong, surrounded with ditches filled with water. They came near to taking it, but our men were afraid to linger too long while the alarm went out across the land, since the Saracens would then assemble from all parts. So our men departed and retuned to Acre with all their loot, safe and sound.141 This is the longest description we have of the fortifications of Qaqun. The reason for the re-establishment of the fortress by Baybars is provided by two sources, one Muslim, the other Frankish. Although they give very different perspectives they seem complementary. According to Ibn al-Furāt, Baybars ordered the restoration of the fortress since he saw it as his responsibility to defend the local population: “his subjects living in those parts needed a place of protection so he decreed that the castle of Qāqūn [Caco] should be restored. This was done and it became a strong place.”142 Burchard of Mount Sion (who visited the area between 1271 and 1285) suggests that the Mamluks “placed a garrison of soldiers here to watch the Pilgrims’ Castle.”143 Ayalon follows Burchard’s suggestion (although he does not mention this source in his discussion). According to Ayalon, because the Mamluks had a relatively poor navy, they established small garrisons in fortified towers as a means of guarding the coast, after they demolished the fortifications of the harbor towns. 144 Although Qāqūn is considerably smaller than al-Raḥba and al-Bīra, it belongs to the same category of advance warning and protection as the fortresses established by the Mamluks along the Euphrates frontier. Baybars combined the two strategies; he settled Türkmen along the coast,145 and garrisoned Qāqūn in order to give the new population a fortified center. The garrison was connected to the capital in Cairo and to Damascus via the barīd. A similar strategy was adopted six hundred years earlier, by the Umayyad governor of Syria, Muʿāwiya b. Abī Sufyān (r. 661–80). After the conquest of the coastal plains, Muʿāwiya brought soldiers from the contingents of the Zuṭṭ (of Indian origin) and the al-Sayābija (of Persian origin) to settle in the region and help strengthen the coast in case a Byzantine fleet tried to re-conquer the region.146 Did Baybars build or simply restore Qāqūn? The tower with its two floors can still be seen today. So far it has only been surveyed and is yet to be excavated.147 It is a complete structure with no obvious repairs or additions on the outer perimeter. Two nineteenth-century descriptions that note the fortress do not describe the tower as part of a larger fortified complex. Guérin reports a large village with remains of a small fortress in its center.148 The tower is best described by Conder and Kitchener who visited the site in 1873: “in the middle of the town is a square tower of small masonry. Some of the arches are pointed … a staircase leads up to the roof … it is about 60 feet square and between 40 and 50 feet high.”149 An aerial photograph taken by the Royal Air Force in 1918 shows the village perched on the hill. The quality of the photograph is poor and the tower can hardly be seen. A faint line may possibly indicate a curtain wall (Figure 4.20).

Mamluk power on display  157

Figure 4.20 Qāqūn, the tower (left) and the village from the southeast (1918)

The clearest reference to the fortress, then, is that in the Gestes des Chiprois which speaks of a tower on a hill surrounded by ditches; since this source describes the assault of 1271, Baybars must have reinforced the site shortly after the Mamluks took it. This has already been suggested by Pringle and Kennedy.150 According to Benvenisti, Qāqūn is “a Mamluk castle built on Crusader foundations.”151 The size and scale of the Mamluk work can only be estimated if the tower and its immediate surroundings are excavated. The fortress of Qāqūn: a free-standing tower The tower of Qāqūn is built on a small hill that rises only slightly above the plain (52m above sea level) (Figure 4.21). Its location 13km east of the Mediterranean enabled its garrison to keep a watchful eye on the shoreline. It is somewhat misleading to describe Qāqūn as a fortress, since it lacks many of the features that we have come to regard as characteristic of a thirteenth-century stronghold (curtain walls, towers, complex gates, and secret passages). It is a square free-standing tower. No external additions can be seen above the ground. We are told that the site was surrounded by ditches filled up with water, but these are difficult to trace, and they may well have filled through lack of maintenance soon after the site lost its military importance, and as houses were built in later centuries along the slopes of the hill. Rising to a height of 8.5m the tower has two spacious floors. The ground floor (14.5 × 17.6m) is comprised of two parallel barrel vaults. The first floor seems to have had a groin-shaped vault, though little evidence remains. The remains of two wide and spacious arrow slits can still be seen on the western side. The size and design of the arrow slits suggests they could have accommodated two archers (Figure 4.22).

158  Muslim Fortresses in the Levant

Figure 4.21 Qāqūn, the tower (2004)

Figure 4.22 Qāqūn, plan of first floor and suggested reconstruction of the tower

Thus, eighteen bowmen could have fought from the first floor. The ground floor is completely devoid of arrow slits. Though the distribution of the arrow slits is fairly generous, the basic square shape of the tower meant that a large area below it was out of the archers’ field of vision. The external walls are 2.58m wide. Their base is slightly wider measuring 2.8m. The porous local beach-stone (Kurkar) is badly eroded.152 The size of the stones is small, but the quality of the masonry seems to have been originally quite high.

Mamluk power on display  159

Figure 4.23 Qāqūn, masonry and construction. Note the columns inserted in the southwest corner

Diagonal lines, generally identified with Crusader masonry, can be seen on many of the stones. Some of the building material is clearly recycled spolia (pillars and fragments of a frieze) from Roman or Byzantine sites. The corners of the tower were built with greater care, using larger and better-hewn stones (Figure 4.23). It is difficult to trace the exact location of the main entrance to the building; the staircase still stood when the site was surveyed by Conder and Kitchner in 1873. The entrance must have been a simple straightforward passage with no angles or turnings, a design that did not provide any form of defense for the people inside. Furthermore,

160  Muslim Fortresses in the Levant since the tower had only one entrance, vanquished defenders would be trapped inside with no means of escape. Considering how strategically important the site was, it is rather strange that Baybars made do with minor repairs and did not enlarge Qāqūn in order to accommodate a bigger garrison. It may well be that an excavation will reveal a more complex fortification that will change the current picture. The tower was put to the test in the winter of 669/1271. The Mamluk garrison managed to hold its ground, though according to the Frankish source if not for the relief force sent by Baybars from Damascus, the Franks would have succeeded in taking the site. The fact that the governor (wālī) and the amir Rukn al-Dīn al-Jāliq were wounded and the ustādār Ḥusām al-Dīn was killed shows how close the Mamluks were to losing the place.153 While Qāqūn was held by the Franks its proximity to the Crusader coastal cities safely guarded its existence. If it came under severe attack it could receive help from Arsuf or Caesarea. A fortified tower was probably all that was necessary if the Franks’ purpose was to guard the inland route, protect the nearby villages and strengthen the status of the small administrative office that represented the Lord of Caesarea. According to Pringle, many such Frankish towers served as residences. It is therefore not surprising that the tower offered such poor defense measures, for it seems to have been designed to suit a landlord or an administrator and to accommodate a garrison that numbered no more than a few dozen men. The responsibilities imposed on the Mamluk garrison at Qāqūn were much greater. Prince Edward’s assault in 1271 demonstrates the vulnerability of this site. The Crusader force, composed of infantry, Hospitallers, Templars and Teutonic knights, as well as knights from Cyprus and England,154 could well have taken Qāqūn if it had been better organized. Although the tower was well built, it represents the simplest form of military architecture. While independent fortified towers were common in the Frankish kingdom,155 this does not seem to have been a form adopted by the Ayyubids or the Mamluks. We do not know of many other sites similar in size and design that were rebuilt by the Mamluks.156 Why was the sultan reluctant to invest in a new fortification even when this appeared to be a real necessity? Possibly most of the responsibility for the protection of the coast fell on the Türkmen tribes that were brought over to settle the area. The sultan may have relied on his army’s ability to reach Qāqūn within a very short time. Thus the garrison would need to survive until a relief force arrived. This strategy was at times pushed to the extreme, as in 1271. Karak Situated in southern Jordan, the large fortress of Karak still dominates the modern town. In recent years the site has undergone considerable renovations and a small part in the southern section has been excavated. Karak cannot be defined as a Mamluk frontier stronghold. It did not protect a capital city, a mountain pass or a river ford. Nor was it rebuilt and maintained in order to keep a watchful eye on

Mamluk power on display  161 any of the Sultanate’s enemies. Its importance derives from its location on the main road between the two capitals of the Sultanate, Damascus and Cairo. Furthermore, this road had served for centuries as the main route for merchant and pilgrim caravans coming from the north and making their way to Mecca and Medina. It was a provincial capital and a refuge for the sultan and members of the royal family. Recent archaeological surveys and research in Trans-Jordan have shown that the area was of considerable economic importance. The fortress at Karak was first built by Paganus Pincerna, Lord of Oultre-Jourdain, in 1142. Ibn Shaddād says that the core of the fortress was built round a monastery.157 In 1177 it passed through marriage to Reynald of Châtillon. According to Milwright as long as the southern part of Trans-Jordan was held by the Franks it served as the agricultural hinterland of Jerusalem.158 Trans-Jordan was conquered after the battle of Ḥaṭṭīn (1187) and Karak together with several other fortresses surrendered (1188–9) and became part of the Ayyubid sultanate. Its first ruler was al-ʿĀdil, Ṣalāḥ al-Dīn’s brother and the future sultan (r. 597/1200–615/1218), who was given Karak and the surrounding region in 1189 as an iqṭāʿ.159 This marked a significant change in the status of the region and of the fortress, which had become the residence of a high-ranking member of the sultan’s family. Al-ʿĀdil remained the governor until 589/1193 and received the news of Ṣalāḥ al-Dīn’s death that year while he was residing at Karak.160 Soon after Karak became an independent principality.161 The town was passed on to Al-ʿĀdil’s son al-Muʿaẓẓam ʿĪsā in 594/1197. He ruled at first under his father’s guardianship, and only became an independent ruler in 615/1218.162 Both the inscriptions and the sources show that it was during his reign that the town and its defenses were extensively built. Al-Nāṣir Dāwūd b. al-Muʿaẓẓam received Karak from al-Kāmil in 625/1228, after losing Damascus. Karak soon became the capital of his principality, which extended over most of Trans-Jordan, and included Jerusalem, Nablus and Bayt Jibrīl.163 In 647/1249 the fortress surrendered to the last Ayyubid sultan of Egypt, al-Ṣāliḥ Ayyūb, and for a brief period Karak fell again under the jurisdiction of Cairo.164 Al-Ṣāliḥ Ayyūb’s death a year later changed matters considerably. Karak’s last Ayyubid ruler was the prince al-Mughīth Fakhr al-Dīn ʿUmar, son of al-ʿĀdil, who had great political ambitions to extend his rule far beyond the territories of Karak. With the arrival of the Mongols in the region he decided to submit to Hülegü. The defeat of the Mongols at ʿAyn Jālūt brought an end to this proposed alliance.165 In 661/1263 al-Mughīth accepted the Mamluk sultan’s invitation to Baysān, where he was immediately executed,166 although initially he was promised safe conduct. Baybars set out to besiege Karak, but it turned out to be a short affair. Karak was soon handed over to Baybars by a delegation that included al-Mughīth’s sons, the qāḍī, the khaṭīb and a group of men representing the town’s population.167 In the course of 74 years of Ayyubid rule in Trans-Jordan Karak had become an important administrative center, although it may have already began to assume this role during the short period of Crusader rule. Karak was to play an important role in early Mamluk politics, but it was not directly involved in the defense of the Sultanate. The town and its fortifications had not been attacked by the Mongols and the fortress did not suffer any destruction.

162  Muslim Fortresses in the Levant Once the Mongols had retreated beyond the Euphrates there was no political entity that cast an eye on the site or wished to control it. The Franks no longer posed a threat to the merchants or pilgrims who used this route. The local Bedouin tribes submitted to the governor, though later, during the reign of al-Ashraf Khalīl, they gained strength and took hold of the nearby fortress of Shawbak. It would seem that in order to secure this main highway the existing fortress would suffice. If a Mamluk army could be sent from Cairo to the Euphrates, it could no doubt reach Karak within a couple of days. And yet Karak received the sultan’s full attention and was among the first fortresses to undergo reconstruction. The work began and ended during Baybars’ reign and was not left to his successor. The major reason for this large investment seems to have been the sultan’s decision to turn Karak together with the nearby fortress of Shawbak into a large regional granary. In his treaty on famines, Maqrīzī reports that, during the severe famine of 695/1295–6, More than twenty thousand ghirarahs [sacks of grain] that were stored in Karak, Shawbak, and [along] the Palestinian littoral were sent to the cities, [although] they had been kept originally as provisions and destined to supply military expeditions.168 This information ties in well with Ibn Shaddād’s description of Baybars’ personal inspection of the granary at Karak immediately after rebuilding the fortress and strengthening the garrison.169 Karak’s location between Cairo and Damascus and the agricultural prosperity of the region no doubt turned it into an important storage site for military expeditions on their way to Syria. It was only during times of acute crisis that wheat stored in southern Jordan was sent to Egyptian cities. There were several other reasons as to why the fortress was enlarged and strengthened, one of them being the fear that the main line of communications between Cairo and Damascus might be severed. Although the Mongols never set foot in TransJordan after 1260, they remained a threat. Other than external enemies, internal threats always existed. The possibility that a Mamluk amir might seize power was a scenario that most Mamluk sultans were aware of. Since Karak had been an independent principality during most of the Ayyubid period, this was a more tangible threat than a renewed Mongol invasion. A partial explanation may be found in the administrative upgrading that Karak underwent during this period. Turning Karak into a Mamlaka, and institutionalizing what was already a regional administrative center in the Ayyubid period may have brought about extensive building.170 The fact that Karak was an established fortress with a large garrison and a well organized regional town no doubt gave it preference over other fortresses along the Darb al-Malik.171 There is also a strong economic argument that should be added to this discussion. Later fourteenth-century sources and new archaeological evidence show that TransJordan had considerable economic potential, not to be relinquished or ignored by the new Mamluk regime. Horses were apparently one of the goods exported from this region. Some of the sultan’s mounts were bought from local tribesman round

Mamluk power on display  163 Karak.172 During most of the Ayyubid period and Baybars’ reign a treasury was kept in the fortress; it is often mentioned by the chroniclers and possibly indicates the importance and wealth of the region.173 According to Ibn ʿAbd al-Ẓāhir, Baybars had large sums of money transferred from Cairo and deposited at Karak;174 Ibn Shaddād only says that the sultan brought his treasury to the fortress.175 While it seems likely that during the Ayyubid period, when Karak was an independent principality, revenues and taxes collected from the region were safely deposited at the treasury,176 it is not quite clear why Baybars should have transferred his treasury or kept an additional treasury at Karak. If indeed Karak served as a regional commercial center with storage space for agricultural goods and large amounts of cash, such functions explain the fortress plan and the Mamluk decision to continue to invest in its defenses. The fortress at Karak clearly illustrates that the concept of large and powerful fortifications does not always correspond to the defense imperatives, but may manifest the need of a new regime to establish its presence and display its power. The provision of law, order, and security helped to maintain and strengthen a regional economy in times of transition from one regime to another. Some differences of opinion The plan of the fortress and the architectural changes Karak underwent while the Ayyubids held it (1189–1263), and later during the first Mamluk decades, is a more difficult problem to resolve than I had at first supposed. The truth of the matter is that unlike al-Ṣubayba, where Mamluk and Ayyubid construction can be differentiated without too much difficulty, at Karak this is not the case and, identifying each phase according to architectural characteristics is not an easy task. Although some inscriptions were found, most are no longer in situ and it is not always clear where exactly they were found and to what buildings they belonged. Several scholars have written about the fortress, though none has explicitly defined the different Muslim phases. The Muslim construction at Karak is generally referred to as belonging to the “Arab period,” as “post-1188”177 or as “later Muslim work.”178 According to Boase and Pringle the southern keep was built by al-ʿĀdil,179 though according to Sourdel it was built by Baybars.180 Milwright claims that some architectural changes were made by the Ayyubids, though this was mainly repair and maintenance work rather than large-scale building.181 Deschamps, who in the early 1930s carried out the most comprehensive survey of the site, concluded on the basis of both the architecture and the contemporary Muslim sources that much of the fortress was repaired by the Ayyubids. The main construction work, however, was done after Baybars had taken Karak, when both the fortress and parts of the town walls were rebuilt. And yet the legend on the citadel plan drawn by Deschamps does not define who built what and he too writes “travaux Francs (1142–1188)” and “travaux Arabes (après 1188).”182 Johns, who studied the town and the region of Karak, says that both the western salient and the southern front were built by Baybars.183 Athough several inscriptions were found and published their original location in the fortress is not always clear. The southern keep has a long band inscription, though its location at the top of the tower, makes it impossible to read (it appears to have

164  Muslim Fortresses in the Levant been almost ignored by most scholars). The two inscriptions that commemorate the urban fortifications of Baybars have been recorded and cited by several authors who have studied this site. They are set on the two large towers inserted along the town fortifications. The first, on the northwest corner tower, was recorded by Luynes as “Tour dite de Malek ed Daher,” the second as “Tour al Banawī,” a semi-circular tower along the southeast section of the town wall.184 None of the inscriptions published to date commemorates Baybars’ work in the fortress. Karak rates among the principal fortresses conquered and rebuilt by the Mamluks in the southern regions of Syria, and is the largest of the fortresses in Trans-Jordan. It has three distinct sections (Figure 4.24). The two later sections were built by Baybars and their construction was completed in his lifetime as attested by the inscription. Although there is more than one Crusader building phase I shall only briefly touch on the subject when analyzing and comparing the various structures and forms used by the Franks, the Ayyubids and the Mamluks. However, it is important to note here that the Crusader plan and building techniques are neither the most sophisticated nor of the finest quality.185 Nevertheless they withstood all three of Ṣalāḥ al-Dīn’s sieges, attesting to the strength of both fortress and garrison.186 The core of the fortress has a trapezoid shape. It is easily distinguished by the dark, hard, coarsely dressed stone.187 The rectangular towers along the eastern and western sides are modest in size. Since they hardly protrude from the curtain wall they provide very poor vision, leaving much of the ground below the wall out of the defenders’ sight. The towers are sparsely distributed and fitted with small and relatively few arrow slits. The vast vaulted halls along the northern side, described by

Figure 4.24 Karak: (1) the heart of the fortress, built by the Franks in two phases (1142–88); (2) the western galleries; (3) the southern keep

Mamluk power on display  165

Figure 4.25 Karak, the Frankish entrance along the northeast corner

Kennedy as “Victorian railway tunnels,” are no doubt the most impressive features of the Frankish building phase.188 An entrance was constructed at the seam between the northeastern tower and the gallery wall (Figure 4.25). The outer defense lines included a glacis that surrounded the walls on the eastern and southern sides, and the wide moat on the north. The moat is described by Ibn Shaddād as being extremely deep, 60 dhirāʿ, which equals 34.8m.189 The depth of the moat and its effectiveness against enemy forces were demonstrated by an accident that occurred while Qalāwūn and his entourage were crossing the bridge above the moat: The bridge of the castle collapsed under them … When the steed felt the bridge collapsing, it thrust forward until it almost trampled the amirs marching before him. Thirty five Mamluks of our lord sultan fell into the fosse. … the sultan dismounted at the gate, sent for tackle and ropes, fetched up everyone who had fallen and ordered them to be treated … for the height of the bridge from which they fell into the fosse was nearly fifty cubits.”190 By the end of the Frankish rule at Karak, the total circumference of the fortress was 730m. It was defended by nine towers distributed along the west, east and south (those along the south were later destroyed by Baybars).

166  Muslim Fortresses in the Levant If the fortress underwent any changes during the Ayyubid period no architectural evidence remains. This is most peculiar, considering that the Ayyubids held Karak for seventy-five years in which the region was often affected by military and political strife. Some evidence of Ayyubid architecture Trans-Jordan was extensively fortified during the Ayyubid period. In addition to existing Crusader fortresses that were taken and reestablished, new fortresses were built at ʿAjlūn (1188–92), al-Ṣalt (1220) and al-Azraq (1236–7).191 As noted above, evidence of Ayyubid construction or even repairs within the fortress grounds is elusive and an archaeological excavation may well be required in order to trace and find conclusive and tangible remains. Thus the only knowledge we have comes from the historical sources and a foundation inscription that may relate to al-ʿĀdil’s work at the fortrss. The most important details, however, are missing. We are not told what building he built and since the inscription was not found in situ we have no means of knowing to which building it belonged. It does, however, state the name of the supervisor of the work, Ṣārim al-Dīn Abū Manṣūr Barghash, who was employed by al-ʿĀdil on a previous construction project at the stronghold of Qalʿat Ṣadr (Sinai) in 578/1183.192 Al-ʿĀdil’s foundation inscription.193

(1) In the name of Allāh, from what has ordered its construction, the noblest lord (2) Al-Malik al-ʿĀdil, the victorious, the helper of Islam, the succor of mankind, Sayf (3) al-Dunyā wal-Dīn, sultan of Islam and the Muslims, suppressor of the infidels and the idolaters, (4) Abū Bakr b. Ayyūb, friend of the Commander of the Faithful. [This construction was done] under the supervision of the noblest amir (5) Ṣārim al-Dīn Abū Manṣūr Barghash al-ʿĀdilī in Jumādā I of the year (?) four (?) … The only surviving figure representing the year of construction is the four. Since al-ʿĀdil (r. 596/1200–615/1218) was already sultan when the inscription was written,194 the year referred to would be 604 or 614, when Karak was governed by his son al-Muʿaẓẓam ʿĪsā. Ibn Shaddād describes a wave of large-scale construction work during the reign of al-Muʿaẓẓam. Though the fortress had new buildings added to it, greater emphasis was laid on the development of the town:

Mamluk power on display  167 5

… and he gave it [the fortress] to his son al-Muʿaẓẓam Sharaf al-Dīn ʿĪsā with all that was in the treasury and its supplies, and he enriched it with buildings and strengthened it. And he brought to it master craftsmen to create a town so that he would need no other. And he planted trees in its villages and ran [water] to them from the springs. Ibn Shaddād states that the work was done in the citadel. He does not mention towers or gates but simply says buildings in the plural. Al-Muʿaẓẓam gathered master craftsmen to carry out the construction work at Karak, as he did when he decided to build the fortress at Mount Tabor. Several sources mention the treasury that al-ʿĀdil kept at Karak. Ibn Wāṣil claims it contained a substantial sum though he does not give an exact figure.196 Sibṭ Ibn al-Jawzī, describing affairs at Karak in 609/1212, says that the fortress and all that was in it, i.e. horses, supplies and money amounted to a million dinars.197 Whatever the treasury contained, according to Ibn Shaddād it was left to al-Muʿaẓẓam. The construction work, whether in the town or the citadel, did not suffer from a shortage of funds. Though written evidence of Ayyubid construction exists, and the fortress was reinforced or repaired during the Ayyubid period, their work is either waiting to be unearthed or was simply swallowed by the large scale Mamluk building phase, leaving no trace at all or no remains above ground. Mamluk construction work When Baybars took Karak it was enlarged by adding an area equivalent to almost half the existing Crusader fortress. A large complex of vaulted galleries was built along the west and a massive keep was built in the south. Although Ibn ʿAbd al-Ẓāhir does not give much information regarding the buildings constructed by Baybars, he opens the chapter concerned with the conquest of Karak with the following comment: 8

And the sultan had taken along with him from Egypt a group of masons, builders, carpenters and craftsmen in order to build al-Ṭūr (Mount Tabor) … and they set out for ʿAyn Jālūt. The sultan made it known that all this was done in order to build a mosque for the sake of al-Karak. It seems a surprise attack against Karak was planned; the sultan was no doubt trying to conceal his true intentions until the army arived at the fortress. The company of skilled workers was intended first and foremost to prepare for the siege and to build manjaniqāt (siege engines). The builders (banā’in) and craftsmen (ṣanāʿ) were

168  Muslim Fortresses in the Levant no doubt to be employed at a later stage, to repair the damage caused during the siege and build up the fortress as the sultan saw fit. In actual fact, the fortress was handed over to Baybars without the necessity of a siege. Karak’s administrative hierarchy seems to have been highly regarded by Baybars. The sultan’s only new appointment was that of the governor ʿAlā al-Dīn Aydakīn al-Shihābī, chosen from amongst his personal Mamluks.199 The western galleries The two levels of barrel-vaulted galleries ran along 200m (Figure 4.26). They are entered via an interesting portal which can be described as a mock main gate. It is set in the center of the western wall (Figure 4.27). The actual entrance is at the bottom of a deep recess with a blind arch. One of the earliest gates of this type is found at Qaṣr Kharana (c. ad 710), an Umayyad palace on the Jordanian steppe.200 It is a form rare in twelfth- and thirteenthcentury military architecture probably because it has few defensive qualities. At Karak it is exposed to enemy fire and its only protection from siege engines is provided by the slope that runs below it. Later civil and religious Mamluk architecture used this form quite frequently.201 Once you entered through the portal you went up a steep staircase; at the top you stood under a domed chamber lit by an oculus. From here it was possible to enter the upper gallery or climb up to the court yard. In addition, another entrance was constructed in the northwest corner of the gallery. A rectangular tower was built at the southern end of the gallery where the slope is moderate and accessible (Figure 4.27). It appears to have been added almost as a second thought. The short band inscription is no longer legible and has been partially eroded by the weather. The few letters that can still be seen are similar in size and design to the letters in Baybars’ inscription on the southern keep. The addition of galleries reinforced the entire western side, though they were most probably built to provide storage room and/or accommodation for garrison soldiers rather than to enhance the defenses of the fortress, already protected by the steep slope. The government granaries may well have been located in the large vaults, where they were kept in

Figure 4.26 Karak, the Mamluk western galleries with their barrel-shaped vaults

Mamluk power on display  169

Figure 4.27 Karak, the western entrance into the Mamluk galleries and the oculus above the inner staircase

sacks and fastened with a seal.202 Before leaving Karak, Baybars himself inspected the granaries (ṣawmaʿa, ) to see all was in order.203 Once he left the granaries became the chief responsibility of the fortress governor. The southern keep The outer defenses underwent one further change during Baybars’ reign. A large tower was built in the south. Its size and height, the unusual thickness of its external walls and the existence of large cisterns at the base endow it with the qualities attributed

170  Muslim Fortresses in the Levant

Figure 4.28 Karak, the Mamluk keep. Note the remains of the Frankish towers destroyed during the keep’s construction.

to keeps; though unlike al-Ṣubayba and Safad at Karak it was incorporated in the external curtain wall. Ibn Shaddād adds a few more details; in his al-Aʿlāq al-khaṭīra he says that Baybars built up (shayyada) the fortress and made it stronger. The moat was dug out, probably meaning cleaned rather than made deeper. He then writes that he, i.e the Sultan, encircled the fortress, using the verb aḥāṭa ( ) which literally means to surround, encompass or enclose.204 The version given by Ibn Shaddād in the Ta’rīkh al-malik al-ẓāhir is quite different. 5

[He i.e Baybars] renewed in the fortress two small towers. [First he] destroyed them and then made them larger and higher. Then he dug the moat and made it more sinuous. This description fits the archaeological evidence. The Mamluk keep was built on the basis of two Frankish towers that guarded the south. The remains of the Frankish towers with their black stones can still be seen below the mamluk keep (Figure 4.28). The Mamluk keep has an unusual shape due to the limited space available at the southern edge of the spur. It is a long and narrow rectangle with cut corners (length 37m, width 16m). The most interesting feature of this structure is its outer walls, measuring 6m, almost two and a half times the thickness of the original Frankish curtain walls. This keep has three floors with an additional open surface at the top. The arrow slits are placed alternately. The upper floors have a denser line of arrow slits than the

Mamluk power on display  171

Figure 4.29 Karak, Baybars’ panther adorning the inscription on the southern keep

lower. The tower is preserved to its full height and a line of merlons pierced with arrow slits can still be seen crowning the top. Cisterns were built at the base, ensuring the water supply. A band inscription ran along the top of the tower. The inscription on each side begins and ends with the sultan’s royal emblem – the panther (Figure 4.29). The inscription is partly eroded and certain sections can no longer be made out. The location at such a height is rather surprising; it could hardly have served for propaganda purposes since passers-by would scarcely have seen it, let alone paused to read it. The inscription mentions Baybars and two of his nisbas, al-Ṣālīḥ and al-Ẓāhir. The date at the end of the inscription is 673/1274–5. The western galleries and the keep are homogeneous in style, construction methods, the type and size of stone used (local soft limestone) and the style of masonry. It seems that the local masonry tradition was followed and although the quality of the Mamluk masonry at Karak is better than that of the Franks, compared to the Mamluk masonry at Safad and al-Ṣubayba, Karak is by far inferior (Figure 4.30).

Figure 4.30 Karak, Mamluk and Frankish masonry

172  Muslim Fortresses in the Levant

Figure 4.31 Karak, the western wall of the fourteenth-century Mamluk palace

One of the most important roles of the fortress was to provide a place of refuge for nobility and royalty. Young heirs whose existence in Cairo was threatened or threatening either fled to Karak or were banished to it. Al-Nāṣir Muḥammad found refuge at the fortress twice (1298–9, 1308–10) before he acceded to the throne and established himself as sultan. According to written sources and the archaeological data the modest residential quarters were built by al-Nāṣir Muḥammad in the early fourteenth century. A subterranean complex (25 × 22m) identified by Brown as a palace was uncovered in the southern part of the fortress. It has an elongated plan, dominated by a central open reception hall. There is an īwān on either side of the central hall and two large rooms branch of it. A small mosque was built to the east (Figure 4.31). In the interior one can still see the remains of fine masonry and plaster.206 This plan had already been well established by the end of the Ayyubid and early Mamluk periods.207 It is difficult to compare the Frankish construction to that of Baybars. Military architecture had advanced considerably since the days of Paganus who first built the fortress in 1142, and even the later Frankish phase dated to the 1160s does not match the Mamluk work at Karak. True to his principles, Baybars left the Frankish curtain walls along the east as they were. Although the Frankish work did not match the standards of thirteenth-century military architecture, other than the two towers that were destroyed, most of the Frankish fortress was left intact and maintained. Before leaving, the sultan appointed a new nā’ib; the garrison was strengthened, salaries paid, storerooms stocked with arms, part of the sultan’s treasury brought to the fortress,208 and the granaries inspected. Assuming that work began in 1263 after the fortress was taken, it took a decade to complete this building project. It seems there was no immediate threat and no one was in a great hurry to complete Karak’s fortifications.

Mamluk power on display  173

Fortresses along the Euphrates border The main routes leading from Mesopotamia to the ports on the Mediterranean passed through the middle course of the upper Euphrates. The reason for preferring the road leading across the upper Euphrates is the considerably short distance from this section of the river to the coast (Map 4.1). There are seven fords along the stretch of the river starting from Carchemish to the modern Ataturk dam (approximately 165 km). Some fords were still in use in the mid twentieth century.209 Most of the river crossings were well established by the Hellenistic and Roman periods. The Roman organization of the Euphrates frontier partly correlates with that of the Mamluk–Īlkhānid frontier. Seleucus I founded the city of Zeugma (c. 300 bc) on a ford 15km north of the medieval fortress of al-Bīra, to safeguard the crossing.210 During the early Roman period this became the main crossing point between Rome and Parthia, and in the middle of the first century ad the tenth legion was based in the town.211 Parker suggests that “A large number of legions would have been required to hold even the vulnerable points on this line, and, doubtless from financial reasons as well as military and political inexpediency, Augustus decided against such a course and

Map 4.1  The middle course of the upper Euphrates

174  Muslim Fortresses in the Levant concentrated his army in Syria while making use of client kings … to act as buffer states between Rome and Parthia.”212 This strategy did not endure and by the second century ad the Roman frontiers were reorganized. The front lines had fortresses protected by cohorts of auxiliaries. The legions were stationed in large camps directly connected by roads to the frontier fortresses.213 The Mamluks’ strategy closely resembles that of the Romans during the second century ad. It seems that since from the outset the sultans had a limited source of manpower they reached the same conclusion. Al-Bīra became an important crossing point only when Zeugma declined, possibly in the early Middle Ages. The advantage of having a crossing at al-Bīra is directly related to the width of the river (180–200m). The wider the river the slower the current, which makes it easier both to cross and to maintain a pontoon bridge. An important point concerns the seasonable floods. Although building a bridge might have been feasible, a solid stone or brick bridges would not have survived the force of the flood. Thus from as early as the Seleucid period until the early twentieth century ferries and pontoon bridges, of various descriptions, were used for crossing.214 The river crossing to which the site owed its importance is mentioned in a number of Muslim sources. Ibn al-Athīr refers to a bridge used by Ṣalāḥ al-Dīn in 1179.215 Bar Hebraeus notes the bridge tied across the Euphrates used by Hülegü’s forces in 1259. In the mid fifteenth century Ibn Shāḥīn al-Ẓāhirī reports that a string of boats tied to one another served as a bridge (jisr mawḍūʿ marākib).216 The first fortress along this section of the river that was taken and rebuilt by the Mamluks was al-Bīra. The fortress is situated in the modern Turkish town of Birecik on the east bank of the Euphrates. The name Birecik derives from the Arabic bīra with a Turkish diminutive, giving the sense of “little fortress.”217 This hardly does justice to the site; although it did not occupy much land, it towered high above its surroundings and was a forbidding and impregnable stronghold. Between 1099 and 1150, al-Bīra officially belonged to the Crusader county of Edessa. Although Edessa fell to Zengi in 1144, he did not manage to conquer al-Bīra. According to Ibn al-Athīr in 577/1178 it was under Mosul’s jurisdiction. The governor of the fortress surrendered to Ṣalāḥ al-Dīn in 578/1179. This was part of a larger campaign waged by Ṣalāḥ al-Dīn during that year in the region of the Euphrates.218 Most of the fortress was probably built by the Ayyubids during the 82 years prior to the Mongol conquest in 1260. However, of the six inscriptions found and recorded by van Berchem, only three are from the citadel and none commemorate the Ayyubid rulers. The earliest inscriptions carries the name of Barka Khan, Baybars’ son and heir.219 On the eve of the Mongol invasion al-Bīra was held by the Ayyubid ruler of Syria, al-Nāṣir Yūsuf.220 Hülegü besieged al-Bīra in Ḍhū al-Hijja 657/December 1259. The garrison fought as best as it could but it was vastly outnumbered by the Mongol army. The fortress fell within two weeks.221 Al-Bīra was re-captured by Aqqush al-Barlī (a Mamluk of al-Nāṣir Yūsuf ’s father) in the winter of 1261. Hülegü sent a message to al-Barlī asking him to submit and offering him al-Bīra as an iqṭaʿ, al-Barlī refused and decided to submit to Baybars. Al-Bīra came under the Sultan’s rule in 660/1262.222 Although al-Bīra is often mentioned in siege accounts, we have few detailed descriptions of the Mamluk construction work or of the town. Ibn Shaddād, Ibn al-Furāt

Mamluk power on display  175

Figure 4.32 Al-Bīra, the fortress in the early twentieth century (left) and 2004 (right)

and Ṣafadī all give brief accounts. Later reports add some important information. The last Mamluk phase of construction work took place under Qaytbāy (r.872/1468– 901/1496), who inspected the fortresses along the Euphrates and personally ordered the repairs of al-Bīra’s fortifications. It seems, however, that Qaytbāy attached more importance to the town’s defenses than to the fortress itself, as attested by three inscriptions along the town’s walls that carry his name.223 Trying to reconstruct the site of al-Bīra The fortress of al-Bīra is in an extremely poor state of preservation, and one can only envy Gertrude Bell and Creswell who visited the site in the early twentieth century while it stood in full grandeur (Figure 4.32). Much of its stone has been taken and recycled by the local people during the last century. The northern part and the entire center of the fortress are almost completely in ruins. An outline of the curtain wall along the western side can still be seen, as well as the tower at the southwest corner and the inner mosque. Along the eastern side a tower still stands to the height of three floors, the façade almost complete, though the inner chambers are partly ruined. The passage carved through the rock along the western side is reasonably well preserved, and even though the upper section has collapsed it is still impressive. In addition to the poor state of conservation our knowledge of the site is hampered by the lack of contemporary Mamluk or Ayyubid texts. A further problem is the lack of a detailed plan.224 This is surprising, since several of the fortresses along the Euphrates have been surveyed and plans have been drawn or sketched by travelers and/or research teams. The reconstruction of the site, therefore, relies heavily on photographs taken by Creswell, a sketchy plan published in 1996, a study of the site I had made in the summer of 2004, the six Mamluk inscriptions, and the few sources that give an account of the sieges. The only information attached to Creswell’s photographs, noted that they were taken in the early years of the twentieth century. There was no other commentary apart from vague dates and a one word description of the structure that appear in

176  Muslim Fortresses in the Levant

Figure 4.33 Al-Bīra, aerial photo (1968)

each photograph. If Creswell had taken notes while he was at the site none were published by him. Since the photographs covered the entire external circumference, it was possible to reconstruct a large part of the fortress.225 Nevertheless, even though the photographs are of excellent quality some details are missing. Al-Bīra as it was seen by Creswell The entire medieval town of al-Bīra, which lay east of the citadel, was walled, hence in order to reach the fortress the town fortifications had to be overcome, which was no small matter (Figures 4.33, 4.34). Creswell dated the fortifications roughly to the eleventh and fifteenth centuries.226 The damage left by the Mongols was repaired soon after al-Bīra was taken by the Mamluks in 1262. Ibn al-Furāt says: “He [Baybars] ordered the frontier town of al-Bīra to be built up.”227 Ibn Shaddad writes: “a number of towers were built in the fortress, a gate was opened [fataḥ, meaning here to construct] facing the Euphrates, its moat was widened, and markets were built in the fortress and in its outskirts [rabaḍ].”228 Al-Ṣafadī says: “the Sultan constructed [ansha’a] the fortress, built [banā] its towers, widened its moat and renewed its mosque.”229 Enlarging the fortress would have been practically impossible due to the restrictions imposed by the topography. Thus the sultan mainly strengthened the existing fortress.

