The Great Seljuqs

This book provides a broad history of the Seljuq Turks from their origins and early conquests in the tenth century, through the rise of the empire, until its dissolution at the end of the twelth century. Where the history of the Seljuqs is usually studied in the context of medieval Persian, Arabic or Islamic history, this book considers the subject from the perspective of Turkish history. Examining the corpus of academic work on the period and how Turkish historiography has interpreted and understood the Seljuqs, the author demonstrates how the Great Seljuq Empire can be considered not only in a historical context, but as the instigator of Turko-Islamic civilization. Rejecting traditional Turkish scholarship, which places Iranian culture and Islam as the civilising elements in the Great Seljuq Empire, the author shows how the nature of nomadic pastoral empires has come under fresh scrutiny, reassessing Seljuq history and the framework within which it has been treated. This book provides a unique insight into the adoption of Turkic expectations to an urban environment that were forged on the Eurasian steppes, showing how the outcome put its stamp on the second millennium throughout the Middle East and the Balkans. It will be an important addition to the literature on medieval Islamic, Turkish and Middle Eastern history.

Aziz Başan completed his PhD at the University of Edinburgh, and is now an independent scholar. Routledge studies in the history of Iran and Turkey Edited by Carole Hillenbrand University of Edinburgh

This series publishes important studies dealing with the history of Iran and Turkey in the period CE 1000–1700. This period is significant because it heralds the advent of large numbers of nomadic Turks from Central Asia into the Islamic world. Their influence was felt particularly strongly in Iran and Turkey, territories which they permanently transformed. The series presents translations of medieval Arabic and Persian texts which chronicle the history of the medieval Turks and Persians, and also publishes scholarly monographs which handle themes of medieval Turkish and Iranian history such as historiography, nomadisation and folk Islam.

Mediaeval Islamic Historiography and Political Legitimacy Bal’ami’s Tarikhnamah A.C.S. Peacock

History of the Seljuq Turks The Seljuq-nama of Zahir al-Din Nishpuri Translated by Kenneth Allin Luther Edited by Edmund Bosworth

The Annals of the Seljuq Turks Selections from al-Kamil fi’l-Ta’rikh of Ibn al-Athir D. S. Richards

Early Mongol Rule in Thirteenth Century Iran A Persian renaissance George Lane

The Ghaznavid and Seljuq Turks Poetry as a source for Iranian history G.E. Tetley

Iranian Elites and Turkish Rulers A history of Isfahan in the Seljuq period David Durand-Guédy

Early Seljuq History A new interpretation A.C.S. Peacock

The Great Seljuqs A history Aziz Başan The Great Seljuqs A history

Aziz Başan First published 2010 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.

© 2010 Aziz Başan 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 Basan, Aziz. The great Seljuqs : a history / Aziz Basan. p. cm. – (Routledge studies in the history of Iran and Turkey) Includes bibliographical references and index. 1. Seljuks – History. 2. Seljuks – Historiography. 3. Turkey – History – To 1453. 4. Turkey – History – To 1453 – Historiography. I. Title. DS27.B37 2010 950’.1 – dc22 2009052455

ISBN 0-203-84923-X Master e-book ISBN

ISBN: 978-0-415-55539-5 (hbk) ISBN: 978-0-203-84923-1 (ebk) For Maureen

Contents

List of maps x Acknowledgements xi Spelling and abbreviations xii

1 Introduction 1 Ottoman influences 1 Republican concerns 5 Why the Seljuqs? 7 The historians and their works 9 The Turkish view and its problems 14 Methodology 20

2 Overview 21 Origins 22 Conquest (985/993–1063) 22 Empire (1063–92) 27 Interregnum (1092–1105) 34 Dissolution (1105–94) 38

3 Origins 45 The Eurasian steppe zone 45 The Oguz 47 The Türkmen 50 The Seljuqs 51

4 Conquest (985/993–1063) 53 Transoxania 53 The Türkmen of Iraq 58 Victory at Dandankan 61 viii Contents Seljuqid sovereignty 63 Ruler of the East and the West 66 Seljuqid discord 67

5 Empire (1063–92) 70 Unification of the empire 71 The Caucasus and Transoxania 73 The war for Anatolia 75 Victory at Malazgirt 78 Contention for the realm 81 Transoxania 82 The Caucasus 83 Anatolia 84 Syria and Palestine 85 The rise of Süleyman-Shah 86 Incorporation of Syria 88 Anatolia after Süleyman-Shah 92 Central Asia 94 Arabian peninsula and the Gulf 94 Coming of the assassins 94

6 Interregnum (1092–1105) 97 Contention for the realm 98 Coming of the Crusaders 100 The rise of Muhammad Tapar 104 Anatolia and Syria 109

7 Dissolution (1105–94) 112 Unruly servants and vassals 113 The Crusaders 116 Transoxania and the Ghaznavids 121 The Seljuqs of Iraq 121 Sultan Mahmud 122 Dissension 127 Defeat at Katvan 131 Sultan Mascud 133 The Oguz rebellion 136 Collapse 138 Contents ix 8 Evaluation 140 Nomadic pastoral organization 143 The Seljuqs’ nobility and aspirations 149 The problems of conquest 151 The conquest of Anatolia 156 The basis of empire 165 The power struggle 180 Common ownership of the realm 183 The role of women 184

9 Conclusion 186

Appendix 1: Maps 188 Appendix 2: The Oguz 197 Appendix 3: Seljuq rulers 198 Bibliography 200 Index 207 Maps

A1.1 Turkic migrations and the Seljuqs 189 A1.2 The Great Seljuq Empire 190 A1.3 Urban centres in Seljuq history 191 A1.4 Raids under Sultan Tuğrul 192 A1.5 Raids under Sultan Alp-Arslan and Afşin 193 A1.6 Emperor Diogenes’ second campaign 194 A1.7 Manuel and Er-Basgan 195 A1.8 Emperor Diogenes’ final campaign 196 Acknowledgements

I would like to thank Professor Carole Hillenbrand for her invaluable guidance throughout. This work would not have been possible without her encouragement and support. There are others I must also thank, as without their help I could not have accessed all the relevant sources. First, I would like to thank the staff at the Librarie de Pera, in Istanbul, who found and delivered all the books by Turkish scholars for the period in question. Secondly, there is my childhood friend Mustafa Bozcaadalı, who acquired copies of Köymen’s articles, and also my good friend Cem Kum, who referred me to relevant works by military historians. Third, there is my colleague Cengiz Tomar, who drew my attention to pertinent articles and theses. Last but not least, I would like to thank my wife Maureen for her moral support whenever I needed it most. Spelling and abbreviations

There are many problems with Turkish spellings. The silent /i/, namely /ı/, pronounced as in –tion, presents a challenge not overcome by /y/ or indeed /ï/. Although Kyrgyz is fine for Kırgız, Osmanly does not render Osmanlı. Likewise /ğ/ is a softening not a hardening of /g/, so that /gh/ as in Oghuz or yoghurt renders neither Oğuz nor yoğurt. Both /ö/ and /ü/ now seem acceptable, /ç/ as in /ch/ and /ş/ as in /sh/ are becoming more so. I have retained all six in proper Turkish nouns, so that care should be taken also with /c/, which is pronounced as /j/. Thus, for example, Sancar should be read as ‘Sanjar’ and Altuncan as ‘Altunjan’. I have given Arabic and Persian names and titles minimal transliteration due to their complexity. Initial hamzas and cayns have been omitted, but they are indicated in the middle of words. There are two abbreviations worthy of note. These concern the I˙slâm Ansiklopedisi (IA) and the Diyanet Vakfı I˙slâm Ansiklopedisi (DVIA). Furthermore, volumes 5 and 12 of the I˙slâm Ansiklopedisi have been split into two. Thus (Eren IA/11: 709ff) refers to Eren’s article on pages 709 following in volume 11 and (Minorsky IA/12/2: 107ff) refers to Minorsky’s article on pages 107 following in volume 12/2. Last but not least, Kaşgarlı Mahmut’s and Ibn Fadlan’s works as well as the Orkhun stone inscriptions are referenced under their editors’ and translators’ names in the Bibliography, Primary sources in Turkish; namely under Atalay, Şeşen and Tekin respectively. 1 Introduction