Mamluk power on display  177

Figure 4.34 Al-Bīra, the town walls and the octagonal tower in the southern corner

Cliffs adorned with curtain walls One of the most striking features of the site is its steep white cliffs. The short ridge is completely detached from the landscape around it. The natural slopes were made vertical by quarrying the soft limestone evenly all around (Figure 4.35). This gave the site much of its initial protection. Since the fortress walls and towers were built at the edge of the cliffs it was practically impossible to mine or sap the foundations. Although it had the protection of the forbidding cliffs, the fortress was also surrounded by a fosse that was filled in modern times and turned into a road.230 The fortress is only 50m above the level of the river and yet it commands the entire landscape in such a way that no matter where an enemy positioned himself he was bound to be at a disadvantage. The fortress measures approximately 300m in length and 60m in width. The layout follows the natural ridge and its shape resembles a kidney. The entire western side parallel to the river is devoid of towers. The height and steepness of the cliffs on this side provided sufficient protection. The remains of a wall and a gate near the river, suggest that the fortress grounds stretched all the way down to the bank. This lower section gave the garrison access to the river, a passage carved in the rock led down to the Euphrates. The first inscription dates to the reign of Baybars’ son al-Malik al-Saʿīd Barka Khan (676/1277–678/1279): it commemorates the rebuilding of the curtain wall:

(1) This blessed curtain wall was renewed in the reign of our lord the Sultan (?) al-Malik (2) al-Saʿīd the learned, the just, the fighter of holy wars Nāṣir al-Dunyā wa’l-Dīn, (3) bringer of justice to those oppressed by oppressors, the eradicator of deviators and rebels Muḥammad Barka Khān (4) son of our lord, Sultan al-Malik al-Ẓāhir, partner of the Commander of the Faithful, may Allah

178  Muslim Fortresses in the Levant perpetuate (5) his rule, under the command of the appointed governorship amir Jamāl al-Dīn, royal governor at al-Bīra. Since the inscription was found at the citadel we can be certain that the rebuilding of the “blessed curtain wall” refers to the fortress and not to the town defenses. No date is given though several clues occur in the inscription as to when and why this curtain wall was rebuilt. Barka Khan’s reign lasted only two years, before his father-in-law Qalāwūn seized power. Renovations could have taken place between 1277 and 1279, though it would be difficult to explain such a late date for the renovations of the wall. A more likely supposition is that the reconstruction work was carried out soon after the Īlkhānid siege of 1275. The appearance of Barka Khan’s name prior to his reign is possible seeing as Baybars appointed him co-sultan as early as 1268 in order to legitimize his succession.232 Another possibility is that the repairs were initiated by Baybars and only completed by Barka Khan, who took credit for the whole work and had his name entered on the inscription. According to Rabbat, Barka Khan acted in the same manner in Cairo: “He [Barka Khan] had his name attached to a number of structures in and around the citadel, but these had all been built by his father.”233 As for the date, the Īlkhānid siege of 673/1275 appears to have caused more damage than the Mamluk sources care to reveal. Although the siege failed, the Īlkhānid army that assembled in front of al-Bīra’s walls numbered 30,000 men and 70 siege machines. The damage may well have been considerable even though the siege lasted only nine days.234 It seems likely that the repairs done by Barka Khan or/ and Baybars were due to damage caused during this siege. The fortress of al-Bīra was constructed in a manner similar to that seen in the Ayyubid phase at al-Ṣubayba, ʿAjlūn, Qalʿat Ṣadr and Mount Tabor: the towers were better built than the curtain walls. When the Mamluks rebuilt the curtain walls they did not change or improve the existing Ayyubid structure. The width of the curtain wall along the western side is approximately 2m while the walls of the southwest

Figure 4.35 Al-Bīra, cliffs adorned with with curtain walls

Mamluk power on display  179 corner tower (F) run between 2.5 and 3.3m. This however must be viewed with some reservations, since there are very few sections where the curtain wall is preserved and can be measured. The plan followed a similar idea to that at ʿAjlūn, where there are only short stretches of curtain wall between the towers. The towers dominated the scene, while the curtain wall played a negligible part. Mighty towers I could identify only two towers when I surveyed the site (towers F and E), though eight square towers can still be seen in the early twentieth-century photographs. What with the high vertical cliffs and the decision to make do with curtain walls of no great thickness, much of the defense depended on the towers. Both Ibn Shaddād and al-Ṣafadī agree that the towers were built by Baybars.235 The masonry is of high quality. Pillars were inserted in the southeast tower in order to strengthen the wall structure (Figure 4.36); the same technique can be seen at the fortifications of Caesarea and the fortress of Shayzar.236 Most of the towers had three floors, each pierced with arrow slits. In addition, archers could be positioned along the slanting base of tower F, where two levels of arrow slits can still be seen on the southern side (Figure 4.37). A chemin de ronde allowed archers to move easily from one position to the other.237 Like Mamluk arrow slits elsewhere during this period, those of al-Bīra are spacious and have a large chamber that could accommodate a pair of archers. The only evidance of a machicoulis is in tower A. Judging from the architectural evidence, and the inscription by al-Nāṣir Muḥammad found in tower F (see below) it seems he enlarged and renovated the entire tower in 1301. The nā’ib’s quarters (?) The large corner tower (F) contains the remains of a rather elaborate building resembling the plan of the governor’s residence found in several fortresses.238 The center of the building is dominated by a square hall with a large room on either side.239 A deep, elongated niche in the southern wall may have served as the miḥrāb. The central hall was covered by a dome decorated with a short inscription. The masonry is of the highest quality, consisting of smooth ashlars with simple decorative patterns created by using a fine mason’s comb. Since the dome has collapsed a substantial part of the building is covered in rubble (Figure 4.38). The inscription found in this building (no longer in situ) and published by van Berchem gives a precise date, stating that it was renewed in 700/1301 by al-Nāṣir Muḥammad during his second reign (1298–1308). Since the sultan was 16 years old, the orders were issued by the ruling junta in al-Nāṣir’s name. This is clearly shown by the mention of the sultan’s helpers in the inscription. Both Creswell and Sinclair identified this building as a mosque.240 However, residences of similar plans and dimensions were built at Qalʿat Ṣahyūn, Qalʿat Najm, Shawbak and Karak; the latter has been ascribed to al-Nāṣir Muḥammad and dates to the early fourteenth century.241

180  Muslim Fortresses in the Levant

Figure 4.36 Al-Bīra, pillars inserted in the lower half of the southeast tower

Figure 4.37 Al-Bīra, arrow slits along the southern glacis

Mamluk power on display  181

Figure 4.38 Al-Bīra, entrance and hall within the large tower on the southwest (F). Note (1) the entrance; (2) the inscription on the ceiling, covered in soot; and (3) the main hall

182  Muslim Fortresses in the Levant The inscription commemorating al-Nāṣir Muḥammad’s work

(1) In the name of Allāh, the Merciful, the Compassionate. This building was renewed in the reign of our lord the sultan al-Malik (2) al-Nāṣir, Abī al-Maʿālá Muḥammad son of the sultan al-Malik al-Manṣūr may God (3) make great his helpers [and this was at] the end of the year seven hundred [1301]. Gates and rock-carved passages The fortress has an unusually large number of entrances: a total of six. While some are built others are cut into the bedrock (Figure 4.39). Some are perched in the steepest section of the cliffs, and only a narrow path cut into the cliff leads to them. The passage in the soft rock along the southwestern side can still be visited today. It is wide (2.55m) and has steps carved into the stone (Figure 4.40). It descends at a moderate angle towards what I presume was once an exit. The upper part has collapsed, and when I was there a stray dog commanded the lower section, growling fiercely at anyone who tried to descend. If you arrived by boat you could anchor and

Figure 4.39 Al-Bīra, the large built entrance in the east, possibly the main gate

Mamluk power on display  183

Figure 4.40 Al-Bīra, rock carved passage in the west

climb up along this lower passage, which was probably meant to be used also as an emergency exit if all was lost (Figure 4.35, 1). The three built gates are located in the southwest, in the east and up on the northern edge. Since no plan was drawn and the photographs do not cover such details, one cannot estimate their size or fully understand their structure. Three-story vaulted halls Al-Bīra’s isolated location and lofty position made it impossible to smuggle in supplies during times of siege. A relatively large part of the fortress on the north-east side was built up with vaulted halls. The photographs show sections with two and three floors (Figure 4.41); this large area probably provided storage space for weapons and food supplies. The third inscription was found at the galleries. It is fragmented and hardly allows one to draw any conclusions, but should be mentioned. Like the nā’ib’s living quarters, it may well date to the second reign of al-Nāṣir Muḥammad, though all we have is the name al-Nāṣir without a date. From the surviving wording it is clear that the galleries were added to the existing fortress, and that they were not merely renewed but rather built anew at the sultan’s order.

184  Muslim Fortresses in the Levant

Figure 4.41 Al-Bīra, vaulted halls along the east and northeast

The fragmented inscription found in the galleries

(1) Our lord, the Sultan the most splendid king … al-Nāṣir ordered the building … The second reign of al-Nāṣir Muḥammad correlates with one of the most trying periods that the Mamluks knew. Ghazan led his armies for three consecutive years and was determined to cross the Euphrates and annex Syria to the Īlkhānid state. Although al-Bīra did not come under direct attack during this period and all the crossings were carried out much further south, the tension along the entire Euphrates frontier may have brought about this wave of building. The water supply Concerning the question of the fortress’s water supply, it is impossible to draw information from the survey of the site or the historical sources. Since al-Bīra is located on the bank of the river a well must have been built within the fortress grounds that reached the water level (this can be seen at Qalʿat al-Rūm); there may have also been cisterns to gather rain water. These were perhaps filled with rubble or buried during one of the several earthquakes that struck the region. There is no indication of their number or whereabouts or indeed of their existence. Much of the evidence necessary to reconstruct the fortress is still missing, but by reconstructing the shell one can get a fairly good picture of what al-Bīra looked like. The construction and layout were well planned, the towers were closely positioned, leaving only short stretches of curtain wall between them. The western side of the fortress was protected by the river and the steep cliffs. It is not surprising that the Mongol forces had difficulties in besieging the fortress. Only after breaching the town fortifications could the fortress be dealt with. Although

Mamluk power on display  185 the town was of no great size, of the four Mongol attempts to conquer the site only the first, led by Hülegü, was successful. Later Īlkhānid armies failed miserably.243 Lack of time was the main cause of failure. Breaching the city walls and moving on to besiege the fortress was an almost impossible undertaking in the time limits imposed by the defenders. Returning to the thirteenth century landscape round Qal ʿat al-Rūm Qalʿat al-Rūm is located on the west bank of the Euphrates, approximately 40km north of al-Bīra, near the modern Turkish village of Halfeti.244 It is built on a spur that is “trapped” between two rivers; the Euphrates flows on the east, and the Merzuman curves round the northern and western slopes (Figure 4.42). Like al-Bīra, the fortress was built on a ford: a diagonal gravel bank is formed at the point where the tributary meets the Euphrates. This shallow section provided a convenient crossing point.245 In 1260 it was one of the five fords chosen by the Mongols to cross the Euphrates on their first attempt to conquer Syria.246 The Birecik dam, completed in 1999, changed the landscape considerably.247 When I first visited the site in the summer of 2004, in order to reach Qalʿat al-Rūm I had to go by boat from the nearby village of Halfeti; it is a forty-minute

Figure 4.42 Qalʿat al-Rūm

186  Muslim Fortresses in the Levant

Figure 4.43 Qalʿat al-Rūm before (left) and after the flooding of the valley (right)

ride through magnificent and dramatic landscape with steep rocky cliffs towering on both sides. Anyone visiting the site today may be somewhat misled by the scene and setting. Because of the reservoir the Euphrates practically reaches the foot of the fortress. The water level during the thirteenth century was much lower and a small valley stretched below the spur all the way down to the river bank (Figure 4.43). The site’s history is rather brief. A survey conducted as part of the Zeugma Project did not find archaeological remains dated earlier than ad 1000.248 The fortress was captured in 1116 by Baldwin of Marʿash and incorporated in the county of Edessa. In 1148 Joscelyn de Courtenay gave Qalʿat al-Rūm to the Armenians, as he found it difficult to hold this region against Türkmen raids. During that year the Catholicos (the Armenian patriarch) was moved to the site. It appears that the initiator of this move was Beatrice, the niece of Joscelyn de Courtenay, who invited Patriarch Grigor II to the fortified town. Qalʿat al-Rūm was officially bought from Joscelyn by the Catholicos in 1150.249 The town was largely populated by Armenians who immigrated to the area during the eleventh century. Sinclair suggests the town was located north and west of the fortress and sections of it were partly enclosed by walls. Although much of the region later became part of the Ayyubid sultanate, Qalʿat al-Rūm remained in the hands of the Armenians.250 Shortly after the Mongol invasion, Hülegü granted King Hetʿum I (r. 1226–69) a large area to the north and the west of Qalʿat al-Rūm, and for a short period the fortress was geographically connected to the Armenian kingdom in Cilicia.251 Following the first treaty signed between Baybars and King Hetʿum (1268) most of the main towns and fortresses in the regions bordering Qalʿat al-Rūm to the north and west were handed over to the Mamluks. Although the territory went back and forth between Armenians and Mamluks during the next two decades, Qalʿat al-Rūm remained in Armenian hands until 1292. The Mamluks’ first attempt to take the fortress was in the spring of 678/1279. Barka Khan sent Qalāwūn (the future sultan) together with Badr al-Dīn Baysarī to Cilicia, his main purpose being to get rid of both these powerful amirs, who he thought threatened his rule. While Qalāwūn headed towards Sīs, Baysarī besieged Qalʿat al-Rūm with the help of a Syrian army commanded by Ḥusām al-Dīn from

Mamluk power on display  187 ʿAyn-Tāb.252 Bar Hebraeus gives a detailed account of the negotiations with the Armenian Catholicos, who decided not to capitulate, though the terms offered were most generous. The sultan hath commanded that thou shalt surrender the fortress in peace, and that thou shalt take thy monks and go and dwell in Jerusalem. And he will give thee villages sufficient to maintain thee. And if thou dost not wish to act thus, but dost wish to go to Cilicia, in this case likewise we will transport thee on our mules and horses with all honour. But if thou resists, and will not surrender [it], God will require the blood of all these Christians from thee.253 The answer of the prelate was short and decisive: I will fight until I die. I cannot be faithful both to God and to the king.254 The Mamluk force was of considerable size – 1,000 mounted men and 4,000 infantry. No siege machines are mentioned and the only siege equipment we hear of are ladders that were used to scale the walls of the town.255 Bar Hebraeus mentions a new wall built by the Armenians; it was meant to guard the city but was quickly overcome by the Mamluk force. Here, however, their assault ended, since they were not able to breach the fortress walls. The siege failed and the entire Mamluk force left after setting fire to the city and raiding the nearby surroundings. The site was evidently well fortified. The strength of the site is attested by the failure of this first siege and the length of the second Mamluk siege in 691/1291–2. In addition to the effort required, the size of the force in both accounts suggests that this may have been one of the most difficult sieges undertaken by the Mamluks in their first three decades. The Armenians did not fall for trickery nor did they favor negotiations. Al-Ashraf Khalīl mobilized 20 siege machines against the defenders; the garrison eventually capitulated after 33 days.256 To celebrate his triumph the sultan insisted on changing the name to Qalʿat al-Muslimīn; the new name, however, never took root and the site continues to appear in later sources as Qalʿat al-Rūm. Once the fortress had fallen the sultan carried out what by then had become almost a routine. He nominated his amir Sanjar al-Shujāʿī to supervise the reconstruction and work began. Al-Shujāʿī who was at the time the governor of Damascus, participated in several building projects conducted during the reign of Qalāwūn.257 According to Maqrīzī the restoration needed was mainly due to damage caused by the Mamluk siege units employed by al-Ashraf Khalīl.258 The victory was further commemorated in an inscription on the Aleppo citadel gate. It sums up the military achievements of al-Ashraf Khalīl: his conquest of Acre and the taking of Qalʿat al-Rūm. The sites are not mentioned by name and the inscription simply refers to the defeated armies of the Franks and Armenians. The Mongols also figure in this inscription though there was no battle or siege in which their armies were directly involved during al-Ashraf Khalīl’s reign.

188  Muslim Fortresses in the Levant The main section of the inscription on the Aleppo citadel gate259

(4) … lord of kings and rulers, Sultan of the (5) army of the monotheists the helper of true faith and proofs, re-newer of justness in the world. [Destroyer] of deviators and the rebels, slayer of infidels and unbelievers, victor (6) over tyrants and renegades, suppressor of the worshipers of the crosses, Alexander of [his] time, conqueror of the cities, defeater of the armies of the Franks, the Armenians and the Mongols, demolisher of Acre and the lands of the coast, re-newer of the noble ʿAbbasid dynasty, helper of the community of Muḥammed Khalīl b. the Sultan al-Malik al-Manṣūr Qalāwūn, may God exalt his supporters, and this was in the year 691/1291–2. Although the siege is described in several sources, none go beyond the military events and no further details are given concerning the layout or general plan of the fortress One of the most detailed modern accounts was written by T. E. Lawrence who visited Qalʿat al-Rūm in July 1911. His letters and diary, later published under the title Oriental Assembly, describe parts of the fortress that have since collapsed. The diary contains his own drawings and a number of photographs.260 A fortress devoid of towers The fortress rises 76m above the Euphrates. This in itself provided ample protection. Qalʿat al-Rūm is rectangular in shape, measuring 230m from north to south and 120m from east to west. The walls follow the contour of the spur and the slopes were hewn and made vertical, as at al-Bīra. This made scaling almost impossible, although the Mamluks eventually were left with no alternative but to climb the walls. On the south, a 30m wide dry moat was dug out of the bed rock (Figure 4.44). The high and forbidding setting makes it difficult to understand where al-Ashraf Khalīl placed his 20 siege machines, and where his sappers dug or hacked a tunnel.261 The towers that are one of the most common features of thirteenth-century fortifications and a Mamluk trademark are almost absent at Qalʿat al-Rūm. The only towers standing today are the gate towers on the west and a single tower at the southwest. The dominant features along the northern and eastern sides are the galleries. Since the western side was well protected by the steep cliff the curtain wall was of no great width. In most cases the Mamluks restored and added new buildings along the circumference or within a fortress as they saw fit. At Qalʿat al-Rūm they seem to have followed the original plan, not adding but merely strengthening and repairing the existing structures, putting their faith in the sheer height and steepness of the location.

Mamluk power on display  189

Figure 4.44 Qalʿat al-Rūm, southern moat

Figure 4.45 Qalʿat al-Rūm, the zigzag wall on the east

Before describing the galleries here, it is noteworthy that the curtain walls do not run in a continuous straight line, but rather short sections protrude zigzagging from the line. This reduces the dead ground below the fortress walls in much the same way as towers would (Figure 4.45). Archers’ galleries are built along the edge of the cliffs (Figure 4.46) they have two levels of cross-vaulted halls (3.5m in width and approximately 55m in length), built of smooth ashlars cut from the local soft limestone. The stone is neatly hewn and well dressed; no traces of mortar are visible. The galleries were originally built by the Armenians and were designed to accommodate a large number of archers. Mamluk repairs can be identified by the different

190  Muslim Fortresses in the Levant

Figure 4.46 Qalʿat al-Rūm, archers’ galleries and vaulted hall. Note the dense line of machicoulis

design of arrow slit, which is spacious, with a large chamber in front. The smaller type is similar to those found in Armenian fortresses, where the arrow slits have no chamber and the angle of fire is narrower (Figure 4.47).262 The Mamluks made several repairs, but where there was no damage they made do with the Armenian architecture. Returning to this peculiar plan where towers are scarce, it seems that this was partly compensated by a dense line of machicolations; thus archers were able to fire down in a vertical line and defend the foot of the cliffs and walls from anyone attempting to scale them. The use of machicolations in this manner is not very common. One of the few remaining examples can be seen at Crac des Chevaliers where the gateway in the east is protected from above by numerous machicolations.263 Otherwise they were often built directly above gates as in ʿAjlūn or Aleppo. Multiple gates Qalʿat al-Rūm had only one main entrance. The postern, on the east, was a steep and narrow staircase carved into the side of the cliff, and could only be used by individuals on foot (Figure 4.48). Once you arrived at the top you entered a gate-tower and were forced to take two sharp turns and maneuver via a narrow corridor before you entered the fortress grounds. When Lawrence visited the site he entered Qalʿat al-Rūm on foot, crossing the Merzuman on some stepping stones. In speaking of crossing this tributary Lawrence mentions the remains of a collapsed stone bridge.264 This brought him to the main entrance located on the west, which is hidden and better protected. There were altogether three bridges along this tributary; two are dated to the medieval period and one to the Roman (Figure 4.42). The main entrance is composed of five individual gates built along the upper part of the slope. Boase mentions an Armenian inscription flanked by a pair of Seleucid lions (?) above the first gate.265 Multiple gate systems are not common in Muslim military architecture and even among the Franks and Armenians they are seldom used. A similar gate structure can be seen at Lampron, an Armenian fortress in the northwest of Cilicia, though most of the gate structure

Mamluk power on display  191

Figure 4.47 Qalʿat al-Rūm, Armenian and Mamluk arrow slits

Figure 4.48 Qalʿat al-Rūm, the eastern gate

192  Muslim Fortresses in the Levant

Figure 4.49 Qalʿat al-Rūm, carved and built entrances along the multiple gate system

has been dated by Edwards to the Byzantine period.266 Crac des Chevaliers’ main entrance is constructed on the same principle, though on a much grander scale. Of the five gates that compose the main entrance at Qalʿat al-Rūm, some are built while others are cut out of the side of the cliff (Figure 4.49). One moved between the gates via a covered passage that was partly built and partly quarried. The gates are fairly wide large enough for carts and mounted men to enter. The second gate-house could accommodate a small guard of up to five men (judging from the number of arrow slits). Wooden doors were fitted in the rock archways, as is attested by the round sockets for the door hinges; this meant that each gate-house and passage could be bolted and act as an independent unit. It was usually the topography, in this case the steepness, of the site that dictated such a complex gate system. Its main advantages were the difficulties it presented to

Mamluk power on display  193 an advancing enemy, which would be halted several times before reaching the fortress grounds. Perhaps this was taken into account during the Mamluk attack of 1292, since the siege report does not mention that the gate was targeted by the army. The Mamluk soldiers eventually entered the fortress by climbing the walls up a long chain tied to the bed rock.267 Water sources Whereas at al-Bīra it was difficult to find the remains of wells or cisterns, at Qalʿat al-Rūm the whole fortress seems to have depended on a central well that reached the river. It was a large, almost bell-shaped structure with a spiral staircase cut along the wall leading down to the water. The water supply was not only plentiful but there was no danger that it would be sabotaged by a besieging army using any of the following methods: The Jihadists … are also permitted to cut water canals and destroy water supplies to prevent the ḥarbis from using them. Poison, blood, or any other material that may spoil the drinking water may be thrown into water supplies or canals in order to force the enemy to capitulate.268 In addition to this well a cistern was dug at the northern edge of the spur (Figure 4.49, 10).269 Water was obviously not a great worry. Buildings along the upper spur of the fortress The southern fortified complex is built at the highest point within the fortress grounds. One can still see the remains of the Armenian patriarchal church that the Mamluks probably converted into a mosque (Figure 4.49, 4a). The carved crosses were removed and two animal figures, possibly panthers, were carved below them. A wall and two towers were built on the western side where the slope is moderate. A second Armenian church lay north of the spur; Sinclair thinks it may have served as a palace during the Mamluk period, but Lawrence suggests it was a mosque.270 Governor’s quarters often included a small private mosque. It seems this model was followed also at Qalʿat al-Rūm.

Mamluk fortresses in Cilicia Acquiring the black fortress Moving west from the region of the Euphrates, 160km as the crow flies, is the fortress of Tall Ḥamdūn, approximately 12km west of the modern Turkish town of Osmaniye. Positioned on a low hill (65m above the surrounding area), it guards the northern passage across the Amanus range. From here one could travel south to the ports along the eastern coast of the present Bay of Iskenderun and west to Ayās (modern Yumurtalik) which was the most important Armenian port connecting the

194  Muslim Fortresses in the Levant Iranian and Central Asian trade routes with the eastern Mediterranean and Europe. The location at an important junction made this a desirable site. Throughout the twelfth century the fortress rotated between the Byzantines, the Franks and the Armenians. John II Comnenus (r. 1118–43), and his successor Manuel Comnenus (r. 1143–80), the Armenian baron Toros II, and the Frankish ruler of Antioch, Bohemond III (r. 1163–1201) played major roles in the history of the fortress during this period.271 The situation did not change much during the first four decades of Mamluk rule. The site was often held by the Armenians, besieged by the Mamluks, lost and then won through diplomatic negotiations. This cycle revolved several times. The Mamluks showed the same keen interest in holding the fortress as did the regional Frankish princes, Byzantine emperors and Armenian kings before them. The following account shows how determined the Mamluk sultan was. During al-Manṣūr Lachin’s reign in 697/1298, a Mamluk army was sent to raid Cilicia and capture Tall Ḥamdūn. It was made clear to the amirs in charge that they were not to return before they seized the fortress. They were further threatened that if they failed they would lose their iqṭāʿāt.272 Besides its economic importance due to its location at the edge of the trade routes, the fortress filled a fairly important place in the Mamluk defenses. The garrison was expected to keep a careful watch on Armenian movements. The Armenians, who were close allies of the Īlkhānids, often joined them on their campaigns into Syria. The possibility of a joint Armenian, European and Īlkhānid front did not escape the Mamluks, and thus their presence in Cilicia was of great importance to them. It is, however, clear from the rotation of the Cilician fortresses back and forth between the Mamluks and the Armenians that the Mamluk strongholds in Cilicia were not as well organized and did not receive the same military support as those along the Euphrates frontier. The archaeological remains to a certain extent display this frequent change of ownership. Sections of what can be seen today were constructed by the Mamluks; those parts are uniform in appearance and the architecture is similar to that found in other Mamluk sites. Both Sinclair and Edwards surveyed the site and recorded the Mamluk phase.273 The Mamluks first captured the fortress in 664–5/1266. The garrison was slain and the fortress set on fire.274 In 1280 the fortress was again in Armenian hands. In 692/1293 al-Ashraf Khalīl received the fortress in a treaty signed by the Armenian king Thoros III (r. 1193–8);275 within five years the fortress was retaken by the Armenians. In 697/1298 the Mamluks besieged Tall Ḥamdūn for almost a month before the Armenian garrison capitulated.276 Later in the same year the Armenians signed a treaty agreeing that the fortress should remain under Mamluk rule. Following the Mongol victory in 1299 the Armenians recovered Tall Ḥamdūn.277 In 703/1303–4 an Egyptian force together with armies from Hama and Aleppo besieged the fortress which fell after terms were negotiated with the Armenian garrison.278 According to Abū’l-Fidā’, the Mamluk forces had it “razed to the ground.” He concludes this short paragraph by saying that he did not take part in this assault since he was in the holy Hijāz. It sounds as if Abū’l-Fidā’ himself was not quite sure how reliable this last piece of information was.279

Mamluk power on display  195 It is difficult to determine when the Mamluks rebuilt Tall Ḥamdūn. On two occasions the fortress was captured and left in ruins (1266 and possibly in 1303–4). The determination displayed by the Mamluks in the last decade of the thirteenth century suggests they may have invested in strengthening the fortress during this period, though none of the sources explicitly say so. The site probably remained in Mamluk hands from 1337, when it was surrendered by Leon IV.280 It is possible that some constructions were added after 1375 when the entire area of Cilicia was annexed by the Mamluk sultanate. The village and fortress defenses The plan and building techniques at Tall Ḥamdūn do not resemble those found in the Armenian fortresses in Cilicia. The fortress had two main building phases, though it is difficult to determine the period of the first phase. Any one of the following is possible: Byzantine, Early Muslim or Frankish. The second phase is almost purely Mamluk.281 While Byzantine and Armenian features can still be seen, the only tangible evidence of an early Muslim presence is the name, “Ḥamdūn,” which suggests it was built by the Ḥamdanid dynasty (293/906–394/1004). It is said that in 173/786 Hārūn al-Rashīd built a fortress above an old Greek fort of black stone.282 Archaeological evidence from this early stage may well be recovered if the site is excavated. Among Mamluk fortresses captured and restored Tall Ḥamdūn rates as fairly modest in size. The core measures 60m in width, while the length is approximately 200m. The southern and eastern sides were protected by two lines of walls; the western side had larger towers and an impressive glacis (Figure 4.50). The little village or rabaḍ (suburb) that grew along the slope adjacent to Tall Ḥamdūn was surrounded by a narrow wall (1.8m) that was linked to the fortress. The towers along it are small and assume an unusual variety of shapes: octagonal, horseshoe, rectangular and semi-circular (Figure 4.51). There is no evidence of arrow slits or galleries; thus the protection provided was only nominal. According to Foss and Winfield this line of prow-shaped towers along the village wall dates to the Byzantine period (late eleventh century) (Figure 4.52).283 One could only enter the fortress by first walking through the village compound. Turning up through the narrow passage you come to the main gate, flanked by two semicircular towers built by the Mamluks to reinforce the existing structure (Figure 4.53). In contrast to Qalʿat al-Rūm which is almost devoid of towers, Tall Ḥamdūn has a dense line of towers along both the inner and the outer curtain walls. Judging by the shapes of the towers and the design of the arrow slits it seems the majority belong to an Armenian phase, probably of the late eleventh or early twelfth century. Horseshoe towers were not part of the Mamluk inventory, but appear to be a design favored in Armenian military architecture.284 Extensive Mamluk repairs were carried out along the archers’ galleries. The large specious chamber in front of the arrow slits serve as a Mamluk trademark. The stirrup at the base increased the angle of fire and allowed archers to aim at objects positioned directly below them (Figure 4.53). By the end of the Mamluk building phase there were thirteen towers along the inner circumference, which measured 380m.

196  Muslim Fortresses in the Levant

Figure 4.50 Tall Ḥamdūn, Mamluk tower and glacis along the western inner wall

Figure 4.51 Tall Ḥamdūn, view from the southwest

Mamluk power on display  197

Figure 4.52 Tall Ḥamdūn, prow-shaped tower along the lower curtain wall and site

plan. Note the Mamluk galleries (1) and the new Mamluk tower along the Armenian walls (n)

Figure 4.53 Tall Ḥamdūn, Mamluk and Armenian arrow slits

198  Muslim Fortresses in the Levant When looking at this semi-concentric section, only one new tower (N) was added by the Mamluks along the inner wall. Although modest in dimensions (5m in diameter), it is designed rather like the large tower at al-Ṣubayba. It has a pentagonal plan with an arrow slit in each wall. A white band adorns the outer wall as at the Mamluk tower at Marqab. The masonry In 1280, during the Mamluk raid on the Armenian kingdom, a number of scouts passed the fortress; when they returned to the camp they described it as “black and huge” (sawādan ʿaẓīman).285 Tall Ḥamdūn was constructed from the local basalt. The masonry is generally coarser than in any of the other sites documented in this study. The only sections of fine smooth ashlar are the frames of the arrow slits (Figure 4.53); the rest of the fortress consists of small coarse building blocks. The Mamluks either recycled building stones or employed local masons.286 In general it is difficult to define traditional Mamluk masonry styles. In many cases the masonry simply changes according to the type of stone and possibly according to regional traditions. Guarding the Syrian Gates Like Tall Ḥamdūn, Baghrās commanded an important pass in the south eastern Amanus known as the Belen pass or Syrian Gates.287 The fortress is situated on one of the lowest points (740m) along the southern range, close to the crossing. The mountain range averages 1820m in height, though its peak, the Miğir (modern Turkish), attains 2262m. Most of the traffic moving between Cilicia and the north Syrian plains down to Aleppo and Antioch would have crossed via this pass. The distances from the Euphrates, which acted as the frontier line to the territory of the Armenian kingdom, and the northern Syrian cities are in fact extremely short. Even by thirteenth-century standards, the fortress of Baghrās was at the threshold of Aleppo and Antioch. The distance to Aleppo is approximately 50km, and to Antioch it was only 26km. During most of the twelfth century Baghrās was officially within the boundaries of the principality of Antioch and together with the fortress of Darbssāk it defended Antioch from the north.288 During the 1130s it passed into the hands of the Templars and remained there almost continuously up until the Mamluks’ arrival in 1268.289 It did not, however, escape the turmoil that swept the region in the late twelfth century and its name appears in the list of fortresses taken by Ṣalāḥ al-Dīn in 584/1188. Ṣalāḥ al-Dīn ordered its reconstruction.290 But soon afterwards the Ayyubids destroyed much of the fortifications for fear the fortress would fall into the hands of the German emperor, Frederick Barbarossa (d. 1190), who was making his way east at the head of a new Crusade. The Crusaders never reached the site. The Armenians were quick to seize the opportunity, and rebuilt what the Ayyubids had destroyed.291 They were to hold on to the fortress for the next 25 years (1191–1216). According to Edwards, the Armenians did not attribute great importance to the castle.292 It is difficult to accept Edward’s conclusion. The fortress figures as a key site,

Mamluk power on display  199 of significant importance to all the political entities that had an interest in the region. Armenians, Franks and later the Mamluks fought over the site continuously; it was never left without a proprietor. It is more than likely that the Templar fortifications were adequate and thus neither the Armenians nor the Mamluks enlarged or added significantly to the existing structure. The Armenian king was eventually forced to return Baghrās and in 1216 he reluctantly handed it back to the Templars. The very fact that he had held it for so many years, resisting pressure from the papal court, shows how important the site was to the Armenians.293 This point is noteworthy since the Armenians’ perception of the site did not change in the following decades. The Mamluks first took possession of Baghrās in 666/1268. The Templars must have realized that they could not continue to hold on to the fortress once Antioch had fallen (1268), and simply handed it to the Mamluks without negotiating terms. Baybars immediately established a garrison and sent orders to arm and stock the place with supplies.294 Like many of the fortresses on the Armenian frontier it was lost and then regained. In 1281 it fell into the hands of the Īlkhānids without a battle. The Mamluk garrison fled and abandoned the site. The Armenians must have received it as a gift since the Īlkhānids very rarely established their own garrisons in strongholds they captured. This arrangement was, however, short-lived. Within four years, Baghrās was returned to the Mamluks following an agreement signed in 1285 between Qalāwūn and Leon II, king of Armenia. From this point onwards it remained in Mamluk hands, and in 1347 it became a stage in the Mamluk barīd on the way to the port of Ayās. Feeble towers and feeble walls There has been some debate as to who built the core of the fortress. It is a good example of a case where neither the style of the architecture nor the masonry can provide secure dating. Lawrence and Edwards date most of the construction to the Templar occupation.295 Müller-Wiener writes: “its compact design is reminiscent of many Armenian Castles.”296 It seems likely that the Templars who held the fortress from the 1130s to 1191 would have built much of it during those years prior to the Armenians’ arrival. The structure has more in common with Frankish fortifications than with Armenian strongholds, which often left much of the natural terrain and exposed bedrock in the center of their fortresses. Much of the fortress is still standing and well preserved. Measuring roughly 90m from north to south and 80m from west to east, it has a peculiarly compact design, as Müller-Wiener observed. The core consists of a large complex of rectangular vaulted halls built on several levels around a courtyard. The quality of the masonry is high, and the decorative designs around the doors and windows resemble Templar work.297 The northwestern side is guarded by a steep gorge and has almost no defensive features. The eastern side is laid out in an interesting fashion. There are two lines of defense along the slope, on two separate terraces (Figures 4.54, 4.55). This plan may be considered as a semi-concentric fortress, similar to that at Tall Ḥamdūn but with considerably fewer towers. Though the Templar work is the most dominant there is evidence

200  Muslim Fortresses in the Levant

Figure 4.54 Baghrās, core of the Templar fortress

Figure 4.55 Baghrās, lower and upper defences along the eastern side. Note the round tower in the front

Mamluk power on display  201

Figure 4.56 Baghrās, the Mamluk tower (C). Note the masonry of the tower wall, and the loose construction of the curtain wall (lower right)

of Armenian construction, identified by Edwards and Sinclair, along the southeast. According to Sinclair the outer galleries were probably built by the Armenians. Ibn-Shaddād, who describes the site, does not mention any reconstruction carried out by Baybars.298 Though the sources are mute, there is evidence of Mamluk building. The scale of construction is small and it may have gone unnoticed.299 The Mamluk approach to Baghrās is a good example of their economical attitude towards military architecture in the rural and frontier areas. Where it was necessary they spared no expense though where one could do without, funds were carefully used and costs were kept low. The tower in the east (C), encased in a new shell, presents some traits that can be identified as Mamluk work (Figure 4.56). The most striking feature is the quality of construction and the relatively well-cut stone. This is particularly striking when compared to the curtain walls. Sinclair suggested that the Mamluk work was executed by Armenian masons. Edwards tends to agree and dates the work on this tower to the reign of Qalāwūn.300 It is not of the highest Mamluk standard but in this remote

202  Muslim Fortresses in the Levant part of the Sultanate it looks distinctly Mamluk and when compared to the work at Tall Ḥamdūn it is most pleasing to the eye. The upper section of the tower has not survived. The curtain walls were left as they were and no attempt was made to improve them by widening or re-facing them. The only other evidence of Mamluk work is the conversion of the chapel, hidden in the center of the fortress, into a mosque.301

Mamluk fortresses: an overview Mamluk military architecture skipped and jumped pass the stage of trial and error and quickly developed a style and technique that could match the finest Frankish fortresses. It is, however, important to remember that the Mamluk volume of work cannot be compared to that of the Ayyubids, the Franks or the Armenians. In contrast to the Mamluks’ grand public and religious buildings, their fortresses were repaired, rebuilt and renewed but seldom enlarged and never constructed from the foundations. When examining the details of Mamluk work it is difficult to find a common denominator that would make the eight fortresses discussed here into a homogeneous architectural group. A certain concept can be seen and followed in all the sites covered by this study, but the work itself varies according to regional building materials, local traditions, and the administrative importance attributed to each site. Rather surprisingly, there appears to have been a hierarchy in the level and scale of construction that has no direct connection with the military and defensive role of the fortress. In certain cases the quality of work evidently depended on the rank and status of the amir who commanded the fortress. The latter argument can be best seen at al-Ṣubayba, which was neither a frontier fortress nor a large administrative center, and yet it displays some of the finest Mamluk military architecture. The fact that it belonged to the nā’ib al-sulṭana set the highest standard. More resources were allocated to the building and a larger force of skilled craftsmen seems to have been employed. The work corresponds to the rank of its owner and is perhaps a symbol of his status. However, one cannot strip al-Ṣubayba of its military functions, even if they were relatively limited. Located close to Damascus, the Sultanate’s second capital, it held a large garrison that could be called upon in times of emergency. The following is a summary of the main architectural characteristics. Masonry Masonry is often an important tool for classification and in some cases can be used to determine chronology. Many scholars maintain that Crusader masonry has distinct characteristics;302 no similar uniformity is apparent in Mamluk masonry. No specific tool-marks are distinctive or unique to Mamluk builders. Some fortresses present an extremely high standard of masonry whereas others, such as Tall Ḥamdūn and Baghrās, show fairly poor work. The masons at Tall Ḥamdūn may well have been local, using a tradition that resulted in inferior work. The use of a harder type of stone (basalt), in this site possibly affected the quality, although well dressed basalt stones can be seen at the Crusader fortress of Belvoir.

Mamluk power on display  203 Gates While the Ayyubids consistently built two types of main gates: the bent gateway and the gate tower,303 the Mamluks in most cases made do with existing structures unless they were badly damaged. The multiple gate system at Qalʿat al-Rūm is a good example. It was probably Armenian-built, and was preserved by the Mamluks. This entrance plan is rare in Muslim fortresses and parallels can be found in Frankish and Armenian work.304 The only exception I found was at Safad. The gate at Safad, ruined in the course of the Mamluk siege, was built anew and a T-shaped plan was drawn, which I have not encountered elsewhere. Secret passages Secret passages were not a novelty, although the scale and the building technique displayed at al-Ṣubayba are indeed unique. A grand secret passage was built at the side of Bīlīk’s tower, using stones of remarkable size. According to the sources a subterranean passage was also built at Safad, though to date there is no archaeological evidence. Moats In general moats were only cleaned out, as Ibn ʿAbd al-Ẓāhir’s account states. We are told that Safad had its moat widened and Karak’s moat possibly had alterations made to it, but these were exceptions. Curtain walls The most outstanding feature in Mamluk military architecture is their treatment of curtain walls. While the Franks saw the walls as the most important part of the defense,305 the Mamluk treatment of curtain walls resembled that of their Ayyubid predecessors. Curtain walls did not play a crucial role in the defense and when fortresses were repaired or rebuilt the walls received little attention. The Mamluks never returned to build walls in the fashion and dimensions of early thirteenthcentury Frankish fortifications. The only evidence we have that walls were heightened is that provided by Ibn ʿAbd al-Ẓāhir.306 Archers’ galleries One way of enhancing firepower without adding monumental towers was by building long galleries. This can be seen at Qalʿat al-Rūm, Tall Ḥamdūn, parts of al-Ṣubayba and the western gallery at Karak. Strictly speaking, archers’ galleries can not be regarded as innovations. Nevertheless, the one built at Qalʿat al-Rūm which displays machicolations and arrow slits crowded together is to some extent an unusual arrangement.