The purpose of this work is to provide a history of the Great Seljuq Empire in a Western language. Although it is a translation, Gary Leiser’s A History of the Seljuqs: I˙brahim Kafesoğlu’s Interpretation and the Resulting Controversy (1988) is the only full length narrative of Seljuqid war and politics in a Western language (see Humphreys 1991: 156). Another aim of this work is to reassess the framework within which the history of the Great Seljuqs has been treated. In the West, the Seljuqs are studied in the context of medieval Persian, Arabic or Islamic history (see Hodgson 1974; Lambton 1987; Frye 1993; Lewis 1993; Kennedy 1994; Morgan 1994a). In Turkey they are considered not only part of Turkish history but also the instigators of Turko-Islamic civilization. Kafesoğlu’s article in the I˙slâm Ansiklopedisi (IA) (IA/10: 353–416) is one of many such works on the Seljuqs by Turkish historians that hold this view. Martin Strohmeier’s Seldschukische Geschichte und türkische Geschicht- swissen schaft – Die Seldschuken im Urteil moderner türkischer Historiker (History of the Seljuqs and Turkish Historiography – Modern Turkish Historians’ Judgement) (1984) has attempted to evaluate the politics and ideologies of Turkish historians on the Seljuqs. Despite first apprising a number of nineteenth- century Ottoman historians, arguably Strohmeier fails in this task. Rather than individual politics, Turkish interpretations of Seljuq history are directly traceable to late Ottoman scholarship. It is the main reason why so many historians in the republican era concentrated initially on the Seljuqs rather than the Ottomans or earlier Turkic empires. As a result, the following sections first discuss developments in nineteenth- century Ottoman scholarship and the republican concerns that arose from it. Next, the Turkish historians on the Seljuqs are introduced, followed by a discussion of the problems their interpretations pose. An outline of ensuing chapters completes the introduction.

Ottoman influences Reşit Pasha (1800–58) is best remembered for drafting the imperial edict of 3 November 1839, the Tanzimat-ı Hayriye Fermanı, which proclaimed that 2 Introduction persons of all religions would be treated equally and that the inviolability of their life, property and honour would be recognized. However, he also founded the Encümen-i dâniş or Ottoman Academy (18 July 1851), whose forty members were charged with the tasks of emphasizing Turkish history rather than that of the Ottomans in an Islamic context, simplifying the Turkish language, and translating western European works on art and science as textbooks (Berkes 1998: 144–7; Eren IA/11: 709ff; Şapolyo 1945: 161; Irmak and Çağlar 1994: 11). The most illustrious member of the academy was Cevdet Pasha (1822–95). Credited with drafting the opening address, he undertook to write a history of Ottoman attempts at reform between the years 1767–8 and 1825–6. Although he included the most important events of each year in chronological order, unlike previous Ottoman chroniclers Cevdet Pasha based his twelve-volume Tarih-i Cevdet (1854) on a wide variety of archival materials. He analysed, compared and criticized in a clear language official and private chronicles, memoirs, diaries, memorandums and reports produced by government officials, as well as legal and treasury documents. Cevdet Pasha is also credited with writing the first Turkish grammar (Berkes 1998: 178; Ölmezoğlu IA/3: 114ff; Irmak and Çağlar 1994: 12–14). Nonetheless, Ottoman historiography did not gain a Turkish emphasis until the last quarter of the nineteenth century, when the innermost reaches of Asia were penetrated by Colonel Prezhevalski, Hedin, Sir Stein, Radlov (Radloff) and Vambéry, to mention but a few. As a result, the eighth-century Kök Türk Orkhun stone inscriptions were discovered in present-day Mongolia and translated, as was Yusuf Has Hajib’s Kutadgu Bilig (1069–70), a ‘mirror for princes’ modelled on Firdausi’s Shah-Nama (1007). Since Guignes’ work in the eighteenth century, knowledge had been building up in western Europe that showed the Turks to have had an extensive and deeply rooted culture in Asia prior to the advent of Islam – one that was closely related to people who had remained outside Islamdom, such as the Finns, Hungarians, Mongolians and Tungus (Minorsky IA/12/2: 107ff; also Avcıoğlu 1979/1: 16). In this intellectual atmosphere, Pan-Turanism, which became synonymous with Pan-Turkism, first emerged as a political concept in Hungary (1839). In reac- tion to Pan-Germanism and Pan-Slavism, initially the term appears to have been a romantic expression of the Hungarian intellectuals’ interest in their ancestral lands and related people. According to them, the Finno-Ugric and Altaic peoples had originated on the steppe between the Caspian Sea and Altay Mountains, the area named as Turan (Ibid.). When Ottoman intellectuals such as Âsım appropriated this concept they differentiated a Lesser Turan for the Turks and a Grand Turan inclusive of the Finns and Hungarians. In particular, Ziya Gökalp avowed that the Turks’ native country was neither Turkey nor Turkistan, but Turan: that grand and eternal place. There were others besides Âsım at the Dar ül-Fünun-u Osman-i (Ottoman Imperial University), which was first established in 1846, then in 1869, 1870–1, 1874–81 and finally once again in 1900, until replaced by the Istanbul University in 1933. For instance, according to Ahmet Vefik Pasha (1823–91), there was Introduction 3 a difference between Turkish and Ottoman in terms of language and history (Tanpınar IA/1: 207ff; also Berkes 1998: 314). Elsewhere, concerned that translations from European works showed Turks in a false light, Süleyman Pasha excluded these from the curriculum in military schools under his jurisdiction and published instead the first and only volume of his Tarih-i Âlem (History of the World) (1876). More importantly, Mustafa Celâlettin Pasha, a Polish immi- grant, argued in his Les Turcs: Anciens et Modernes (1869) that linguistically and racially the Europeans were related to the Turks. This is noteworthy not so much from the point of view of Pan-Turkism, but because it foreshadowed the ideas that went into the Turkish Historical Thesis more than half a century later (Berktay 1983: 29; Timur 1994: 138–43; Berkes 1998: 316–17). Presented at the First Turkish Historical Congress held in Ankara in 1932, this proclaimed the Turks were not of a yellow but a brachycephalic white race; Turkish history did not consist merely of the Ottoman; the Turks were the nation that had dispersed culture to all other nations from Central Asia (Birinci Türk Tarih Kongresi, 1932). On the whole, however, during this period Ottoman intellectuals were concerned with the establishment of a constitutional monarchy, not Pan-Turkism. Swayed in particular by Montesquieu’s ideas, the Young Ottomans (Yeni Osmanlılar), thought to have been founded in 1865, advocated the separation of powers through the constitution of an elected parliament to which the administration, namely the Servants of the Porte, would be held answerable (Berkes 1998: 208–14, 304–13; Mardin 1992a: 31–45; Kuran 1945). Without a doubt, the most influential Young Ottoman was Nâmık Kemâl (1840–88), who voiced his wide ranging ideas through various publications, both at home and abroad (Tasvir-i Efkâr, Tercümân-ı Ahvâl, Hürriyet, I˙bret). Kemâl’s ideas on constitutional monarchy, which he tried later to reconcile with Shari law (Berkes 1998: 218–22; Çavdar 1995: 30), were less than welcome. Posted to Erzurum and forbidden from publishing, he went into self-exile in Paris (1867). Although continuing to publish while abroad, on his return at first he desisted (1870). This did not last; public excitement over his fervently nationalistic play Vatan yahut Silistre (My Country or Silistria) caused him to be arrested and exiled (1873). Although subsequently pardoned (1876), he was arrested again, largely due to his involvement and opposition to Sultan Abdülhamid II (1875–1909) in the drafting of the Constitution of 1876 (Berkes 1998: 223–50), and after five months in prison he was posted to the Aegean islands where he died. While in Rhodes, Kemâl started a history of the Ottoman Empire (1299–1919), which attacked the Byzantinist view of Ottoman history. This was finally published in 1908–10, some twenty years after his death (Akün IA/9: 54ff), coincidental with Mehmet Fuat Köprülü’s two works that criticized the same; Osmanlı Müesseselerine Tesiri (The Influence of Byzantine Institutions on Ottoman Institutions) and Les Origines de l’Empire Ottoman, published in 1931 and 1934. While, with others, Nâmık Kemâl helped determine the tenor of Turkish historiography, Şemseddin Sâmi (1850–1904) continued to establish the scholastic standard set by Cevdet Pasha. Gainfully employed in the palace and decorated 4 Introduction on several occasions by Sultan Abdülhamid II, Sâmi did not belong to either the Young Ottomans or their successors, the Young Turks. A journalist by profes- sion, he published educational pamphlets (Cep Kütüphânesi or Pocket Library) and encyclopaedic articles in various weekly magazines until his death. His main opus as an encyclopaedist was the Kamus al-a’lam. This was based on Bouillet’s Dictionnaire Universel d’Histoire et de Géographie, which he expanded consid- erably by adding his own articles and others from different western European sources. More specifically, there are his dictionaries: the Kamus-ı Fransevi, Dictionnaire Turc-Français, Kamus-ı Türki and the incomplete Kamus-ı Arabi. As a lexicographer, Sâmi was well aware of the differences in the Turkic language group, pointing to Central Asia for the origins of Turkish literature. He was the first to translate the Orkhun stone inscriptions and Yusuf Has Hajib’s Kutadgu Bilig into ‘contemporary’ Turkish. Originally from Albania, Sâmi is also credited with the modern Albanian alphabet (Akün IA/11: 411ff; also Berkes 1998: 320). Sâmi’s influence is likewise easily traceable. Not only the concept but also the name Cep Kütüphânesi continues to be used to this day. His encyclopaedia, the Kamus al-a’lam clearly foreshadowed the I˙slâm Ansiklopedisi. The Hayat maga- zine’s Büyük Türk Sözlüğü, published in the 1950s, was based on Kamus-ı Türkî. Sâmi took care to simplify written Turkish, thus also beginning the movement to ‘cleanse’ it of Persian and Arabic vocabulary and grammar, a task taken up by the now defunct Turkish Language Institute (Türk Dil Kurumu). It must also be noted that although strongly criticized at the time, Sâmi’s method of translating word for word is still adhered to rigidly in Turkey. Last but not least, any assessment of Ottoman influences on the development of historiography in the Republic of Turkey must include Yusuf Akçura (1876–1935), who was born in Simbirsk on the Volga River south of Kazan. His father appears to have died soon afterwards (1878). Although his mother moved to Istanbul (1883), she kept in touch with her husband’s family and her own through regular visits to Kazan. One of these relations was Ismail Gasprinski, her sister’s husband, who thought the Tatar community in Crimea would survive the Slav yoke only if Muslims throughout Russia united, or at least co-operated. Akçura, on the other hand, came to believe that being a Turk was more important than being a Muslim. He appears to have been affected by his father’s brother, who had an extensive library and spoke several European as well as Turkic languages. During one of his visits, Akçura stayed for some time with the nomadic pastoral Başkurt, whose language and customs showed little outside influence. Later, at the Military Academy in Istanbul, Akçura avidly followed the writings of Âsım and other adherents of Pan-Turkism. Having joined the Young Turks while at the academy, Akçura was arrested and imprisoned (1896). The following year he was tried along with others and exiled to North Africa. Although pardoned (1898), he was not allowed to return and escaped to Paris via Tunis with some friends (1899). In Paris, he enrolled in the L’École de Science Politique and contributed to the Young Turk publication Meşveret, as well as the journal Şura-yı Ümmet, which was published in Cairo (Akçura 1976: 1–18; Georgeon 1996: 15–17, 20–2, 28). Introduction 5 After graduation (1903), Akçura returned to Simbirsk where he wrote the article he is best remembered for, Üç Tarz-ı Siyaset (Three Kinds of Policy). The said piece, which established the theoretical basis of Turkish nationalism, was first published in Cairo in the periodical Türk (April–May 1904). Akçura argued that Pan-Islamism was impractical given the West’s almost complete hegemony over Islamdom (1976: 21–3, 31–3). On the other hand, according to Akçura, Pan- Turkism was not as yet a viable alternative. National consciousness had remained slow to take root among Turkic people (1976: 23–4, 33–4). Nevertheless, his preference for the latter policy was already clear. In his university thesis he had argued that despite Chinese, Persian, Arabic, Byzantine and Buddhist influences, Turkic people had retained not only their languages but also many common cultural traits. These were a patriarchal family structure, the common ownership of land, a focusing of political power in the person of an elected ruler whose sovereignty was curtailed by customary law, an aristocratic class, a tendency for statecraft and an exceptional religious tolerance (Georgeon 1996: 29). Soon after the restoration of the 1876 Constitution by the Young Turks (23 July 1908), Akçura returned to Istanbul. He became an outspoken leader of the nationalist movement although he did not join the Young Turks’ Society for Union and Progress (Ibid. 60–72). In 1919, however, he became a founding member of the first political party to include the name Türk, namely the Millî Türk Fırkası (National Turkish Party) (Tunaya 1952: 383, 418, 441–2). After the occupation of Istanbul by the Allies (March 1920), Akçura joined Atatürk in Anatolia and fought at Sakarya (January and March 1921). These two battles reversed the Greek invasion of western Anatolia. On the establishment of the Republic of Turkey (29 October 1923), Akçura became a Member of Parliament for Istanbul for the Republican People’s Party. In 1932, Atatürk appointed Akçura to head the Society for the Study of Turkish History. Akçura was a founding member. As a result, that same year he chaired the First Turkish Historical Congress (Georgeon 1996: 126–8), which had a lasting effect on Turkish historiography.

Republican concerns Atatürk’s concern in setting up the Society for the Study of Turkish History was to draw attention away from the Islamic Ottoman past in order to better focus Turkish patriotism on its independent republic in Anatolia and eastern Thrace. But in his efforts to consolidate the Turkish nation-state, Atatürk could not ignore the West’s perception of the Turks’ place in history. As early as 1735 de Linné categorized people by their physical and moral characteristics. Accordingly, those blond and blue eyed were superior to those with yellow skin, namely the melancholic, mean and ruthless Asiaticus. Basing his views on men of letters, such as Belon, who had travelled in the Ottoman Empire, de Linné classified the Turks in the former category. Nearly half a century later, in 1775, Lavater argued that the Turks’ physiognomy showed a mixture of aristocratic Anatolian and crude Tatar blood – a downgrading to be sure. However, as indicated by his contemporary, the craniologist Camper, who, 6 Introduction as de Linné, thought the Turks to be of the white race, scientific opinion appears to have been divided at this time. In 1839, another craniologist, Morton, finally classified the Turks under the Mongols while agreeing with Lavater that they had inter-married with the Aryan and Semitic races. In 1879, coincident with growing interest in the whereabouts of the Indo-Europeans’ ancestral lands, the anthropologist de Mortillet claimed that the Neolithic Revolution, namely the domestication of animals and plants, was the product of a brachycephalic people. By tracing the domestication of animals, he proposed that these people had originated in the Caucasus (Timur 1994: 120–5). However, ‘civilization’ had already been correlated with ‘race’ by men of letters such as de Gobineau (Essai sur l’Inégalité des Races Humaines) (1853–5). Another such work, Leon Cahun’s Introduction à l’Histoire de l’Asie (1896), which was translated by Âsım, considered both the Turks and Mongols to be half- Chinese, though derisory of scientific claims to that effect. Be that as it may, in his opinion the Turks and Mongols had not progressed beyond being intermediaries between China and Persia. This was because of their self-serving greed, which had prevented them from appropriating either civilization. Given that they preferred to believe rather than comprehend, by championing Islam against a Christian Europe, according to Cahun, from the Seljuqs onward, Turks had dissipated their genius for courage, obedience, directness and common sense – without ever becoming aware of what they were doing (Ibid. 135–6; also Berkes 1998: 315). Cahun’s portrayal of the Turks as noble savages seems to have been overlooked by adherents of Pan-Turkism in favour of his attempt to demonstrate through a comparison of Kutadgu Bilig and the Great Seljuq Sultan Malik-Shah’s vizier Nizam al-Mulk’s Siyasat-Nama, that the origins of the Turks’ national character lay in Central Asia not Islamdom (Ibid. 137). There is also his view of the Turks as a martial élite born to rule (Ibid. 136). Particularly as, at the time, the Ottoman intellectuals’ overriding concern remained the salvation of their imperial polity, this may have encouraged them to pick out suitable ideas. It is clear that the Ottomans were not blind to European claims of racial superiority; quite the opposite. They were beginning to despair of this attitude well before the First World War (Temo 1939: 170–2; also Berkes 1998: 308, 313–14 and Mardin 1992a: 274). It is all the more ironic, therefore, that ideas such as those of Morton and Cahun became so widely accepted in the West as to appear in French textbooks used in Turkish schools during the late 1920s. Atatürk’s adopted daughter Afetinan drew his attention to textbooks at her school that classified the Turks, racially, as second-class Mongoloids, referring to them as an ‘inva- sive barbaric tribe’ (Avcıoğlu 1979/1: 20; also Berktay 1983: 51). As a result, Atatürk, who was already involved in how history was being taught (Ibid. 19), now set to work with the express purpose of bringing to light ancient Turkish civilization (Inan 1981: 194–8). Nevertheless, it is clear that he was not satisfied with the initial drafts of the Turkish Historical Thesis presented at the First Turkish Historical Congress held in 1932 (Avcıoğlu 1979/1: 21–7; also Timur 1994: 144–50 and Çavdar 1995: 299–322). Certainly, Köprülü (1890–1966) and Zeki Velidi Togan (1890–1970), both of whom were to gain Introduction 7 international recognition, took the presenters of the thesis to task (Birinci Türk Tarih Kongresi 1932: 42–7, 79, 82–3, 167–76, 308–20, 369–76). Significantly, Pittard was the honorary chairman of the Second Turkish Historical Congress (1937). Having studied the Turks in the Balkans in 1911 he had come to the conclusion that their skulls were mesocephalic, a cross between brachycephalic and dolichocephalic (Timur 1994: 125–6). In a paper presented at the congress he put forward the view that the citizens of the Republic of Turkey were the inheritors of all the cultures present prior to the conquest of Anatolia and the Balkans (Avcıoğlu 1979/1: 28, 30–1; Timur 1994: 147–8). Finally, this was also Atatürk’s perception (Avcıoğlu 1979/1: 27–32; Frye 1996: 4). Despite xenophobic excesses, under the guidance of historians such as Köprülü and Togan (who eventually returned), Turkish historiography continued to build on the scholarly standards established during the final century of the Ottoman Empire. In particular, Köprülü’s successful application of Akçura’s thesis to the origins of Ottoman institutions (Georgeon 1996: 29; Berktay 1983: 35) focused Turkish historians on the Seljuqs. Three of them, Sümer, Turan and Köymen, were Köprülü’s assistants or students (Berktay 1983: 90).