204  Muslim Fortresses in the Levant Table 4.2  Towers added and renewed in Mamluk fortresses1 Name of fortress

Towers built

Towers renewed

Al-Ṣubayba Safad

6 new 1 central round tower

Ḥisn al-Akrād3 ʿAkkār Baalbak Baghrās Shawbak6 Karak Qalʿat al-Rūm Tall Ḥamdūn Ḥisbān

2 new towers built4 Towers added5 1 1 3 2 new towers7 ?

2 enlarged Towers rebuilt along the curtain wall2

Al-Bīra Marqab Shawbak

Tower unearthed during the excavations8 New towers were built9 2 new towers10 3 towers carry inscriptions commemorating the Mamluk sultan Lājīn (1297–8)11

1 enlarged Enlargement of southern tower

Notes 1 The information in the first two columns is based on archaeological finds and on inscriptions. 2 Ṣalāḥ al-Dīn Khalīl b. Aybeg Ṣafadī, al-Wāfī bi’l-wafayāt, ed. H. Ritter et al. (Wiesbaden, 1980), vol. 10, 341. 3 P. Deschamps, Les Chateaux des Croisés en Terre Sainte, Le Crac Des Chevalliers, Album (Paris, 1934), Plan 2; T. Biller et al., Der Crac des Chevaliers (2006, Berlin), 378. 4 Ṣafadī, Wafayāt, vol. 10, 342. 5 Ibid. 6 R. M. Brown, “Summary report of the 1986 excavations in late Islamic Shobak,” ADAJ 32 (1988): 225–54, 227. 7 Ṣafadī, Wafayāt, vol. 10, 341; Ibn Shaddād, Ta’rīkh al-malik al-ẓāhir, ed. A. Khuṭayṭ (Wiesbaden, 1983), 351–2. 8 B. J. Walker, “Mamluk investment in Southern Bilād al-Shām in the eighth/fourteenth century: the case of Ḥisbān,” JNES 62 (2003): 241–61, 250–1. 9 Ibid. 10 H. Kennedy, Crusader Castles (Cambridge,1994), 171, fig. 23. 11 Brown, “Shobak,” 227, fig. 2.

Towers Mamluk military architecture stands in sharp contrast to the neat uniformity of the Franks. Frankish towers and curtain walls advanced and improved simultaneously. Although there was no strict canon, the proportions between towers and curtain walls were evenly balanced throughout the Crusader period. This rule of repair and renew was by and large broken when it came to the construction of towers. Towers are the one Mamluk architectural structure that was added in an almost methodical manner to many of the fortresses they had conquered (Table 4.2). Safad presents an interesting case. Both Ibn Shaddād and al-Ṣafadī describe the use of huge stones – hirqilī stones. The architect at Safad was obviously eager

Mamluk power on display  205 to test new ideas. The scale of his work shows that he was neither an amateur nor a beginner. This in turn leads us to an interesting aspect of this survey – the short phase of Mamluk experimental work. Although one can pinpoint its beginning, it seems to have fizzled out within a short space of time. Again my best examples are from the round tower at Safad, and al-Ṣubayba, where Bīlīk’s tower was constructed of huge stones. The other example is the large secret passage that coils round Bīlīk’s tower. Though the use of very large building stones is rare it had been tried out earlier in the thirteenth century by al-Malik al-Ẓāhīr Ghāzī, when he built the bāshūra at Aleppo. The round tower, whether successful or not, was never reproduced elsewhere. One should constantly bear in mind that much of the work was dictated by the damage caused by Mongols or Mamluk sieges. On the whole, genuinely new ideas were few and there was little desire to change and enlarge existing constructions if they functioned. Preference was given to round towers, but they were hardly a novelty by the mid thirteenth century. Both the scale and the standard of construction of twelfth and early thirteenth century Frankish towers and keeps such as those at Chastel Pelerin, Chastel Blanc (Safita), Gibelet (Jubayl), and Margat appear similar to Mamluk work. In comparison to the Ayyubid towers at Mount Tabor, al-Ṣubayba and even at ʿAjlūn, the Mamluks present higher standards of construction and more sophisticated plans. The free-standing round tower at Safad, the southern keep at Karak and the octagonal tower along the southern curtain wall at al-Ṣubayba display some of the finest Mamluk military architecture. They were built with great care and precision, on a grand scale that is not found in rural and frontier fortresses of the Ayyubid period. They were designed to provide ample storage place and accommodate a greater number of bowmen. Above all, towers were placed at very close intervals. Where the distances appeared too great and the curtain wall was of poor quality the Mamluks inserted new towers and “stitched” them to the existing curtain walls. This can be seen quite clearly at al-Ṣubayba, Tall Ḥamdūn and Baghrās. The scale of building The Mamluk scale of building and investment differs significantly from one region to the other. The scale of construction at Karak and al-Ṣubayba and probably at Safad too was considerable and the quality of work higher than that of the Cilician fortresses. Although the fortresses along the Euphrates were maintained at the highest level, it seems that Baybars’ initial investment in the construction of al-Bīra and al-Raḥba was limited. The inland fortresses of Karak, Safad and al-Ṣubayba, with their large garrisons and forbidding fortifications, were never put to the test by either the Franks or the Mongols. And yet the scale of building was significantly larger than along the frontiers. It appears that the important achievements of the Mamluks’ defense were not manifested in their military architecture, but rather in their military organization and their ability to find and maintain the balance between the garrisons on the frontier and the central Mamluk army. This fine balance could only be achieved by a strong centralized regime.

5 Military architecture versus political and military organization

Examining the development of Ayyubid fortifications in light of the political and military situation in the late twelfth and early thirteenth centuries reveals some aspects of the Muslim concept of defense within the political frame of a loose federation. Individual rulers of varying rank each built according to his own immediate needs and capabilities. Nevertheless, fortresses constructed in different regions by different rulers share a number of architectural features. One can see the growing importance the Ayyubids attributed to towers and the steady decline in both quality of construction and width of curtain walls in rural fortifications. Above all they made no attempt to imitate or use as a model the grand concentric fortresses built by the Franks. Ayyubid military architecture clearly reacted and developed according to the gradual decline in Frankish military power. By the mid thirteenth century, although the Franks were still raiding neighboring territories, they avoided large scale confrontations and seldom besieged enemy towns. Siege campaigns were carried out only when a new Crusade arrived and the Kingdom’s army was reinforced. Local Frankish rulers could not recruit the necessary manpower or afford to lose large numbers of men. The superiority of Muslim siege warfare that began to make itself felt in the last quarter of the twelfth century continued well into the early Mamluk period. Siege campaigns carried out along the coastal plains, Syria and on the banks of the Euphrates clearly display the Mamluks’ high abilities in this field. Siege machines of all sizes were dismantled, transported over great distances and re-assembled in the designated camps, or built from local materials when such were available. It was due to this continuous superiority in Muslim siege warfare that Mamluk military architecture followed lines similar to those of the Ayyubids. Although influenced by their predecessors, Mamluk work cannot be seen as an amalgamation of Ayyubid and Frankish military architecture. In general, one may describe their construction during this period as an act that resembles fine tuning. The arrival of the Mongols disrupted the balance of power in the region and led to considerable changes in the concept of defense. It did not, however, bring about any substantial innovations in the field of military architecture. The Mamluks arrived at the very end of a long period during which fortresses had been evolving and changing at a rapid pace. They were quick to learn and to make the necessary adjustments. But,

Military architecture versus political and military organization  207 while the last Frankish wave of fortifications in the late twelfth and early thirteenth centuries included the construction of concentric fortresses, the significant widening of curtain walls, the erection of monumental towers as well as the designing of large scale moats and glacis,1 fear of the Mongols or of the Franks found no real tangible expression in Mamluk military architecture along the frontiers or in the rural areas of the Sultanate. The standard of building at Safad, Karak, al-Ṣubayba and Qalʿat al-Rūm is outstanding in terms of the scale of the towers and their large galleries. But to a large extent it seems that what existed was more than adequate. The passive role played by the Franks and the inferior siege warfare deployed by the Īlkhānids did not drive or challenge Mamluk architects to make dramatic changes in fortifications. This short answer, however, provides only a partial explanation. The longer and more complex answer to the question why the Mongol threat did not trigger significant developments in fortress building lies in the political and military organization of the Mamluk sultanate, which differed greatly from that of the Ayyubids and the Franks. Crusader and Ayyubid fortifications were built by a large number of bodies each possessing its own military force. Among the Franks there were the king, the Military Orders, princes, counts and lesser noblemen. Many of the Ayyubid fortresses were built by a similar assembly: sultans, princes and highranking amirs. One of the most interesting aspects of Mamluk architecture is its strong expression of a powerful and centralized government run by a sultan and a small entourage of mostly military men. Holt defined the sultan’s formal rule as “a despotic monarchy” though he claimed that in practice it was “a veiled oligarchy of the great amirs.”2 The latter part of this statement cannot be supported when we look at the defense policies practiced during the second half of the thirteenth century. Mamluk fortifications and strategy display the abilities of a centralized regime. The responsibilities concerning fortifications were left solely to the sultan, who relied heavily on his own army quartered in Cairo, and the governors of urban citadels and rural and frontier fortresses. The governors were carefully chosen from among the sultan’s most loyal amirs. When a new sultan ascended to power he almost immediately replaced the fortress governors with his own amirs. In some fortresses a selected number of men from his own core of private Mamluks were left to establish or join the fortress garrison.3 The Ayyubid system was different. Following the conquest of Safad and Tiberias by Ṣalāḥ al-Dīn, the two towns were given as an iqṭāʿ to one of his most trustworthy amirs, Sa‘ūd b. Mubārak b. Tamīrak. After Sa‘ūd’s death his lands and title were inherited by his son.4 In Syria the muqṭaʿs (iqṭāʿ holders) were responsible for the defense of their lands. According to Michaudel, “delegation of power was accompanied by delegation of defensive decision making in which Ṣalāḥ al-Dīn … only rarely interfered.”5 Fortresses in the early Mamluk period were never permitted to be bequeathed by governors to a member of their own family. The case of al-Ṣubayba ,which was given to Bīlīk al-Khaznadār as private property, was almost unheard of.6 Strongholds were considered the property of the sultan. Any changes in their architectural layout or in the composition of their garrisons required the sultan’s approval. In quite a few cases funds for the reconstruction of fortresses came from the sultan’s treasury. A unique appointment is created during the reign of the sultan Lachin (r.

208  Muslim Fortresses in the Levant 696/1296–698/1298); the amir Sayf al-Dīn Quṭlū-Bek was sent by the sultan to oversee the finances of all the fortresses in the region of Aleppo.7 According to Keegan, wherever there is a strong central government one almost invariably finds the construction of strategic defenses at the periphery. Keegan cites a number of examples: the Roman Limes, the Egyptian fortresses on the Nubian border built in the second millennium bc, and the fortifications of the Persian Empire in the fourth century bc that stood against the armies of Alexander the Great.8 The idea is supported by Tracy, though his study deals explicitly with urban fortifications.9 The concept that a strong central government builds fortifications along the periphery of its state fits the general pattern of Mamluk defenses. When Baybars began to rebuild strongholds he worked from the periphery to the center. The first fortresses rebuilt and garrisoned were those along the Euphrates. The main defensive idea behind rebuilding existing inland fortresses was to help secure the new Mamluk authority.10 The Mamluks had hardly been in power in Egypt for a decade before Hülegü’s invasion, and Mamluk rule in Syria only began in 1260 after the Mongols were defeated at the battle of ʿAyn Jālūt. Thus they were still trying to establish their own regime. This was partly achieved by the presence of Mamluk governors and garrisons in strongholds such as Safad and Karak that were officially turned into provincial administrative centers soon after they had been taken. Mamluk military architecture was directly influenced by the structure of the Mamluk army. Well-constituted professional armies existed in the region prior to the Mamluk period. The two best examples are the military orders,11 and the mamluk forces that served in the Ayyubid army as early as Ṣalāḥ al-Dīn’s reign. The military orders were well trained, disciplined and fairly mobile. They were, however, settled in their own fortresses and urban quarters and were not mobile in the sense that they could be sent to central or remote parts of the kingdom at the king’s command: “The military orders were not fully subject to any Latin magnate, either secular or ecclesiastical, and they dearly prized their independence.”12 Their interests often clashed with those of the king and his barons and they could not be summoned or employed in the same manner as the Mamluk army under Baybars or any of his immediate successors.13 According to Smail, “No Christian ruler of the twelfth century had an army at his disposal which met his needs in full … Frankish rulers and in particular the Kings of Jerusalem, did not enjoy full authority as commanders of their military forces.”14 The main flaw of the Frankish military system, Smail thinks, was the inability to maintain both a strong field army and adequate garrisons.15 This is where the Mamluks’ system differed greatly from that of the Franks. It was not the military architecture but rather the organization and fine balance between the field armies and the garrisons that enabled them to defend their Sultanate successfully. Although the Ayyubid armies had a core of mamluks, their number and standards of military training can not be compared to those of the Mamluk armies in the second half of the thirteenth and early fourteenth centuries.16 Ṣalāḥ al-Dīn had full control over his armies, but winter campaigns were difficult to organize, for most of his amirs demanded leave at the beginning of winter.17

Military architecture versus political and military organization  209 Since the entire army was under the Mamluk sultan’s command the organization of relief forces for strongholds could be implemented throughout the Sultanate.18 The Franks were well aware that fortresses largely depended on reinforcements. When the fortress of Vadum Iacob was besieged by Ṣalāḥ al-Dīn in 1179, the king managed to recruit a large Frankish army. The fortress fell because the Muslim besieging force was quick, and organizing the relief force was in this case a slow procedure.19 The king’s army left Tiberias for Vadum Iacob (approximately 40km), but the fortress fell before they had covered half the distance. By the mid thirteenth century the king’s weak position and the quarrels among the various factions in the Crusader kingdom made it increasingly difficult to organize a large relief force for besieged strongholds. When Baybars besieged Caesarea, Arsūf, Haifa and Safad no Frankish force in the kingdom could afford to offer help.20 The Ayyubids suffered from similar problems although it seems that their failure to organize reinforcement stemmed from disunity and lack of a central authority rather than shortage of manpower. There were several cases where strongholds were dismantled because there was no possibility of ensuring the support of a field army in times of siege. After he took Baghrās (1188) Ṣalāḥ al-Dīn ordered its demolition, realizing that the Crusader army would pass through the region. Mount Tabor was built and then destroyed (1219) after learning that the Crusader armies had reached the Nile. In the same year al-Muʿaẓẓam ordered the destruction of Safad, the curtain walls of Jerusalem, and Tibnīn because he feared they would all be retaken by the Franks.21 Rather than refortify and reinforce the garrisons, these fortresses were abandoned and left in ruins. If the key to a fortress’s existence indeed lay in the ability of the ruler to provide a relief force, this was best understood and implemented by the Mamluks.22 Though by no means an innovative strategy, it could not be fully exploited by the Ayyubids or the Franks, for neither regime possessed a sufficiently tight control over its army. The structure of the Mamluk regime and army enabled it to guarantee reinforcement to almost all its strongholds, although, as has been shown, certain regions were given preference over others. The frontier with the Armenians came second to the frontier with the Īlkhānids. During Baybars’ reign the Mamluk army probably numbered between 30,000 and 40,000 horsemen. The sultan’s force of Royal Mamluks numbered 4,000 men, who were chosen, bought, trained, armed and paid by him and could be called upon at his will. During the Baḥrī period their number never exceeded 10,000. High-ranking amirs bought and trained their own mamluks and the rulers of the Syrian cities, who had fully submitted to Baybars, maintained their own forces. Nomadic tribes such as the Bedouin (ʿurbān) and the Türkmen were often recruited, as well as refugees from the Īlkhānid state (wāfidiyya).23 During the first three decades, the several components of the Mamluk army were subject only to the sultan’s command. This unique military system and the urban-oriented social structure caused the weight to shift from the fortresses to the field armies. A greater emphasis was laid on creating an independent and mobile army. It is important to keep in mind that the Mamluks had few military allies during this period.

210  Muslim Fortresses in the Levant Though Mamluk defense still relied on strong fortresses and the performance of their garrisons, the sultan’s field armies carried a greater share of responsibility. This had a direct influence on the development of military architecture. The investment in fortifications decreased; no new fortresses were built, few architectural innovations can be seen and the scale of building was limited. To this one may add the conclusions of Chapter 2, which show that although the Mongol threat constantly hovered above the sultanate, Mongol siege warfare capabilities were inferior to those known in the Eastern Mediterranean. The establishment of the Īlkhānid state did not change matters; as its siege abilities along the Mamluk frontier slowly declined there was no real call for new ideas within the field of Mamluk military architecture.

Was there a grand Mamluk defense strategy? The question of frontier policy and the existence of a grand strategy in the Roman Empire is a controversial subject amongst scholars in the field of military history.24 The same question can be asked with regard to the early Mamluk period. Was there a planned defense strategy? Was the location of each and every fortress thought out? The historical sources show that Baybars followed a well-thought-out strategy, which was maintained by his successors. It seems though that to a certain extent the numerous fortresses already scattered across the region, their location and size, greatly influenced and helped form this strategy, if not fully dictate it. On the whole the number of fortresses the Mamluks rebuilt and maintained was smaller than that held by the Franks. If a fortress was not a true necessity (administrative, military or economic), and might become a burden on the treasury and the field army, Baybars ordered its destruction. Fortresses along the coast and strongholds such as Mount Tabor, Ṣafītha, ʿAmūdayn, al-Tīna,25 Qalʿat al-Qurayn (Montfort) and others were sacked and deserted. Ibn Shaddād lists 22 fortresses in the vicinity of Aleppo that were taken by Baybars and abandoned. Several accounts show the sultan weighing the pros and cons of a site before deciding whether to keep the fortress and restore it, or demolish it.26 Much emphasis was laid on setting a communication system between fortresses along the frontier and the centers of government. The barīd developed step by step with the Mamluk conquests. Key fortresses along the frontier, the main roads, and administrative centers were almost immediately connected to the postal route. Fortresses on the Euphrates had three means of sending messages in times of an emergency. Al-Bīra and al-Raḥba are the only fortresses that al-ʿUmarī mentions as having dovecotes, fire and smoke signals as well as the barīd services.27 The existence of a well-organized strategy is manifested in the selection of sites, the relatively modest scale of building, the establishment of the barīd, the organization of relief forces, the constant concern for frontier garrisons and fortress maintenance, the choice of governors and the composition of the garrisons. Although this grand strategy had its faults it served the Mamluks well. Mamluk amirs occasionally revolted and tried to form their own quasi-principalities, and the Īlkhānid armies managed to cross the Euphrates, for frontiers cannot be sealed, and yet the Mamluk sultanate outlived all three of its most prominent enemies: the Franks, the Īlkhānids, and the Armenians.

Military architecture versus political and military organization  211 The Mamluks learnt from their predecessors’ mistakes and appear to have changed their priorities. Mamluk power was first and foremost displayed in its centralized regime and superior military organization. Military architecture came second. During this period architecture meant to display power was mostly constituted by the many monumental public and religious buildings in the large urban centers. Expenditure on grand and lavish projects such as mosques, madrasas, hospitals, bathhouses, markets, and caravansaries may well have surpassed the funds allocated to military architecture.

Epilogue Though the thirteenth century defense system had proved itself, it was not maintained throughout the Mamluk period. During the late fourteenth and the fifteenth centuries Mamluk strategy changed considerably in response to geopolitical movements in the region with the fall of the Īlkhānid state, the Armenian kingdom in Cilicia and the rise of new Türkmen principalities. According to Har-El: “The Mamluks, who were no longer able to garrison those places, agreed to turn them over to the Türkmen chieftains, hoping in return to gain their assistance in the defence of the frontier.”28 Maqrīzī describes the Türkmens thus: “The Türkmens were the equivalent of fortifications protecting the country … And if the sultan delegated them to fight a war they acquiesced to his order and they went ahead in obeisance and prostration.”29 It seems, however, that the changes in defense can be partly attributed to changes in the nature of the Mamluk sultanate and the standing of the sultan. The internal chaos that struck the Sultanate soon after the death of al-Nāṣir Muḥammad (741/1340) led to the gradual deterioration of fortifications along the eastern frontiers, and slackness among the garrisons who were no longer directly supervised by the sultans. Towards the end of the fourteenth century a large Central Asian nomadic army was to invade the Middle East. The new defense strategy failed miserably, leaving much of northern and central Syria in ruins. News of Temür’s (Tamerlane) arrival at Tabrīz and the conquest of Baghdad reached Barqūq’s court as early as 795/1392–3.30 In addition, the sultan received a warning from the Jalayirid ruler of Iraq. Internal politics made it difficult to act, and the threat was probably not taken seriously. The latter would appear to have been the case. According to Ibn Khaldūn, Barqūq admitted that he feared the Ottoman sultan more than he feared Temür.31 Reports of Temür’s conquests and atrocities must have been confirmed by the early 1390s. Barqūq set out for Syria in 797/1394–5 and personally saw to the appointment of new governors in three of the towns along the Euphrates frontier (Qalʿat al-Rūm, Rūha (Edessa) and Malaṭya).32 It is difficult to determine whether this move made by the sultan was directly connected with turbulent events that were to follow, or if it was a routine shuffle of governors intended to prevent powerful amirs from rebelling. The doubt concerning this act surfaces due to the fact that some of the towns (Tarsus, Safad, Tripoli and Karak) that had their governors replaced were not directly involved in the defense of the frontier. Whatever the case, no other preparations were made in anticipation of Temür’s invasion; neither fortresses nor garrisons were strengthened. Temür eventually invaded

212  Muslim Fortresses in the Levant Syria in 803/1400–1, a year after Barqūq’s death. Matters were left to a handful of powerful amirs, since Barqūq’s heir, al-Malik al-Nāṣir Faraj b. Barqūq was only eleven years old (b. 791/1389; first reign 801/1399–808/1405). Temür’s armies marched from the north, taking a different route than Hülegü’s armies 140 years earlier. The Mamluk defense had lost all ability to hold back an invading army. Whatever changes Barqūq had made among the governors, it did not prepare them to resist the coming Mongol invasion. No alert was sounded and the garrisons of the inland towns of Sivas, Malaṭya, Bahasnā and ʿAyn Tāb were overrun.33 According to Ibn Taghrī Birdī, “No one in Egypt made preparations for war against Temür.”34 Temür’s armies headed straight to the central Syrian cities. The Syrian Mamluk armies were defeated outside Aleppo and the city was left in ruins and abandoned by its inhabitants. The Egyptian army eventually left Cairo in an attempt to save Damascus from a similar fate, but to no avail. The army withdrew before the battle began. The inhabitants of Damascus were tortured and massacred, the city plundered and large parts of it thoroughly destroyed.35 At the beginning of his account, Ibn Taghrī Birdī sums up the reasons for the Mamluk failure: “neglect and inattention marked affairs at Cairo because of the absence of centralized authority and because of the varied opinions that were held.”36 Matters did not improve later in the century. A description of al-Bīra and Qalʿat al-Rūm dating to the last quarter of the fifteenth century gives a clear idea of the state of affairs along the Euphrates. In 877/1472 the Aqquyunlu, led by Uzun Ḥasan (d. 1478/883) took several of the fortresses and towns along the river, invested al-Bīra and arrived at the outskirts of Aleppo,37 but soon retreated. The Mamluk sultan Qaytbāy (r.872/1468–901/1496) set out in 882/1477 on a four-month tour of his sultanate, the Euphrates being one of the regions visited. The journey was organized in a somewhat pretentious manner. The sultan was accompanied by a fairly large staff, and received by the governors at each site. While some fortresses were inspected by the sultan himself others were visited only by his staff. A detailed itinerary was written by Badr al-Dīn Abū’l-Baqāʿ Ibn Yaḥyā Ibn Jīʿān, deputy to Qaytbāy’s personal secretary. It was titled Al-qawl al-mustaẓraf fī safar mawlānā al-malik al-ashraf (“The elegant report recounting the voyage of our lord the esteemed monarch”).38 Some of the fortresses along the Euphrates were inspected. According to Ibn Jīʿān al-Bīra’s houses and curtain wall were destroyed by the Aqquyunlu leader Ḥasan Bak (1457–1478).39 The sultan’s entourage stayed at al-Bīra where they were received by the nā’ib. A short account is given of Qalʿat al-Rūm, described as a large town. A group of rebels settled in the fortress, their strength steadily growing as people from al-Bīra and ʿAyn Tāb flocked to it. While al-Bīra was reinforced, nothing was done to regain Qalʿat al-Rūm. The sultan did not personally visit the site, and it appears that only a small detachment was sent. Qaytbāy saw to the repairs of al-Bīra’s town fortifications. His work is commemorated in three inscriptions set along the walls and gates.40 Though Ibn al-Iyās only briefly mentions the works of Qāytbāy at al-Bīra (887/1482) he adds a few interesting details which show the financial advantages of having the sultan initiate and sponsor fortification projects:

Military architecture versus political and military organization  213

… in this year the sultan ordered the restoration of the curtain wall at al-Bīra, and he created the finest buildings and spent his money on its [i.e. the curtain wall] construction.41 Creswell had described the rebuilding of the fortifications in a humorous way: “Qaytbāy was so pleased with his work that he gave himself a new title – Sāḥib al-Qilāʿ al-Rūmīyya – Master of the fortresses of Rūm,” which appears in an inscription in Cairo.42 This is the last account of Mamluk construction work along the Euphrates frontier before the Ottoman conquest. Though the Sultanate’s administrative structure survived through several long periods of internal strife, the defense of the periphery in general and the Euphrates frontier in particular could not maintain itself throughout such uncertain periods that could last for a substantial length of time. The governors of towns and fortresses in remote districts largely relied on a strong central government to provide a field army in times of enemy raids or full scale invasions. Funds for the upkeep of garrisons and maintenance of fortifications for the most part seem to have depended on the Sultanate’s treasury. Even if resources were partially taken or supplemented from local taxes, this required an order from the central authority. Large-scale repairs and renovations were initiated by the Sultan and supervised by his closest and most able amirs. Fortresses may have been reasonably maintained when the defense of the frontier was put in the hands of the Türkmen chieftains, though it is doubtful if they had the resources or whether their political interests were always identical to those of the Sultanate. In times of trouble, the frontier defenses again became the sultan’s direct responsibility and in both cases noted above the sultan indeed intervened. Though Barqūq and Qaytbāy showed concern, the changes they ordered along the eastern frontier were negligible. Barqūq’s nominal preparations resulted in a devastating invasion. This, however, did not prompt the Mamluks to act. No lesson was learnt and almost seventy years later the Aqquyunlu managed to conquer a large section of the middle Euphrates and once again the enemy’s army arrived at Aleppo. The success of the defense system largely depended on a centralized authority, the sultan’s political strength and the extent of his control of the army. If the sultan did not have all of these, the Sultanate’s frontiers were only truly secure in times of peace.

Notes

Introduction 1

Survey group: Prof. R. Ellenblum, Prof. R. Amitai, Y. Shapira and K. Raphael.

1  Ayyubid fortresses in the late twelfth and early thirteenth centuries   1 Petersen, A., Dictionary of Islamic Architecture (London and New York, 1996), 89.   2 Creswell, K. A. C., Fortifications in Islam Before a.d. 1250, Proceedings of the British Academy 38 (London, 1952), 112–16. Badr al-Jamālī was a general of Armenian origin who commanded the army (r. 1074–94).   3 William Archbishop of Tyre, The History of Deeds Done Beyond the Sea, trans. E. A. Babcock and A. C. Krey (New York, 1976), vol. 2, book 13, 9 (Tyre), book 17, 219 (Ascalon); Ellenblum, R., Crusader Castles and Modern Histories (Cambridge, 2007), 243–4.   4 I was not able to visit Qalʿat Ṣadr, as it is currently in an Egyptian military zone. The study of the fortress has been done from maps, photographs and articles.   5 Ibn Wāṣil, Jamāl al-Dīn Muḥammad b. Sālim, Mufarrij al-kurūb fī akhbār banī ayyūb (Alexandria, 1953), vol. 3, 215–16.   6 Yāqūt al-Rūmī, Muʿjam al-buldān (Beirut, 1957), vol. 3, 397.   7 Johns, C. N., “Medieval ʿĀjlūn,” QDAP 1 (1931): 21–33.   8 Dangles P. et al., Le Chateau d’Āğlūn/Qal’at ar-Rabad Jordanie, Report intermédaire de la mission menée en mars et avril 2002, I.N.R.A.P. 2002.   9 Battista, A. and Bagatti, B., La Fortezza Saracena Del Monte Tabor (ah 609–15; ad 1212– 18) (Jerusalem, 1976). 10 Ellenblum, R., “Who built Qalʿat al-Ṣubayba ?,” DOP 43 (1989): 103–12; Amitai, R., “Notes on the Ayyubid inscriptions at Al-Ṣubayba (Qalʿat Nimrod),” DOP 43 (1989): 113–19. 11 Hartal, M., The Al-Ṣubayba (Nimrod) Fortress, Towers 11 and 9 (Jerusalem, 2001), 3–107. 12 Barthoux, J., “Description d’une forteresse de Saladin découverte au Sinai,”Syria 3 (1922): 43–57. 13 Wiet, G., “Les inscriptions de la Qalʿah Guindi,” Syria 3 (1922): 57–65. 14 Tamari, S., “Darb al-Hajj in Sinai: an historical-archaeological study,” Accademia Nazionale dei Lincei Memorie 25 (1982): 431–525; ibid., “Two further inscriptions from Qalʿat Al-Ṣadr,” in Studies in Memory of Gaston Wiet, ed. M. Rosen-Ayalon (Jerusalem, 1977), 261–5. 15 Mouton, J. M., “Autour des inscriptions de la forteresse de Ṣadr (Qalʿat al-Ğundī) au Sinaï),” Annales Islamologiques 28 (1994): 30–57; Mouton, J. M. and Sāmi Sāliḥ ʿAbd al-Mālik, “Les decors animaliers de la forteresse de Ṣadr (Qalʿat al-Ğundī),” Annales Islamologiques 28 (1994): 60–9; Mouton, J. M. and Sāmī Ṣāliḥ ʿAbd al-Mālik, “La route

216  Notes de Saladin (ṭarīq Ṣadr wa Ayla) au Sinaï,” Annales Islamologiques 30 (1996): 42–70; Mouton, J., “Sadr au Sinaï, la forteresse de Saladin,” Archéologie 410 (2004): 28–39. 16 Creswell, Fortifications, 89–125; ibid., “Archeological researches in the Citadel of Cairo,” Bulletin de l’Institut français ďarchéologie orientale 23 (1924): 89–158. The Citadel of Homs was partly studied by King, G. R. D., “Archaeological fieldwork at the Citadel of Homs, Syria: 1995–1999,” Levant 34 (2002): 39–58; King, D. J., “The defences of the Citadel Damascus: a great Mohammedan fortress of the time of the Crusades,” Archaeologia 94 (1951): 57–96; Chevedden, P. E., The Citadel of Damascus. PhD diss., University of Los Angeles, California, 1986, unpublished. 48–81. 17 Creswell, Fortifications, 89–125. 18 Kennedy, H., Crusader Castles (Cambridge, 1994), 180–5. 19 Korn, L., Ayyubidische Architektur in Ägypten und Syrien: Bautätigkeit im Kontext von Politik und Gesellschaft 564–658/1169–1260 (Heidelberger Orientverlag, 2004). 20 Hillenbrand, C., The Crusades: Islamic Perspectives (New York, 2000), 467–509. 21 Ibid., 467. 22 The only fortress that may have been built prior to the Ayyubid period is Qalʿat Ṣadr. Tamari, S., “Darb al-Hajj in Sinai: an historical-archaeological study,” Accademia Nazionale dei Lincei Memorie 25 (1982): 448, n. 4. Mount Tabor had a fortified monastery which was built by the Franks. 23 Faroqhi, S., Pilgrims and Sultans: The Hajj under the Ottomans 1517–1683 (London and New York, 1994), 54. 24 Kennedy, Castles, 98; Ellenblum, Modern Histories, 189; Chevedden, P. E., “Fortifications and the development of defensive planning during the Crusader period,” in The Circle of War in the Middle Ages, eds. D. J. Kagay and L. J. A. Villalon (Woodbridge, UK, 1999), 34.   The Hungarian scholar Erik Fügedi reached a very different conclusion while researching fortresses of the first half of the thirteenth century in Hungary. According to Fugedi: “Innovations in castle building during the thirteenth century were not triggered by advances in military technology, but rather by social development, enhanced by the Mongol invasion of Hungary in 1241.” Fügedi, E., Castles and Society in Medieval Hungary (1000–1437) (Budapest, 1986), 42. 25 Marshall, C. J., Warfare in the Latin East, 1192–1291 (Cambridge, 1996), 93. For a totally different opinion, emphasizing the importance of the superiority of the Frankish field army during the 12th century see Ellenblum, Modern Histories, ch. 10, especially 161–4. 26 On the economic situation of the Ayyubid sultanate during the first two decades after Ṣalāḥ al-Dīn’s death, see the short paragraph in Humphreys, R. S., From Saladin to the Mongols: The Ayyubids of Damascus 1193–1260 (Albany, 1977), 132. According to Ehrenkreutz, the financial condition of Egypt in the mid-twelfth century, based on numismatic studies, archaeological finds and historical sources, was far from one of bankruptcy. Ehrenkreutz, A., Saladin (New York, 1972), 17–18, 102–3. 27 Ayalon, D., “Studies in the structure of the Mamluk army,” BSOAS 15 (1953): 448–76; Humphreys, R. S., “The emergence of the Mamluk army,” SI 46 (1977): 148–52; Humphreys, Saladin, 304; Lev, Y., Saladin in Egypt (Leiden, Boston and Cologne, 1999), 141–58. 28 The development of Karak and Safad into mamlakas will be noted in Chapter 4. 29 ʿImād al-Dīn al-Isfahanī, Ḥurūb Salāḥ al-Dīn wa-fatḥ bait al-maqdis (Beirut, 2003), 62; Sibṭ Ibn al-Jawzī, Mir’āt al-zamān fī ta’rīkh al-aʿyān (Hyderabad, 1952), vol. 8, pt. 1, 394. 30 Chevedden, “Development,” 36–8. The subject has been discussed by Chevedden in several articles: Chevedden, P. E., “Artillery in late antiquity,” in The Medieval City under Siege, eds. I. Corfis and M. Wolfe (Woodbridge, 1995), 131–73; ibid., “The trebuchet: recent reconstructions and computer simulations reveal the operating principles of the most powerful weapon of its time,” Scientific American (1995, July): 80–6; Michaudel, “Islamic military architecture,” 106, 109, 112. Kennedy seems to support this idea although he is slightly cautious: Kennedy, Castles, 9. The same attitude is adopted by Yovitchitch in his article:

Notes  217

31 32 33 34 35 36 37 38 39 40

41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63

“The Tower of Aybak in ʿAjlun Castle: an example of the spread of an architectural concept in early thirteenth century Ayyubid fortifications,” in Military Architecture in Greater Syria, ed. H. Kennedy (Leiden and Boston, 2006), 225–42. See especially p. 236. Chevedden, “Artillery,” 131–73; ibid., “The trebuchet,” 66–71; ibid., “Development,” 36–8. Ellenblum, Modern Histories, 187–98. Ehrenkreutz, Saladin, 36–7, 42–3, 48. Riley-Smith, J., The Crusades: A Short History (London, 1990), 84; Lev, Saladin, 163. Lev, Saladin, 163. Marshall, Warfare, 256. Ibid., ch. 6, especially 255–6; Hillenbrand, Crusades, 467. Ibn al-Athīr, ʿIzz al-Dīn, Al-Kāmil fī al-Ta’rīkh, ed. C. J. Tornberg (Beirut, 1966), vol. 12, 322. Ibid., vol. 12, 323–6; Prawer, J., A History of the Latin Kingdom of Jerusalem, 2nd edn (Jerusalem, 1963), vol. 2, 142 (Hebrew). Ibn al-Furāt, Ayyubids, Mamlukes and Crusaders: Selections from the Tā’rīkh al-duwal wa’lmulūk of Ibn al-Furāt, ed. and trans. U. and M. C. Lyons; introduction and notes J. S. C. Riley-Smith (Cambridge, 1971), 26–8. Cited in Ellenblum, Modern Histories, n. 47, pp. 136, 137. Ellenblum, Modern Histories, ch. 9. Dawkins, R. and Krebs, J. R., “Arms races between and within species,” Proceedings of the Royal Society of London. Ser. B 205 (1979): 492. Dawkins and Krebs, “Arms races,” 493. Ibid., 489. Marshall, C. J., “The French regiment in the Latin East, 1254–91,” Journal of Medieval History 15 (1989): 301–7. Ellenblum, “Al-Ṣubayba,” 103–12. Johns, “ʿAjlūn,” 23. Johns refers at length to the Banū ‘Awf and their growing strength in the region, but suggests that one of the main reasons for the construction was the Frankish forts south of ʿAjlūn – Karak and Shawbak. Ibn Shaddād al-Ḥalabī, ʿIzz al-Dīn Muḥammad b. ʿAlī, al-Aʿlāq al-khaṭīra fī dhikr umarā al-shām wa’l-jazīra (Damascus, 1972), 86–7. Al-Qalqashandī, Shihāb al-Dīn Aḥmad, Ṣubḥ al-aʿshā fī Ṣināʿat al-inshā’ (Beirut, 1987), vol. 4, 89. Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 87. Johns, “ʿAjlūn,” 23. Ibn al-Athīr, Kāmil, vol. 12, 22–3, ʿImād al-Dīn, Ḥurūb, 167–9. Ibid., vol. 12, 300. Marshall, Warfare, 65. Benvenisti, M., The Crusaders in the Holy Land (Jerusalem, 1976), 297. There is no historical evidence of the Mamluk conquest. Tamari, “Darb al-Hajj,” 448, fn. 4. The architecture and its origins will be discussed in detail later in this chapter. Abū Shāma, Shihāb al-Dīn ʿAbd al-Raḥmān b. Ismāʿīl, Kitāb al-Rawḍatayn (Beirut, 1997), vol. 3: 21, 105, 107, 288, 339–40. Ibid., vol. 2, 88. Al-Maqrīzī, Taqī al-Dīn Aḥmad b. ʿAlī, Kitāb al-sulūk li-maʿrifat duwal al-mulūk (Beirut, 1997), vol. 1, 75. Ibid., vol. 1, 193. Ibid., vol. 1, 199–200. Yāqūt al-Rūmī (d. 626/1229) reports in a short passage in the Muʿjam al-Buldān that the fortress is in ruins. He began to write this work in 615/1218–19; the final draft dates to 625/1228, and the fortress fell into decay some time during those ten years. Yāqūt al-Rūmī, Muʿjam al-Buldān (Beirut, 1957), vol. 3, 397.