Why the Seljuqs? Köprülü maintained that a nation’s history could not be studied properly until set in its correct historical perspective. In his opinion, given Turkic hegemony over Islamdom during the second millennium, just as Islamic history cannot be eval- uated without an understanding of Turkic history, so medieval Turkish history cannot be understood outside an Islamic context. For instance, in order to study Islamic jurisprudence it is not sufficient to understand the theory of Islamic law, one must also evaluate the origins and development of Turkic legal institutions beside those of the Arabs and Iranians (Köprülü 1940: xvii–xx). Köprülü thought that the distinguishing characteristic of a comparative methodology should be the historian’s ability to disregard thousands of reported incidents that have no bearing on the historical evolution of a society, because these are either of a secondary nature or totally irrelevant (Köprülü 1994: 25). That said, Köprülü did not approve of the selection of historical facts to justify dogmatic interpretations (Ibid. xxi–xxii). On the contrary, it is his open admiration of Febvre and hence Bloch, thus the historians associated with the Annales d’Histoire Economique et Sociale (Ibid. xix), which pinpoints his influences (Berktay 1983: 84). His insistence on the importance of social factors, that is, religious, legal and economic as well as structural characteristics, distinguished him not only from doctrinaire philosophers and political historians, but also from empiricists such as his contemporary Ömer Lütfü Barkan (Ibid. and 1992: 109–84; see also Köprülü 1994: 24). Köprülü initially researched the origins and development of Turkish literature. Türk Edebiyatı Tarihinde Usül (Method in the History of Turkish Literature) was published in 1913 (see Köprülü 1966: 3–47), Türk Edebiyatında I˙lk Mutasavvıflar (Early Mystics in Turkish Literature) followed in 1918 (Köprülü 1993), and later there was Türk Edebiyatı Tarihi (History of Turkish Literature) (also Köprülü 8 Introduction IA/12/2: 530–65). But it was his application of Akçura’s thesis to the question of whether or not the Ottomans had taken their institutions from the Byzantines in Bizans Müesseselerinin Osmanlı Müesseselerine Tesiri (The Influence of Byzantine Institutions on Ottoman Institutions), which was published in 1931, that set the standard (Köprülü 1981). According to Köprülü, in western European historiography, the nomadic clan situated near the Byzantine Empire’s marches by the Seljuqs of Anatolia, which went on to found the Ottoman Empire, consisted of simple shepherds. Since they had only recently accepted Islam under Osman, the Ottomans must have acquired the civilizing elements of statecraft from their non-Muslim subjects, namely the Byzantines – a process consolidated by the conquest of Constantinople. Köprülü’s argument was diametrically the opposite. He proposed that with the exception of kapudan paşa (grand admiral), which he dated to the seventeenth century, Ottoman titles and institutions were derived from the Seljuqs of Anatolia and hence the Great Seljuqs in Iran and Iraq. Consequently, they were in the main Sasanian, not Byzantine; although Köprülü thought some were solely Turkic. He pointed to the Anadolu and the Rumeli Beylerbeyi, the governors general of Anatolia and the Balkans, which correlated to the bipartite principle of Right and Left Wing seen in Turkic polities, the latter being the senior and in the East as Turkic tent open- ings faced south (Çandarlıoğlu 2003: 95; see also Pritsak IA/6: 212). Similarly, the Ottoman concept of world dominion and the symbol of the crescent, both of which Western historians thought had been taken from the Byzantines were, according to Köprülü, essentially Turkic – although not uniquely so. The former had also been part of Abbasid ideology and the Sasanians had employed the latter. There was also the drum and banner, which had been symbols of Kök Türk sovereignty, so that they could not have been taken from Islamdom. As for the Ottoman tımar system, in Köprülü’s opinion this was developed by the Seljuqs and did not derive from the Islamic iqtac. He argued that under the Seljuqs lands and a set amount of their revenues were apportioned among the military on a hereditary basis, whereas previously the military had been paid quar- terly. Last but not least, in Köprülü’s estimation the office of defterdar (minister of finance or director of provincial finances) was originally Ilkhanid. In short, Köprülü’s argument was that the Abbasid caliphate and hence the Seljuqs and their successor states, namely the Khwarazmshah, the Ayyubid, the Mamluk and the Ilkhanid had all adopted and modified Sasanian and, therefore, Irano-Islamic institutions. He left it for others to determine how much these had been influenced by earlier Roman as well as Byzantine institutions and practices. Not surprisingly, in his subsequent work, Les Origines de l’Empire Ottoman, which consisted of three lectures given at the Sorbonne in Paris and was first published in French the following year in 1934, Köprülü argued that the origins of the Ottoman Empire must be sought in the socio-economic and political mosaic of Anatolia during the thirteenth and fourteenth centuries, which consisted of various Turkic polities, and not that of the Byzantine Empire. Köprülü did not stop there. Over the next decade, mainly in papers published in the scholarly journal Belleten (see Köprülü 1938 and 1941), he defined the Introduction 9 main thrust of the argument Turkish scholars have applied in their interpretations of Seljuq history. According to Köprülü, the Turks entered Islamdom with their tribal customs and laws. Although some of these changed after they established Turko-Islamic states, it was natural for many others to continue under an Islamic veneer. In his opinion, the most important of these was the Turkic leaders’ tradition of law- making. Consequently, because their empire was founded by former affiliates of the Kınık and other Oguz lineages rather than by Turkic mamluks (slave soldiers), as in the case of the Ghaznavids, the Seljuqs could not command the loyalty of the Türkmen without their chieftains’ fealty, which meant that at first customary rather than Shari laws prevailed among the Türkmen. As their empire took on an Islamic character, however, the Seljuqs allied instead with the indigenous aristocracies who became their civil administrators and broke up the Türkmen, settling them on the marches or on smaller individual or clan iqtacs rather than large tribal holdings. The policy was supported by the autocratic traditions of the Great Seljuqs’ viziers, namely Kunduri and Nizam al-Mulk. Although in his opinion only partially successful, Köprülü maintained that this policy nevertheless succeeded in strengthening central government because it lessened the influence Türkmen chieftains could wield over the Seljuqs, whom they regarded as first among equals (primus inter pares) (Berktay 1983: 68–73). Köprülü never ceased to be critical of western European historiography. In his introduction to Barthold’s I˙slâm Medeniyeti Tarihi (History of Islamic Civilization), he noted that the forces driving Islamic studies during the nineteenth century had ceased to be Biblical studies or missionary activity. Western powers needed to learn the languages and religion of their subjects in order to administer them and to sell their goods. The Eurocentric histories that resulted were racially biased, invariably prejudging the Turks’ historical role as solely military and destructive (Köprülü 1940: xxviii–xxix, 149–50 and 1981: 23; also Berktay 1983: 14–15). A memorandum to the Turkish delegation at the Sèvres Peace Talks confirms this (23 June 1919). According to the Allies, the Turks had ravaged and destroyed the lands they had conquered in Christendom and in Islamdom, because it was not in their nature to develop in peace what they had won in war (Berktay 1992: 138–9). In sum, by challenging western European historical conceptions of the Ottoman Empire, Köprülü not only directed Turkish scholars to the study of the Seljuqs but also established a framework for this. It is, however, an oversimplification to call him the father of modern Turkish historiography, as Strohmeier and Leiser do. This denies the importance not only of nineteenth-century Ottoman intellectual concerns but also of Togan’s and Uzunçarşılı’s outstanding contributions to Turkic and Ottoman historiography respectively.