218  Notes 64 Ibn Shaddād, Aʿlāq , vol. 2, pt. 2, 86. 65 Abū’l-Fidā’, al-Malik al-Muʿayyad ʿImād al-Dīn Ismāʿil b. ʿAlī., al-Mukhtaṣar fi ta’rīkh al-bashar (Beirut, 1997), vol. 2, 206. 66 Humphreys, Saladin, 144. 67 Abū’l-Fidā’, Mukhtaṣar, 206. 68 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 89. 69 Chevedden, Damascus, 63. 70 Twain, M., The Innocents Abroad (New York, 1962), 397. 71 On the importance of Safforie during the Crusader period see Ellenblum, R., Frankish Rural Settlement in the Latin Kingdom of Jerusalem (Cambridge, 1988), 205. 72 Jezrael (Parvum Gerinum) was held by the Order of the Templars during the years proceeding the battle of Ḥaṭṭīn. Pringle, D., Secular Buildings in the Crusader Kingdom of Jerusalem (Cambridge, 1997), 56. 73 La Fève (Castrum Fabe), taken over by the Templars in the early 1170s, was well built and manned; the garrison was of considerable size numbering between 50 and 60 knights. To that one should add sergeants and foot soldiers. The fortress also served as a storage site for arms, tools and food. It was used twice as an assembly place for the army of the whole kingdom. Barber, Knighthood, 87; Kedar, B. Z. and Pringle, D., “La Fève: a Crusader castle in the Jezreal valley,” IEJ 35 (1985):164–79. 74 Daniel of Kiev, The Pilgrimage of the Russian Abbot Daniel in the Holy Land 1106–1107 a.d., trans. from Russian by C. W. Wilson (London, 1985), PPTS, vol. 4, 66–7; Raba, J., Russian Travel Accounts on Palestine (Jerusalem, 1986), 63 (Hebrew). For a detailed account of the Crusader sources and the description of the Monasteries see Pringle, D., The Churches of the Crusader Kingdom of Jerusalem (Cambridge, 1998), vol. 2, 64–8. 75 William of Tyre, Deeds, vol. 2, 495; This same attack is described by Ibn al-ʿAdīm, but he does not describe the place. Kamāl al-Dīn Ibn al-ʿAdīm, Zubdat al-Ḥalab min ta’rīkh Ḥalab, ed. Sāmī al-Dahhān (Damascus, 1951), vol. 3, 73–4; Ellenblum, Modern Histories, 136, 169. 76 Ibn Wāṣil, Mufarrij, vol. 3, 215. 77 Abū Shāma, Rawḍatayn, vol. 3, 315, 318, 320. According to Pringle Mount Tabor was raided in June 1187 a month before the battle of Ḥaṭṭīn. But no damage was caused to the monasteries. Pringle, Churches, vol. 2, 66. 78 Qalqashandī, Ṣubḥ, vol. 4, 109. 79 Humphreys, Saladin, 75–9. 80 Ibn al-Athīr, ʿIzz al-Dīn ʿAlī, al-Kāmil fī’l-ta’rīkh, ed. C. J. Tornberg (Beirut, 1966), vol. 12, 194–5. 81 Ibn al-ʿAdīm, Ta’rīkh Ḥalab, vol. 3, 819. 82 Marshall, Warfare, 19. 83 Humphreys, Saladin, 136–7. 84 Ibn Wāṣil, Mufarrij, vol. 3, 216; Ibn al-Athīr, Kāmil, vol. 12, 300. 85 Benvenisti, Crusaders, 359. 86 Ibn Wāṣil, Mufarrij, vol. 3, 215–16. 87 Jacques de Vitry, History of Jerusalem a.d. 1189, trans. from Latin by A. Stewart (London, 1886), PPTS, vol. 11, 119. 88 Ibid., 119. 89 Riley-Smith, Crusades, 144–7. 90 nafṭ: thick conical clay pots with a short and narrow neck, containing an inflammable liquid and a wick, similar to a Molotov Cocktail. The wick was lit and the pot was thrown at the enemy. There are several questions concerning this peculiar weapon that have yet to be answered. In the case of Mount Tabor the Franciscan excavation had unearthed over twenty of those vessels in the fortress grounds. See Battista and Bagatti, Monte Tabor, 119–42. Ayalon says that “the use of naphtha by the Muslims reached its peak during the period of the Crusades.” See Ayalon, D., Gunpowder and Firearms in the Mamluk Kingdom, 2nd edn (London, 1978), 10–11.

Notes  219 91 Sibṭ Ibn al-Jawzī, Mir’āt, vol. 8, pt. 2, 584–5. 92 Ibn al-Athīr, Kāmil, vol. 12, 322. 93 Ibn al-Wāṣil, Mufarrij, vol. 3, 215; Sibṭ Ibn Jawzī, Mir’āt, vol. 8, pt. 1, 585; Ibn al-Athīr, Kāmil, vol. 12, 323. 94 Oliver of Paderborn, cited in Ellenblum, Modern Histories, n. 47, pp. 136, 137. 95 Ibn Wāṣil, Mufarrij, vol. 3, 215. 96 The Crusader armies only left Egypt twenty months later at the end of August 1221. Riley-Smith, Crusades, 147–9. 97 Ibn Wāṣil, Mufarrij, vol. 3, 212. 98 Abū ‘l-Fidā’, Mukhtaṣar, vol. 12, 208. 99 Maqrīzī, Sulūk , vol. 1, part 1, 176. 100 Johns, “ʿAjlun,” 24, 28. 101 Ibn Wāṣil, Mufarrij, vol. 3, 202 102 Ibid., 216. 103 Battista and Bagatti, Monte Tabor, 72–3. 104 Quran, Sura 2: 262–3. Text, translation and commentary by Abdullah Yusuf Ali (Cambridge, MA, 1946), vol. 2. 105 Battista and Bagatti, Monte Tabor, 92. 106 Battista and Bagatti, Monte Tabor, 87. From first line to third line see Quran, sura II: 262. 107 Amitai, “Ayyubid Inscriptions,” 114, n. 9. 108 Battista and Bagatti, Monte Tabor, 95. 109 Quran, Sura 48: 1–3, vol. 2. 110 Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 260. This title is on the whole quite rare in inscriptions. It appears in an inscription from Bayt Hanūn dated 1239. Sharon, M., Corpus Inscriptionum Arabicarum Palaestinae (Leiden, Boston and Cologne, 1999), vol. 2, 99. During the early Mamluk period the rank of the amir Isfahsalar was probably equal to that of the amir tablkhānā. Humphreys, “Mamluk Army,” 147–82. See especially p. 174. 111 Battista and Bagatti, Monte Tabor, 99. 112 This could mean command or governorship – walaya, . 113 Battista and Bagatti, Monte Tabor, 101. 114 Cytryn-Silverman, K., The Road Inns (Khāns) of Bilad al-Shām during the Mamluk Period (1260–1516): An Architectural and Historical Study, PhD Diss, The Hebrew University of Jerusalem 2004, 24–5. Unpublished 115 Ibn Khallikān, Wafayāt al-A‘yān (Beirut, 1988), vol. 3, 494; Littman, E., “Aybak,” EI 2 1: 780. 116 This corresponds to the dismissal of ʿIzz al-Dīn Usāma, the first governor of ʿAjlūn, by al -ʿĀdil as mentioned earlier in this chapter. 117 According to Ayalon the word khādim indicates that the person referred to was a eunuch. Ayalon, D., Eunuchs, Caliphs and Sultans: A Study in Power Relations (Jerusalem, 1999), 200–84. 118 Ayalon, Eunuchs, 176, 179. 119 Battista and Bagatti, Monte Tabor, 82. The last half of the fifth row can hardly be read as the letters are not legible. 120 Husām al-Dīn Lu’lu’ appears in inscriptions 1, 3, 4, 5, 6, and 8, from which I cite only four. Battista and Bagatti, Monte Tabor, 72–102. 121 An amir kabir in the Ayyubid period could be simply a senior officer. Humphreys, R. S., “The Emergence of the Mamluk Army,” SI 45 (1977): 86–9; Ayalon, D., “ Studies in the structure of the Mamluk army – III,” BSOAS 16 (1954): 81–2. Later in the Mamluk period it was given to amirs of the rank of 100 (i.e. who had a permanent force of 100 Mamluk horsemen, and commanded 1000 men in the battlefield). Humphreys, R. S., “The emergence of the Mamluk army (conclusion),” SI 46 (1977): 175–7. 122 Battista and Bagatti, Monte Tabor, 116–117.

220  Notes 123 Ayalon, Eunuchs, 294; Ayalon, D., “The eunuchs in the Mamluk sultanate,” in The Mamluk Military Society (London, Variorum, 1979), III, 279. 124 Battista and Bagatti, Monte Tabor, 106. 125 Irwin, R., The Middle East in the Middle Ages: The Early Mamluk Sultanate 1250–1382 (London, 1986), 39. 126 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 86. 127 Ellenblum, “Al-Ṣubayba,” 10, 3–112. 128 Amitai, “Ayyubid inscriptions,” 114–15. 129 Amitai, “Ayyubid inscriptions,” 114. 130 Al-Malik al-Sa‘īd Fakhr al-Dīn Ḥasan b. al-ʿAzīz was later offered an iqṭāʿ in Egypt by the sultan al-Ṣalīḥ Ayyūb and therefore left Banias. Humphreys, Saladin, 223, 290, 292. 131 Amitai, “Ayyubid inscriptions,” 114–15. 132 Hartal, Al-Ṣubayba, 93; Amitai, “Ayyubid inscriptions,” 118–19. 133 Wiet, G., “Les inscriptions de la Qalʿah Guindi,” Syria 3 (1922), 62–3. 134 Ibid., 62–3. 135 Ibid., 62. 136 Combe, E. et al., eds, Répertoire chronologique d’épigraphie arabe, Publications de l’Institute François d’Archéologie orientale (Cairo, 1939), vol. 10, 276, inscription no. 3800 A. 137 There is no actual reference or mention of the qualifications of the figure who supervised the building. The term Muhandis found occasionally in the sources seems to refer to a land surveyor, engineer or contractor. Behrens-Abouseif, D., “Muhandis, Shād, Mu’allim – notes on the building craft in the Mamluk period,” DI 72 (1995): 295. 138 At the fortress of Belvoir, stone was brought from three different quarries, one as far as 20 km away. Ellenblum, R., “Construction methods in Frankish rural settlements,” in The Horns of Ḥaṭṭin, ed. B. Z. Kedar (Jerusalem, 1992), 173. 139 Vitruvius, The Ten Books on Architecture, trans. M. H. Morgan (Cambridge, MA and London, 1926), 49–50. 140 An account of the site and an architectural analysis are given by Ellenblum, Modern Histories, 258–74. 141 The fortress passed into the hands of the Franks along with many other fortresses in the Galilee in 1240–41, after a treaty was signed between the Ayyubid Sultan, al-Ṣalaḥ Ayyūb and Richard of Cornwall. Prawer, Latin Kingdom, vol. 2, 271.   It seems, however, that the Franks did not rebuild the fortress during this period. If any restoration work was done it was probably carried out from 1255 onwards under the supervision of the Hospitallers. 142 Barbican meaning the outer defense of a fortress. Bastion, a projecting part of the fortification. 143 Battista and Bagatti, Monte Tabor, 86. 144 Abū Shāma, Rawḍatayn, 108–9; Battista and Bagatti, Monte Tabor, 40 145 Edwards, R.W., The Fortifications of Armenian Cilicia (Washington, DC, Dumbarton Oaks Research Library and Collection, 1986), 18–24. 146 Johns, “ʿAjlun,” 26–8. 147 Ellenblum, “Construction,” 169–89. 148 Tapered curtain wall stones can be seen at al-Ṣubayba. But many of the stones do not comply with this rule. 149 Edwards, Fortifications, 22. 150 Raphael, K. and Tepper, Y., “The archaeological evidence from the Mamluk siege of Arsūf,” MSR 9/1 (2005): 85–100, fig. 7. 151 Johns, “ʿAjlūn,” 29. 152 The siege of Nicea 1097, Gesta Francorum et aliorum Hierosolimitanorum, ed. R. Hill (London 1962), 2, VIII, 14; the siege of Tyre 1112, William of Tyre, book 13, ch. 6, 10–11; the siege of Ascalon 1153, ibid., bk. 17, ch. 27, 226; and at the siege of Arsuf 1101, ibid., book 17, ch. 23, 220–2.

Notes  221 153 Duffy, C., Fire and Stone: The Science of Fortress Warfare 1660–1860 (London and Vancouver, 1975), 62. 154 The short space of time between the first and second phase should be regarded with some suspicion, it seems more than possible that al-ʿAzīz ʿUthmān’s initial idea was to build along the whole spur. If this was the case the second phase and first stages should be presented as one. It is possible that the construction was delayed due to the fear that the German Emperor and the Ayyubid Sultan, al-Malik al-Kāmil, would attack before the fortress was completed. I would like to thank Professor Ehud Netzer of the Department of Archaeology at the Hebrew University who suggested this idea in a conversation we had on methods of construction in the winter of 2005. 155 Barthoux, “Forteresse de Saladin,” 48. 156 Shapira, Y., “The tower in the Crusader kingdom of Jerusalem,” MA thesis, The Hebrew University of Jerusalem, 2000, 49 (Hebrew). 157 Ibn ʿAbd al-Ẓāhir, Muḥyī al-Dīn ʿAbd Allāh, Tashrīf al-ayyām wa’l-ʿuṣūr fī sīrat al-malik al-manṣūr, ed. M. Kāmil (Cairo, 1961), 255. 158 Barthoux, “Forteresse de Saladin,” 52. It is quite likely that the passage was sealed at an earlier period or that its construction was never completed. 159 Wiet, “Qalʿah Guindi,” 59. Barthoux, “Forteresse de Saladin,” 50–1. 160 This seems to be the main difficulty, since it hardly seems possible that garrisons stabled their horses outside the fortress. 161 Genequand, D., “Umayyad castles; the shift from Late Antique military architecture to early Islamic palatial buildings,” in Military Architecture in Greater Syria, ed. H. Kennedy (Leiden and Boston, 2006), 3–25. See especially pages 13–20. 162 Hartal, Al-Ṣubayba, 9–17. 163 Amitai, “Ayyubid Inscriptions,” 113–19. 164 The size of this gate tower is somewhat obscure, since there is no apparent correlation between the measurements given in the plan in Bagatti and Batistta’s publication and the measurements given in the text (20 × 15m). Battista and Bagatti, Monte Tabor, 46. As the tower no longer exists it is impossible to ascertain the true size of the construction. I decided to follow the plan, and the gate tower measurements of 12.5 × 17.5m. 165 A French team comprised of architects, an archaeologist and a historian conducted a comprehensive study of the gate and its development. Dangles,“Āğlūn,” 3–4. 166 Fleming, J., Honour, H. and Pevsner, N., Dictionary of Architecture (London, 1991), 278. 167 For similar examples of gate towers see Belvoir, Caesarea, Chastel Pelerin, Beaufort, Saone, Crac des Chevaliers, and Margat. 168 For a detailed description of this site see Ellenblum, Modern Histories, ch. 16. 169 The fortress of Vadum Iacob was never completed. It is in fact a building site, a fortress under construction. The fortress is being excavated by a team from the Hebrew University headed by Professor Ellenblum from the Department of Geography, Hebrew University of Jerusalem. 170 Barthoux, “Forteresse de Saladin,” 52. 171 Kennedy, Castles, 17–19; Edwards, Fortifications, 14. 172 Shapira, Tower. Chevedden gives an extensive survey emphasizing the increase in the size of towers during the twelfth and thirteenth centuries. Chevedden, Damascus, 277–80. See also Benvenisti, Crusaders, 288–9; Pringle, D., “Towers of Crusader Palestine,” Château Gaillard 16 (1994): 335–50. 173 Barthoux, “Forteresse de Saladin,” 52. 174 Tamari, “Inscriptions,” 263–5; Wiet, “Qalʿah Guindi,” 59. 175 Barbé et al., “Habonim,” 34–8; Benvenisti, Crusaders, 329–31; Pringle, Secular Buildings, 58, fig. 31. 176 Pringle, Secular Buildings, 72; Benvenisti, Crusaders, 326–7. 177 Benvenisti, Crusaders, 128. It seems that the walls and towers were partly rebuilt and repaired by Richard Coeur de Lion before they were destroyed again under the treaty signed in the summer of 1192 between the Franks and the Ayyubids.

222  Notes 178 The term manjanīq is used here as a generic name for a variety of machines used for hurling stones. 179 Barthoux, “Forteresse de Saladin,” 52. 180 Kennedy, Castles, pl. 36; Pringle, “Towers,” 339. 181 Pringle, “Towers,” 339. 182 At Qalʿat Ṣadr, the L-shaped tower at the northeastern corner was a later addition by Ṣalāḥ al-Dīn. 183 Mesqui, J., Châteaux forts et fortifications en France (Paris, 1997), 25. 184 Brown, R. A., English Castles (London, 1976), 10, 106. 185 Jones P. N. and Renn, D., “The military effectiveness of arrow loops: some experiments at White-Castle,” Château Gaillard 9–10 (1982): 445–56. 186 The date of construction is stated in an inscription found in situ in tower one. Amitai, “Ayyubid inscriptions,” 113–14. 187 This was first observed by Kennedy. Kennedy, Castles, 184. 188 Hartal, Al-Ṣubayba, 94–6. 189 Very few towers have survived and it is almost impossible to derive all the measurements directly from the archaeological remains. My own measurements are partly taken on the site and partly based on the plan published by Bagatti and Battista and drawn by Berlutzi. There is, however an earlier plan, dated 1881, drawn by the German surveyor Lendle, who was able to take measurements of what was left of the original gate tower. This was completely rebuilt in 1897 by the Franciscans. 190 Chevedden, Damascus, 277. 191 Ibid., 285–6. 192 Shapira, Tower, 13–14. 2  Mongolian siege warfare and the defense of Mamluk fortresses 1

Smith, J. M. Jr., “Mongol manpower and Persian population,” JESHO 18 (1975): 271���������������������������������������������������������������������������������� –99; ibid., “Mongol society and military in the Middle East: antecedents and adaptations,” in War and Society in the Eastern Mediterranean 7th–15th centuries, ed. Y. Lev (Leiden, NY and Cologne, 1997), 249–66; Allsen, T., Mongol Imperialism: The Policies of the Grand Qan Möngke in China, Russia and the Islamic Lands 1251–1259 (Berkeley, 1987), 189–93. For a comprehensive study on Mongol warfare see May, T., The Mongol Art of War (Yardley, PA, 2007), especially pp. 77–9. 2 Hsiao, C. The Military Establishment of the Yuan Dynasty (Cambridge, MA, 1978), 12. 3 The Secret History of the Mongols: A Mongolian Epic Chronicle of the Thirteenth Century, trans. and ed. with historical and philological commentary by I. de Rachewiltz (Leiden, 2004), vol. 1. 207; vol. 2. 1012–1015. Batu (d. 1255/6) was the son of Jochi, the eldest son of Chinggis Khan. 4 Smith, J.M. Jr., “Demographic considerations in Mongol siege warfare,” Archivum Ottomanicum 13 (1993–4): 332–3. 5 Sinor, D., “On Mongol strategy,” Inner Asia and its Contacts with Medieval Europe (London, Variorum 1977), XVI, 239–47. 6 Morgan, D. O., “The Mongol armies in Persia,” DI 56 (1979): 86–7. 7 Allsen, T.T., “The circulation of military technology in the Mongolian empire,” in Warfare in Inner Asian History, ed. N. Di Cosmo (Leiden, 2002), 265–93; Morgan, D., The Mongols (Oxford, 1986), 91; Martin, D. H., The Rise of Chingis Khan and his Conquest of North China (Baltimore, MD, 1950), 30–1. 8 Allsen, “Military technology,” 265–93. 9 Hsiao, Yuan Dynasty, 133, fn. 79. 10 Ibid., 133, fn. 79. This information was drawn from the Persian historian Jūzjanī, who wrote Tabakát-i-Násiri. 11 Allsen, “Military technology,” 278–9. The question of gunpowder and gunpowder makers is discussed below.

Notes  223 12 Rashiduddin Fazlullah, Jami‘u’t-tawarikh, 3 vols, ed. and trans. by W. M. Thackston (Cambridge, MA, 1998–9), vol. 2, 478. 13 Rossabi, M., Khubilai Khan (Berkeley, Los Angeles and London, 1988), 87; Carrington, L. Goodrich and Fềng Chia-shềng, “The early developments of firearms in China,” Isis 36 (1945–6): 118. Hsiao, Yuan Dynasty, 134. The source he cites is the Yuan-shih chishih pen-mo (‘Records of the history of the Yuan from the beginning to the end’), reprint (Peking: Chung-hua shu-chu, 1979), 4544–5. 14 The superior Myriarchy of Muslim Trebuchet Operators and Military Artisans (hui-hui p’ao-shou – chün-chiang shang wan-hu-fu). Farquhar, D. M., The Government of China under Mongolian Rule: A Reference Guide (Stuttgart, 1990), 259. 15 Martin, Rise of Chingis, 118–19; Fennell, J., The Crisis of Medieval Russia 1200–1304 (London and New York, 1983), 80. 16 Mission to Asia: Narratives and Letters of the Franciscan Missionaries in Mongolia and China in the Thirteenth Century, trans. by a nun of Stanbrook Abbey, ed. C. Dawson (New York, 1955), 37–8. 17 Dawson, Mission, 37. 18 Allsen, T. T., Culture and Conquest in Mongol Eurasia (Cambridge, 2001), 24. 19 Martinez, P. “Some notes on the Īl-Xānid army,” Archivum Eurasiae Medii Aevi 6 (1986): 146. 20 Allsen, Imperialism, 207. 21 Martinez, “Army,” 146–7. 22 Smith, “Mongol manpower,” 271–8. 23 Allsen, Imperialism, 203–206. 24 The term catapult is a generic name that includes all siege machines that do not use gunpowder. De Vries, K., Medieval Military Technology (Peterborough, Ontario, 1998), 127. 25 Needham, J., Science and Civilization in China, vol. 5, part 6: Military Technology: Missiles and Sieges (Cambridge, 1994), 213; Świętosławski, W., Arms and Armour of the Nomads of the Great Steppe in the Times of the Mongol Expansion (12th–14th Centuries) (Łodź, 1999), 70. 26 Franke, H., “Siege and defense of towns in medieval China,” in Chinese Ways in Warfare, eds. F. A. Kierman and J. K. Fairbank (Cambridge, MA, 1974), 169, 195. 27 An illustration of the counterweight trebuchet can be found in the Jami‘u’t-Tawarikh, Tabriz, Iran 1306–14, currently in ms. Edinburgh University, 20. A copy can be seen in Nicolle, D., Arms and Armour of the Crusading Era 1050–1350: Islam, Eastern Europe and Asia, 2nd edn (London and Mechanicsburg, PA, 1999), 242, fig. 626m. 28 Allsen, “Military technology,” 270; Yuan-shih chi-shih pen-mo (‘Records of the history of the Yuan from the beginning to the end’), reprint (Peking:Chung-hua shu-chu,1979), 4544–5. In actual fact only the Yuan-shi tells about the two Muslim engineers sent to the court of Qubilai Khan. They were both received with great honor and each has his biography written in the official history of the Yuan dynasty. 29 Raphael, K., “The al-Mansuri’s mangonel stone balls: some new evidence from the Mamluk siege of Akko (1291),” in ʿAkko, the Excavations of 1991–1998 II, The Later Periods, eds E. Stern and D. Syon (IAA Reports), forthcoming. 30 Świętosławski, Arms and Armour, 70–1. 31 Franke, “Siege and defense,” 169. 32 Aconite is the plant from which aconitine, a particularly potent poison, is extracted.The Merck Index, 9th edn, 15, 113, defines “Aconitine” as coming from “Aconitum Napellus L., Ranunculaceae and other aconites.” 33 Franke, “Siege and defense,” 166. 34 Braudel, F., The Structures of Everyday Life (London, 1981), 272–3; Knapp, R. G., The Chinese House (Oxford, 1994), 46–7. 35 Franke, “Siege and defense,” 159. 36 Knapp, R. G., China’s Walled Cities (Oxford, 2000), 25, 33.

224  Notes 37 Sen-Dou Chang, “The morphology of walled capitals,” in The City in Late Imperial China, ed. G. W. Skinner (Stanford, 1977), 77. 38 Franke, H., Krieg und Krieger in Chinesischen Mittelalter (12.bis 14. Jahrhundert), Band. 81 (Stuttgart, 2003), 212–13. Although there is evidence of wooden city towers, wood was on the whole not a common building material in fortifications in China. Farmer, E. L., “The hierarchy of Ming city walls,” in City Walls: The Urban Enceinte in Global Perspective, ed. J. D. Tracy (Cambridge, 2000), 485. 39 ʿAlā-al-Dīn Atā-Malik Juvainī, History of the World-Conqueror. 2 vols, trans. J. A. Boyle (Cambridge, MA, 1958), vol. 1, 106. 40 Boas, A. J., Crusader Archaeology (London, 1999), 218–219. 41 Franke, “Siege and defense,” 193; Knapp, Walled Cities, 15–26. See especially p. 20. 42 Knapp, Walled Cities, 4. 43 Di Cosmo, N., Ancient China and its Enemies: The Rise of Nomadic Power in East Asian History (Cambridge, 2002), 138–40. 44 Knapp, Walled Cities, 6. 45 Parker, G., “The artillery fortress as an engine of European overseas expansion, 1480– 1750,” in City Walls: The Urban Enceinte in Global Perspective, ed. J. D. Tracy (Cambridge, 2000), 388, n. 6. 46 Goodrich and Chia-shềng, “Firearms,” 114–23; Allsen, “Military technology,” 275. Khan, I. A., “The role of the Mongols in the introduction of gunpowder and firearms in South Asia,” Gunpowder: The History of an International Technology, ed. J. Buchanan (Bath, 1996), 33–44; D. H. Martin,”The Mongol army,” JRAS (1943), 67; De Vries, Military Technology, 143; Chase, K., Firearms: A Global History (Cambridge, 2003), 58; Needham, J. Science and Civilization in China, vol.5, part 7: The Gunpowder Epic (Cambridge, 1986), 117, 163. 47 Needham, J. Science in Traditional China: A Comparative Perspective (Hong Kong, 1981), 39. 48 Needham, Gunpowder Epic, 325, n. F. 49 Ibid., 573, n. C. 50 According to Martin gunpowder was used during the siege of Kaifeng in 1233–4. Martin, “Mongol army,” 67. 51 Khan, ”Firearms,” 33–44. 52 Allsen, “Military technology,” 275, 279. Allsen relies on Yuanshi, ch. 98, p. 2514., and on Hsiao,Yuan Dynasty, 80. 53 Franke, Krieg und Krieger, 213; May, Art of War, 141. 54 Ayalon, D., Gunpowder and Fire Arms in the Mamluk Kingdom, 2nd edn (London, 1978), 9–44. 55 Juwaynī, World-Conqueror (Boyle), vol. 1, 7. 56 Juwaynī, ‘Alā’ al-Dīn ‘Aṭā Malik, Ta‘rīkh –i jahān-gushā, ed. M. M. Qazvini (Leiden, 1912–37), 96. 57 Rashīd al-Dīn Faḍl Allāh Abū ‘l-Khayr, Jāmi‘ al-tawārīkh, ed. B. Karimi (Teheran, 1959), vol. 2, 686. 58 I would like to thank Prof. Shaul Shaked of the Hebrew University of Jerusalem for translating and analyzing the Persian sources and Prof. Michal Biran of the Hebrew University of Jerusalem Department of Asian Studies and Middle Eastern History for her most valuable help in translating and analyzing of the Chinese sources and solving this enigma. 59 Yuanshi, Beijing: Zhonghua shuju, 1978, ch. 98, p. 2514. 60 Needham, Science, vol. 5, part 6, 198. 61 Hong, Weapons, 260–7. 62 Willey, P., The Castles of the Assassins (London, 1963), 166. 63 Juvainī, World-Conqueror, vol. 2, 630–1. 64 Manually operated siege machines were employed alongside counterweight siege machines. 65 France, J. Western Warfare in the Age of the Crusades (Ithaca, New York, 1999), 123.

Notes  225 66 The largest number of nafṭ vessels were found during the Franciscan excavations at Mt Tabor. Picherilo, M., “A survey of the fortifications at Mt. Tabor,” Qardom 20 (1982): 59–64. [Hebrew]. They have been dated to the Frankish siege of 1217. I would like to thank Dr G. Avni and J. Zeligman for the information on nafṭ vessels found in their excavations at Bait Shean and Ramla (yet to be published). 67 Ibn al -Athīr, Kāmil, vol. 12, 393. 68 Juwaynī, World–Conqueror, vol. 1: 176–7. This point has been emphasized by Smith: “Mongol warfare was distinguished not so much by its skill and aptitude as by its scale and persistence.” Smith, “Mongol manpower,” 272. 69 D’Ohsson, C., Histoire des Mongols (Amsterdam, 1834) (reprinted in Tientsin China, 1940), vol. 1: 289; Martin, Rise of Chingis Khan, 31; May, T., “Genghis Khan’s secrets of success,” Military History (July/Aug., 2007), 44. D’Ohsson and Martin do not reveal their source. Other than Juwaynī, I could not find a primary source that describes this siege. 70 Martinez, “Army,” 148–9, 99–116. 71 Rashīd al-Dīn, Jāmī‘ al-tawārīkh (Thackston), vol. 2, 507–8, in the cases of Mayyāfarqīn and Mārdīn. 72 Ibid., vol. 2, 484. The Ismā‘īli fortresses of Maymundiz and Gird Kūh are located in the region south of the Caspian Sea. 73 Ibid., vol. 2, 507–8. 74 Ibid., vol. 2, 481. 75 Järgä was a hunting method in which riders surrounded an animal and slowly decreased the diameter of the circle until the animal was caught and killed. Ibid., vol. 3, 768. 76 Waṣṣāf. B. Spuler, A History of the Mongols Based on Eastern and Western Accounts of the Thirteenth and Fourteenth Centuries, trans. from German by H. and S. Drummond (London, 1972); Geschichte Wassafs, ed. in Persian and trans. from German by Josef von Hammer-Purgstall (Vienna, 1856), vol. 1, 68–75. 77 Ibid., 68–75. 78 Rashīd al-Dīn, Jāmī‘ al-tawārīkh (Thackston), vol. 2, 503; Humphreys, Saladin, 349. 79 Rashīd al-Dīn, Jāmī‘ al-tawārīkh (Thackston), vol. 2, 500–501. 80 Juvainī, Ta‘rīkh -i jahān-gushā (‘History of the world-conqueror’), translated and cited by Boyle, J. A. in “The death of the last Abbasid Caliph: a contemporary Muslim account,” JSS, 6, 2 (1961): 160; The Travel of Ibn Baṭṭūta a.d. 1325–1354, trans. H. A. R. Gibb (Cambridge, 1962), vol. 2, 326. Samarqand’s recovery took almost a century and a half. It revived in the last quarter of the 14th century. Crowe, Y., “Samarkand,” EI2 8: 1033. Aleppo, taken in January 1260 recovered only a century later. Humphreys, Saladin, 349. 81 While Baghdad was of great importance due to its place in the Muslim world and the fact that it was the seat of the Caliph, its geographical location was of considerable importance because of its winter pasture that served the Īlkhāid army. Thus, the city was restored at the order of the Mongols. The fortifications of the Assassins were of no use to the Mongols and were left in ruins or destroyed immediately after the siege. 82 Rashīd al-Dīn, Jāmī‘ al-tawārīkh (Thackston), vol. 2, 507–9; Patton, D., Badr al-Din Lu’lu’, Atabeg of Mosul, 1211–1259 (Seattle, 1991), 63. 83 Hama was ruled by an Ayyubid prince but had a Persian vicegerent nominated by Hülegü. Humphreys, Saladin, 350. 84 Chase, Firearms, 12. 85 Smail, R. C., Crusading Warfare (1097–1193), 2nd edn, (Cambridge, 1995), 215. 86 Ellenblum, R., “Frankish Castles in the Crusader Kingdom of Jerusalem: the third generation,” in Le Concile de Clermont et la premiére Croisade, ed. M. Balard (Paris, 1996), 522–3. 87 A detailed description of the fortress is given in Chapter 4. 88 Thorau, P., The Lion of Egypt, trans. P.M. Holt (London & New York, 1992), 64; Humphreys, Saladin, 348.

226  Notes 89 Amitai-Preiss, R., Mongols and Mamluks: The Mamluk-Īlkhānid War 1260–1281 (Cambridge, 1995), 60–1, 203–4. The following Mongol sieges have been described in great detail by Thorau and Amitai. 90 Ibn ʿAbd al-Ẓāhir, Muḥyī al-Dīn ʿAbd Allāh, al-Rawḍ al-zāhir fī Sīrat al-Malik al-ẓāhir, ed. ʿA. ʿA. al-Khuwayṭir (Riyad, 1976), 225; Baybars al-Manṣūrī al-Dawādār, Rukn al-Dīn, Mukhtār al-Akhbar ta’rīkh al-dawla al-ayūbiyya wa-dawlat al-mamalīk al-baḥriyya ḥattā 652 h, ed. A. S. Ḥamdān (Cairo, 1993), 29; al-Yūnīnī, Quṭb al-Dīn Mūsā b. Muḥammad, Dhayl mir’āt al-zamān fī ta’rīkh al-aʿyān (Hyderabad, 1954–61), vol. 2, 318. 91 Ibn ʿAbd al-Ẓāhir, al-Rawḍ, 226–8. 92 Der Nersessian, S., “The Armenian chronicle of the Constable Sampad or of the “Royal Historian,” DOP 13 (1959), 163. 93 Ibid., 112. 94 Thorau, Baybars, 223; Amitai-Preiss, Mongols, 128–31. 95 Ibn ʿAbd al-Ẓāhir, Rawḍ, 405–8; A short version is found in Baybars al-Manṣūrī, Ta’rīkh, 55. Waṣṣāf says the Mongol force was twice the size of the Mamluk army. Thorau, Baybars, 223. 96 Yūnīnī, Dhayl, vol. 3, 114; Ibn Shaddād, Ta’rīkh, 124–6; Ibn al-Furāt, Ta’rīkh, vol. 7, 41. Although Baybars was residing at Damascus when he received the news (November 1275) of the Mongol attack he did not manage to get to the fortress. 97 Yūnīnī, Dhayl, vol. 3, 115;Thorau, Baybars, 236; Amitai-Preiss, Mongols, 136–7. 98 Yūnīnī, Dhayl, vol. 4, 457. 99 Ibn ʿAbd al-Ẓāhir, Tashrīf, 76–7; Yūnīnī, Dhayl, vol. 4, 45–6; Northrup, L. S., From Slave to Sultan: The Career of Al-Manṣūr Qalāwūn and the Consolidation of Mamluk Rule in Egypt and Syria (678–689 a.h./1279–1290 a.d.) (Wiesbaden, 1998), 101–2. 100 Yūnīnī, Dhayl, vol. 4, 90–1. 101 Rashīd al-Dīn, Jāmī‘ al-tawārīkh (Thackston), vol. 3, 544–5; Boyle, J. A., The Cambridge History of Iran (Cambridge, 1968), vol. 5, 363; Amitai-Preiss, Mongols, 184–97. 102 Rashīd al-Dīn, Jāmī‘ al-tawārīkh (Thackston), vol. 3, 644–5. 103 Al-Nuwayrī, Shhab al-Din Aḥmad b. ʿAbd al-Wahhāb, Nihāyat al-arab fī funūn al-adab (Cairo, 1992), vol. 31, 384–88; Amitai, R., “Whither the Īlkhānid army ? Ghazan’s first campaign into Syria (1299–1300),” in Warfare in Inner Asian History (500–1800), ed. N. Di Cosmo (Leiden, 2002), 221–64. 104 Al-Yūnīnī’s Dhayl Mir’āt al-zamān, in Guo, L., Early Mamluk Syrian Historiography (Leiden, 1998), 151–7; Maqrīzī, Sulūk,. vol. 1 pt. 3, 882–902; Amitai, R., “The Mongol occupation of Damascus in 1300: a study of Mamluk loyalties,” in The Mamluks in Egyptian and Syrian Politics and Society, eds. M. Winter and A. Levanoni (Leiden and Boston, 2004), 21–41. 105 Maqrīzī, Sulūk, vol. 1 pt. 3, 906–8; Nuwayrī, Nihāyat, vol. 31, 413–4; Yūnīnī, Dhayl (Guo), vol. 1, 108; Rashīd al-Dīn, Jāmī‘ al-tawārīkh (Thackston), vol. 3, 641; Abū’lFidā, The Memoirs of a Syrian Prince: Abū’l-Fidā, Sultan of Hamā 672–732 (1273–1331), trans. P.M. Holt (Wiesbaden, 1983), 38; Ibn Taghrībirdī, Nujūm, vol. 32, 24–5; Maqrīzī, Sulūk, vol. 1 pt. 3, 928–33. It took the army 48 days to march from al-Ḥila to al-Bīra. 106 Rashīd al-Dīn, Jāmī‘ al-tawārīkh (Thackston), vol. 3, 655; Nuwayrī, Nihāyat, vol. 32, 24–5. 107 Biran, M., Qaidu, 1–3; Morgan, D., Medieval Persia 1040–1797 (London and New York, 1986), 63; Boyle, Iran, 393. 108 Rashīd al-Dīn, Jāmī‘ al-tawārīkh (Thackston), vol. 3, 656; Nuwayrī, Nihāyat, vol. 32: 28–30. 109 Boyle, Iran, 403. 110 Abū’l-Fidā’, Syrian Prince (Holt), 62–3. 111 Ibid., 62; Al-Maqrīzī, Taqī al-Dīn Ahmad b. ʿAli, Kitāb al-sulūk li-ma’rifat duwal al-mulūk, ed. M. M. Ziyāda, and S. ‘A-F. Āshūr (Cairo, 1934–73), vol. 2, pt. 1, 115. 112 Abu’l-Fidā’, Syrian Prince (Holt), 62; Boyle, Iran, 403; Amitai-Preiss, R., “New material from the Mamluk sources for the biography of Rashīd al-Din,” in The Court of the Il-khans 1290–1340, eds. J. Raby and T. Fitzherbert (Oxford, 1996), Oxford Studies in Islamic Art XII, 29–31.