The historians and their works The first Turkish history of the Seljuqs was Mükremin Halil Yınanç’s Selçuklular Devri Türkiye Tarihi (The Seljuqid Period of Turkey’s History), which was published in 1944 towards the end of the Second World War. Yınanç wrote that due 10 Introduction to the war he was not able to gain access to all the relevant documents as some of these were in Europe. He did not, however, either discuss or reference these documents in a clear and systematic fashion. His book contains a general list of primary sources in the introduction and more specifically at the end of each chapter. As a result, it is difficult to ascertain which primary or secondary sources he used to arrive at his conclusions. For this reason Yınanç’s history of the Seljuqs has not been referenced except where other Turkish scholars have done so. Zeki Velidi Togan (1890–1972) did not write a history of the Seljuqs, but he does refer to them in his Umumi Türk Tarihi’ne Giriş (Introduction to the General History of the Turks) and Oğuz Destanı – Reşideddin Oğuznâmesi, Tercüme ve Tahlili (The Oguz Epic – Translation and Analysis of Rashid al-Din’s Oguz-Nama). A protégé of Sir Aurel Stein, Togan was a Başkurt who had studied at Kazan University and had done research work at the Petersburg Academy before the First World War. Politically active, he was instrumental in establishing the Başkurt army, which sided with the Bolsheviks. For a time Togan was also a member of the Committee of Six, which included Lenin, Stalin and Trotsky, but as a leader of the Başkurt independence movement he later came into conflict with the Bolsheviks and fled to Turkey (1924) (Togan 1969). There he became an advisor to the Ministry of Education and then Professor of Turkish at Istanbul University. When asked to teach the Turkish Historical Thesis he resigned and went to Vienna where he studied medieval history under Professor Dopsch. He gained his doctoral degree with a thesis on the Arabic text of Ibn Fadlan’s journey to the Urals in a manuscript he had discovered in Mehshed (Koestler 1952: 212–13). Togan returned to Turkey a year after the Second Turkish Historical Congress (1937) and was appointed Professor of Medieval Turkish History at Istanbul University, where he remained. Since the historical framework he proposed in Umumi Türk Tarihi’ne Giriş was based on an internationally acclaimed standard of scholarship, Turkish scholars could not ignore his views on the Seljuqs, which downgraded their role. Togan thought Selçuk Beg was the descendant of a tent pole sharpener, which was challenged by Kafesoğlu in his Selçuklu Ailesinin Menşei Hakkında (On the Origins of the Seljuq Family). Togan also argued that the Mongols unified Turkic people, enabling them to become established in the Middle East (1981: 182ff), a view with which Sümer agreed (1992: 7). According to Togan, without a further influx under the Mongols, the Türkmen could have become assimilated by the Iranians and the Greeks (1981: 206–22). Togan extended his analysis to include the Ottomans. According to him, it was Timur’s defeat of Sultan Bayezid I that enabled the consolidation of the Ottoman Empire. Otherwise, the Türkmen would have become spread too thinly across the Middle East and the Balkans, possibly also Italy, and the Ottoman Empire would have lost its Turkish character (Ibid. 347–50), a point with which Turan by and large also agreed (1993a). İbrahim Kafesoğlu (1914–84), Togan’s most vociferous critic, was not only familiar with Islamic sources on the Seljuqs but also western European, Hungarian and Russian research on Inner Asia. During the Second World War, Kafesoğlu briefly studied with A. Alföldi, Gy. Németh, D. Ligeti and L. Ràsonyi in Budapest, being forced to return to Istanbul due to the Russian occupation. Back in Turkey Introduction 11 he became Yınanç’s assistant and thus involved in Seljuq history. His doctoral thesis was published as Sultan Melikşah Devrinde Büyük Selçuklu I˙mparatorluğu (The Great Seljuq Empire at the Time of Sultan Malik-Shah). Subsequently, his thesis for assistant professorship, Harezmşahlar Devleti Tarihi (History of the Khwarazmshah State), was also published as a book in 1956. After a brief stint at Erzurum’s Atatürk University, Kafesoğlu returned to Istanbul, whereupon Togan’s death he was appointed to the latter’s chair (Leiser 1988: 14). Besides numerous other articles in the IA, Kafesoğlu wrote several works on the origins of the Turks and their culture, as well as schoolbooks that exalted the Turks’ role in history, which have allowed him to be accused of distorting history (Strohmeier 1984: 163–82; see also Akpınar 1993: 48–9). Except for the overreach- ing conclusion with regard to the Seljuqs’ role in world history, however, such criticism is not sustainable with regard to his work on the Seljuqs’ family origins, Sultan Malik-Shah’s reign, or the Khwarazm-shah, at least not without a compre- hensive assessment of his use of primary sources. The views Kafesoğlu is criticized for by Strohmeier and Akpınar alike adhere closely to those held by the Vienna Diffusionist School. Whatever their merit, in principle these differ little from many other historians’ views, such as René Grousset’s, who claimed that like the Romans, the Turko-Mongols were one of the few governing races on earth (1997: xxx). Osman Turan (1914–78), Köprülü’s assistant, is not only the most popular Turkish historian on the Seljuqs today but also the most criticized (Strohmeier 1984: 151–62; also Akpınar 1993: 47). His most popular work is a history of the Seljuqs in Anatolia, Selçuklular Zamanında Türkiye (Turkey at the Time of the Seljuqs). His other work on Anatolia is Doğu Anadolu Türk Devletleri Târihi (History of the Turkish States in Eastern Anatolia). More importantly, however, he authored a highly controversial overview of Seljuq civilization, Selçuklular Târihi ve Türk- I˙slâm Medeniyeti (Seljuq History and Turko-Islamic Civilization), a 500-page work originally rejected by the editorial board of the IA due to its tardiness and length (Leiser 1988: 158ff; see also Akpınar 1993: 46). When Kafesoğlu’s much shorter article was published instead (1964–5), Turan accused him of plagiarism in the scholarly journal Belleten and again in the preface to the second edition of his book. Besides numerous earlier publications and articles, which are in the main still highly regarded (Ibid. 137–9; also Strohmeier 1984: 151–62), Turan wrote two further books entitled Türk Cihân Hâkimiyeti Mefküresi Tarihi (History of the Turkish Concept of World Dominion) and Selçuklular ve I˙slâmiyet (The Seljuqs and Islam). It is the ideological aspects of these and his evaluation of the Seljuqs in the second half of Selçuklular Târihi ve Türk-I˙slâm Medeniyeti, which Turan kept revising and adding to, that have tarnished his reputation. Unlike Köprülü, who was a founding member of the Democrat Party and served as Foreign Minister, Turan can be said to have allowed his politics to distort his scholarly work. Having entered Parliament in 1954 as a member of the Democrat Party, Turan was imprisoned for 18 months after the coup d’état of 27 May 1960. He was tried by the junta’s military tribunal and, like his mentor Köprülü and other Democrat MPs, cleared of violating the constitution. Like many of his political colleagues, Turan was of the view that the materialistic Western 12 Introduction values adopted since the Tanzimat period were at odds with Islam, which he maintained had enabled the Turks’ many brilliant periods in history. Turan considered the reforms effected under the Republic of Turkey needlessly harsh and the cause of its moral crisis. Nevertheless, he did not advocate a return to purely Islamic values. Instead of being adopted wholesale, he advocated adapting Western science and industry to Turko-Islamic values. His criticism of Atatürk’s reforms in various articles caused him to be expelled from the Turkish Historical Institute. When he eventually left politics, because he deemed the successor to the Democrat Party too Left Wing, he was refused a university posting and thus also left out of academic circles (Köymen’s biog- raphy in Turan 1979: 13–36; Akpınar 1993: 47). Mehmet Altay Köymen (1915–93) concerned himself solely with the history of the Seljuqs in Khurasan and Iraq, the so-called Great Seljuqs. However, only the first and last volumes of his proposed five-volume history were published, namely Büyük Selçuklu I˙mparatorluğu – Kuruluş Devri (The Great Seljuq Empire – Foundation Period) and Büyük Selçuklu I˙mparatorluğu – I˙kinci I˙mparatorluk Devri (The Great Seljuq Empire – 2nd Empire Period). According to Köymen, his books Tuğrul Bey ve Zamanı (Tuğrul Beg and His Time), Alp-Arslan ve Zamanı (Alp-Arslan and His Time) and associated articles in the I˙slâm Ansiklopedisi (IA), the Ankara University DTCF’s Tarih Araştırmaları Dergisi (DTCF Journal of Historical Research) and the Selçuklu Araştırmaları Dergisi (Journal of Seljuqid Research) were to provide the basis for his proposed second and third volumes. Similarly, although it required revision and expansion in light of new material, Kafesoğlu’s work on Sultan Malik-Shah was to provide the basis for the fourth volume. The problem with this framework is that the proposed series did not include the reigns of Sultan Berk-Yaruk and Sultan Tapar, which Köymen nevertheless covered in his Selçuklu Devri Türk Tarihi (The Seljuqid Period of Turkish History), an overview of the Great Seljuqs. Strohmeier also criticized Köymen. He thought Köymen had allowed his politics to influence him when he claimed the Oguz justified Atatürk’s precept of a classless society (1984: 137–51). Köymen replied in writing, saying that Strohmeier would better occupy himself criticizing his academic works on the Seljuqs, rather than trawling through publications on current affairs; in his opinion this was what a qualified historian did (1986: 23–4). Faruk Sümer (1924–95) wrote Oğuzlar (Türkmenler) – Tarihleri, Boy Teşkilatı, Destanları (The Oguz (Türkmen) – History, Organization, Epics), Yabanlu Pazarı – Selçuklular Devrinde Milletlerarası Büyük Bir Fuar (Foreigners’ Market – A Large International Fair During the Seljuq Period) and Selçuklular Devrinde Doğu Anadolu’da Türk Beylikleri (The Turkish Beyliks in Eastern Anatolia During the Seljuq Period), as well as numerous articles in both the IA, various Turkish periodicals, and the second edition of the Encyclopaedia of Islam. The first I˙slâm Ansiklopedisi is a translation, revision and expansion of the first edition of the Encyclopaedia of Islam, commissioned by the Ministry of Education. Sümer’s magnum opus Oğuzlar, which was first published in 1965, is of considerable importance. This is because despite its brevity on the Seljuqs it not only Introduction 13 excludes material incorporated by his contemporaries but on one occasion also criticizes poor scholarship, a rarity in Turkish historiography. Sümer’s student Ali Sevim has since written several studies of the Seljuqs in Syria and Palestine, the most notable being Suriye-Filistin Selçuklu Devleti Tarihi (The History of the Seljuqs of Syria and Palestine). His reviews of the Türkmen chieftains’ raids into Anatolia under the Seljuqs’ direction, Ünlü Selçuklu Komutanları – Afşin, Atsız, Artuk ve Aksungur (Famous Seljuq Commanders – Afşin, Atsız, Artuk and Aksungur), and of Süleyman-Shah, Anadolu Fatihi Kutalmışoğlu Süleymanşah (The Conqueror of Anatolia, Süleyman-Shah), however, are comparatively lightweight, as is his Selçuklu-Ermeni I˙lişkileri (Seljuq-Armenian Relations). There are also Erdoğan Merçil’s Kirman Selçukluları (Seljuqs of Kirman) and recently Abdülkerim Özaydın’s Sultan Berkyaruk Devri Selçuklu Tarihi (485– 498/1092–1104) (Seljuq History during the Reign of Sultan Berkyaruk) and Sultan Muhammed Tapar Devri Selçuklu Tarihi (498–511/1105–1118) (Seljuq History during the Reign of Sultan Muhammad Tapar). Mention must also be made of Işın Demirkent’s two-volume Urfa Haçlı Kontluğu (The County of Edessa). The first volume covers 1098–1118 and the second 1118–48. Recent research appears to be on the Seljuqs and Mongols in Anatolia. There are Ahmet Yaşar Ocak’s studies, namely Selçukluların Dini Siyaseti (1040–1092) (The Seljuqs’ Religious Policies (1040–1092)), Babailer I˙syanı – Aleviliğin Tarihsel Altyapısı (The Babai Uprising – Historical Foundations of Alawism) and Bektaşi Menâkıbnâmelerinde I˙slam Öncesi I˙nanç Motifleri (Pre- Islamic Beliefs in Bektashi Legends), which directly concern Seljuq history, and still others he has written on aspects of religious life during pre- and Ottoman times. Other works include Tuncer Baykara’s Türkiye Selçuklularının Sosyal ve Ekonomik Tarihi (Social and Economic History of the Seljuqs of Turkey), Ali Üremiş’ Türkiye Selçuklularının Doğu Anadolu Politikası (Policies of the Seljuqs of Turkey in East Anatolia) and Muammer Gül’s Doğu ve Güneydoğu Anadolu’da Moğol Hakimiyeti (Mongol Rule in East and Southeastern Anatolia). One exception is Osman Özgüdenli’s Ortaçağ Türk-I˙ran Tarihi Araştırmaları (Researches in Medieval Turco-Iranian History), which analyses aspects of Great Seljuq rule. However, this cannot hide the fact that the main body of Turkish scholarly writing on the Seljuqs is now quite dated. Kafesoğlu wrote his works on the Seljuqs’ origins, Sultan Malik-Shah and the Khwarazm-shah in the first half of the 1950s, his article on the Seljuqs in the IA having been published in 1965, when he was already far more concerned with Eurasian history. Turan published all his work on the Seljuqs between 1965 and 1971. Köymen, on the other hand, published in a steady stream over four decades, from the 1950s into the 1980s. Nevertheless, all but one of his major works on the Seljuqs was published in the 1970s. The policy of the Council for Higher Education, which insists all theses presented cover new ground, is the main reason why the above-mentioned historians’ works have not been reassessed in newer studies. The one exception is the journalist Doğan Avcıoğlu. His five-volume evaluation of the history of the Turks, Türklerin Tarihi, concerns itself solely with the Seljuqs 14 Introduction from the latter part of the third volume. However, having, on his own admission, taken a Marxist approach, Avcıoğlu spent all five volumes discussing whether or not nomadic pastoral cultures and Islamdom were ‘feudal’ and the applicability of Karl Marx’s ‘Asiatic Mode of Production’ to Turkic polities. Unlike Gordlevski, who also applied Marxist theory in his book on the Seljuqs of Anatolia, which has been translated into Turkish as Anadolu Selçuklu Devleti (The Anatolian Seljuq State), Avcıoğlu based his conclusions only partly on primary sources. Given that his use of secondary sources was also highly selective, Avcıoğlu’s work failed to shed any new light on the Seljuqs or their historians, at best airing certain questions of political history at issue between Turkish intellectuals during the Cold War. Aside from the above works of Turkish historians, there are also those Western scholars whose major works are not only available in Turkish but also are frequently referred to, namely Barthold (whose Turkistan Down to the Mongol Invasion dates from 1900), Grousset and Cahen. Last but not least, in this context mention must also be made of the sinologist Eberhard, who published some of his books first in Turkish and, therefore, is regarded here as belonging also to Turkish historiography. Eberhard is referenced with regard to early Inner Asian history.