Notes  227 113 Ibn al-Qalanisi, The Damascus Chronicle of the Crusades, extracted and translated from the Chronicle of Ibn al-Qalanisi, trans. H. A. R. Gibb (London, 1932), 56–7. 114 Ibid., 78–9. 115 Amitai-Preiss, Mongols, 202–5; on Mongol imperial ideology see Amitai-Preiss, R., “Mongol imperial ideology and the Ilkhanid war against the Mamluks,” in The Mongol Empire and its Legacy, eds R. Amitai-Preiss and D. Morgan (Leiden, 1999), 57–72. 116 A brief hint of a change of policy can be found in the reign of the Mamluk sultan al-Ashraf Khalīl (689/1290–693/1293); see Chapter 4. 117 Abū ‘l-Fidā, Syrian Prince (Holt), 36. 118 ʿUmar Ibn Ibrāhīm al-Awasī al-Anṣarī, A Muslim Manual of War, Tafrīj al-kurūb fī tadbīr al-ḥurūb, ed. and trans. G. T. Scanlon (Cairo, 1961), 43–4. 119 Ibid., 120–2. 120 Ibid., 25. 121 Ibid., 27. 122 Ibid., 25. 123 Ibid., 95 (Arabic section). 124 Ibid., 120–1. 125 Ayalon, D., “Egypt as a dominant factor in Syria and Palestine during the Islamic period,” in Outsiders in the Lands of Islam: Mamluks, Mongols and Eunuchs (London, Variorum, 1988), II, 35. 126 Ayalon, D., “The Muslim city and the Mamluk military aristocracy,” in Studies on Mamluks of Egypt (1250–1517) (London, Variorum, 1977), VII, 319–29. 127 Thorau, Baybars, 138–9; Ayalon, D., “Le régiment Baḥriyya dans l’armée Mamelouke,” in Studies on Mamluks of Egypt (1250–1517), (London, Variorum, 1977), III, 133–41. 128 Ibn al-Furāt, Ta’rīkh (Lyons), 96–95; ibid, “al-Baḥriyya,” EI2 1: 944–5. 129 Amitai, R. “Foot soldiers, militiamen and volunteers in the early Mamluk army,” Texts, Documents and Artifacts, ed. C. F. Robinson (Leiden, 2003), 246; Ibn ʿAbd al-Ẓāhir, Tashrīf, 80–1; Northrup, Qalāwūn, 132; for a detailed explanation of the term jarjya see Chapter 3, note 110. 130 Ayalon, D., “The auxiliary forces of the Mamluk Sultanate,” DI 65 (1988): 15–16; Benvenisti, Crusaders, 199. 131 Amitai, “Foot Soldiers,” 239–40. 132 Al-Yūnīnī, Dhayl (Guo), vol. 1, 103. 133 Abu’l-Fidā’, Syrian Prince (Holt), 63. 134 Kennedy, Castles, 164. 135 Abu’l-Fidā’, Syrian Prince (Holt), 63. 136 Al-Yūnīnī, Dhayl (Guo), vol. 1, 166. 137 Ibid., vol. 1, 170-171. 138 Leonard, R. A. A Short Guide to Clausewitz, On War (London, 1967), 30. 139 Verbruggen, J. F. The Art of Warfare in Western Europe during the Middle Ages, (Woodbridge, 1997), 119. 140 Ibid., 108. 141 Paterson W. F., “The Archers of Islam,” Journal of the Economic and Social History of the Orient 9 (1996): 83; Foley, V., Palmer, G. and Soedel W. “The Crossbow,” Scientific American 252 (January, 1985): 85. 142 Latham, J. D. and Paterson, W. F., Saracen Archery, An English Version and Exposition of a Mameluke Work on Archery (ca. A.D. 1368). (London, 1970), xxxv, 137. 143 This will be further discussed and demonstrated in the following chapters. 144 Rashīd al-Dīn, Jami‘u’t-Tawarikh (Thackston), II, 507. 145 Chevedden, Damascus, 8-9; Chevedden, “Development,” 33-43. 146 Ambraseys, N and Melville, C. P. “An analysis of the Eastern Mediterranean earthquake of 20 May 1202,” in Historical Seismograms and Earthquakes of the World, eds. W. H. K. Lee, H. Meyers and K. Shimazaki (London, 1998), 181-200.

228  Notes 3  Laying the foundations 1 2 3

4

5 6

7 8 9

10 11 12 13 14 15 16 17 18 19 20 21

Morgan, D., The Mongols (Oxford, 1986), 84–6; Martin, D. H. “The Mongol army,” JRAS (1943): 46–85. Smith, J. M. Jr. “Ayn Jālūt: Mamluk success or Mongol failure?,” HJAS 44 (1984): 307–45, 314. Barīd, the official Mamluk postal service reorganized by Baybars. During the thirteenth century it was used primarily for the Sultanate’s political and military needs. In later centuries it also served commercial traffic. Sourdel, D., “Barīd,” EI2 1: 1045–6; Silverstein, A. J., Postal Systems in the Pre-Modern Islamic World (Cambridge, 2007), 165–85. Ayalon, D., “Studies in the structure of the Mamluk army – I: The army stationed in Egypt,” BSOAS 15 (1953): 203–28. [Rpt. in Studies on the Mamluks of Egypt (1250– 1517), London, Variorum,1977, no. I]; ibid, “Studies in the structure of the Mamluk Army – II: The Ḥalk. a,” BSOAS 15 (1953): 448–76. [Rpt. in Studies on the Mamluks of Egypt (1250–1517), London, Variorum,1977, no. I]. The discussion here and throughout the study does not deal with the Sultanate’s southern border, i.e. the region of Sudan or the western desert, today the modern border between Egypt and Libya. Abū’l-Fidā’, The Memoir of a Syrian Prince Abū’l-Fidā, Sultan of Hamā 672–732 (1273– 1331), trans. P. M. Holt (Wiesbaden, 1983), 29. See Map 3.5 showing the fortresses in Armenian territory that were conquered by the Mamluks, and shuttled back and forth between the Mamluks and the Armenians. Al-ʿAynī, Badr al-Dīn Maḥmūd b. ʿAlī, ʿIqd al-jumān fī ta’rīkh ahl al-zamān (Cairo, 1992), vol. 4, 300–1. Abū’l-Fidā’, al-Mukhtasar fī ta’rīkh al-bashar (Istanbul 1869–70), vol. 2, pt. 4, 122–123. A comprehensive discussion on the difficulties concerning the definition of borders in the eleventh to thirteenth centuries in the Crusader Kingdom of Jerusalem can be found in Ellenblum, R., Crusader Castles and Modern Histories (Cambridge, 2007), ch. 9: “Borders, Frontiers and Centers,” 118–45. Mission to Asia: Narratives and Letters of the Franciscan Missionaries in Mongolia and China in the Thirteenth Century, trans. by a nun of Stanbrook Abbey, ed. C. Dawson (New York, 1955), 35–36. Lewis, C. G., “A survey of the Euphrates valley from Deir-Ez-Zor to Aleppo,” Geographical Journal 59 (1922): 453–7. Brawer, M., Atlas Carta of the Middle East (Jerusalem, 1990), 105–6 [Hebrew]. Eddé, A. M., La principauté ayyoubide d’Alep (Stuttgart, 1999), fig. 64; Historical Atlas of Islam, 2nd rev. edn, ed. H. Kennedy (Leiden, 2002), 22–3. The details are based on a reading of four topographical maps of the region of the upper Euphrates. The maps were drawn by the Russian General Staff in 1983, scale 1:500,000. The distances are measured according to aerial lines. Thus they are approximations. Somogyi, J., “Ada-Dhahabis record of the destruction of Damascus by the Mongols in 699–700/ 1299–1301,” in Ignace Goldziher Memorial Volume, eds. S. Lowinger and J. Somogyi (Budapest, 1948), pt. 1, 353–86. Amitai-Preiss, R., Mongols and Mamluks: The Mamluk-Īlkhānid War 1260–1281 (Cambridge, 1995), 202. Edwards, R. W., The Fortifications of Armenian Cilicia (Washington, 1986), 9. Sinclair, T. A., Eastern Turkey: An Architectural and Archaeological Survey (London, 1990), vol. 4, 229. Der Nersessian, S., “The Kingdom of Cilician Armenia,” in A History of the Crusades, gen. ed. K. M. Setton, II: The Later Crusades (Madison, 1969), 635. A short description of the mountain passages is given in Har-El, S., Struggle for Domination in the Middle East: The Ottoman Mamluk War 1485–91 (Leiden, New York and Cologne, 1995), 52–4. The topographic description is based on a map prepared by the French

Notes  229 22 23 24 25 26 27 28 29 30 31 32 33

34 35 36 37

38

39 40 41 42 43 44 45

army in 1942, Syrie et Liban, Carte dressée et publiée par le Service Geographique, des Forces Françaises Libres au Levant, 1:500,000 (1942). Amitai, R., “The conquest of Arsuf by Baybars: political and military aspects,” MSR 9/1 (2005): 61–83. Riley-Smith, J., The Atlas of the Crusades (New York and Oxford, 1991), 115; Barag, D., “A new source concerning the ultimate borders of the Latin Kingdom of Jerusalem,” IEJ (1970): 29, 197–217; Ellenblum, Modern Histories, 177–86. Marshall, C., Warfare in the Latin East, 1191–1291 (Cambridge, 1996), 93, 242. Amitai, R., “Edward of England and Abagha Ilkhan,” in Tolerance and Intolerance, eds M. Gervers and J. M. Powell (Syracuse, 2004), 75–80. Amitai-Preiss, R., “The aftermath of ʿAyn Jālūt: the beginning of the Mamlūk-Ilkhānid cold war,” Al-Masaq 3 (1990): 9–10, n. 48. Nicolle, D., Arms and Armour of the Crusading Era 1050–1350: Islam, Eastern Europe and Asia, 2nd edn (London and Mechanicsburg, PA, 1999), 67–9. Edwards, Fortifications, 67–9, 37–50. Molin, K., Unknown Crusader Castles (London and New York, 2001), 152–3. Lewis, B., The Assassins (New York, 1968), 116–117. Melville, C., “Sometimes by the sword, sometimes by the dagger: the role of the Ismaʿilis in Mamluk-Mongol relations in the 8th/14th century,” in Medieval Ismaʿili History and Thought, ed. F. Daftary (Cambridge, 1996), 247–63. Ibn Baṭṭūta, Rihlat Ibn Baṭṭūta, ed. Talal Harb (Beirut, 1992), 95. Ibn al-Furāt, Ta’rīkh, MS, Vienna, fol. 73b, cited in Ayalon, D., “The Mamluks and naval power: a phase of the struggle between Islam and Christian Europe,” in Proceedings of the Israel Academy of Sciences and Humanities 1/8 (1967): 1–12. [Rpt. in Studies on the Mamluks of Egypt (1250–1517), London, Variorum, 1977, no. VI], 12; al-Maqrīzī, Kitāb al-Sulūk li-ma’rifat duwal al-mulūk (Beirut, 1997), vol. 1, 531. Anonymous author, De constructione castri Saphet, cited in Kennedy, H., Crusader Castles (Cambridge, 1994), appendix, 196. Lane, E. W., Arabic–English Lexicon (London, 1984) bk. 1, pt. 1: 586–7; Dozy, R., Supplément aux Dictionnaires Arabes, vol. 1, 297; vol. 2, 404; Wehr, H., A Dictionary of Modern Written Arabic, Arabic–English, ed. J. M. Cowan (London and Beirut), 183, 787. Ibn Shaddād al-Ḥalabī, ʿIzz al-Dīn Muḥammad b. ʿAli. Al-Aʿlāq al-khaṭīra fī dhikr umarā al-shām wa’l-jazīra: Ta’rikh ḥalab, ed. Y. Z. ʿAbārh, (Damascus, 1991), vol. 1, pt. 2, 135–9. This notion concerning the timing of reconstruction was borrowed from HowardJohnston, J. D., “Procopius: Roman defences north of the Taurus and the new fortress of Citharizon,” in The Eastern Frontier in the Roman Empire, Proceedings of the Colloquium held at Ankara in September 1988, eds D. H. French and C. S. Lightfoot (1989), BAR International Series 553, vol. 1, 219. Boyle, J. A. (ed.), The Cambridge History of Iran (Cambridge, 1968), vol. 5, 403; Abū’lFidā’, Syrian Prince (Holt), 63; Amitai-Preiss, R., “New material from the Mamluk sources for the biography of Rashīd al-Din,” in The Court of the Il-khans 1290–1340, eds. J. Raby and T. Fitzherbert (Oxford, 1996), Oxford Studies in Islamic Art XII,” 29–31. Ibn ʿAbd al-Ẓāhir, Rawḍ, 93. The fortress of Shumaimis is approximately 25km southeast of Hama. The Arabic word used in the source is nāṣafa ( ), the literal meaning of which is to share equally, go halves. Wehr, Dictionary, 971. Ibn al-Furāt, Ayyubids, Mamluks and Crusaders: Selections from the Tārīkh al-Duwal wa’lMulūk of Ibn al-Furāt, 2 vols, trans. Lyons, U. and M. C. (Cambridge, 1971), 166. Ibn al-Furāt, Ta’rīkh (Lyons), 68. A detailed description of the Mongol invasion routes is given in Chapter 2. This site has never been identified. Bar Hebraeus, The chronography of Grighor Abū‘l Faraj, the son of Aaron, the Hebrew physician commonly known as Bar Hebraeus. Being the first part of his political history of the world, ed. and trans. from the Syriac by Budge, E. A.W. (Amsterdam, 1976), vol. 1, 435.

230  Notes 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

65 66 67 68 69 70 71 72 73 74 75 76

Boyle, Iran, 388–389. Boyle, Iran, 393. Ibid., 403. Lane, Lexicon, book 1, pt.1: 961–2; Wehr, Dictionary, 356–7; Ibn Shaddād, Aʿlāq, vol. 1, pt. 1, 89. Ibn Shaddād, Aʿlāq, vol. 1, pt .1, 88–9. Ibid., vol. 1, pt. 1, 89. Abū’l-Fidā’, Syrian Prince (Holt), 18; Sauvaget, J., “Ḥalab” EI2, 3: 85–9; Nicola, Z., Urban Life in Syria (Beirut, 1953), 85; Lapidus, I. M., Muslim Cities in the Later Middle Ages (Cambridge, 1988), 14–15. Irwin, R., “The supply of money and the direction of trade in thirteenth century Syria,” in Coinage in the Latin East: The Fourth Oxford Symposium and Monetary History, eds P. W. Edbury, and D. M. Metcalf, BAR International Series, 77 (1980), 73–5. Abū’l-Fidā’, Syrian Prince (Holt), 18–19. Qalqashandī, Shihāb al-Dīn Aḥmad, Ṣubḥ al-aʿshā fī ṣināʿat al-inshā’ (Beirut, 1987), vol. 4, 143; Tonghini, C., Qalʿat Jabar Pottery: A Study of a Syrian Fortified Site of the Late 11th-14th Centuries (Oxford, 1998), 22. Ibn Shaddād, al-Aʿlaq,143; Amitai, R., “An Arabic inscription at al-Ṣubayba (Qalʿat Namrud) from the reign of sultan Baybars,” in M. Hartal, The Al-Subayba (Nimrod) Fortress, Towers 11 and 9 (Jerusalem, 2001), 113. Ellenblum, “Ṣubayba,” 103–12. Jackson, P., “The crisis in the Holy Land in 1260,” English Historical Review 95 (July 1980): 502, n. 5. Ibn ʿAbd al-Ẓāhir, Rawḍ, 162–4. Ibid., 164. Ibid., 93. Ibn Wāṣil, Jamāl al-Dīn Muḥammad b. Sālim, Mufarrij al-kurūb fī akhbār banī ayyūb (Alexandria, 1953), vol. 3, 215–16. Crusader Syria in the Thirteenth Century the Rothelin Continuation of the History of William of Tyre with part of Eracles or Acre text, Crusader Texts in Translation, trans. by J. Shirley (Brookfield Vermont, 1999), 140, n. 6. Humphreys, R. S., From Saladin to the Mongols: The Ayyubids of Damascus 1193–1260 (Albany, 1977), 136–7, 159; Ibn Wāṣil, Mufarrij, vol. 3, 215–16. For a detailed discussion concerning the building of Mount Tabor and the final decision to destroy it see ch.1 in Burchard of Mount Sion, A Description of the Holy Land, trans. from Latin by A. Stewart (London, 1896), PPTS, vol. 12, 43; Pringle, Churches, 68. Ibn al-Furāt, Ta’rīkh (Lyons), 66. Irwin, R., “The Mamluk conquest of the County of Tripoli,” in Crusade and Settlement, ed. P. W. Edbury (Cardiff, 1985), 249. Ayalon, D., “The Mamluks and naval power: a phase of the struggle between Islam and Christian Europe,” in Studies on the Mamluks of Egypt (1250–1517) (London, Variorum, 1977), VI, 5. Ayalon, “Naval power”, 6–9. Ibn al-Furāt, Ta’rīkh (Lyons), 72. Ibid., 72. Ibid., 101. Drory, J., “Founding a new Mamlaka: some remarks concerning Safed and the organization of the region in the Mamluk period,” in The Mamluks in Egyptian and Syrian Politics and Society, eds. M. Winter and A. Levanoni (Leiden and Boston, 2004), 163–87. Johns, “ʿAjlun,” 30. Ibn ʿAbd al-Ẓāhir, Rawḍ, 297–9; Maqrīzī, Sulūk, vol. 1, pt. 2, 565–6. Bar Hebraeus, Chronography, vol. 1, 448; Ibn ʿAbd al-Ẓāhir, Rawḍ, 307–13; Maqrīzī, Sulūk, vol. 1, pt. 2, 568–9. Maqrīzī does not give the exact number of fortresses or their names.

Notes  231 77 78 79 80 81

Maqrīzī, Sulūk, vol. 1, pt. 2, 568. Ibid., vol. 1, pt. 2, 568. Ibn al-Furāt, Ta’rīkh (Lyons), 126. Ibn Shaddād, Aʿlāq, vol. 1, pt. 2, 71. Major, B. “Medieval cave fortifications of the Upper Orontes Valley (a preliminary report),” in Military Architecture in Greater Syria, ed. H. Kennedy (Leiden and Boston, 2006), 256–7. 82 Kennedy, Castles, 84. 83 Ellenblum, Modern Histories, chs 15–16. 84 Chevedden, P. E., “Fortifications and the development of defensive planning during the Crusader period,” in Circle of War in The Middle Ages: Essays on Medieval Military and Naval History, eds. D. J. Kagay, and L. J. Andrew Villalon (Woodbridge, UK, 1999), 33–43. 85 Raphael, K., “Archers during the Crusader period,” MA thesis. The Hebrew University of Jerusalem (2002) [Hebrew]. 86 Ibn Shaddād, Aʿlāq, vol. 1, pt. 2, 417; Maqrīzī, Sulūk, vol. 1, pt. 2, 525. 87 Ibn ʿAbd al-Ẓāhir, Rawḍ, 375–7; Barber, Knighthood, 79, n. 56. 88 Ibn ʿAbd al-Ẓāhir, Rawḍ, 379–80. 89 Barber, Knighthood, 158–160. 90 Ibid., 385–6. 91 In the two defeats of the battles of La Forbie (1244) and Mansura (1250) the Franks lost a large number of men. It seems they could not rebuild their army without considerable help from Europe. According to Marshall there was a strategic change during the mid thirteenth century and the Franks clearly preferred defending themselves from within their fortresses to heading out into the open battlefield. Marshall, Warfare, 182. 92 A description of naft containers is given in Chapter 2. 93 Amitai, “Arsuf,” 85–101. Raphael, K. and Tepper, Y. “The archaeological evidence from the Mamluk siege of Arsūf,” MSR 9/1 (2005): 85–100. 94 Ibn Baṭṭūṭa, Riḥla, 95. 95 Al-ʿUmarī, Aḥmad b. Yaḥyā ibn Faḍl Allāh, al-Taʿrīf fī al-muṣṭalaḥ al-sharīf (Cairo, 1894), 196. 96 Maqrīzī, Sulūk, vol. 1, pt. 2, 564. 97 Ragette, F., Baalbek (Princeton, NJ, 1980), 76–8; Chevedden, P. E., The Citadel of Damascus. PhD diss., University of Los Angeles, California, 1986, unpublished, 48–81. 98 See Chapter 2, nn. 98–99. 99 Abū‘l-Fidā’, al-Mukhtaṣar fī ta’rīkh al-bashar, eds. M. Zinhum and Y. al-Saʿīd (Cairo, 1999), vol. 4, 57; Abū‘l-Fidā, Syrian Prince (Holt), 37. 100 Boase, T. S. R. “The History of the Kingdom,” in The Cilician Kingdom of Armenia, ed. T. S. R. Boase (Edinburgh and London, 1978), 32–3. 101 Abū‘l-Fidā, Syrian Prince (Holt), 18; Lapidus, Muslim Cities, 14–15; Nicola, Urban Life, 85; Sauvaget, J. “Aleppo,” EI2, 3, 87–8. 102 Chevedden, Damascus, 90–2. 103 Amitai-Preiss, Mongols, 187. 104 Northrup, L. S. From Slave to Sultan: The Career of Al Mansur Qalawun and the Consolidation of Mamluk Rule in Egypt and Syria (678–689a.h./1279–1290 a.d.) (Wiesbaden, 1998), 113–14. 105 Irwin, “County of Tripoli,” 249. 106 Northrup, Qalāwūn, 127–130. Concerning the slave trade and the shift from the Asian land route to the naval route see: Ehrenkreutz, A., “Strategic implications of the slave trade between Genoa and Mamluk Egypt in the second half of the thirteenth century,” in The Islamic Middle East 700–1900: Studies in Economic and Social History, ed. A. L. Udovitch (Princeton, NJ, 1981), 335–45. 107 Northrup, Qalāwūn, 152. 108 Abū ‘l-Fidā’, Syrian Prince (Holt), 12.

232  Notes 109 Ibn ʿAbd al-Ẓāhir, Tashrīf, 80. 110 Ibid. 80, the origin of the word jarjya ( ) meaning ‘troops’ may be Mongolian, the changing to in Arabic. Doerfer, G., Türkische und Mongolische Elemente im Neupersischen, vol. 1 (Wiesbaden, 1963), 293. I would like to thank Professor M. Biran for her help. The word jurājīr in Arabic also means ‘troops’. Steinglass, F., Arabic-English Dictionary (London, 1884), 227. Baḥriyya in this context is used to note Syrian fortress garrisons. The original Baḥriyya performed garrison duties especially during the reign of Qalāwūn. Ayalon, D. “Baḥriyya,” EI2 1: 944–5. 111 Molin, Unknown Castles, 22. 112 Ibn ʿAbd al-Ẓāhir, Tashrīf, 89–90. 113 Humphreys, Saladin, 9. 114 Irwin, “Direction of trade,” 79. See Table 3.1 in this chapter. 115 Irwin, “Direction of trade,” 80. 116 Ibid., 77. 117 Ibn al-Furāt, Muḥammad b. ʿAbd al-Raḥīm, Tarīkh Ibn al-Furāt, eds N. Izzedin, and C. K. Zurayk (Beirut, 1939), vol. 8, 81; Maqrīzī, Sulūk, vol. 1, pt. 3, 748. Jubail and the fortresses in the near vicinity were part of a fief that belonged to a Frankish lord by the name of Sir Guy. He was later given Jubail as an iqṭāʿ by Qalāwūn. 118 Ibn al-Furāt, Ta’rīkh al-duwal wa’l-mulūk, vol. 8, 81. 119 Ibid., 81. 120 Maqrīzī, Sulūk, vol. 1, pt. 3, 746–8. 121 Irwin, “County of Tripoli,” 247. 122 Ibn ʿAbd al-Ẓāhir, Tashrīf, 147–50. 123 Nicola, Urban Life, 85. 124 Chevedden, Damascus, 92–5. 125 Ragette, Baalbek, 76–78. 126 Al-ʿAynī, Badr al-Dīn Maḥmūd b. ‘Ali, ʿIqd al-jumān fī ta’rikh ahl-al-zamān, ed. M. M. Amīn (Cairo, 1990), vol. 3, 151. 127 On the conquest of Qalʿat al-Rūm, Maqrīzī, Sulūk, vol. 1, pt. 3, 777–8. 128 Répertoire chronologique d’épigraphie arabe, eds J. Combe, J. Sauvaget, et G. Wiet (Cairo, 1943), vol. 13, no. 4956. 129 Stewart, A. D., The Armenian Kingdom and the Mamluks: War and Diplomacy during the Reigns of Hetum II (1289–1307) (Leiden, 2001), The Medieval Mediterranean, vol. 34. 74. 130 Bar Hebraeus, Chronography, vol. 1, 461. 131 Rashīd al-Dīn, Jami‘u’t-Tawarikh (Thackston), vol. 3, 58. 132 Abū’l-Fidā’, Syrian Prince (Holt), 19. 133 Al-Nuwayrī, Shihāb al-Dīn Aḥmad b. ʿAbd al-Wahhāb. Nihāyat al-arab fī funūn al-adab, (Cairo, 1992), vol. 31, 228, 231; Honigmann, E, and Bosworth, C. E., “Rūm K. al’esi,” EI2: 8, 607. 134 Nuwayrī, Nihāyat, 231. This council is also described by The Templar of Tyre (born c. 1255), The ‘Templar of Tyre’: Part III of the ‘Deeds of the Cypriots,’ trans. P. Crawford (Aldershot, 2003), 122. 135 Al-Nuwayrī, Shihāb al-Dīn Ahmad b. ʿAbd al-Wahhāb, Nihāyat al-arab fī funūn al-adab (Cairo, 1992), vol. 31, 228, 231; Honigmann, E, and Bosworth, C. E., “Rūm K. al’esi,” EI2: 8, 607. 136 Boyle, Iran, 372–3. 137 Al-ʿAynī, ʿIqd, vol. 3, 164. 138 Ibid., 165. 139 Ibid., 164. 140 Humphreys, Saladin, 129–30, 310. 141 Der Nersessian, S., “The Armenian chronicle of the Constable Smpad or of the “Royal Historian,” DOP 13 (1959): 142–68, 162–3; Amitai-Preiss, “ ʿAyn Jālūt,” 10–12. 142 Atiya, A. S., The Crusade in the Later Middle Ages, 2nd edn (New York, 1965), 41, 46. 143 Runciman, S., A History of the Crusades (London, 1994), vol. 3, 333–7.

Notes  233 144 Schein, S., Fideles Crucis: The Papacy, the West and the Recovery of the Holy Land 1274– 1314 (Oxford, 1991), 256. 145 ����������������������������������������������������������������������������������� Tyerman, C. J., “Marino Sanudo Torsello and the lost Crusade: lobbying in the fourteenth century,” Transactions of the Royal Historical Society 32 (1982): 65. 146 If Marino Sanudo could have witnessed the results of Peter I, King of Cyprus’ attack on Alexandria in 1365 and the effect it had on the Italian merchant communities and the Mamluks, he might have abandoned the entire idea of an economic embargo and of the conquest of Egypt. The economic damage that both sides suffered was devastating. The interdependency of the Mamluks and the Italian merchants, however, led to a relatively fast recovery; trading was resumed within a couple of years, though the volume of trade and profit never returned to that of the thirteenth century. Ashtor, E., Levant Trade in the Later Middle Ages (Princeton, NJ, 1983), 88–91. 147 Since the Mamluks did not have a merchant fleet, most of the trade was conducted by foreigners. Thus their role was similar to that of the Armenians, who acted as intermediary providing traders with commercial services. Labib, S., “Egyptian commercial policy in the Middle Ages,” in Studies in the Economic History of the Middle East from the Rise of Islam to the Present Day, ed. M. A. Cook (London, New York and Toronto, 1970), 65, 73. 148 The following peace treaty signed by Qalāwūn in 684/1285 with the Armenian king Leon gives an idea of the importance of the Armenian territory to the Mamluk slave trade: “Provided also that King Leon shall not forbid anyone, whether a merchant or not, who imports slaves, slave girls, horses, mules and all kinds of goods, to bring them to the Sultan’s territory: nor shall he detain them, nor permit anyone to detain them. He shall open the way for them to import the slaves, slave girls, horses, mules and all kinds of goods, all nationalities of slaves and nationalities of slave girls of their various kinds and not detain any of them.” Holt, P. M., Early Mamluk Diplomacy (1260–1290): Treaties of Baybars and Qalāwūn with Christian Rulers (Leiden, 1995), Islamic History and Civilization, vol. 12, 102; Ayalon, D., “Mamlūk,” EI2 6: 315. 149 Schein, Fideles Crucis,156–7. 150 Housley, N., The Avignon Papacy and the Crusades 1305–1378 (Oxford, 1986), 16. 151 Schein, Fideles Crucis, 195–6. 152 Ibid., 214 153 Marino Sanuto, Liber secretorum fidelium crucis super Terrae Sanctae recupratione et conservatione, intro. by J. Prawer (new edn, 1972), pp. i–xiii; Atiya, Crusade, 116–26. 154 Tyerman, “Marino Sanudo,” 63. 155 Luttrell, A. T., “The Hospitallers’ interventions in Cilician Armenia: 1291–1375,” in The Cilician Kingdom of Armenia, ed. T. R. S. Boase (Edinburgh and London, 1978), 126. 156 Tyerman, “Marino Sanudo,” 65–7. 157 Schein, Fideles Crucis, 195. 158 Hayton, La flor des estoires de la Terre d’orient, Recueil des Historiens des Croisades, Documents Arméniens, tome II, Documents Latins et Français relatifs à l’Arménie (Paris, 1900), bk IV, 248–9; Hetoum, A Lytell Cronycle, ed. G. Burger (Toronto, 1988), 81–5; Luttrell, “Hospitallers,” 123–4. 159 Luttrell, “Hospitallers,” 123. 160 Amitai-Preiss, Mongols, 94–105. 161 Yūnīnī, Dhayl mir’āt al-zamān (Hyderabad, 1955), vol. 2, 343–4. 162 Stewart, Armenian Kingdom, 50. 163 Smpad (Der Nersessian), 167. 164 Holt, Diplomacy, 102. 165 Ibn Shaddād, Aʿlāq, vol. 1, pt. 2, 342–3. 166 Bar Hebraeus, Chronography, 445. 167 Smpad (Der Nersessian), 165; Bar Hebraeus, Chronography, 446. 168 Holt, Diplomacy, 93; Bar Hebraeus, Chronography, 446. 169 Ibn ʿAbd al-Ẓāhir, Rawḍ, 280; Bar Hebraeus, Chronography, 446; Abū‘l-Fidā’, Syrian Prince (Holt), 29; Thorau, Baybars, 174.

234  Notes 170 Ibn ʿAbd al-Ẓāhir, Rawḍ, 327–9; Amitai-Preiss, Mongols, 119; Stewart, Armenian Kingdom, 47–9. 171 Maqrīzī, Sulūk, vol. 1, pt. 2, 568–9. 172 Ibn ʿAbd al-Ẓāhir, Rawḍ, 417; Amitai-Preiss, Mongols, 131–2. 173 This new fortress is called Gadariats and it does not appear in earlier or later treaties. Smpad, (Der Nersessian), 166. Ibn ʿAbd al-Ẓāhir, Rawḍ, 432. 174 Bar Hebraeus, Chronography, 453–4. 175 Holt, Diplomacy, 102. 176 Al-ʿAynī, ʿIqd, vol. 3, 150–2. 177 Yūnīnī, Dhayl (Guo), vol. 1, 110; Abū‘l-Fidā’, Syrian Prince (Holt), 27–8. 178 The fortress of Nujayma is yet to be identified. 179 Edwards situates Hāmīs 12km southeast of the modern city of Osmaniye. Edwards, Fortifications, 110–13. 180 Yūnīnī, Dhayl (Guo), vol. 1, 110 181 Abū‘l-Fidā’, Syrian Prince (Holt), 29. Sarfandkar is identified as an Armenian fortress by the name of Savranda. Al-Naqīr is synonymous with a stronghold that appears in the Armenian sources as Nghir and is identified as the Armenian site Mancilik (Kalesi) situated on the western slopes of the Anti-Taurus. Hajar Shaghlan is identified with Çalan, an Armenian fortress in the northern section of the Amanus range. Edwards has some reservations concerning this identification. Kuwayrā has not been identified and there are very few clues as to its location. Edwards, Fortifications, 99, 185; Stewart, Armenian Kingdom, 121. 182 Abū‘l-Fidā’, Syrian Prince (Holt), 39–40. 183 Ibid., 45. 184 Ibid., 44–5. 185 Ibid., 67–8. 186 Ibid., 79. 187 Ibid., 80. 188 Edwards, Fortifications, 77–83. 189 Abū‘l-Fidā’, Syrian Prince (Holt), 82 190 Edwards, Fortifications, 77–83. 191 Ibid., 32–3. 192 Methods of construction will be discussed in detail in Chapter 4. 193 Though the fortress is of considerable importance due to its location on the main route that leads south to Ayās, it is seldom mentioned in the sources. The nine-month siege conducted by the Mamluks in 1375 is recordnotesed only by the Armenian Chronicler. Edwards, Fortifications, 125, n. 17. 194 Ibid., 39–41. 195 Ibid., 41. 196 Ibid., 42. 197 Marco Polo, The Travels of Marco Polo, trans. R. Latham (London, 1958), 15; Chahin, M., The Kingdom of Armenia (New York, 1987), 289–90; Yule, H., The Book of Sir Marco Polo the Venetian (London, 1871), 41–3. 198 Ashtor, Trade, 57–9. 199 Ibid., 56–7. 200 Jackson, P. The Mongols and the West, 1221–1410 (Harlow, England, 2005), 295–298 201 Morgan, Persia, 83–84. 202 Ashtor, Trade, 63, 101. 203 Bar Hebraeus, Chronography, 465. 204 Abū‘l-Fidā’, Syrian Prince (Holt), 44–5. 205 Har-El, Ottoman-Mamluk War, 39–59. 206 Ayalon, D., “The Muslim city and the Mamluk military aristocracy,” in Studies on the Mamluks of Egypt 1250–1517 (London, Variorum,1977), VII, 311–29. See especially pages 319–20.

Notes  235 207 Humphreys, R. S., “The politics of the Mamluk Sultanate: a review essay,” MSR 9/1 (2005): 223. 208 Cahen, C., “Ayyubids,” EI2 1: 796–807; Humphreys, R. S., “Ayyubids, Mamluks and the Latin East in the thirteenth century,” MSR 2 (1998): 4–6. 209 According to Humphreys there was no notion of solidarity in the Ayyubid federation and in their policy towards the Frankish infidels. They seldom managed to unify their ranks and confront the Franks. Humphreys, “Ayyubids,” 7. 210 Humphreys, R. S., “Politics and architectural patronage in Ayyubid Damascus,” in The Islamic World, eds C. E. Bosworth, C. Issawi, R. Savory, and A. L. Udovitch (Princeton, NJ, 1989), 163. 211 Even in Islamic cities that had citadels, these usually did not serve as the residence of the ruler until the eleventh century. Most rulers built and lived in palaces outside the urban centers. It was only towards the end of the eleventh century that rulers moved to live in urban citadels. Bacharach, J. L., “Administrative complexes, palaces and citadels, changes in the loci of medieval Muslim rule,” in The Ottoman City and Its Parts, eds I. Bierman, R. A. Abou-Al-Haj and D. Preziosi (New York, 1991), 111–28. 212 The fortresses listed by Morray are Sahyun, Blantus, Shayzar, Qal’at Munqidh, Shughar and Bakās in the Orontes valley. Morray, D., An Ayyubid Notable and his World (Leiden, New York and Cologne, 1994), 3. 213 Humphreys, Saladin, 114–15; Tonghini, Qal ʿat Jabar, 21. 214 Maqrīzī, Sulūk, vol. 1, 149–50. 215 Ibid., vol. 1, 114–15. 216 Streck, M., “Biredjik,” EI2 1: 1233–4. 217 Humphreys, Saladin, 310. 218 Humphreys, “Latin East,” 3. 219 Edwards, Fortifications, 42. 220 Northrup, Qalāwūn, 212 221 Ibid., 113. 222 A map showing a partial distribution of Ayyubid fortresses can be seen in Eddé, Ayyubide, fig. 64. 223 For a detailed map of the Crusader fortresses see Kennedy, Castles, 1. 4  Mamluk power on display 1 2

3

4 5

Ellenblum, R., Crusader Castles and Modern Histories (Cambridge, 2007), 231–57; Marshall, C. J., Warfare in the Latin East, 1192–1291 (Cambridge, 1996), 98–111; Kennedy, H., Crusader Castles (Cambridge, 1994), 120–79. Deschamps, P., Les Châteaux des Croisés en Terre-Sainte: La Défense du Royame de Jérusalem (Paris, 1939), vol. 1, ch. 2, “Kérak et les châteaux de la Terre Oultre le Jourdain,” 35–98, ch. 5: “Le Chateau de Subeibe près de Banyas,” vol. 2, 147–74 (plans and plates). Ibid., Les Chateaux des Croisés en Terre-Sainte: le Carac des Chevaliers (Paris, 1934), 2 vols. De Vries, B., “Ḥiṣbān in the Ayyubid and Mamluk periods,” in Hesban after 25 Years, eds D. Merlin and L. T. Geraty (Berrien Springs, MI, Andrews University, 1994), 151–66; Walker, B. J. and LaBianca, Ø. S., “The Islamic Qusūr of Tall Ḥiṣbān: preliminary report on the 1998 and 2001 seasons,” ADAJ 47 (2003): 449; Walker, B. J. “Militarization to nomadization: the middle and late Islamic periods,” NEA 62/4 (1999): 212; Walker, B. J., “Mamluk investment in Southern Bilād al-Shām in the eighth/fourteenth century: the case of Ḥiṣbān,” JNES 62 (2003): 241–61. Brown, R. M., “Summary report of the 1986 excavations in late Islamic Shobak,” ADAJ 32 (1988): 225–54; ibid., “Excavations in the 14th century Mamluk palace at Karak,” ADAJ 33 (1989): 287–304. Edwards, R.W., The Fortifications of Armenia Cilicia (Washington, DC, Dumbarton Oaks Research Library and Collection, 1986)�������������������������������������������������� ; ibid.,������������������������������������������ “Bagras and Armenian Cilicia: a reassessment,” Revue des Etudes Arméniennes 17 (1983): 415–35.