The Turkish view and its problems The first generation of Seljuq scholars wrote detailed historical overviews that aid the task of identifying their interpretations. Of these Köymen’s Selçuklu Devri Türk Tarihi (The Seljuqid Period of Turkish History) is by far the superior. One reason is its structure. Köymen reviewed Great Seljuq history on the basis of the Seljuqs’ relationships, such as with the caliphate and the Türkmen. Having listed his primary and secondary sources after each section, at the end of the book Köymen also assessed available primary sources. Köymen’s overview was published in 1963, before either Kafesoğlu’s article in the IA or Turan’s Selçuklular Târihi ve Türk-I˙slâm Medeniyeti. In comparison to Köymen, both Kafesoğlu’s and Turan’s approaches make it difficult to establish a meaningful connection between their assessments and their preceding historical narratives. The problem becomes more acute the further one progresses into their evaluations, which take up half their works respectively. One reason is that both authors included material more suited to historians of literature and the arts, not to mention economic history. More importantly however, unlike Kafesoğlu and Turan, Köymen did not attempt to define the Seljuqs’ impact on Islamdom, let alone world history, a different topic altogether. Like Köymen, Sümer did not offer any overarching interpretations; however, he treated the history of the Seljuqs as a defining moment for the Oguz. According to him, the Seljuqs’ conquests enabled the Oguz to settle in the area covered by present-day Iran, Azarbayjan, northern Iraq and Syria, and Turkey (1992: 68–120). Subsequent historians on the Great Seljuqs, namely Merçil, Sevim and Özaydın have sufficed with brief summaries tagged on to their specific studies. Introduction 15 Any comments they have made in these remain very much in line with their predecessors’ assessments, which remained well within Akçura and Köprülü’s guidelines. As will be remembered, according to Akçura, Turkic polities were based on a patriarchal family structure, common ownership of land, an aristocratic class that elected a ruler whose sovereignty was curtailed by customary law, a tendency for statecraft, and an exceptional religious tolerance. To these precepts Köprülü added world dominion, which he noted was not unique to Turkic polities. With respect to the Seljuqs, Köprülü proposed that given the Islamic character of their empire, they had allied with the indigenous aristocracies, breaking up the Türkmen and settling them on the marches. Even though this was not wholly successful, according to Köprülü, it served to break the Türkmen chieftains’ hold on the Seljuq sultans, whom they considered first among equals rather than absolute rulers. Köymen agreed; he thought the Seljuqs had come under the influence of their subjects and that as a result their empire had acquired an Irano-Islamic character. He noted that from the time of Sultan Alp-Arslan the Seljuqs ceased to hold councils (kengeş) whereby they chose their leader, determined policy and apportioned their lands. Nonetheless, according to Köymen, the Seljuqs’ concept of sovereignty remained Turkic. Based on the common ownership of land, it was the main reason for their succession battles. However, the Great Seljuq Empire did not break up along dynastic lines; in Köymen’s opinion it fractured first along tributary lines. This was because, whether Türkmen begs (nobility) or mamluks (slave soldiery) of Turkic origin, the commanders shared sovereignty with the Seljuqs they were affiliated with. As a result, in time they were able to usurp the Seljuqs’ authority wherever it weakened and establish their own; first on the western borders and then throughout the Great Seljuq Empire after the defeat at Katvan (1141) (1963: 11, 58–96, 97–157). Köymen divided the Türkmen into two categories: those who went to Anatolia and those who did not. Although the latter became incorporated in their empire as military, Köymen thought that the Türkmen remained at odds with the Seljuqs. He noted that particularly in the early days the Türkmen sided with dynastic contenders who promised them that Iranian viziers would not be appointed without their approval (1963: 158–65). As for the Seljuqs’ relationship to the Abbasids, Köymen considered this to be political. In his view, despite Tuğrul Beg’s initial attempt to gain caliphal approval after the Seljuqs’ victory at Dandankan and legitimize their sovereignty over Khurasan, it was Caliph al-Qa’im’s attempts to improve Tuğrul Beg’s dealings with the Twelver Shicite Buyids that led to the establishment of for- mal relations. According to Köymen, matters improved immeasurably when the Buyid commander in Baghdad, al-Basasiri, proclaimed allegiance to the Fatimids and in return received financial and military support. Invited by Caliph al-Qa’im, Tuğrul Beg arrived in Baghdad with a large army and terminated the Buyid state, imprisoning its leader (1055). During Tuğrul Beg’s stay the caliph not only married Çağrı Beg’s daughter but also proclaimed Tuğrul Beg sultan and ‘Ruler of the East and the West’ (1058). The latter title was wholly Turkic and marked the 16 Introduction separation of religious affairs from worldly ones, changing Islamic statecraft such that it became secular (1963: 168–205). Nonetheless, according to Köymen, whenever they could the Abbasids treated the Seljuqs as their protectors, that is, as emirs or generals. Matters came to a head when Sultan Mahmud (1118–31) refused to deal with the local ruler of Hilla. Caliph al-Mustarshid (1118–35) took the field and although he was defeated from this point the Abbasid caliphs gained in power, playing one Seljuqid off against another, finally achieving a brief period of independence upon the collapse of the Great Seljuq Empire (Ibid.). In contrast to Köymen, Turan blamed the wars of succession and replacement of the Türkmen by mamluks on the Seljuqs’ attempt to centralize power in the Iranian tradition. He thought Tuğrul Beg the instigator of this process, as did Sümer. Although Kafesoğlu strongly disagreed, along with Sümer and Turan he remarked on Tuğrul Beg’s declared willingness to rule in accordance with local institutions from when he first entered Nishapur (1038). According to Kafesoğlu, the fact that Tuğrul Beg’s viziers from this date on were Iranians confirmed this. Sümer also remarked that Tuğrul Beg threatened suicide when Çağrı Beg wanted to sack Nishapur on grounds that unless they were rewarded the Türkmen would find new leaders. Having evacuated the city, Tuğrul Beg is said to have personally compensated its residents (Turan 1993a: 234–7; see also Sümer 1992: 82, 87, 91 and Kafesoğlu IA/10: 386–7). According to Turan, after Sultan Tuğrul, Seljuqid adherence to local institutions was pursued vigorously by Sultan Alp-Arslan’s and Sultan Malik-Shah’s renowned vizier, Nizam al-Mulk. Nevertheless, he argued that the Turkic principle of common ownership prevented the Great Seljuq Empire from becoming autocratic in the Iranian or Greek traditions. Turan considered common ownership the basis of what he termed Turkic feudalism, pointing to the Kök Türk (552–740) and the Karakhanids (840–1212) as mature examples. So that, having failed to centralize into an autocratic empire, according to Turan, under Great Sultan Sancar the Seljuqs reverted to kind (1993a: 85–7, 234–7). Turan saw the Seljuqs’ relationship with the Abbasids as part of the struggle between Sunnis and Shicites. He argued that Tuğrul Beg was proclaimed sultan only after he promised Caliph al-Qa’im he would open the pilgrimage or hajj routes. Turan used this to explain the size of the army with which Tuğrul Beg came to Baghdad (1055), as well as the time lapse between his arrival and his investiture (1058). He pointed out that Tuğrul Beg spent the time in between to drive al-Basas- iri and the Fatimids from most of Mesopotamia. However, he failed to comment on subsequent relations between the Seljuqs and the Abbasids (1993a: 91–4). Kafesoğlu differed from both Köymen and Turan. Although he did not deny that common ownership prevented a clear line of succession, he argued that each Turkic dynast thought he was a rightful heir because they had the same blood. Given their further belief that victory in combat signified divine favour, according to Kafesoğlu, Turkic succession tended to be determined on the