236  Notes 6 7

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

Several fortresses that are not part of this study are shown on Figure 4.1 in order to give the reader a better idea of the extent of Mamluk fortifications. Vitruvius spent most of his life in the military. Julius Caesar employed him as a military engineer; he specialized in building fortifications and siege machines. Vitruvius’ Ten Books on Architecture, trans., prefaced & annotated by R. Reich (Tel-Aviv, 1997), 12–14. [Hebrew]; Boardman, J., Griffin, J. and Murray O., The Oxford History of the Classical World (London, 1988), 772. Rabbat, N., “Perception of architecture in Mamluk sources,” MSR 6 (2002): 155–76. See especially 158–9. Ibn ʿAbd al-Ẓāhir, al-Rawḍ al-ẓāhir fī sīrat al-malik al-ẓāhir, ed. ʿA. ʿA. al-Khuwayṭir (Riyad, 1976), 93. Al-Ṣubayba did not have a moat around it and ʿAjlūn’s walls were not in need of any repair since they did not suffer any damage at the hands of the Mongols. These two sites are discussed in greater detail in this chapter. Ibn ʿAbd al-Ẓāhir, Rawḍ, 165. Ibn ʿAbd al-Ẓāhir, Rawḍ, 265. Ibn al-Furāt, Ayyubids, Mamluks and Crusaders: Selections from the Tā’rīkh al-duwal wa’lmulūk of Ibn al-Furāt, ed. and trans. U. and M. C. Lyons; Introduction and notes J. S. C. Riley-Smith (Cambridge, 1971), 96. Sourdel, D. “Ibn Shaddād,” EI2 3: 933. Rabbat, “Perception,” 164. The account of the reconstruction work carried out at Safad is given in more detail below. Cahen, C., “Ibn al Furāt,” EI2 3: 768–9. Qāqūn – Ibn ʿAbd al-Ẓāhir, Rawḍ, 275; Qāqūn – Ibn al-Furāt, Tarikh (Lyons), 101; Safad – Ibn ʿAbd al-Ẓāhir, Rawḍ, 278. Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 69. Ibn ʿAbd al-Ẓāhir, Rawḍ, 93. Shāfī’ Ibn ʿAlī, Shāfī’ Ibn ʿAlī’s Biography of the Mamluk Sultan Qalāwūn, ed. P. B. Lewicka (Warsaw, 2000). Orientalia Polona, 379. Ibn ʿAbd al-Ẓāhir, Muḥyī al-Dīn ʿAbd Allāh, Tashrīf al-ayyām wa’l-ʿuṣūr fī sīrat al-malik al-manṣūr, ed. M. Kāmil (Cairo, 1961), 96. Ibn Shaddād, al-Ḥalabī, Al-Aʿlāq al-khaṭīra … vol. 2, pt. 2: Ta’rīkh lubnān, al-urdunn wa-filasṭīn, ed. S. Dahhan (Damascus, 1963), 146. Ibn ʿAbd al-Ẓāhir, Rawḍ, 384. al-Qalqashandī, Shihāb al-Dīn Aḥmad, Ṣubḥ al-a’shā fī ṣinā’at al-inshā’ (Beirut, 1987), vol. 4, 206. Ellenblum, R., “Who built Qalat al-Subayba?,” DOP 43 (1989): 103–12; Amitai, R. “Notes on the Ayyubid inscriptions at al-Ṣubayba (Qalʿat Nimrod),” DOP 43 (1989): 113–19, 114; Ibn ʿAbd al-Ẓāhir, Rawḍ, 375. Ibn Shaddād, Aʿlāq, vol. 2, 69. Ibn ʿAbd al-Ẓāhir, Rawḍ, 93; al-Ṣafadī, Ṣalāḥ al-Dīn Khalīl b. Aybeg, al-Wāfī bi’l-wafayāt, ed. H. Ritter et al. (Wiesbaden, 1980), vol. 10, 341. For Ayyubid inscriptions see ch. 1. Ibn Shaddād, Ta’rīkh, 357; Baybars al-Manṣūrī al-Dawādār Rukn al-Dīn, Kitāb al-tuḥfa al-mulūkiyya fī ‘l-dawla al-turkiyya, ed. ‘A. S. H. Ḥamdān (Cairo, 1987), 57. Ibn Shaddād, Ta’rīkh, 354. Ibid., 357, 353 The secret passages at Safad are mentioned by Ibn ʿAbd al-Ẓāhir, Rawḍ, 285; al-Maqrīzī, Taqī al-Dīn Aḥmad b. ʿAlī, Kitāb al-sulūk li-maʿrifat duwal al-mulūk (Beirut, 1997), vol. 1, pt. 2, 563. Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 150; ibid., Ta’rīkh, 353. Lane, E. W., Arabic–English Lexicon (New York, 1955), bk. 1, pt. 4, 1369. Ibn Shaddād, Aʿlāq, vol. 2, 89. al-Dimashqī, Shams al-Dīn Abū ʿAbd Allāh Muḥammad B. Abī Ṭālib al-Anṣārī al-Sufī, Nukhbat al-dahr fī ‘ajā’ib al-barr wa’l-baḥr, ed. M. A. F. Mehren (Saint Pétersbourg, 1866), 210.

Notes  237 35 al-Ṣafadī, Ṣalāḥ al-Dīn Khalīl b. Aybeg, Al-Wāfī bi’l-wafayāt, ed. H. Ritter (Wiesbaden, 1980), vol. 10, 342. 36 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 63 37 al-Ṣafadī, vol. 10, 342. 38 See Chapter 1. 39 Amitai, “Inscriptions at al-Ṣubayba,” 110. 40 Oliver of Paderborn, Historia Damiatina, 169–71; Benvenisti, M., The Crusaders in the Holy Land (Jerusalem, 1976), 176. 41 Benvenisti, Crusaders, 203. 42 Ibn Shaddād, Ta’rīkh, 351–9; Amitai, “Inscriptions at al-Ṣubayba,” 113, Ellenblum, “Al-Ṣubayba,” 110–11. 43 For a detail of the events that lead to the destruction see Chapter 1. 44 Ibn ʿAbd al-Ẓāhir, Tashrīf, 89–90. 45 Ibn ʿAbd al-Ẓāhir, Rawḍ, 93. 46 Ibn Shaddād, Ta’rīkh, 242. 47 Ibn ʿAbd al-Ẓāhir, Tashrīf, 129 48 Ayalon, D., “ Studies in the structure of the Mamluk army – III,” BSOAS 16 (1954): 63–4; Humphreys, R. S. From Saladin to the Mongols: The Ayyubids of Damascus 1193– 1260 (Albany, 1977), 456, n. 34; Irwin, R., The Middle East in the Middle Ages: The Early Mamluk Sultanate 1250–1382 (London, 1986), 38. 49 Ayalon, “Structure III,” 64. 50 Irwin, Middle East, 38. 51 See the case of Ṣārim al-Dīn Barghash who supervised the building of both Qalʿat Ṣadr and Karak: Chapter 1. 52 Ibn ʿAbd al-Ẓāhir, Rawḍ, 377. 53 Ibid., 385. 54 Ibid., 238. At the end of the siege of Arsūf he received a village in the region as freehold. R. Amitai-Preiss, “The Mamluk officer class during the reign of sultan Baybars,” in War and Society in the Eastern Mediterranean, 7th–15th Centuries, ed. Y. Lev (Leiden, 1997), 296. Aybak al-Afram’s extraordinary wealth is described by Ibn Taghrī Birdī, Abū ‘l-Mahāsin, Al-Nujūm al-zāhira fī muluk miṣr wa’l-qāhira (Cairo, 1930–56.), vol. 8, 80–1 55 Ibn ʿAbd al-Ẓāhir, Rawd, 231. 56 Ibn Shaddād, Ta’rīkh, 257. 57 Maqrīzī, Sulūk, vol. 1, part 3, 763. 58 This event is mentioned by Maqrīzī in a roundabout way when he describes the murder of al-Ashraf Khalīl; Aybak al-Afram’s name appears as one of the amirs who were ill-treated by this sultan. Maqrīzī, Sulūk, vol. 1, pt. 1, 792. 59 Ibn Taghrī Birdī, Nujūm, vol. 8, 80. 60 I would like to thank Mr Asad Daʿbūs, the manager of the site, for showing me around the fortress, for his patient explanations, and several interesting conversations which were of great help in my research. 61 Maqrīzī, Sulūk, vol. 1, pt. 2, 441; al-Yūnīnī, Quṭb al-Dīn Mūsā b. Muḥammad, Dhayl mir’āt al-zamān fī ta’rīkh al-aʿyān (Hyderabad, 1954–61), vol. 2, 69–71; Amitai-Preiss, R., Mongols and Mamluks: The Mamluk–Īlkhānid War 1260–1281 (Cambridge, 1995), 32, 45. 62 Ellenblum, “Al-Ṣubayba,” 110–11; Amitai, “Al-Ṣubayba,” 114. 63 Ibn Shaddād, Ta’rīkh, 354. 64 Ṣafadī, Wafayāt, vol. 10, 365; Maqrīzī, Sulūk, vol. 1, pt. 2, 603. 65 Irwin, Middle East, 39. 66 Ṣafadī uses the term Iqṭāʿ, Wafayāt, vol. 10, 366; Ibn Shaddād uses milk, Aʿlāq, vol. 1, pt. 1, 144. This matter is discussed at length in Amitai, “Al-Ṣubayba,” 109–23. See especially 114–15. 67 Ibn Shaddād, Aʿlāq, vol. 1, pt. 1, 145. 68 Amitai, “Al-Ṣubayba,” 114.

238  Notes 69 Ṣafadī, Wafayāt, vol. 10, 366; Irwin, Middle East, 62; Amitai-Preiss, Mongols, 179–80. 70 Mayer, L. A., Islamic Architects and Their Works (Geneve, 1956), 18–26. 71 Behrens-Abouseif, D., “Muhandis, Shād, Muʿallim – note on the building craft in the Mamluk period,” DI 72 (1995), 295; Amitai, “Al-Ṣubayba,” 116. 72 Rabbat, N., “Architects and artists in Mamluk society: the perspective of the sources,” Journal of Architectural Education 52 (1998): 32. 73 Mayer, Islamic Architects, 27. 74 Behrens-Abouseif, D., Islamic Architecture in Cairo: An Introduction (Leiden and New York, 1989), 18. 75 Irwin, Middle East, 55; Amitai, “Al-Ṣubayba,” 109–23. See especially 114–15. 76 Humphreys, R. S., “The emergence of the Mamluk army (conclusion),” SI 46 (1977): 175–7. 77 Ibn Shaddād, Ta’rīkh, 354. 78 Apart from the urban fortifications, round towers can be seen at Arsuf, Crac des Chevaliers, Margat. 79 Keen, M., “The changing scene: guns, gunpowder and permanent armies,” in Medieval Warfare, ed. M. Keen (Oxford, 1999), 273–91. 80 Hartal, M., The Al-Ṣubayba (Nimrod) Fortress, Towers 11 and 9 (Jerusalem, 2001), 69. 81 Ibid., 23. 82 Ibn Shaddād, Aʿlāq, vol. 1, pt. 1, 82. 83 Hartal, Al-Ṣubayba, 36–7. 84 Excavations at the citadel: Damati, E., “ Safed,” ESI 4 (1985): 98; ibid., “Safed,”, ESI 5 (1986): 93–4; ibid., “Safed Citadel,” ESI, 1988/89, vols 7–8, 159–60; ibid., “Safed Citadel,” ESI 9 (1989/90): 13. More recently excavations were conducted by Barabé, H. and Damati, E., “La forteresse médievale de Safed: données récentes de l’archéologie’,” Crusades 3 (2004): 171–3. I would like to thank Haim Barabé for his help and explanations concerning the plan of the fortress and some of the debates concerning the dating of the gate structure he excavated. 85 Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 88; Amitai-Preiss, R., “Ṣafad,” EI2 8: 757. 86 A full translation is given in Kennedy, Castles, 190–8. The most comprehensive examination of this source is by R. B. C. Huygens, De constructione castri Saphet: construction et functions d’un chateau fort franc en Terre Sainte (Amsterdam, Oxford and New York, 1981); Pringle, D., “Reconstructing the Castle of Safad,” PEQ 117 (1985): 139–49. 87 Dunlop, D. M., “Dimashqī,” EI2 2: 291. 88 When he returned he was employed as a clerk in the administration of Safad. Rosenthal, F., “Al-Ṣafadī,” EI2 8: 759. 89 Ṣafadī, Wafayāt, vol. 10, 341. 90 Maqrīzī’s account is an almost exact copy of that given by Ibn al-Furāt who probably copied his directly from Ibn ʿAbd al-Ẓāhir, Rawḍ, 280–1; Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 92–5; Maqrīzī, Sulūk, vol. 1, pt. 2, 562–3. 91 Tracy, J. D., “To wall or not to wall: evidence from medieval Germany,” in City Walls: The Urban Enceinte in Global Perspective, ed. J. D. Tracy (Cambridge, 2000), 71–88, table 3.5, p. 84. 92 Benvenisti, Crusaders, 202. Prawer does not rule out the possibility that it may have been built in 1142. Prawer, J., The Crusaders: A Colonial Society, 2nd edn (Jerusalem, 1985), 366. [Hebrew]. 93 Runciman, S., A History of the Crusades: The Kingdom of Jerusalem and the Frankish East 1100–1187 (London, 1994), vol. 2, 306. 94 Runciman, Crusades, 382. 95 Ibn Shaddād, Aʿlāq vol. 2, pt. 2, 147; Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 88. 96 Ibn al-Athīr, ʿIzz al-Dīn, Al-Kāmil fī al-Ta’rīkh, ed. C. J. Tornberg (Beirut, 1966), vol. 12, 87. 97 Mayer, H. E., “Two unpublished letters on the Syrian earthquake of 1202,” in Medieval and Middle Eastern Studies in Honor of Aziz Suryal Atiya, ed. S. A. Hanna (Leiden, 1972),

Notes  239 295–310; Ellenblum, R., Marco, S., Agnon, A., Rockwell, T. and Boas, A., “Crusader castle torn apart by earthquake at dawn, 20 May 1202,” Geology 26 (1998): 303–96. 98 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 147; Ibn Shaddād, Ta’rīkh, 353. 99 Prawer, Crusaders, 149; Runciman, Crusades, 183; Humphreys, Saladin, 266–67. 100 Kennedy, Castles, 195; Ellenblum, Modern Histories, 246–255. 101 Ibn ʿAbd al-Ẓāhir, Rawḍ, 265; Kennedy, Castles, 128. 102 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 149; Baybars al-Manṣūrī, Kitāb al-tuḥfa al-mulūkiyya fī l’dawla al-turkiyya, ed. A. R. S. Hamdān (Cairo, 1987), 57; Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 95; Ibn ʿAbd al-Ẓāhir gives a different account, see Rawḍ, 260–1. 103 Ibn ʿAbd al-Ẓāhir, Rawḍ, 281, 285; Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 101–2. 104 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 150. 105 Baybars al-Manṣūrī, Tuḥfa, 57. Accounts of the siege are given also by Ibn ʿAbd al-Ẓāhir, Rawd, 257–8; Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 92–3. 106 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 149. 107 Ṣafadī, Wafayāt, vol. 10, 341. Ṣafadī probably borrowed the term “hirqilī” from Ibn Shaddād. 108 Wehr, H., A Dictionary of Modern Written Arabic, ed. J. M. Cowan, 4th edn (Wies­baden, 1994), 1026; Steingass, F., Arabic–English Dictionary (London, 1884), 1173. 109 Sharoni interprets Hiraqlī as being of exceptional/unique strength. Sharoni, A., The Comprehensive Arabic–Hebrew Dictionary (Tel-Aviv, 1987), vol. 3, 1375. 110 Ibn Shaddād, Aʿlāq, vol. 1, pt. 1, 82; Tabbaa explaines bāshūra in this particular case as a fortified gate. Tabbaa, Y., Constructions of Power and Piety in Medieval Aleppo (Pennsylvania Park, PA., 1997), 73. 111 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 150; ibid., Ta’rīkh, 353. 112 Ibn Shaddād, Ta’rīkh, 353. 113 Ibn ʿAbd al-Ẓāhir, Rawḍ, 285; Maqrīzī, Sulūk, vol. 1, pt. 2, 563. 114 Cited in Kennedy, Castles, 195. 115 Pringle, “Safad,” 140. 116 Hartal, Al-Ṣubayba, 24–30. 117 Ṣafadī, Wafayāt, vol. 10, 341; dhirāʿ ( ) in Syria = 0.68m; in Egypt = 0.58m. 118 Dimashqī, Kitāb Nukhbah, 210. 119 Ibn Shaddād, Ta’rīkh, 353. 120 Ṣafadī, Wafayāt, vol. 10, 341. 121 Dimashqī, Kitāb Nukhbah, 210. 122 Duffy, C., Fire and Stone: The Science of Fortress Warfare 1660–1860 (London and Vancouver, 1975), 36. 123 Ibn Shaddād, Ta’rīkh, 353. 124 Dimashqī, Kitāb Nukhbah, 210. 125 Ibn ʿAbd al-Ẓāhir, Rawḍ, 287; Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 105; Ibn al-Furāt appears to be quoting Ibn ʿAbd al-Ẓāhir. 126 Pringle has suggested that the tower was originally built by the Franks. Pringle, “Safad,” 148. 127 Ibn Shaddād, Ta’rīkh, 353. 128 Ibn ʿAbd al-Ẓāhir, Rawḍ, 285–287; Maqrīzī, Sulūk, vol. 1, pt. 2, 563. 129 The Qur’ān, Sura 21:105. Text, translation and commentary by Abdullah Yusuf Ali (Cambridge, MA, 1946), vol. 2 130 Qur’ān, Sura, 58:22. 131 Ibn ʿAbd al-Ẓāhir, Rawḍ, 285–287. I would like to thank Dr Miriam Goldstein from Bar Ilan University for her help in translating. 132 Blair, S. S., “Decoration of city walls in the medieval Islamic world: the epigraphic message,” in City Walls; The Urban Enceinte in Global Perspective, ed. J. D. Tracy (Cambridge, 2000), 488–529, see especially 503, 525. 133 Ibn ʿAbd al-Ẓāhir, Tashrīf, 80. 134 Ibn ʿAbd al-Ẓāhir, Rawḍ, 163.

240  Notes 135 There seems to be some confusion concerning the mosque. Al-Ṣafadī says that it was newly built while Ibn ʿAbd al-Ẓāhir and Ibn al-Furāt say the church was turned into a mosque. Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 101; Ibn ʿAbd al-Ẓāhir, Rawḍ, 275; Ṣafadī, Wafayāt, vol. 10, 341; Ibn Shaddād, Ta’rīkh, 352; al-Maqrīzī, Hiṭaṭ, vol. 4, 248. Cytryn-Silverman, K. The Road Inns (Khans) of Bilad al-Sham during the Mamluk Period (1260–1516): An Architectural and Historical Study, PhD diss.; The Hebrew University of Jerusalem, 2004, unpublished, Gazetteer, 171–4. The site is mentioned by two other chroniclers, though they refer to the Frankish attack and do not add to our knowledge of the fortress, Baybars al-Manṣūrī, Tuḥfa, 73; Yūnīnī, Dhayl, vol. 2, 467–8. 136 Qalqashandī, Shihāb al-Dīn Aḥmad, Ṣubḥ al-a’shā fī ṣinā’at al-inshā’ (Beirut, 1987), vol. 4, 206. 137 Qalqashandī, Ṣubḥ al-a’shā fī ṣinā’at al-inshā’ (Cairo, 1985), vol. 14, 379, 393. 138 Qalqashandī, Ṣubḥ (Beirut 1987), vol. 4, 104. 139 William of Tyre, bk. 12, 21, 546; Fulcher of Chartres, A History of the Expedition to Jerusalem 1095–1127, trans. F. R. Ryan, ed. H. S. Fink (University of Tennessee, 1969), bk. III, ch. 28, 1, 241–2; Ellenblum, Modern Histories, 170. 140 Ibn ʿAbd al-Zāhir, Rawd, 396–7, Maqrīzī, Sulūk, vol. 1, pt. 2, 600–1. 141 The ‘Templar of Tyre’: Part III of the ‘Deeds of the Cypriots’, trans. P. Crawford (Aldershot, 2003), 68; on the attack of Edward of England see Amitai, R., “Edward of England and Abagha Ilkhan,” in Tolerance and Intolerance, eds M. Gervers and J. M. Powell (Syracuse, 2004), 75–80; Thorau, Baybars, 209; Marshall, Warfare, 206–7. 142 Ibn al-Furāt, Ta’rīkh (Lyons), 101. 143 Burchard of Mount Sion, A Description of the Holy Land (London, 1897), PPTS, vol. 12, 94–5. 144 Ayalon, D., “The Mamluks and naval power: a phase of the struggle between Islam and Christian Europe,” in Proceedings of the Israel Academy of Sciences and Humanities 1/8 (1967): 8–9. 145 Ibn ʿAbd al-Ẓāhir, Rawḍ, 294; Maqrīzī, Sulūk, 1, pt. 2, 51; Poliak, A. N., Feudalism in Egypt, Syria, Palestine, and the Lebanon, 1250–1900 (London, 1939), 9; Ayalon, “Naval power,” 10; Thorau, Baybars, 188; Amitai-Preiss, Mongols, 70. R. Irwin, “Iqṭā’ and the end of the Crusader states,” in The Eastern Mediterranean Lands in the Period of the Crusades, ed. P. M. Holt (Warminster, 1977), 71. In later decades groups of Wāfidiyya were also settled along the Syro-Palestinian coast. This policy was established at the end of the Ayyubid period. Ayalon, D., “The Wafidiyya in the Mamluk kingdom,” in Studies on the Mamluks of Egypt (1250–1217) (London, Variorum, 1977), II, 94, 99–100. 146 Elad, A., “The cities of Palestine during the early Arab period (604–1066 a.d.) according to Arabic sources,” Cathedra 8 (1978): 16 [Hebrew]; Masarwa, Y., “Early Islamic military architecture: the birth of the Ribat on the Palestinian coast,” Al-ʿUsur al-Wusta (October 2008): 38. 147 Pringle, D., The Red Tower (al Burj al Aḥmar) Settlement in the Plain of Sharon at the Time of the Crusaders and Mamluks a.d. 1099–1516 (London, 1986), 58–71. 148 Guérin, V., Description géographique, historique et archéologique de la Palestine: Seconde partie-Samarie (Paris, 1875), vol. 2, 241–2. 149 Condor, C. R. and Kitchner, H. H., The Survey of Western Palestine: Memories of the Topography, Orography, Hydrography and Archaeology (London, 1881), vol. 2, 153. 150 A discussion concerning the tower and its role in the Crusader period can be found in Kennedy, Castles, 36–7. 151 Benvenisti, Crusaders, 198–9. 152 Kurkar: fossilized sand dune. 153 Ibn ʿAbd al-Ẓāhir, Rawḍ, 396–7. According to Maqrīzī, the governor of the fortress fled. Maqrīzī, Sulūk, vol. 1, pt. 2, 600; al-ʿAynī, ʿIqd, vol. 2, 91; Ibn Shaddād gives a strange report and says that the fortress was destroyed by Bīlīk al-Khaznadār. This information is not supported by any other chronicler. Ibn Shaddād, Ta’rīkh, 252; Pringle, Red Tower, 62.

Notes  241 154 Marshall, Warfare, 193, 206. Ibn ʿAbd al-Ẓāhir mentions that 500 Franks were counted among the dead. Ibn ʿAbd al- Ẓāhir, Rawḍ, 397. 155 Pringle, D., “Towers of Crusader Palestine,” Château Gaillard 16 (1994): 335–50. 156 Of the eighty Crusader towers listed by Pringle very few were rebuilt or occupied as military posts during the Mamluk period. The tower at the foot of Mount Tabor in the village of Dabbūriyya was rebuilt by al-Muʿaẓẓam ʿĪsā in 1213, but was destroyed by Baybars in 1263. The tower at Baisan was turned into a sugar mill during the Mamluk period; Burj al-Aḥmar was probably destroyed by the Mamluks in 1265. Kurdana was initially destroyed by Baybars in 1267, though later it was restored and the mill was shared by Mamluks and Hospitallers. Pringle, D., Secular Buildings in the Crusader Kingdom of Jerusalem (Cambridge, 1997), 25, 39, 64. 157 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 69–70. 158 Milwright, M., “Central and southern Jordan in the Ayyubid period: historical and archaeological perspective,” JRAS 3 (2006), 26. On the early settlement of southern Jordan by the Franks see Walmsley, A., “ Fatimid, Ayyubid and Mamluk Jordan and the Crusader interlude,” in The Archaeology of Jordan, eds B. MacDonald, R. Adams and P. Bienkowski (Sheffield, 2001), 518–20. 159 Ibn Shaddād, Aʿlaq, vol. 2, pt. 2, 73. 160 Ibn al-Athīr, Kāmil, vol. 2, 95–6; Ibn Wāṣil, Mufarrij, vol. 5, 317. 161 Walmsley, “Jordan,” 521, 529, 531; Humphreys, Saladin, 285. 162 Humphreys, Saladin, 384. 163 The towns mentioned above were taken from al-Nāsīr Dāwūd when al-Ṣāliḥ Ayyūb rose to power. Ibn Wāṣil, Mufarrij, vol. 5, 364; Humphreys, Saladin, 206. 164 Humphreys, Saladin, 296–7. 165 Amitai-Preiss, Mongols, 19–21, 76; Amitai-Preiss, R. “Hülegü and the Ayyubid Lord of Transjordan (more on the Mongol Governor of al-Karak),” in Archibum Eurasiae Medii Aevi, 9 (1995–7), ed. Th. T. Allsen, P. B. Golden, A. P. Martinez and Th. S. Noonan (Wiesbaden, 1997), 5–16. 166 Ibn ʿAbd al-Ẓāhir, Rawḍ, 150; Yūnīnī, Dhayl, vol. 1, 358; Amitai-Preiss, Mongols, 34; According to al-Yūnīnī, al-Mughīth was imprisoned in the Cairo citadel, where he was probably murdered by the sultan’s men. Thorau, Baybars, 136, n. 13. 167 Ibid., 163. 168 Al-Maqrīzī, Taqī al-Dīn Aḥmad b. ʿAli, Ighāthat al Ummah bi-Kashf al-Ghummah (Cairo, 1999), 28–9; Allouche, A., Mamluk Economics: A Study and Translation of al-Maqrīzī, Ighātha (Salt Lake City, 1994), 44. 169 Ibn Shaddād, Ta’rīkh, 252. 170 Although the basic administrative structure was maintained throughout the Mamluk period there were many changes in the jurisdictions and territorial boundaries of Karak, which appear to have shifted and changed quite often. “Rivalries among officers, up­risings and revolts, and tribal relations may have influenced a sultan’s or amīr’s decision to promote one administrative center over another.” Walker, “Ḥisbān,” 242. 171 After the fortress surrendered, Baybars summoned the administrative clerks at Karak and saw to their salaries. He did not make any changes and decided to maintain their payrolls. Ibn ʿAbd al-Ẓāhir, Rawḍ, 124. For the administrative structure at Karak under al-Mughīth see Milwright, “Jordan,” 8. 172 Johns, J., “The longue durée: state and settlement in southern Transjordan across the Islamic centuries,” in Village, Steppe and State: The Social Origins of Modern Jordan,” eds E. L. Rogan and T. Tell, (London and New York, 1994), 17–19; Walker, “Mamluk investment”, 241–8. On rearing of horses see 243–4; Walmsley, “Fatimids”, 542–3; Irwin, Middle East, 185–6. 173 Ibn Wāṣil, Mufarrij, 276; Sibṭ Ibn al-Jawzī, Mir’āt, vol. 8, pt. 2, 561. 174 Ibn ʿAbd al-Ẓāhir, Rawḍ, 165. 175 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 79. 176 During al-ʿĀdil’s reign another treasury was kept at Qaʿlat Jaʿbar, located on the Euphrates on the southeast side of the modern Asad Dam.

242  Notes 177 Müller-Wiener, Castles, 47. 178 Kennedy, Castles, 47. 179 Boase, Castles, 70; Pringle, D. “Crusader castles in Jordan,” in The Archaeology of Jordan, eds B. MacDonald, R. Adams and P. Bienkowski (Sheffield, 2001), 678. 180 Sourdel, D., “Al-Karak,” EI2 4: 609. 181 Milwright, “Jordan,” 24–5. 182 Deschamps, P. Les Châteaux des Croiés en Terre Sainte. vol. II: La défense du royaume latin de Jérusalem (Paris, 1939), 75–7; Châteaux des de Kerak (Paris, 1934), vol. 1, Plan 1: Kerak Châteaux des et enceinte de la ville. 183 Johns, “Southern Transjordan,” 17. 184 Luynes, H. T., Voyage d’exploration à la Mer Morte, à Petra et sur la rive gauche du Jourdain (Paris, 1874), vol. 2, 199. 185 Ellenblum, Modern Histories, 178–9. 186 Ṣalāḥ al-Dīn laid siege to Karak in 1183, 1184 and 1188. The shortest of those sieges lasted four weeks; the longest went on for a year and a half, according to Ibn Shaddād al-Ḥalabī, Aʿlāq, vol. 2, pt. 2, 72, 71, 73. 187 Flint of the Mashash formation. 188 Kennedy, Castles, 50. 189 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 69. See n. 69. 190 Abū’l Fidā’, Syrian Prince (Holt), 48. 191 Walmsley, “Fatimid,” 521. 192 Ṣārim al-Dīn Abū Manṣūr Barghash al-ʿĀdilī does not appear in the biographies of Ibn Khallikān. 193 Combe et al., Répertoire chronologique, vol. 10, 276, inscription 3800A. 194 Gibb, H. A. B., “al-ʿĀdil,” EI2 1: 197–8. 195 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 73. The word ḥiṣn does not occur though the description is given in the masculine. When the author refers to the town as madīna he uses the feminine. 196 Ibn Wāṣil, Mufarrij, 276. 197 Sibṭ Ibn al-Jawzī, Mir’āt, vol. 8, pt. 2, 561. Ibn Wāṣil, mentions the treasury again when he describes the taking of the fortress by al-Saliḥ Ayyūb b. al-Kāmil in 644/1247. Ibn Wāṣil, Mufarrij, vol. 5, 364. 198 Ibn ʿAbd al-Ẓāhir, Rawḍ, 162–3. 199 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 79. 200 Ettinghausen, R., Grabar, O. and Jenkins-Madina, M., Islamic Art and Architecture 650–1250, 2nd edn (New Haven and London, 2001), 39, fig. 39. 201 Similar but more elaborate portals can be seen in Cairo at the mosque of al-Nāṣir Muḥammad (1318–35), the complex of Ḥasan built in 1356, the complex of Qa’itbāy (1472–4) and the mosque of Aqbughā al-Utrush (1399–1410). Blair, S. S. and Bloom, J. M., The Art and Architecture of Islam 1250–1800 (New Haven, CT, 1995), 74–90, figs.107, 110, 117. 202 A letter appointing a new nā’ib to the fortress of Ṣalḥad in 679/1280, contains a long list of the governor’s responsibilities. Among them are the fortress granaries, which were directly supervised by him. Ibn al-Furāt, Nāṣir al-Dīn ʿAbd al-Raḥmān b. Muḥammad , Ta’rīkh al-duwal wa’l-mulūk, vol. 7, ed. Zurayk, Q. (Beirut, 1942), 192. I would like to thank Dr Y. Frankel from the Department of Middle Eastern History at Haifa University for drawing my attention to this information. 203 Ibn Shaddād, Ta’rīkh, 252. 204 Ibn Shaddād, Aʿlāq, vol. 2, pt. 2, 79; Lane, Lexicon, bk. 1, pt. 2, 671. 205 Ibn Shaddād, Ta’rīkh, 351–352; Ṣafadī, Wafayāt, vol. 10, 341. Ṣafadī was probably following Ibn Shaddād. 206 Brown, “Karak,” 287–304. 207 ʿAli Ibrahim, “Residential architecture,” 53. 208 Ibn ʿAbd al-Ẓāhir, Rawḍ, 163–4.

Notes  243 209 Comfort, A., Abadie-Reynal, C. and Ergeç, R. “Crossing the Euphrates in antiquity: Zeugma seen from space,” Anatolian Studies 50 (2000):107–21. 210 The name Zeugma means “link” or “bridge” in Greek. Comfort et al., “Euphrates,” 101–2. 211 Keppie, L., The Making of the Roman Army, from Republic to Empire (London, 1998), 196. 212 Parker, H. M. D.,The Roman Legions (Oxford, 1928, 1980), 126. 213 Parker, Roman Legions, 259; Stark, F., Rome on the Euphrates (London, 1966), 252. 214 Comfort et al., “Euphrates,” 111, 107–8. 215 Ibn al-Athīr, Kāmil, 475 216 Al-Ẓāhirī Khalīl b. Shāhīn, Kitāb zoubdat Kashf al-Mamālik, ed. P. Ravaisse (Paris, 1884), 51. 217 Streck, M., “Biredjik,” EI2 1: 1233. 218 Ibn al-Athīr, Kāmil, vol. 11, 475, 482–4; ʿAbd al-Qādir ibn Muḥammad, Al-Dāris fī tārīkh al-Madāris (Beirut, 1990), vol. 2, 141. The latter mentions al-Bīra at the end of a list of sites conquered by Ṣalāḥ al-Dīn in 1179; Ehrenkreutz, Saladin, 176. 219 Several fragments of inscriptions were found in a survey conducted by M. Meryem and A. Durukan in 1997–8. Durukan, A. ., The Cultural Heritage in the Towns of Birecik, Halfeti, Suruç, Bozova and Rumkale, ed. A. Durukan, trans from Turkish A. T. Mitchell, M. Kadiroğlu Leube and A. Durukan (Ankara, 2003), 204–6. 220 Humphreys, Saladin, 310. 221 Bar Hebraeus, Chronography vol. 1, 435; Thorau, Baybars, 64; Humphreys, Saladin, 348. 222 Ibn ʿAbd al-Ẓāhir, Rawḍ, 133–5; Ibn al-Furāt, Ta’rīkh (Lyons), 68; Irwin, Middle-East, 46; Amitai-Preiss, Mongols, 61. 223 Streck, M., “Biredjik,” EI2 1: 1233–4; van Berchem, M., Arabische Inschiften aus Syrien Mesopotamien und Kleinasien: In Beiträge zur Assyriologie und semitischen Sprachwisssenschaft (Leipzig, 1909), VII, 1, 102–5. 224 Kürkçüoğlu, Cihat A., Birecik Monografisi (Ankara, 1996). 225 The photographs are kept in the Creswell Photographic Archives, Ashmolean Museum, Oxford. 226 Part of Creswell’s dating relied on inscriptions along the towers and gates, some of which can be seen in the photographs though they were not the main focus of the survey. 227 Ibn al-Furāt, Ta’rīkh (Lyons), vol. 2, 68. 228 Ibn Shaddād, Ta’rīkh, 358. 229 Ṣafadī, Wafayāt, vol. 10, 342. 230 Durukan, Cultural Heritage, 199. 231 Van Berchem, AISM, VII, 1, 101. 232 Ibn ʿAbd al-Ẓāhir, Rawḍ, 338; Maqrīzī, Sulūk, bk. 1, pt. 2, 515–16; Thorau, Baybars, 152–3, 196. 233 Rabbat, Citadel of Cairo, 134–5. 234 Yūnīnī, Dhayl, vol. 3, 114; Ibn Shaddād, Ta’rīkh, 124–6; Ibn al-Furāt, Ta’rīkh, vol. 7, 41. Although Baybars was residing at Damascus when he received the news (November 1275) of the Mongol attack he did not set out to the fortress. 235 Ibn Shaddād, Ta’rīkh, 358; Ṣafadī, Wafayāt, vol. 10, 342. 236 Tonghini, C., Donato, E., Montevecchi, N. and Nucciotti, M., “The evolution of masonry technique in Islamic military architecture: the evidence from Shayzar,” Levant 35 (2003), 179–212. See especially pages 190–3, figs 12, 18, 24. 237 Durukan, Cultural Heritage, 197–202. 238 The palace at Karak and at Qalʿat Najm have the same dimensions, both measuring 40 × 40m. 239 Sinclair, Eastern Turkey, vol. 4, 157–8. 240 Creswell’s interpretation is given beside the photographs in the Ashmolean Museum photographic archives, photographs, 6589–90. Sinclair, Turkey, vol. 4, 157. 241 Tabba, Constructions, 84–92; Brown, “Karak,” 287–304. 242 van Berchem, AISM, VII, 1, 102.

244  Notes 243 In 1259 the fortress was taken by Hülegü after a siege that lasted two weeks. The sieges of 1264–5, 1272 and 1274 all failed. A detailed discussion is presented in Chapter 3. 244 Known as Ranculat in the Frankish sources. The Armenians called it Hromgla. Qalʿat al-Rūm is the Arabic version found in contemporary twelfth- and thirteenth-century sources. The modern Turkish name is Rumkale. Stewart, A., “Qalʿat al-Rūm/Hrmogla/ Rumkale and the Mamluk siege of 691 ah/1291ce,” in Military Architecture in Greater Syria, ed. H. Kennedy (Leiden and Boston, 2006), 270–1. 245 Comfort et al., “Euphrates,” 113. 246 Bar Hebraeus, Chronography, vol. 1, 435. 247 The dam was completed in December 1999. The archaeological survey conducted in the area prior to the flooding included several archaeological teams from various universities around the world; it was named “Project Zeugma” after the classical town of Zeugma. The finds were published in two main articles that will be referred to in the following pages. 248 Comfort et al., “Euphrates,” 113. 249 Der Nersessian, S., “The Kingdom of Cilician Armenia,” in A History of the Crusades, gen. ed. K. M. Setton, II: The Later Crusades (Madison, 1969), 641–2; Boase, “Gazetteer,” 166–7; Sinclair, Eastern Turkey, vol. 4, 209–11. MacEvitt, C., “Christian authority in the Latin East: Edessa in Crusader history,” in The Medieval Crusade, ed. S. J. Ridyard (Woodbridge, 2004), 71–83; Stewart, “Qalʿat al-Rūm,” 271–80. 250 Sinclair, Eastern Turkey, vol. 4, 215. 251 Der Nersessian, “Kingdom of Cilician Armenia,” 653; Stewart, “Qalʿat al-Rūm,” 272. 252 This assault is only briefly mentioned by Baybars al-Manṣūrī, Tuḥfa, 88. 253 Bar Hebraeus, Chronography, vol. 1, 461. 254 Ibid. 255 Ibid. 256 Maqrīzī, Sulūk vol. 1, pt. 3, 778; Nuwayrī, Nihāyat, vol. 31, 226–7. 257 Rabbat, Citadel of Cairo, 148. 258 Ibid. 259 Sobernheim, M., “Die arabischen Inschriften von Aleppo,” DI 15 (1926): 176. 260 Lawrence, T. E., Oriental Assembly, ed. A. W. Lawrence (London, 1939), 28–37. 261 Maqrīzī, Sulūk, vol. 1, pt. 3, 778. 262 Edwards, Fortifications, see pl. 213b Sis and pl. 256b Tumlu. 263 Toy, S., A History of Fortifications, from 3000 b.c. to a.d. 1700, 2nd edn (London, 1966), 100. 264 Lawrence, Oriental Assembly, 30. 265 Lawrence, A. W., “The Castle of Baghras,” ed. T. S. R. Boase, The Cilician Kingdom of Armenia (Edinburgh and London, 1978), 166. 266 Edwards, Fortifications, 180–1. 267 Maqrīzī, Sulūk, vol. 1, pt. 3, 778. 268 Shaybānī, Muḥammad ibn al-Ḥasan, Kitāb al-Siyar al-Kabīr (Hyderabad, ah 1335), vol. 3, 212–13, cited in Khadduri, M., War and Peace in the Law of Islam (Baltimore and London, 1969), 106. Shaybānī was a jurist of the Hanafi school (b. 132/750). 269 Sinclair, Eastern Turkey, vol. 4: 170–2. 270 Sinclair, Eastern Turkey, vol. 4: 17; Lawrence, Oriental Assembly, 31–2. 271 Müller-Wiener, Castles, 71–72; Edwards, Fortifications, 244. 272 Maqrīzī, Sulūk, vol. 1, pt. 3, 839. 273 Sinclair, Eastern Turkey, vol. 4: 326–7. 274 Ibn ʿAbd al-Ẓāhir, Rawḍ, 280; Baybars al-Manṣūrī, Tuḥfa, 58; Thorau, Baybars, 174. 275 The raid, led by the governor of Aleppo, passed the fortress but did not attack it. Ibn ʿAbd al-Ẓāhir, Tashrīf, 31–2, 273. 276 Maqrīzī, Sulūk, vol. 1, pt. 3, 839. 277 Der Nersessian, “Kingdom of Cilician Armenia,” 657. 278 Al-ʿAynī, ʿIqd, 300–2.

Notes  245 279 Abū’l Fidā’, Syrian Prince (Holt), 44. According to Nuwayrī the fortress was not destroyed; Nuwayrī, Nihāyat, vol. 32, 76. 280 Boase, “Gazetteer,” 183; Edwards, Fortifications, 244–5. 281 Edwards, Fortifications, 245. 282 Ibid., 244. 283 Foss, C. and Winefield, D., Byzantine Fortifications: An Introduction (Pretoria, 1986), 23, 30. 284 Edwards, Fortifications, 14. 285 Ibn ʿAbd al-Ẓāhir, Tashrīf, 31 286 Sinclair, Eastern Turkey, vol. 4: 328. 287 Mentioned in Crusader sources as Gaston. 288 Molin, Unknown Castles, 185. 289 Lawrence, “Baghras,” 43–6; Barber, Knighthood, 35, 79. 290 Ibn al-Athīr, Kāmil, 18–19; Ibn Shaddād, Aʿlāq, vol. 1, pt. 1, 413. Ibn Shaddād gives a detailed historical account from 31/953 to the Mamluk conquest. Maqrīzī, Sulūk, vol. 1, pt. 1, 100. 291 Ibn Wāṣil, Mufarrij, vol. 2, 268–9; The Old French Continuation of William of Tyre, 1184–97, in P. W. Edbury, The Conquest of Jerusalem and the Third Crusade: Sources in Translation (Aldershot, 1998), 87. The German emperor died on the way while crossing Anatolia. 292 Edwards, “Bagras,” 431–2. 293 Barber, Knighthood, 121–2; Molin, Unknown Castles, 184. 294 Ibn Shaddād, Aʿlāq, vol. 1, pt. 1, 414. 295 Lawrence, “Baghras,” 82–3; Edwards, “Bağras,” 431–2. 296 Müller-Wiener, Castles, 48. 297 Edwards, “Bağras,” 431. 298 Ibn Shaddād, Aʿlāq, vol. 1, pt. 1, 414. 299 Sinclair, Eastern Turkey, vol. 4, 269. 300 In 684/1285 Qalāwūn and the Armenian King Leon signed a peace treaty meant to last for ten years, ten months and ten days, in which Baghrās together with other fortresses in Cilicia was handed over to the Mamluks. Ibn ʿAbd al-Ẓāhir, Tashrīf, 95–6; Edwards, “Bağras,” 420; Sinclair, Eastern Turkey, vol. 4, 269. 301 The conversion is mentioned in a short account by Ṣafadī, Wafayāt, vol. 10, 342. 302 Boas, Archaeology, 184–7; Ellenblum, “Construction,” 170–2; Kennedy, Castles, 65 303 Creswell, Fortifications, 39. 304 Boas, Military Orders, 166–8. 305 Ibid., 173. 306 Ibn ʿAbd al-Ẓāhir, Rawḍ, 93. 5  Military architecture versus political and military organization 1 2 3 4 5 6

Ellenblum, R., Crusader Castles and Modern Histories (Cambridge, 2007), 239–56. Holt, P. M., “The structure of government in the Mamluk sultanate,” The Eastern Mediterranean Lands in the Period of the Crusades, ed. P. M. Holt (Warminster, 1977), 46. Ibn Shaddād al-Ḥalabī, ʿIzz al-Dīn Muḥammad b. ʿAlī, al-Aʿlāq al-khaṭīra fī dhikr umarā al-shām wa’l-jazīra (Damascus, 1972), vol. 2, pt. 2, 150. Ibid., 147. Michaudel, B., “The use of fortifications as a political instrument by the Ayyubids and the Mamluks in Bilad al-Sham and Egypt (twelfth and thirteenth centuries),” MSR 11/1 (2007): 56. Irwin mentions a cases in the 690s/1290s when the area of al-Ṣubayba and its surroundings was given as mulk by al-Ashraf Khalīl to a favorite amir of his, Baydarā al-Ashrafī. R. Irwin, “Iqṭā’ and the end of the Crusader states,” in The Eastern Mediterranean Lands in the Period of the Crusades, ed. P. M. Holt (Warminster, 1977), 68.

246  Notes 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

24

25 26 27

28

29 30

Al-Yūnīnī, Dhayl Mirʿāt al-zamān, in L. Guo, Early Mamluk Syrian Historiography, (Leiden, 1998), 2 vols, vol. 1: 118 Keegan, J. A. History of Warfare (New York, 1993), 142–6. Tracy, J. D., “To wall or not to wall: evidence from medieval Germany,” in City Walls; the urban enceinte in global perspective, ed. J. D. Tracy (Cambridge, 2000), 71–88, 72. Keegan, Warfare, 142, A central authority struggling to secure itself will often use strongholds. Smail has a long and detailed chapter on recruitment in the Latin armies and the defects of the military establishment. Smail, R. C., Crusading Warfare, 1097–1193, 2nd edn (Cambridge, 1995), 87–106. Ibid., 102–3. Ibid., 98. Ibid., 97. Ibid., 104. Lev, Y., Saladin in Egypt (Leiden, Boston and Cologne, 1999), 154–7. Ibid., 160. Amitai-Preiss, R., Mongols and Mamluks: The Mamluk–Īlkhānid War 1260–1281 (Cambridge, 1995), 233–5. Ellenblum, Modern Histories, 270–3. Marshall, C. J., Warfare in the Latin East, 1192–1291 (Cambridge, 1996), 93, 131, 136–9. Sibṭ Ibn al-Jawzī, Mir’āt al-zamān fī ta’rīkh al-aʿyān (Hyderabad, 1952), vol. 8, pt. 2, 597–8, referring to affairs in 615/1218. Amitai-Preiss, Mongols, 76–7. Ibid., 71–3; Ayalon, D. “Auxiliary Forces,” 13–37, Ayalon, D. “Studies on the structure of the Mamluk army – I the army stationed in Egypt”, BSOAS 16 (1953): 204; Ayalon, D., “Studies on the structure of the Mamluk army – II, The Ḥalk. a”, BSOAS 16 (1953): 448–66; Ayalon, “Wafidiya,” 90. Isaac, B., The Limits of Empire: The Roman Army in the East (Oxford, 1993), 14, 372–6; ibid., “Luttwak’s ‘Grand Strategy’ and the eastern frontier of the Roman Empire,” in The Eastern Frontier of the Roman Empire, eds D. H. French and C. S. Lightfoot, BAR international series, 553 (1989), pt. 1, 231–4. See n. 99, Parker, H. M. D., The Roman Legions (Oxford, 1928), 259; Stark, F., Rome on the Euphrates (London, 1966), 252; Keppie, Roman Army, 191–6. Ibn al-Furāt, Ta’rīkh (Lyons), 99. ʿAmūdain and al-Tīna are located in Cilicia and belonged to the Teutonic Knights and the Templars. Ibn ʿAbd al-Ẓāhir, Muḥyī al-Dīn ʿAbd Allāh, Tashrīf al-ayyām wa’l-ʿuṣūr fī sīrat al-malik al-manṣūr, ed. M. Kāmil (Cairo, 1961), 80. ʿUmarī mentions postal stages at the following fortresses: al-Bīra, Qalʿat al-Muslimin (i.e. Qalʿat al-Rūm) and Baghrās, ʿUmarī, Taʿrīf, 195. Fortresses that had dovecotes: Safad, Qāqūn, Qalʿat al-Rūm, al-Bīra and al-Raḥba. Al-ʿUmarī, Aḥmad b. Yaḥyā ibn Faḍl Allāh, al-Taʿrīf fī al-muṣṭalaḥ al-sharīf (Cairo, 1894), 197. Al-Bīra and al-Raḥba could also use fire signals if they were attacked at the night. ʿUmarī, Taʿrīf, 199–200. Amitai-Preiss, Mongols, 75; Silverstein, A. J., Postal Systems in the Pre-Modern Islamic World (Cambridge, 2007), 172, 176–9. Har-El, S., Struggle for Domination in the Middle East: The Ottoman–Mamluk War 1485–91 (Leiden, New York and Cologne, 1995), 39; “those places” i.e. towns and fortresses east of the Amanus range and the Anti-Taurus as well as the area of the Jayḥān and Sayhan rivers. Al-Maqrīzī, Taqī al-Dīn Aḥmad b. ʿAlī, Kitāb al-sulūk li-maʿrifat duwal al-mulūk (Beirut, 1997), vol. 3, pt. 1, 347. Translation taken from Massoud, S. G., “Al-Maqrīzī as a historian of the reign of Barqūq,” MSR 7/2 (2003): 128. Maqrīzī, Sulūk, 3, pt. 2, 787; Ibn-Taghrī Birdī, Jamāl al-Dīn Abū al-Mahāsin, Al-Nujūm al-zāhira fī mulūk Miṣr wa-l-Qāhira (Cairo 1956), vol. 12, 261.

Notes  247 31 Har-El, Ottoman–Mamluk War, 68. 32 Maqrīzī, Sulūk, 3, pt. 2, 824. Maqrīzī does not explain what led to the replacement of this group of governors. Ibn-Taghrī Birdī gives a similar account but mentions fewer towns in which governors were replaced. Ibn-Taghrī Birdī, Nujūm, vol. 12, 59. 33 Ibn-Taghrī Birdī, History of Egypt 1382–1469 a.d. (pt. II, 1399–1411 a.d.), tran. W. Popper (Berkeley and Los Angeles, 1954), vol. 14, 33–5. 34 Ibid., vol. 14, 34. 35 Ibid., vol. 14, 39–51; Manz, B. F., The Rise and Rule of Tamerlane (Cambridge, 1989), 73. 36 Abū ’l-Maḥāsin Ibn-Taghrī Birdī, al-Nujūm al-zāhira fī mulūk miṣr wal-Qāhira, ed. W. Popper (Berkeley, 1920–3), vol. 6, pt. 2, 47; Ibn-Taghrī Birdī, History (Popper), vol. 14, 35. 37 The first full-scale Aqquyunlu attacks on the Mamluk towns and fortresses along the Euphrates took place during al-Ashraf Barsbay’s reign in 830/1427. Following the death of the Aqquyunlu leader Qara ʿUthmān in 839/1435 the Mamluks enjoyed almost forty years of quiet before the attacks were resumed. Woods, J. E., The Aqquyunlu Clan, Confederation, Empire, rev. and expanded edn (Salt Lake City, 1999), 50, 54, 116–17; Har-El, Ottoman–Mamluk War, 96–9. 38 Petry, C. F., Twilight of Majesty: The Reigns of the Mamluk Sultans al-Ashraf Qāytbāy and Qānsūh al-Ghawrī in Egypt (Seattle and London, 1993), 11–12. 39 Ibn Jīʿān, Al-Qawl al-mustẓaraf, 67. Ḥasan Bak is in fact the Aqquyunlu leader Uzun Ḥasan, his full title being Uzun Ḥasan b. ʿAlī Bayandur Abū al-Nāṣir. 40 Ibn Jīʿān Badr al-Dīn Abū’l-Baqā’ Ibn Yaḥyā, Al-Qawl al-mustẓaraf fī safar mawlānā al-malik al-ashraf (Bairut, 1984), 67–8; Streck, M., “Biredjik,” EI2 1: 1233–4 41 Ibn Iyās, Muḥammad Ibn Aḥmad, Badā’iʿ al-zuhūr fī waqa’iʿ al-duhūr, ed. M. Mostafa (Wiesbaden, 1963), vol. 3, 196. 42 Creswell, K. A. C., “Archaeological researches in the Citadel of Cairo,” Bulletin de l’Institut français d’Archéologie orientale 23 (1924): 125, n. 1.

Bibliography

Primary sources Arabic sources ʿAbd al-Qadir Ibn Muḥammad, al-Dāris fī tā’rīkh al-madāris, ed. Ibrāhīm Shams al-Dīn (Beirut, 1990). Abū’l-Fidā’, al-Malik al-Muʾayyad ʿImād al-Dīn Ismāʿīl b. ʿAlī, al-Mukhtaṣar fī ta’rīkh al-bashar (Istanbul, 1869–70). —— al-Mukhtaṣar fi ta’rīkh al-bashar (Beirut, 1997), vol. 2. —— al-Mukhtaṣar fī ta’rīkh al-bashar, eds M. Zinhum and Y. al-Saʿīd (Cairo, 1999), vol. 4. —— The Memoir of a Syrian Prince Abū’l-Fidā, Sultan of Hamā 672–732 (1273–1331), trans. P. M. Holt (Wiesbaden, 1983). Abū Shāma, Shihāb al-Dīn ʿAbd al-Raḥmān b. Ismāʿīl, Kitāb al-rawḍatayn fī akhbār al-dawlatayn (Beirut, 1997), vols 2–3. —— Tarājim rijāl al-qarnayn al-sādis wa’l sābiʿ al-maʿrūf bi’l-dhayl ʿalā al-rawḍatayn, ed. M. al-Kawtharī (Cairo, 1947). al-ʿAynī, Badr al-Dīn Maḥmūd b. ʿAlī, ‘Iqd al-jumān fī ta’rīkh ahl-zamān, ed. M. M. Amīn (Cairo,1988–1992), vols 2–4. Baybars al-Manṣūrī al-Dawādār, Rukn al-Dīn, Kitāb al-tuḥfa al-mulūkiyya fī l’-dawla al-turkiyya, ed. ʿA-R. S. Ḥamdān (Cairo, 1987). —— Mukhtār al-akhbār ta’rīkh al-dawla al-Ayyūbiyya wa-dawla al-mamalīk al-baḥriyya ḥattā 752 H., ed. ʿA-R. S. Ḥamdān (Cairo, 1993). al-Dimashqī, Shams al-Dīn Abū ʿAbd Allāh Muḥammad b. Abī Ṭālib al-Ansārī al-Ṣūfī, Kitab al-Nukhbat al-dahr fī ‘ajā’ib al-barr wa’l-baḥr, ed. M. A. F. Mehren (Saint Pétersbourg, 1866). Ibn ʿAbd al-Ẓāhir, Muḥyī al-Dīn ʿAbd Allāh, Tashrīf al-ayyām wa’l-ʿuṣūr fī sīrat al-malik al-manṣūr, ed. M. Kāmil (Cairo, 1961). —— al-Rawḍ al-ẓāhir fī sīrat al-malik al-ẓāhir, ed. ʿA. ʿA. al-Khuwayṭir (Riyad, 1976). Ibn al-ʿAdim, Kamāl al-Dīn, Zubdat al-Ḥalab min ta’rīkh Ḥalab, ed. Sāmī al-Dahhān (Damascus, 1951), vol. 3. Ibn al-Athīr, ʿIzz al-Dīn ʿAlī, al-Kāmil fī’l-ta’rīkh, ed. C. J. Tornberg (Beirut, 1966), vol. 12. Ibn Baṭṭūṭa, Riḥla Ibn Baṭṭūṭa, ed. T. Ḥarb (Beirut, 1992). The Travels of Ibn Baṭṭūta a.d. 1325–1354, trans. H. A. R. Gibb (Cambridge, 1962), 2 vols. Ibn al-Dawādārī Abū Bakr b. ʿAbd Allāh, Kanz al-durar wa Jāmī al-ghurar, ed. H. R. Roemer (Cairo, 1971). Ibn al-Furāt, Nāṣir al-Dīn ʿAbd al-Raḥmān, b. Muḥammad, Tā’rīkh ibn al-furāt, eds N. Izzedin and C. K. Zurayk (Beirut, 1939), vols 7–8. —— Ayyubids, Mamluks and Crusaders: Selections from the Tā’rīkh al-duwal wa’l-mulūk of Ibn al-Furāt, ed. and trans. U. and M. C. Lyons; Introduction and notes J. S. C. Riley-Smith (Cambridge, 1971).

Bibliography  249 Ibn Iyās, Muḥammad Ibn Aḥmad, Badā’iʿ al-zuhūr fī waqa’ iʿ al-duhūr, ed. M. Mostafa (Wiesbaden, 1963). Ibn Jīʿān, Badr al-Dīn Abū’l-Baqā’, Yaḥyā, Al-qawl al-mustaẓraf fī safar mawlānā al-malik al-ashraf (Beirut, 1984). Ibn Khallikān, Wafayāt al-a‘yān (Beirut, 1988), vol. 3. Ibn al-Qalānisī, The Damascus Chronicle of the Crusades, extracted and trans. from the Chronicle of Ibn al- Qalānisī, trans. H. A. R. Gibb (London, 1932). Ibn Shaddād, al-Ḥalabī, ʿIzz al-Dīn Muḥammad b.ʿAlī, al-Aʿlāq al-khaṭīra fī dhikr umarā’ al-shām wa’l-jazīra, vol. 1, pt. 1 and 2: Tā’rīkh ḥalab, ed. Y. Z. ʿAbbāra (Damascus, 1991). —— al-Aʿlāq al-khaṭīra fī dhikr umarā’ al-shām wa’l-jazīra (Damascus, 1972). —— Al-Aʿlāq al-khaṭīra … vol. 2, pt. 2: Ta’rīkh lubnān, al-urdunn wa-filasṭīn, ed. S. Dahhan (Damascus, 1963). —— Ta’rīkh al-malik al-ẓāhir, ed. A. Khuṭayṭ (Wiesbaden, 1983). Eddé-Terrasse, A. M, ‘Izz al-Dīn Ibn Šaddād, Description de la Syrie du Nord, traduction annotée de al-Aʿlāq al-Ḥaṭīra fī Ḍikr Umarā’ al-Šam wal-Ğazīra (Damascus, 1984). Ibn Taghrī Birdī, Abū ‘l-Maḥāsin, al-Nujūm al-zāhira fī mulūk miṣr wa’l-qāhira (Cairo, 1938– 72), vols 8, 12, 16. —— al-Nujūm az-zāhira fī mulūk miṣr wa’l-qāhira, ed. W. Popper (Berkeley, CA, 1920–3), vol. 6, pt. 2. —— History of Egypt 1382–1469 a.d. (Part II 1399–1411 a.d.), trans. from the Arabic annals of Abū ‘l-Maḥāsin Ibn Taghrī Birdī by W. Popper (Los Angeles and Berkeley, 1954), University of California Publications in Semitic Philology, vol. 14. Ibn Wāṣil, Jamāl al-Dīn Muḥammad b. Sālim, Mufarrij al-kurūb fī akhbār banī ayyūb (Alexandria, 1953), vol. 3. —— Mufarrij al-kurūb fī akhbār banī ayyūb, ed. G. al-Dīn al-Shayyal (Cairo, 1957), vol. 2. ʿImād al-Dīn al-Iṣfahānī, Ḥurūb Ṣalāḥ al-Dīn wa-fatḥ bayit al-maqdis (Beirut, 2003). al-Maqrīzī, Taqī al-Dīn Aḥmad b. ʿAli. Kitāb al-sulūk li-ma’rifat duwal al-mulūk, eds M. M. Ziyāda, and. S. ʿA.-F. Āshūr (Cairo, 1934–73), 4 vols in 12 parts. —— Kitāb al-Sulūk li-ma’rifat duwal al-mulūk (Beirut, 1997), vol. 1. —— A History of the Ayyubid Sultans of Egypt, trans. R. J. C. Broadhurst (Boston, 1980). —— Ighāthat al Ummah bi-Kashf al-Ghummah (Cairo, 1999). Allouche, A., Mamluk Economics: A Study and Translation of al-Maqrīzī, Ighātha (Salt Lake City, 1994) al-Nuwayrī, Shihāb al-Dīn Aḥmad b. ʿAbd al-Wahhāb, Nihāyat al-arab fī funūn al-adab (Cairo, 1992), vols 31, 32. al-Qalqashandī, Shihāb al-Dīn Aḥmad, Ṣubḥ al-aʿshā fī ṣināʿat al-inshā’ (Beirut, 1987), vol. 4. —— Ṣubḥ al-aʿshā fī ṣināʿat al-inshā’ (Cairo, 1985), vol. 14. al-Ṣafadī, Ṣalāḥ al-Dīn Khalīl b. Aybeg, al-Wāfī bi’l-wafayāt, ed. H. Ritter et al. (Wiesbaden, 1980), vol. 10. Shāfī’ Ibn ʿAlī, Biography of the Mamluk Sultan Qalāwūn, ed. P. B. Lewicka (Warsaw, 2000). Orientalia Polona. Sibṭ ibn al-Jawzī, Mir’āt al-zamān fī ta’rīkh al-aʿyān (Hyderabad, 1952), vol. 8, pt. 1. ʿUmar Ibn Ibrāhīm al-Awasī al-Anṣārī, A Muslim Manual of War, Tafrīj al-kurūb fī tadbīr al-ḥurūb written by ‘Umar Ibn Ibrahim al-Awasī al-Ansarī, ed. and trans. G. T. Scanlon (Cairo, 1961). al-ʿUmarī, Aḥmad b. Yaḥyā ibn Faḍl Allāh, Taʿrīf fī al-muṣṭalaḥ al-sharīf (Cairo, 1894–5). Yāqūt al-Rūmī, Muʿjam al-Buldān (Beirut, 1957), vol. 3. al-Yūnīnī Quṭb al-Dīn Mūsā b. Muḥammad, Dhayl mir’āt al-zamān fī ta’rīkh al-aʿyān (Hyderabad, 1954–61), 4 vols. —— Dhayl Mirʿāt al-zamān, in L. Guo, Early Mamluk Syrian Historiography (Leiden, 1998), 2 vols. al-Ẓāhirī, Khalīl b. Shāhīn, Kitāb zubdat Kashf al-mamālik, ed. P. Ravaisse (Paris, 1884). The Quran, text, translation and commentary by Abdullah Yusuf Ali (Cambridge, MA, 1946)

250  Bibliography Western sources Anon., De constructione castri Saphet, in H. Kennedy, Crusader Castles (Cambridge, 1994), appendix 190–8. Burchard of Mount Sion, A Description of the Holy Land, trans. from Latin by A. Stewart (London, 1896), PPTS, vol. 12. Crusader Syria in the Thirteenth Century: The Rothelin Continuation of the History of William of Tyre with Part of the Eracles or Acre Text, Crusader Texts in Translation, trans. by J. Shirley (Brookfield, VT, 1999). Fulcher of Chartres, A History of the Expedition to Jerusalem 1095–1127, trans. F. R. Ryan, ed. H. S. Fink (University of Tennessee, 1969). Gesta Francorum et aliorum Hierosolimitanorum, ed. R. Hill (London 1962). Jacques de Vitry, The History of Jerusalem a.d. 1189, trans. from Latin by A. Stewart (London, 1896), PPTS, vol.11, pt. 2. Marino Sanuto, Liber secretorum fidelium crucis super Terrae Sanctae recuperatione et conservatione, intro. J. Prawer (Jerusalem,1972). Marco Polo, The Travels of Marco Polo, trans. R. Latham (Harmondsworth, 1958). The Old French Continuation of William of Tyre, 1184–97, in trans. P. W. Edbury, The Conquest of Jerusalem and the Third Crusade: Sources in Translation, trans. P. W. Edbury (Aldershot, 1998). The Templar of Tyre, “Part III of the Deeds of the Cypriots,” trans. P. Crawford (Aldershot, 2003). Vitruvius, The Ten Books on Architecture, trans. M. H. Morgan (Cambridge, MA and London, 1926). —— Vitruvius’ Ten Books on Architecture, trans., prefaced and annotated R. Reich (Tel-Aviv, 1997). [Hebrew] William Archbishop of Tyre, A History of The Deeds Done Beyond the Sea, trans. E. A. Babcock and A. C. Krey (New York, 1943). Yule, H., The Book of Ser Marco Polo the Venetian (London, 1871). Eastern Christian sources Bar Hebraeus, The chronography of Grighor Abū‘l Faraj, the son of Aaron, the Hebrew physician commonly known as Bar Hebraeus. Being the first part of his political history of the world, ed. and trans. from the Syriac by E. A. W. Budge (Amsterdam, 1976), vol. 1 Daniel of Kiev, Pilgrimage of the Russian Abbot Daniel in the Holy Land 1106–1107 a.d., trans. from Russian by C. W. Wilson (London, 1985), PPTS, vol. 4. Der Nersessian, S., “The Armenian Chronicle of the Constable Smpad or of the “Royal Historian,” DOP 13 (1959): 142–68. Hetoum, A Lytell Cronycle, ed. G. Burger (Toronto,1988). Persian sources Juwaynī, ʿAlā’ al-Dīn ʿAṭā Malik, Ta‘rīkh-i-jahān-gushā, ed. M. M. Qazwīnī (Leiden and London, 1911–1937) Juwaynī, The History of the World – Conqueror, ed. and trans. J. A. Boyle (Cambridge, MA, 1958), 2 vols. Rashīd al-Dīn Faḍal Allāh, Jāmiʿ al-tawārīkh, ed. B. Karimi (Teheran, 1959), vol. 2. —— The Successors of Genghis Khan, trans. J. A. Boyle (New York and London, 1971). —— Jāmiʿ al-tawārīkh, ed. and trans. W. M. Thackston (Cambridge, MA, 1998–9), 3 vols. Spuler, B., A History of the Mongols Based on Eastern and Western Accounts of the Thirteenth and Fourteenth Centuries, trans. from German by H. and S. Drummond (London, 1972). Waṣṣaf, Geschichte Wassafs, ed. in Persian and trans. into German by Josef von HammerPurgstall (Vienna, 1856), vol. 1.

Bibliography  251 Miscellaneous Mission to Asia: Narratives and Letters of the Franciscan Missionaries in Mongolia and China in the Thirteenth Century, trans. by a Nun of Stanbrook Abbey, ed. C. D’Ohsson (London, 1980). The Secret History of the Mongols: A Mongolian Epic Chronicle of the Thirteenth Century, trans. and ed. with historical and philological commentary by I. De Rachewiltz (Leiden, 2004), 2 vols. Yuanshi (Beijing, 1978). Atlases Brawer, M., Atlas Carta of the Middle East (Jerusalem, 1990). [Hebrew] Kennedy, H., ed., Historical Atlas of Islam, 2nd ed., rev. (Leiden, 2002). Pourkamal, M., ed. Historical Atlas of Iran (Teheran, 1971). Riley-Smith, J., The Atlas of the Crusades (New York and Oxford, 1991). Dictionaries Doerfer, G., Türkische und Mongolische Elemente im Neupersischen, vol. 1 (Wiesbaden, 1963). Dozy, R., Supplément aux Dictonnaires Arabes (Beyrouth, 1991), 2 vols. Lane, E. W., Arabic–English Lexicon (New York, 1955). Sharoni, A., The Comprehensive Arabic–Hebrew Dictionary (Tel-Aviv, 1987), 3 vols. Steingass, F., Arabic–English Dictionary (London, 1884). Wehr, H., A Dictionary of Modern Written Arabic, ed. J. M. Cowan, 4th edn (Wiesbaden,1994). Modern studies ʿAbd al-Malik, S. S., “Les Mosquées du Sinaï au Moyen Âge,” in Le Sinaï durant l’Antiquité et le Moyen Âge: 4000 ans d’histoire pour un désert, eds D. Valbelle, C. Bonnet (Paris, 1998), 171–6. ʿAli Ibrahim, L., “Residential architecture in Mamluk Cairo,” Muqarnas 2 (1984): 47–59. Allsen, T., Mongol Imperialism: The Policies of the Grand Qan Möngke in China, Russia and the Islamic Lands 1251–1259 (Los Angeles, 1987). —— Culture and Conquest in Mongol Eurasia (Cambridge, 2001). —— “The circulation of military technology in the Mongolian Empire,” in Warfare in Inner Asian History (500–1800), ed. N. di Cosmo (Leiden and Boston, 2002), 265–93. Ambraseys, N. N. and Melville, C. P., “An analysis of the Eastern Mediterranean earthquake of 20 May 1202,” in Historical Seismograms and Earthquakes of the World, eds W. H. K. Lee, H. Meyers and K. Shimazaki (London, 1998), 181–200. Amitai-Preiss, R., “In the aftermath of Ayn Jālūt: the beginning of the Mamlūk–Ilkhānid Cold War,” Al-Masāq 3 (1990): 1–21. Amitai-Preiss, R., Mongols and Mamluks: The Mamluk–Ilkhanid War 1260–1281 (Cambridge, 1995). Amitai-Preiss, R., “Ṣafad,” EI2 8 (1995): 757–9. Amitai-Preiss, R., “New material from the Mamluk sources for the biography of Rashid al-Din,” in The Court of the Il-khans 1290–1340, eds J. Raby and T. Fitzherbert, Oxford Studies in Islamic Art XII (Oxford, 1996), 23–37. Amitai-Preiss, R., “The Mamluk officer class during the reign of Sultan Baybars,” in War and Society in the Eastern Mediterranean, 7th–15th Centuries, ed. Y. Lev (Leiden, 1997), 267–300. Amitai, R., “An Arabic inscription at al-Ṣubayba (Qalʿat Namrud) from the reign of Sultan Baybars,” in M. Hartal, The Al-Ṣubayba (Nimrod) Fortress, Towers 11 and 9 (Jerusalem, 2001), 109–23.

252  Bibliography Amitai, R., “Notes on the Ayyubid inscriptions at Al-Ṣubayba (Qalʿat Nimrod),” DOP 43 (1989): 113–19. Amitai-Preiss, R., “Hülegü and the Ayyubid Lord of Transjordan (more on the Mongol Governor of al-Karak),” in Archivum Eurasiae Medii Aevi, 9 (1995–7), eds Th. T. Allsen, P. B. Golden, A. P. Martinez and Th. S. Nonan (Wiesbaden, 1997), 5–16. Amitai-Preiss, R., “Mongol imperial ideology and the Ilkanid war against the Mamluks,” in The Mongol Empire and its Legacy, eds R. Amitai-Preiss and D. Morgan (Leiden, 1999), 57–72. Amitai, R., “Whither the Ilkhanid army ? Ghazan’s first campaign into Syria (1299–1300),” in Warfare in Inner Asian History (500–1800), ed. N. di Cosmo, (Leiden and Boston, 2002), 221–64. Amitai, R., “Foot soldiers, militiamen and volunteers in the early Mamluk army,” in Texts, Documents and Artifacts, ed. C. F. Robinson (Leiden, 2003), 232–49. Amitai, R,. “Edward of England and Abagha Ilkhan,” in Tolerance and Intolerance, eds M. Gervers and J. M. Powell (Syracuse, 2004), 75–80. Amitai, R. “The Mongol occupation of Damascus in 1300: a study of Mamluk loyalties,” in The Mamluks in Egyptian and Syrian Politics and Society, eds M. Winter and A. Levanoni (Leiden and Boston, 2004), 21–41. Amitai, R. “The conquest of Arsuf by Baybars: political and military aspects,” MSR 9/1 (2005): 61–83. Ashtor, E., Levant Trade in the Later Middle Ages (Princeton, NJ, 1983). Atiya, A. S., The Crusade in the Later Middle Ages (New York, 1965). Ayalon, D. “Egypt as a dominant factor in Syria and Palestine during the Islamic period,” in Egypt and Palestine: A Millennium of Association (868–1948) (Jerusalem and New York, 1948), 17–47. [Rpt. in Outsiders in the Lands of Islam: Mamluks, Mongols and Eunuchs, London, Variorum, 1988, no. II.] —— “The Wafidiya in the Mamluk kingdom,” in Islamic Culture 25 (1951): 89–104. [Rpt. in Studies on the Mamluks of Egypt (1250–1517), London, Variorum, 1977, no. II.] —— “Studies in the structure of the Mamluk army – I: The army stationed in Egypt,” BSOAS 15 (1953): 203–28. [Rpt. in Studies on the Mamluks of Egypt (1250–1517), London, Variorum, 1977, no. I] —— “Studies in the structure of the Mamluk army – II: The Ḥalk. a,” BSOAS 15 (1953): 448–76. [Rpt. in Studies on the Mamluks of Egypt (1250–1517), London, Variorum, 1977, no. I] —— “Studies in the structure of the Mamluk army – III,” BSOAS 15 (1954): 57–90. [Rpt. in Studies on the Mamluks of Egypt (1250–1517), London, Variorum, 1977, no. I.] —— “Baḥriyya,” EI2 1 (1960): 944–5. —— “Notes on the Furūsiyya exercises and games in the Mamluk sultanate,” in Studies in Islamic Civilization and History = Scripta Hierosolymitana, ed. U. Heyd (Jerusalem), 9 (1961): 31–62. [Rpt. in The Mamluk Military Society, London, Variorum, 1979, no. II.] —— “The Mamluks and naval power: a phase of the struggle between Islam and Christian Europe,” in Proceedings of the Israel Academy of Sciences and Humanities 1/8 (1967): 1–12. [Rpt. in Studies on the Mamluks of Egypt (1250–1517), London, Variorum, 1977, no. VI.] —— “The Muslim city and the Mamluk military aristocracy,” in Proceedings of the Israel Academy of Sciences and Humanities 2 (1968): 311–29. [Rpt. in Studies on Mamluks of Egypt (1250–1517), London, Variorum, 1977, no. VII.] —— “Le Régiment Baḥriyya dans l’armée Mamelouke,” in Studies on the Mamluks of Egypt (1250–1517) (London, Variorum, 1977), III, 133–41. —— “The eunuchs in the Mamluk sultanate,” in Studies in Memory of Gaston Wiet (Jerusalem, 1977), 267–95. [Rpt. in The Mamluk Military Society London, Variorum, 1979, no. III.] —— Gunpowder and Firearms in the Mamluk Kingdom, 2nd edn (London, 1978). —— “The auxiliary forces of the Mamluk sultanate,” DI 65 (1988): 13–37. [Rpt. in Islam and the Abode of War, London, Variorum, 1994, no. VI.] —— “Mamlūk,” EI2 6 (1991): 314–31.

Bibliography  253 —— “Karak and (Shawbak) in the Mamluk period,” Qadmoniot 111 (1997): 44–52. [Hebrew] —— Eunuchs, Caliphs and Sultans: A Study in Power Relations (Jerusalem, 1999). Bacharach, J. L., “Administrative complexes, palaces and citadels: changes in the loci of medieval Muslim rule,” in The Ottoman City and Its Parts, eds I. Bierman, R. A. Abou-Al-Haj and D. Preziosi (New York, 1991), 111–28. Barbé, H. and Damati, E., “La forteresse médiévale de Safed: données récentes de l’archéologie,” Crusades 3 (2004): 171–3. —— “Le Chāteau de Safed: sources historiques, problématique et premiers resultats des recherches archéologique,” in Archéologie et culture; la fortification au temps des Croisades, eds N. Faucherre et al. (Rennes, 2004), 77–93. Barbé, H., Avisar, M. and Lehrer, Y., “Habonim,” Hadashot Arkheologiyot. Excavations and Surveys in Israel 114 (2002): 34–8. [Hebrew] Barber, M., The New Knighthood: A History of the Order of the Temple (Cambridge,1995). Battista, A. and Bagatti, B., La Fortezza Saracena del Monte Tabor (ah 609–15; ad 1212–18) (Jerusalem, 1976). Barthoux, J., “Description d’une forteresse de Saladin découverte au Sinai,” Syria 3 (1922): 43–57. Behrens-Abouseif, D., Islamic Architecture in Cairo: An Introduction (Leiden and New York, 1989). —— “Muhandis, Shād, Mu’allim: notes on the building craft in the Mamluk period,” DI 72 (1995), 293–309. Benvenisti, M., The Crusaders in the Holy Land (Jerusalem, 1976). Berchem, M. van, Arabische Inschriften aus Syrien, Mesopotamien und Kleinasien, in Beiträge zur Assyriologie und Semitischen sprachwisssenschaft (Leipzig, 1909), VII, 1. Biller, T. et al., Der Crac des Chevaliers (2006, Berlin). Biran, M., Qaidu and the Rise of the Independent Mongol State in Central Asia (Richmond, 1997). Blair, S. S., “Decoration of city walls in the medieval Islamic world: the epigraphic message,” in City Walls: The Urban Enceinte in Global Perspective, ed. J. D. Tracy (Cambridge, 2000), 488–529. Blair, S. S. and Bloom, J. M., The Art and Architecture of Islam 1250–1800 (New Haven, NJ, 1995). Boas, A. J., Crusader Archaeology (London, 1999). —— Archaeology of the Military Orders (London and New York, 2006). Boase, T. R. S. Castles and Churches of the Crusading Kingdom (London, 1967). —— “The history of the Kingdom,” in The Cilician Kingdom of Armenia, ed. T. S. R. Boase (Edinburgh and London, 1978), 1–33. —— “Gazetteer,” in The Cilician Kingdom of Armenia, ed. T. S. R. Boase, (Edinburgh and London, 1978), 145–85. Boyle, J. A., “The death of the last Abbasid caliph: a contemporary Muslim account,” JSS 6/2 (1961): 145–61. —— The Cambridge History of Iran (Cambridge, 1968), vol. 5. Bradbury, J., The Medieval Siege (Bury St Edmunds, 1992). Braudel, F., The Structures of Everyday Life (London, 1981). Boardman, J., Griffin, J. and Murray O., The Oxford History of the Classical World (London, 1988). Brown, R. A., English Castles (London, 1976). Brown, R. M., “Summary report of the 1986 excavations in late Islamic Shobak,” ADAJ 32 (1988): 225–54. —— “Excavations in the fourteenth-century Mamluk palace at Karak,” ADAJ 33 (1989): 287–304. Cahen, C. “Ayyubids,” EI2 1 (1960): 796–807. —— “Ibn al Furāt,” EI2 3 (1979): 768–9. Chain, M., The Kingdom of Armenia (New York,1987).

254  Bibliography Chase, K., Firearms: A Global History (Cambridge, 2003). Chevedden, P. E., The Citadel of Damascus, PhD diss., University of Los Angeles, 1986, unpublished. —— “Artillery in late antiquity,” in The Medieval City under Siege, eds I. Corfis and M. Wolfe (Woodbridge, 1995), 131–73. —— “The trebuchet: recent reconstructions and computer simulations reveal the operating principles of the most powerful weapon of its time,” Scientific American (July, 1995): 66–71. —— “Fortifications and the development of defensive planning during the Crusader period,” in The Circle of War in the Middle Ages, eds D. J. Kagay and L. J. A. Villalon (Woodbridge, 1999), 34–43. Combe, E. et al., eds. Répertoire chronologique d’épigraphie arabe, Publications de l’Institut Français d’Archéologie orientale (Cairo, 1931, 1943), vols 10, 13. Comfort, A. and Ergeç, R., “Following the Euphrates in antiquity: north–south routes around Zeugma,” Anatolian Studies 51 (2001): 19–49. Comfort, A., Abadie-Reynal, C. and Ergeç, R., “Crossing the Euphrates in antiquity: Zeugma seen from space,” Anatolian Studies 50 (2000): 99–126. Conder, C. R. and Kitchener, H. H., Survey of Western Palestine: Memories of the Topography, Orography, Hydrography and Archaeology (London, 1881), vol. 2. Creswell, K. A. C. “Archaeological researches in the Citadel of Cairo,” In Bulletin de l’institut Français d’archéologe orientale 23 (1924):89–158. —— Fortifications in Islam before a.d. 1250, Proceedings of the British Academy 38 (London, 1952). Crowe, Y., “Samarkand,” EI2 8 (1995): 1031–8. Cytryn-Silverman, K., The Road Inns (Khans) of Bilad al-Sham during the Mamluk Period (1260–1516): An Architectural and Historical Study, PhD diss., The Hebrew University of Jerusalem, 2004, unpublished. Damati, E. “Safed,” ESI 4 (1985): 98. —— “Safed,” ESI 5 (1986): 93–4. —— “Safed Citadel,” ESI 7–8(1988/9): 159–60. —— “Safed Citadel,” ESI 9 (1989/90): 13. Dangles, P. et al., Le Chateau d’Āğlūn/Qal ʿat ar-Rabad Jordanie, Report intermédiaire de la mission menée en mars et avril, I.N.R.A.P., 2002. Dawkins, R. and Krebs, J. R., “Arms races between and within species,” Proceedings of the Royal Society of London, Ser. B 205 (1979): 489–511. De La Croix, H., Military Considerations in City Planning: Fortifications (New York, 1972). Der Nersessian, S., “The kingdom of Cilician Armenia,” in A History of the Crusades, gen. ed. K. M. Setton (Philadelphia, 1962), vol. 2, 630–59. Deschamps, P., Les Châteaux des Croisés en Terre Sainte I: Le Crac des Chevaliers (Paris, 1934), 2 vols. —— Les Châteaux des Croisés en Terre Sainte II: La défense du royaume latin de Jérusalem (Paris, 1939), 2 vols. De Vries, B., “Hisban in the Ayyubid and Mamluk periods,” in Hesban after 25 Years, eds D. Merlin and L. T. Geraty (Berrien Springs, MI, Andrews University, 1994), 151–66. De Vries, K., Medieval Military Technology (Peterborough, Ontario, 1998). D’Ohsson, C., Histoire des Mongols (Amsterdam, 1834; reprinted in Tientsin, China, 1940), vol. 1. Di Cosmo, N., Ancient China and its Enemies; The Rise of Nomadic Power in East Asian History (Cambridge, 2002). Drory, J., “ Founding a new Mamlaka: some remarks concerning Safed and the organization of the region in the Mamluk period,” in The Mamluks in Egyptian and Syrian Politics and Society, eds M. Winter and A. Levanoni (Leiden and Boston, 2004), 163–87. Duffy, C., Fire and Stone: The Science of Fortress Warfare 1660–1860 (London and Vancouver, 1975). Dunlop, D. M., “Dimashqī,” EI2 2 (1965): 291.

Bibliography  255 Durukan, A. et al., The Cultural Heritage in the Towns of Birecik, Halfeti, Suruç, Bozova and Rumkale, ed. A. Durukan, trans. from Turkish by A. T. Mitchell, M. Kadiroğlu Leube and A. Durukan (Ankara, 2003). Eddé, A. M., La principauté ayyoubide d’Alep (Stuttgart, 1999). Edwards, R. W., “Bagras and Armenian Cilicia: a reassessment,” Revue des Etudes Arméniennes 17 (1983): 415–35. —— The Fortifications of Armenia Cilicia (Washington, DC, Dumbarton Oaks Research Library and Collection, 1986). Ehrenkreutz, A., Saladin (New York, 1972). —— “Strategic implications of the slave trade between Genoa and Mamluk Egypt in the second half of the thirteenth century,” in The Islamic Middle East, 700–1900: Studies in Economic and Social History, ed. A. L. Udovitch (Princeton, NJ, 1981), 335–47. Elad, A., “The cities of Palestine during the early Arab period (604–1066 a.d.) according to Arabic sources,” Cathedra 8 (1978): 156–78. [Hebrew] Ellenblum, R. Frankish Rural Settlement in the Latin Kingdom of Jerusalem (Cambridge, 1988). —— “Who built Qalat al-Subayba?,” DOP 43 (1989): 103–12. —— “Construction methods in Frankish rural settlements,” in The Horns of Hattin, ed. B. Z. Kedar (Jerusalem, 1992), 168–89. —— “Frankish castles in the Crusader Kingdom of Jerusalem: the third generation,” in Le concile de Clermont et la premiére Croisade, ed. M. Balard (Paris, 1996), 518–51. —— Crusader Castles and Modern Histories (Cambridge, 2007). Ellenblum, R., Marco, S., Agnon, A. et al. “Crusader castle torn apart by earthquake at dawn, 20 May 1202,” Geology 26 (1998): 303–96. Esper, T., “Military self-sufficiency and weapons technology in Muscovite Russia,” Slavic Review 28 (1969), 185–208. Ettinghausen, R., Grabar, O. and Jenkins-Madina, M., Islamic Art and Architecture 650–1250 (New Heaven, NJ, and London, 2001). Farmer, E. L., “The hierarchy of Ming city walls,” in City Walls: The Urban Enceinte in Global Perspective, ed. J. D. Tracy (Cambridge, 2000). Faroqhi, S., Pilgrims and Sultans: The Hajj under the Ottomans 1517–1683 (London and New York, 1994). Farquhar, D. M., The Government of China under the Mongolian Rule: A Reference Guide (Stuttgart, 1990). Fennell, J., The Crisis of Medieval Russia 1200–1304 (London and New York, 1983) Fleming, J., Honour, H. and Pevsner, N., Dictionary of Architecture (London, 1991). Foley, V., Palmer, G. and Soedel, W., “The crossbow,” Scientific American 252 (January 1985): 80–6. Foss, C. and Winefield, D., Byzantine Fortifications: An Introduction (Pretoria, 1986). France, J., Western Warfare in the Age of the Crusades (Ithaca, NY, 1999). Franke, H., “Siege and defense of towns in medieval China,” in Chinese Ways in Warfare, eds F. A. Kierman and J. K. Fairbank (Cambridge, MA, 1974). Franke, H., Krieg und Krieger in Chinesischen Mittelalter (12.bis 14. Jahrhundert), band 81 (Stuttgart 2003). Fügedi, E., Castles and Society in Medieval Hungary (1000–1437) (Budapest, 1986). Garbois, A. “The city of Baniyas and the castle of Ṣubeibeh,” in The Crusaders in their Kingdom 1099–1291, ed. B. Z. Kedar (Jerusalem, 1988), 148–68. [Hebrew] Genequand, D., “Umayyad castles; the shift from late antique military architecture to early Islamic palatial buildings,” in Military Architecture in Greater Syria, ed. H. Kennedy (Leiden and Boston, 2006), 3–25. Gibb, H. A. R., “al-ʿĀdil,” EI2 1 (1960): 197–8. Goodrich, L. Carrington and Fềng Chia-shềng, “The early developments of firearms in China,” Isis 36 (1945–6): 114–23. Grabar, O., “The architecture of power: places, citadels and fortifications,” in Architecture of the Islamic World, its History and Social Meaning, ed. G. Michell (New York, 1984), 48–79.

256  Bibliography Guérin, V., Description géographique, historique et archelogique de la Palestine, seconde partie – Samarie. (Paris, 1875), Tomb 2. Guilmartin, J. F., Gunpowder and Galleys (Cambridge, 1974). Halperin, C. J., Russia and the Golden Horde (Bloomington, IN, 1985). Har-El, S., Struggle for Domination in the Middle East: The Ottoman–Mamluk War 1485–91 (Leiden, New York and Cologne, 1995). Hartal, M., The Al-Ṣubayba (Nimrod) Fortress, Towers 11 and 9 (Jerusalem, 2001). Hartog de, L., Russia and the Mongol Yoke (London, 1996). Hess, A. C. “The evolution of the Ottoman seaborn empire in the age of oceanic discoveries 1453–1525,” American Historical Review 75 (1970): 1892–1919. Hillenbrand, C., The Crusades: Islamic Perspectives (New York, 2000). Hillenbrand, R., Islamic Architecture (New York, 1994). Holt, P. M., “The structure of government in the Mamluk sultanate,” in The Eastern Mediterranean Lands in the Period of the Crusades, ed. P. M. Holt (Warminster, 1977), 44–61. —— Early Mamluk Diplomacy (1260–1290):Treaties of Baybars and Qalāwūn with Christian Rulers (Leiden, 1995), Islamic History and Civilization Studies and Texts, vol. 12. Hong, Y., ed., Weapons in Ancient China (New York and Beijing, 1992). Honigmann, E. and Bosworth, C. E., “Rūm K. al’esi,” EI2 8 (1995): 607–8. Housley, N., The Avignon Papacy and the Crusades 1305–1378 (Oxford, 1986). Howard-Johnston, J. D., “Procopius; Roman defences north of the Taurus and the new fortress of Citharizon,” in The Eastern Frontier in the Roman Empire, Proceedings of the Colloquium held at Ankara in September 1988, eds D. H. French and C. S. Lightfoot, BAR International Series 553, vol. 1 (1989), 203–34. Hsiao, Chi-ching, The Military Establishment of the Yuan Dynasty (Cambridge, MA, 1978). Humphreys, R. S., “The emergence of the Mamluk army,” SI 45 (1977): 147–79. —— From Saladin to the Mongols: The Ayyubids of Damascus 1193–1260 (Albany, 1977). —— “Politics and architectural patronage in Ayyubid Damascus,” in The Islamic World from Classical to Modern Times: Essays in Honor of Bernard Lewis, eds C. E. Bosworth, C. Issawi, R. Savory and A. L. Udovitch (Princeton, NJ, 1989), 151–74. —— “Ayyubids, Mamluks and the Latin East in the thirteenth century,” MSR 2 (1998): 1–17. —— “The politics of the Mamluk sultanate: a review essay,” MSR 9 (2005): 221–31. Iqtidar Alam Khan, “The role of the Mongols in the introduction of gunpowder and firearms in South Asia,” in Gunpowder: The History of an International Technology, ed. B. J. Buchanan (Bath, 1996), 33–44. Irwin, R., “Iqtā’ and the end of the Crusader States,” in The Eastern Mediterranean Lands in the Period of the Crusades, ed. P. M. Holt (Warminster, 1977), 62–77. —— “The supply of money and the direction of trade in thirteenth-century Syria,” in Coinage in the Latin East: The Fourth Oxford Symposium and Monetary History, eds P. W. Edbury and D. M. Metcalf, BAR International Series, 77 (1980), 73–104. —— “The Mamluk conquest of the County of Tripoli,” in Crusade and Settlement, ed. P. W. Edbury (Cardiff, 1985), 246–9. —— The Middle East in the Middle Ages: The Early Mamluk Sultanate 1250–1382 (London, 1986). Isaac, B., “Luttwak’s “Grand Strategy“ and the eastern frontier of the Roman Empire,” in The Eastern Frontier of the Roman Empire, eds D. H. French and C. S. Lightfoot, BAR International Series 553 (1989), pt. 1, 231–4. —— The Limits of Empire: The Roman Army in the East (Oxford, 1993). Issac, B., Jackson, P., “The crisis in the Holy Land in 1260,” English Historical Review 95 (1980): 481–513. Jackson, P., The Mongols and the West, 1221–1410 (Harlow, 2005). Johns, C. N., “Medieval ʿĀjlūn,” QDAP 1 (1931): 21–33. —— “The Citadel of Jerusalem,” QDAP 14 (1950): 163–70. Johns, J., “ The longue durée: state and settlement in southern Transjordan across the Islamic centuries,” in Village, Steppe and State: The Social Origins of Modern Jordan, eds E. L. Rogan and T. Tell (London and New York, 1994), 1–31.

Bibliography  257 Jones, P. N. and Renn, D. “The military effectiveness of arrow loops: some experiments at White-Castle,” Château Gaillard, 9–10 (1982): 445–56. Kedar, B. Z. and Pringle, D., “La Fève: a Crusader castle in the Jezreal valley,” IEJ 35 (1985):164–79. Keegan, J., A History of Warfare (New York, 1993). Keen, M., “The changing scene: guns, gunpowder and permanent armies,” in Medieval Warfare, ed. M. Keen (Oxford, 1999), 273–91. Kennedy, H., Crusader Castles (Cambridge, 1994). Keppie, L., The Making of the Roman Army: From Republic to Empire (London, 1998). Khadduri, M., War and Peace in the Law of Islam (Baltimore and London, 1969). King, D. J., “The defences of the Citadel Damascus: a great Mohammedan fortress of the time of the Crusades,” Archaeologia 94 (1951): 57–96. King, E. J., The Rule, Statutes and Customs of the Hospitallers 1099–1310 (London, 1934). King, G. R. D., “Archaeological fieldwork at the Citadel of Homs, Syria 1995–1999,” Levant 34 (2002): 39–58. Knapp, R. G., The Chinese House (Oxford, 1994). —— China’s Walled Cities (Oxford, 2000). Korn, L., Ayyubidische Architektur in Ägypten und Syrien: Bautätigkeit im Kontext von Politik und Gesellschaft 564–658/1169–1260 (Heidelberg, 2004). Kürkçüoğlu, Cihat A., Birecik Monografisi (Ankara, 1996). Labib, S., “Egyptian commercial policy in the Middle Ages,” in Studies in Economic History of the Middle East from the Rise of Islam to the Present Day, ed. M. A. Cook (London, New York and Toronto, 1970), 63–77. Lapidus, I. M., Muslim Cities in the Later Middle Ages (Cambridge, 1988). Latham, J. D. and Paterson, W. F., Saracen Archery: An English Version and Exposition of a Mameluke Work on Archery (ca. a.d. 1368) (London, 1970). Lawrence, A. W. “The Castle of Baghras,” in The Cilician Kingdom of Armenia, ed. T. S. R. Boase (Edinburgh and London, 1978), 34–83. Lawrence, T. E. Oriental Assembly, ed. A. W. Lawrence (London, 1939). —— Crusader Castles, ed. D. Pringle (Oxford, 1988). Leonard, R. A., A Short Guide to Clausewitz, On War (London, 1967). Lev, Y., Saladin in Egypt (Leiden, Boston and Cologne, 1999). Levanoni, A., A Turning Point in Mamluk History: The Third Reign of al-Nāṣir Muḥammad Ibn Qalāwūn (1310–1341) (Leiden, New York and Cologne, 1995). —— “Al-Maqrīzī’s account of the transition from Turkish to Circassian Mamluk sultanate: history in the service of faith,” in The Historiography of Islamic Egypt (c. 950–1800), ed. H. Kennedy (Leiden, Boston and Cologne, 2001), 93–105. Lewis, C. G., “A survey of the Euphrates valley from Deir-Ez-Zor to Aleppo,” Geographical Journal 59 (1922): 453–7. Lewis, B., The Assassins (New York,1968). Littmann, E., “Aybak,” EI2 1 (1960): 780. Luttrell, A. T., “The Hospitallers’ interventions in Cilician Armenia: 1291–1375,” in The Cilician Kingdom of Armenia, ed. T. R. S. Boase (Edinburgh and London, 1978). Luynes, H. T., Voyage d’exploration à la Mer Morte, à Petra et sur la rive gauche du Jourdain (Paris, 1874), vol. 2. Major, B., “Medieval cave fortifications of the upper Orontes valley (a preliminary report),” in Military Architecture in Greater Syria, ed. H. Kennedy (Leiden and Boston, 2006), 251–68. Manz, B. F., The Rise and Rule of Tamerlane (Cambridge, 1989). Marshall, C. J., “The French Regiment in the Latin East, 1254–91,” Journal of Medieval History 15 (1989): 301–8. —— Warfare in the Latin East, 1192–1291 (Cambridge, 1996). Martin, D. H., “The Mongol army,” JRAS (1943): 46–85. —— The Rise of Chingis Khan and his Conquest of North China (Baltimore, 1950). Martinez, P., “Some notes on the Īl-Xānid army,” Archivum Eurasiae Medii Aevi 6 (1986): 129–242.

258  Bibliography Masarwa, Y., “Early Islamic military architecture: the birth of the Ribat on the Palestinian coast,” Al-ʿUsur al-Wusta (October 2008): 36–42. Massoud, S. G., “Al-Maqrīzī as a historian of the reign of Barqūq,” MSR 7/2 (2003): 119–36. May, T., The Mongol Art of War (Yardley, PA, 2007). —— “Genghis Khan’s secrets of success,” Military History (July/Aug., 2007), 42–9. Mayer, H. E., “Two unpublished letters on the Syrian earthquake of 1202,” in Medieval and Middle Eastern Studies in Honor of Aziz Suryal Atiya, ed. S. A. Hanna (Leiden, 1972), 295–310. Mayer, L. A., Islamic Architects and Their Works (Geneve, 1956). Meinecke, M., “Ṣalkhad,” EI2 8 (1995): 994–6. Melville, C., “ ‘Sometimes by the sword, sometimes by the dagger’: the role of the Ismaʿilis in Mamluk–Mongol relations in the 8th/14th century,” in Medieval Ismaʿili History and Thought, ed. F. Daftary (Cambridge, 1996), 247–63. Mesqui, J., Châteaux forts et fortifications en France (Paris, 1997). Meyvaert, P., “An unknown letter of Hulagu, Il-Khan of Persia, to King Louis IX of France,” Viator 11 (1980): 245–60. Michaudel, B., “The development of Islamic military architecture during the Ayyubid and Mamluk reconquest of Frankish Syria,” in Military Architecture in Greater Syria, ed. H. Kennedy (Leiden and Boston, 2006), 106–21. Michaudel, B., “The use of fortifications as a political instrument by the Ayyubids and the Mamluks in Bilād al-Shām and Egypt (twelfth and thirteenth centuries),” MSR 11/1 (2007): 55–67. Milwright, M., “Central and southern Jordan in the Ayyubid period: historical and archaeological perspectives,” JRAS 3 (2006): 1–27. Molin, K., Unknown Crusader Castles (London and New York, 2001). Morgan, D. O., “The Mongol armies in Persia,” DI 56 (1979): 81–96. Morgan, D., The Mongols (Oxford, 1986). —— Medieval Persia 1040–1797 (London and New York, 1988). Mouton, J. M., “ Shayzar,” EI2 9 (1997): 410–11. —— “Sadr au Sinaï, la forteresse de Saladin,” Archéologie, 410 (2004): 28–39. Mouton, J. M. et al., “La route de Saladin (ṭarīq Ṣadr wa Ayla) au Sinaï,” Annales Islamologiques 30 (1996): 42–70. Muir, W., The Mameluke or Slave Dynasty of Egypt 1260–1517 (Edinburgh, 1896). Musil, A., The Middle Euphrates (New York ,1927). Müller-Wiener, W., Castles of the Crusaders, trans. J. Brownjohn (London, 1966). Needham, J., Science in Traditional China: A Comparative Perspective (Hong Kong, 1981). —— Science and Civilization in China, vol. 5, pt. 7: The Gunpowder Epic (Cambridge 1986). —— Science and Civilization in China, vol. 5, pt. 6: Military Technology: Missiles and Sieges (Cambridge, 1994). Nicola, Z., Urban Life in Syria (Beirut, 1953). Nicolle, D., Arms and Armour of the Crusading Era 1050–1350: Islam, Eastern Europe and Asia, 2nd edn (London and Mechanicsburg, PA, 1999). Northrup, L. S., From Slave to Sultan: The Career of Al Mansur Qalawun and the Consolidation of Mamluk Rule in Egypt and Syria (678–689 a.h./1279–1290 a.d.) (Wiesbaden, 1998). Paillet, J., Le Chateau de Rahba, PhD diss., Universite de Lyon 2, 1983, unpublished. Parker, G., “The artillery fortress as an engine of European overseas expansion, 1480–1750,” in City Walls: The Urban Enceinte in Global Perspective, ed. J. D. Tracy (Cambridge, 2000), 386–416. Parker, H. M. D., The Roman Legions (Oxford, 1928). Paterson, W. F., “The archers of Islam,”Journal of the Economic and Social History of the Orient 9 (1966): 69–87. Patton, D., Badr al-Dīn Lu’lu’, Atabeg of Mosul, 1211–1259 (Seattle, 1991). Petersen, A., “Two forts on the medieval Hajj route in Jordan,” ADAJ 35 (1991): 347–57. —— Dictionary of Islamic Architecture (London and New York, 1996).

Bibliography  259 Petry, C. F., Twilight of Majesty: The Reigns of the Mamluk Sultans al-Ashraf Qaytbay and Qansuh al-Ghawri in Egypt (Seattle and London, 1993). Poliak, A. N., Feudalism in Egypt, Syria, Palestine, and the Lebanon, 1250–1900 (London, 1939). Powell, J. M., Anatomy of a Crusade 1213–1221 (University of Pennsylvania, Philadelphia, 1986). Prawer, J., A History of the Latin Kingdom of Jerusalem, 2nd edn, 2 vols (Jerusalem, 1963). [Hebrew] —— The Crusaders: A Colonial Society, 2nd edn (Jerusalem, 1985). [Hebrew] Pringle, D., “Reconstructing the castle of Safad,” PEQ 117 (1985): 139–49. —— The Red Tower (al Burj al Aḥmar) Settlement in the Plain of Sharon at the Time of the Crusaders and Mamluks a.d. 1099–1516 (London, 1986). —— “Towers of Crusader Palestine,” Château Gaillard 16 (1994): 335–50. —— Secular Buildings in the Crusader Kingdom of Jerusalem (Cambridge, 1997). —— The Churches of the Crusader Kingdom of Jerusalem (Cambridge, 1998), vol. 2. —— “Crusader castles in Jordan,” in The Archaeology of Jordan, eds. B. MacDonald, R. Adams and P. Bienkowski (Sheffield, 2001), 677–84. Raba, J., Russian Travel Accounts on Palestine (Jerusalem, 1986). [Hebrew] Rabbat, N. O. The Citadel of Cairo: A New Interpretation of Royal Mamluk Architecture (Leiden, New York and Cologne, 1995). —— “Architects and artists in Mamluk society: the perspective of the sources,” Journal of Architectural Education 52 (1998), 30–7. —— “Perception of architecture in Mamluk sources,” MSR 6 (2002): 155–76. Ragette, F., Baalbek (Princeton, NJ, 1980). Raphael, K., “Archers in the Crusader Kingdom,” MA thesis, The Hebrew University of Jerusalem (2001). [Hebrew] —— “The Al-Mansūrī’s mangonel stone balls: some new evidence from the Mamluk siege of Akko (1291),” in ʿAkko, the Excavations of 1991–1998: II, The Later Periods, eds E. Stern and D. Syon (IAA Reports). Forthcoming. Raphael, K. and Tepper, Y., “The archaeological evidence from the Mamluk siege of Arsūf,” MSR 9/1 (2005): 85–100. Raudzens, G., “War-winning weapons: the measurement of technological determinism in military history,” Journal of Military History 54 (1990): 403–33. Rex Smith, G., Medieval Muslim Horsemanship: A Fourteenth-Century Arabic Manual (London, 1979). Riley-Smith, J., The Crusades: A Short History (London, 1990). Rosenthal, F., “Al-Ṣafadī,” EI2 8 (1995): 759–60. Rossabi, M., Khubilai Khan (Berkeley, Los Angeles and London, 1988). Runciman, S., A History of the Crusades (London, 1994), vol. 3. al-Sarraf, S., “Mamluk Furūsīya literature and its antecedents,” MSR 8/1 (2004): 141–200. Sauvaget, J., “Ḥalab,” EI2 3 (1979): 85–90. Schein, S., Fideles Crucis: The Papacy, the West and the Recovery of the Holy Land 1274–1314 (Oxford, 1991). Sen-Dou Chang, “The morphology of walled capitals,” in The City in Late Imperial China, ed. G. W. Skinner (Stanford, 1977), 75–100. Serjeant, R. B., The Portuguese off the South Arabian Coast: Hadrami Chronicles (Beirut, 1974). Shapira, Y., “The tower in the Crusader Kingdom of Jerusalem,” MA thesis, The Hebrew University of Jerusalem (2000). [Hebrew] Sharon, M., Corpus Inscriptionum Arabicarum Palaestinae (Leiden, Boston and Cologne, 1999), vol. 2. Silverstein, A. J., Postal Systems in the Pre-Modern Islamic World (Cambridge, 2007). Sinclair, T. A., Eastern Turkey: An Architectural and Archaeological Survey, 4 vols (London, 1990). Sinor, D., “On Mongol strategy,” Inner Asia and its Contacts with Medieval Europe (London, Variorum, 1977), XVI, 238–49.

260  Bibliography Smail, R C., Crusading Warfare (1097–1193) (Cambridge, 1956). —— Crusading Warfare (1097–1193), 2nd edn (Cambridge, 1995). Smith, J. M. Jr. “Mongol manpower and Persian population,” JESHO 18 (1975): 271–99. —— “Ayn Jālūt: Mamluk success or Mongol failure?,” HJAS 44 (1984): 307–45. —— “Demographic considerations in Mongol siege warfare,” Archivum Ottomanicum 13 (1993–4): 229–334. —— “Mongol society and military in the Middle East: antecedents and adaptations,” in War and Society in the Eastern Mediterranean 7th–15th Centuries, ed. Y. Lev (Leiden, NY, and Cologne, 1997), 249–66. Sobernheim, M., “Die arabischen Inschriften von Aleppo,” DI 15 (1926): 162–210. Somogyi, J., “Adh-Dhahabi’s record of destruction of Damascus by the Mongols in 699–700/1299–1301,” in Ignace Goldziher Memorial Volume, eds S. Lowinger and J. Somogyi (Budapest, 1948), 353–86. Sourdel, D., “Barīd,” EI2 1 (1960): 1045–6. —— “Karak,” EI2 4 (1978): 609. —— “Ibn Shaddād,” EI2 3 (1979): 933. Stark, F., Rome on the Euphrates (London, 1966). Stewart, A. D., The Armenian Kingdom and the Mamluks: War and Diplomacy during the Reign of Hetʿum II (1289–1307), The Medieval Mediterranean, vol. 37 (Leiden, 2001). —— “Qalʿat al-Rūm/Hrmogla/Rumkale and the Mamluk siege of 691 ah/1291ce,” in Military Architecture in Greater Syria, ed. H. Kennedy (Leiden and Boston, 2006), 269–80. Streck, M., “Biredjik,” EI2 1 (1960): 1233–4. Świętosławski, W., Arms and Armour of the Nomads of the Great Steppe in the Times of the Mongol Expansion 12th–14th centuries (Łodź, 1999). Tabba, Y., Constructions of Power and Piety in Medieval Aleppo (Pennsylvania State Univesity, 1977). Tamari, S., “Two further inscriptions from Qalʿat Al-Ṣadr,” in Studies in Memory of Gaston Wiet, ed. M. Rosen-Ayalon (Jerusalem, 1977), 261–5. —— “Darb al-Hajj in Sinai: an historical-archaeological study,” Accademia Nazionale dei Lincei Memorie 25 (1982): 431–525. —— “Darb al-Hajj, the Qal’a of Nakhl,” Hamizraḥ Haḥadash 23 (1983): 141–67. [Hebrew] Thorau, P., The Lion of Egypt, trans. P. M. Holt (New York, 1987). Tonghini, C., Qalʿat Jabar Pottery: A Study of a Syrian Fortified Site of the Late 11th–14th Centuries (Oxford, 1998). Tonghini, C., Donato, E., Montevecchi, N. and Nucciotti, M., “The evolution of masonry technique in Islamic military architecture: the evidence from Shayzar,” Levant 35 (2003): 179–212. Toy, S., A History of Fortifications, from 3000 b.c. to a.d. 1700. 2nd ed. (London,1966). Tracy, J. D. “To wall or not to wall: evidence from medieval Germany,” in City Walls; the urban enceinte in global perspective, ed. J. D. Tracy (Cambridge, 2000), 71–88. Twain, M., The Innocents Abroad (New York, 1962). Tyerman, C. J., “Marino Sanudo Torsello and the lost Crusade: lobbying in the fourteenth century,” Transactions of the Royal Historical Society 32 (1982): 57–73. Verbruggen, J. F., The Art of Warfare in Western Europe during the Middle Ages (Woodbridge, 1997). Walker, B. J., “Militarization to nomadization: the middle and late Islamic periods,” NEA 62/4 (1999): 202–32. —— “Mamluk investment in Southern Bilād al-Shām in the eighth/fourteenth century: the case of Ḥisbān,” JNES 62 (2003): 241–61. Walker, B. J. and LaBianca, Ø. S., “The Islamic Qusūr of Tall Hisbān: preliminary report on the 1998 and 2001 seasons,” ADAJ 47 (2003): 443–71. Walmsley, A., “Fatimid, Ayyubid and Mamluk Jordan and the Crusader interlude,” in The Archaeology of Jordan, eds B. MacDonald, R. Adams and P. Bienkowski (Sheffield, 2001), 515–59.

Bibliography  261 Whitcomb, D., “The town and name of ‘Aqaba: an inquiry into the settlement history from an archaeological perspective,” in Studies in the History and Archaeology of Jordan, eds. G. Bisheh, M. Zaghloul, and I. Kehrberg (Amman, 1997), vol. 6, 359–63. Wiet, G., “Les inscriptions de la Qal’ah Guindi,” Syria 3 (1922): 58–65. Willey, P., The Castles of the Assassins (London, 1963). Woods, J. E., The Aqquyunlu Clan: Confederation, Empire (Salt Lake City, 1999). Yovitchitch, C., “The tower of Aybak in ‘Ajlun castle: an example of the spread of an architectural concept in early thirteenth-century Ayyubid fortifications,” in Military Architecture in Greater Syria, ed. H. Kennedy (Leiden and Boston, 2006), 225–42. Zaki, A. R. “Military literature of the Arabs,” Cahiers d’histoire égyptienne 7 (1955):149–169.

Index

Abagha 53, 56, 69,70, 84, 94,111, 112 Acre 86, 133 al-ʿĀdil 11, 17–18, 19, 20, 79, 132, 161, 166, 122 al-Afḍal 16–17 ʿAjlūn 5, 7, 11,13,19, 27, 29–30, 34, 37–8, 40, 41, 44–5, 131 Aleppo 14, 65, 69, 70, 83, 84, 85, 93–4, 103, 113, 121 122 149, 212 Amān (guaranty of safety) 71, 96–7, 148, Amanus gate (Darband Arslān) 85 Amir jāndār 132–3 Amir kabir 137 Amir ṭablkhāna 132 Anavarza (al-Zarb) 112, 116 Anonymous French writer 146 Antioch 96 Aqquyunlu 212, 213 Arbela (Irbil) 65, 66 Archers 51, 74, 75, 76, 78, 139, 157–8,179, 203 Architectural terminology 129–30 Arms race 10 Army Armenian 87–8 Ayyubid 8, 208 Frankish 87 Īlkhanīd 87 Mamluk 120–1 Arzanjān 117 Arrow slits 45–6, 78, 139–40, 157, 170–1, 179, 189–90,195 Arsūf 31 al-Ashraf Khalīl 103–4, 108, 114, 122, 124, 187 ʿAthlit (Chateau Pelerin) 18, 50, 86,131 ʿAyn Tāb 70 Ayās 83, 89, 110, 116,117, 118, 119 ʿAzīz al-Dawla Rayḥān 25

al-ʿAzīz Haṣan b.ʿUthmân 25, 94 Baalbek 99 Bab al-Malik 85 Bader al-Dīn Lu’lu’ (ruler of Mosul) 66, 78 Baghrās 70, 85, 97, 113, 198–202 Baghdad 65, 104 Bahasnā 112, 114, 115, 129 Baḥriyya 75 Bakhīmh 83 Bānās 83 Banū ʿAwf 11 Barīd 82, 83, 120, 155, 156, 199, 210 Barqūq al-Malik al-Nāṣir Faraj b. 74, 211–13 Bāshūra 21–2, 28, 148–9 Baybars 68, 69, 75, 81, 87–99, 112, 119, 120, 124, 131, 156, 161 Belen mountain pass 70, 85 Belvoir (Kawkab) 11, 17, 45 Berke Khan 104, 177, 186 Bīlīk al-Khaznadār 96, 134–5, 137 al-Bira 51, 68, 70, 72, 76, 82, 83, 90, 93, 174–185, 212–13 Bedouin 11, 108, 133, 162, 209 Boniface VIII 109 Borders 82, 83–4, 134 Bows 77–8 Caesaria 133, 160 Chaghatai 72 Charles IV 109, 110 Chastle Blank (Ṣafītha) 97 Chinggis Khan 52, 53, 58 Cilicia (Armenian Kingdom) 82, 85, 99, 108–120, 193–202 Clement V 109 Concentric fortress 50, 147, 198, 199, 205, 207 Constantine (Armenian King) 82 Counterweight trebuchet 9, 56–7, 62, 63, 79

Index  263 County of Tripoli 86, 103 Curtain walls 22, 28–9, 39–43, 58, 138, 177–8, 103 Dabbūriya 16 Daniel Rrussian traveler 16 Damascus 14, 23, 70, 74, 77, 84, 91, 94, 103, 121, 212 Darb al-Ḥājj 12 Darbassāk 70, 85, 112,113 Dayr Kūsh (Dargous) 97 Dubbīn 97 Dulkadir (Buffer state) 120 Earthquake (1202) 79 Edward I 87, 155 Fifth Crusade 2, 18–19 Fidenzio of Padua 109 Francesco Balducci Pegolotti 117 Fredrick II 2, 11 Gaben (Gabon) 117 Gargar 115 Gates 34–9, 150, 182, 190–3, 195, 203 Genoese merchants 118 Ghazan 70, 71, 74, 84, 85, 182–3 Gibelacar (Ḥiṣn al-ʿAkkār) 97 Gird Kūh 65 Glacis 32, 130, 149 Golden Horde 72 Gunpowder 59–61 Hajar Shaghlān 83, 100, 114 Hama 69, 70, 84 Hāmīs 114 Ḥamūs 100, 112 Hārūnyah 83 Hetʿum (Arminian chroniler) 110 Hetʿum I (Arminian king) 111, 112, 186 Hetʿum II (Arminian king) 83, 114 al-Ḥilla 71, 93 Hirqilī stones 149, 153 Ḥiṣn al-Akrad (Crac des Chevaliers) 50, 79, 97, 99, 102, 120 Homs 69, 70, 113, 121 Hospitallers 97, 100 Hülegü 2, 53, 54–5, 61, 62, 64–66, 68, 93, 114, 186 Hunin 96 Ḥusām al-Dīn Lu’lu’ 24 Ibn ʿAbd al-Ẓāhir 91, 127–8 Ibn Shaddād 128

Inflammable materials 57–8, 62 Innocent III 18 Innocent IV 54 ʿIzz al-Dīn Usāma 11, 13, 24 ʿIzz al-Dīn Aybak 13, 23, 24, 121 ʿIzz al-Dīn Aybak al-Afram 108, 132–3 Ismāʿīli fortresses 89, 98 Jayḥān River 82–3, 116, 117 John of Plano Carpini 54 John XXII 109, 110 Juwaynī ʿAlā al-Dīn ʿAṭā Malik 60–1 Al-Kahf 98 Kakhtā 113, 115 Karak 75, 95, 123, 127, 160–173 Karanjil 112 Kāsīsh 8 Kaysariye 117,119 Keep 50, 150–1, 169–71 Khoy (Xoy) 117 Al-Khawābī 98 Kingdom of Jerusalem 86, 109, 122 Kuwayrā 83, 114 Latrines 145, 152 Lachin 82, 114, 115 Latakia 101 Lateran Council IV 18 Leon II 113, 199 Leon III 113 Leon VI 110 Luis IX 2 Machicolations 38, 130, 190, 203 Al-Malik al-Kāmil 11 Al-Malik al-Salih Ismāʿīl 147 Malatya 115 Mamluk garrissons 74–5, 85, 86 Mamluk navy 96, 120 Manīqa 99 Marʿash 76, 83, 114 Maraqīya 100, 132 Marqab (Margat) 76, 101 Marzabān 112 Marino Sanudo 109–10 Masonry 29–31, 149, 158–9, 171, 198, 202 Maymundiz 64 Mayyāfarqīn 64, 66, 78 Mesis (Momsuestia) 112, 113,118 Mi’mār 135–7 Möngke 53 Moats 32–4, 129, 165, 203

264  Index Mount Tabor 5, 7, 9, 14–24, 26, 27, 28–9, 34, 36, 39, 41, 45, 47–8, 95, 131 al-Muʿaẓẓam ʿIsa 13, 18, 19, 20, 132, 161, 166–7 Mubārīz al-Dīn 9, 25 Al-Mughīth Fakhr al-Dīn 161 Muhandis 135 Nafṭ 18, 61, 62–3, 71 Naib (fortress governor) 75 Al-Naqīr 100, 115 Al-Nāṣir Yūsuf 13 Al-Nāṣir Muḥammad 83, 115, 172, 179, 182, 183, 184 Al-Nāṣir Dāwūd b al-Muʿaẓẓam 161 Nujayma 114 Ögedei 52–3 Oliver of Paderborn 18, 131 Öljaitü 71, 93, 110 Oshin 110, 116 Otto of Grandson 110–11 Philip IV 109 Philip of Taranto 109 Philip VI 109, 110 Port Bonnel (Ḥiṣn Rūsrūs) 97 Portcullis 35 Posterns 35, 38, 190 Principality of Antioch 86, 103 Qadmūs 99 Qaidu 27 Qaitbāy 212–13 Qalʿat Jaʿ bar 70, 84, 93, 122 Qalʿat Najm 93, 122 Qalʿat al-Qurayn (Montfort) 44, 97 Qalʿat al-Rūm 82, 94, 103–4, 114, 115, 185–93, 212 Qalʿat Ṣadr (Qalʿat Jundī) 5, 7, 11–13, 25, 27, 33, 35, 41, 43 Qalʿat Shaqīf (Beaufort) 96, 97 Qalāwūn 76, 94, 99–103, 113, 120, 122, 123–4, 132 Qāqūn 87, 97, 120, 127, 155–60 Quarries 27 Qubilai Khan 53, 56–7 al-Quṣayr (Cursat) 97 Qutuz 2, 134 Quwara pass (Demir Kapi) 85 Raban 112 al-Raḥba 51, 69, 70, 71, 72, 73, 76, 79, 82, 84, 90, 91, 93, 122

Ramadan (buffer state) 129 Rashīd al-Dīn 60, 71 Richard of Cornwall 2, 11, 147 La Roch de Roussel 97 al-Ruṣāfa 98 Safad 75–6, 89, 96, 99, 120, 127, 131, 146–55 al-Saʿīd Ḥasan b. al-ʿAziz ʿUthman 134 Ṣāliḥ al-Dīn 1, 13, 16, 35, 121, 209 al-Ṣāliḥ Ayyūb 11 Ṣalkhad 23, 121 Sarfadkār 83, 100, 114 Al-Ṣalt 91 Sahyūn 99 Ṣarkhad 91 Ṣārim al-Dīn Barghash al-ʿĀdilī 26 Secret passages 130, 150, 203 Shawbak 122, 108, 133 Shayzar 91 Shīḥ al-Ḥadīd 112 Shumaymish 91, 131 Siege Warfare 6–7 Frankish 9–10, 18, 27, 50, 63 Īlkhānid 57, 64, 66–73 Mongol 34, 53–66 Muslim 9, 27, 50, 63, 98, 116, 187 Sis 112, 113, 114, 115, 116 Sivas 117, 118 Storage 22, 183 al-Ṣubayba Ayyubid period 5, 7, 13, 25, 26–7, 32, 35, 39, 41, 47, 91, 94, 120 Mamluk period 121, 131, 134–146 Tabrīz 117, 118 Tall Bashir 113 Tall Ḥamdūn 76, 82, 83, 100, 112, 114, 116 193–8 Talamis (Cavea Balmīs) 97 Tancred 15–16 Tarsūs 113 Templars 97, 147–8, 198–9 Templar of Tyre 155–6 Temür (Tamerlane) 74, 84, 211–12 Teutonic Order 97 Theobold of Champagne 2 Thoros III 194 Towers 43–9, 129, 138–45, 149, 150–3, 157–60, 179, 195–8, 201–2, 204–5 Trade 85, 100, 109, 117–19 Treaties Mamluk -Armenian (666/ 1268) 112, 186

Index  265 Mamluk-Frankish (669–1272) 102 Mamluk-Armenian (1282, 1286) 100 Mamluk-Armenian (684/ 1285) 113 Mamluk-Armenian (687/ 1289) 114 Mamluk-Armenian (692/1293) 114 Mamluk-Armenian (697/ 1298) 82–3, 114 Tripoli 100, 102 Tumlu 116 Türkmen 76, 113, 116, 119, 186, 211 ʿUlayqa 98 ʿUmar Ibn Ibrāhīm al-Anṣārī 74 Ustādār 23

Ustādh 134–5 Uzun Ḥasan 212 Vadum Iacob 28, 29, 40–1, 131, 209 Venetian merchants 109, 118 Warehouse 21, 28, 118 Water cisterns 24,145–6, 153, 171, 184, 193 Yoshmut 66 Yuanshi 60–61 Zeugma 173, 174, 176