battlefield (see İnalcık 1993b: 37–69). Kafesoğlu further argued that the Turkic concepts of nobility and divine favour were both necessary for world dominion. Consequently, Introduction 17 in his view, the Great Seljuq Empire was based on an aristocratic warrior class rather than the principle of common ownership. According to Kafesoğlu, despite the existence of a diwan or council of ministers through which the Seljuqs taxed their realms, dynasts, Türkmen begs, the heads of local dynasties, vassals and commanders alike reported directly to the sultan (IA/10: 386–7, 390, 397). Like Turan, Kafesoğlu established a link between the Great Seljuqs and the Kök Türk in his arguments. He pointed to Great Sultan Sancar’s claim that sovereignty had passed to him from his father and that God had placed the world under his authority (see also Turan 1993a: 189 and Köymen 1954: 219, 222), remarking that this claim was almost identical to those made by Kök Türk rulers on the Orkhun stone inscriptions (see Tekin 1988: 3, 29). With regard to the Türkmen, Kafesoğlu followed Köprülü. He stated that the Seljuqs had been forced to relegate the Türkmen to a secondary role because of the extent of their conquests, which necessitated Islamic elements to be introduced into Turkic statecraft. In defining the latter, once again Kafesoğlu referred to the Orkhun stone inscriptions, as did both Turan (1993a: 72) and Köymen (1976: 20). In the inscription erected for him, the Kök Türk Bilge Kagan states that he expects loyalty in return for having provided food and clothing (see Tekin 1988: 43). Kafesoğlu noted that the concept had not changed by Karakhanid times. In Yusuf Has Hajib’s Kutadgu Bilig (1069–70), a Turkic noble was still expected to be benevolent and correct in behaviour (also Köymen 1976: 20). According to Kafesoğlu, this contrasted starkly with the Iranian emperors or Abbasid caliphs whose authority had been unconditional. Consequently, having replaced the Türkmen with mamluks, as Köymen also pointed out, Kafesoğlu was of the opinion that the Seljuqs redirected the Türkmen to Anatolia; a point with which Turan also agreed at length (1993a: 72–6). According to Kafesoğlu, this policy not only resolved the ‘difficulties’ caused by the Türkmen’s seasonal migrations, but also helped to undermine the Byzantine Empire in preparation for conquest (IA/10: 386–7, 390, 397). Kafesoğlu’s view of the Great Seljuqs’ relationship with the Abbasid caliphs was similar to Köymen’s. Despite the reverence of Sunni Muslims, he thought the Abbasid caliphs were concerned almost solely with regaining worldly power and to this end spent their time plotting against the Great Seljuqs, even during the Crusades. According to Kafesoğlu, the military confrontations that arose from the caliphs’ constant attempts to undermine the authority of the sultans whose rule they legitimized had a negative effect on the populace and contributed to the collapse of the Great Seljuq Empire (IA/10: 397). Sümer’s views with regard to Seljuqid succession and the ‘Türkmen problem’ were similar to Köymen’s. According to Sümer, despite adopting many Islamic institutions, Seljuqid succession continued to be treated in terms of family inheritance. This enabled the commanders to usurp their masters’ authority and finally establish their own dynasties by killing them. As for the ‘Türkmen problem’, Sümer was of the opinion that this arose because the Türkmen expected to continue plundering the realms they had helped conquer. Sultan Tuğrul and Sultan Alp-Arslan, on the other hand, wanted their depredations to end. The Türkmen 18 Introduction were angered by this, blaming the Seljuq sultans’ haughtiness on their Iranian viziers and the mamluk commanders who had replaced them. As a result the Türkmen either sided with dynastic members who challenged the right of claimants to the throne or migrated to the western marches of the Great Seljuq Empire (1992: 94, 96–9). Last but not least, Köymen and Kafesoğlu noted that the Great Seljuqs had been patriarchal, aristocratic and, from the outset, intent on establishing their sovereignty. In support they pointed to Tuğrul and Çağrı Beg’s uncle Arslan Yabgu. After his capture by Sultan Mahmud, Arslan Beg is said to have sent a message advising his brothers not to give up their ambition of conquering Khurasan, because a state could not be left to the likes of Sultan Mahmud who was no more than the son of a slave. In a letter said to have been sent by Tuğrul Beg to Caliph al-Qa’im, Tuğrul Beg linked the superiority of Seljuq sovereignty over that of the Ghaznavids to the Seljuqs’ stature (1043) (Köymen 1979: 88–9, 1976: 35; Kafesoğlu IA/10: 386–7). The Turkish historians’ adherence to Akçura’s precepts and Köprülü’s concept of world dominion when defining Seljuq ideology, not to mention their employment of Köprülü’s framework when explaining the Great Seljuqs’ relationship to the Türkmen and their indigenous subjects, makes it necessary to evaluate both Akçura’s and Köprülü’s interpretations. First and foremost, the Khazar kagan is known to have been sacred, and the Uygur and Kırgız rulers are thought likewise to have been sacred. While the Kök Türk kagans appear to have been highly active, there is some evidence of the sanctity of early Kök Türk kagans as well, which may help to explain the Khazar case; the Khazar kagans claimed descent from the Kök Türk imperial clan, the A-shih-na. The Karakhanid rulers also claimed descent from the A-shih-na, but they were not sacred. However, they do not appear to have been elected either. As in the Kök Türk case, kagans came from the same noble house. Consequently, it is difficult to make a case for Akçura’s precept regarding the election of Turkic leaders or for the curtailment of their sovereignty by customary law. On the other hand, Ibn Fadlan’s detailed observations of the Oguz uphold Akçura’s precepts. Despite this, Turkish historians on the Seljuqs have ignored his travelogue, even though the Seljuqs were not the only Oguz lineage to establish an empire. After them came the Ottoman (1299–1923), Akkoyunlu (1378–1508) and Safavid (1501–1736) empires. The problem for them appears to have been that the Oguz did not have an empire at the time the Seljuqs broke away. Consequently, they felt obliged to refer to the Kök Türk and the Karakhanids, and hence to the Orkhun stone inscriptions and Yusuf Has Hajib’s Kutadgu Bilig. Ibn Fadlan was part of an embassy sent in 922 by Caliph al-Muqtadir (908–32) in response to a request by Almış, the (Volga) Bulgar ruler, who had just converted to Islam (Şeşen 1975: 39). His account was discovered by Togan (1923), who then edited and translated it into German for his doctoral thesis (1939). As it precedes Selçuk Beg by a generation, Ibn Fadlan should be regarded as much a contemporary of the Seljuqs as Yusuf Has Hajib and Mahmud of Kashgar. Although Turkish historians do not appear to have been troubled by it, Köprülü’s framework is far more unsatisfactory than Akçura’s precepts. Introduction 19 Pastoralism is practicable from Khurasan to Anatolia, but settlements become more frequent and grow in number and density westward. It is the tensions this created that mark out the Great Seljuq Empire. The socio-economic organization that nomadic pastoralism necessitated on the Eurasian steppes met with resistance from predominantly agricultural populations residing in urban centres, particularly those dependent on irrigation. The institutions developed by the latter lie at the opposite extreme to nomadic pastoralism. In order to rule their indigenous subjects the Seljuqs employed local nobility rather than their own. They did not ally with them. They retained those Türkmen willing to remain their affiliates, dispersing them as their military to regions they could nomadize. Other Türkmen were incorporated as mamluks with their begs as commanders. It is because the Seljuqs’ military continued to reflect all the martial traditions inherent to warrior élites on the Eurasian steppes that they were fatally weakened. These traditions required that a lord earned the fealty of his military by example. More importantly, the Seljuqs’ indigenous civilian administrators resisted military domination. Consequently, although the Great Seljuqs’ model of government may have become the norm throughout Islamdom, it did not stem from the adoption of Irano-Islamic institutions. Rather, it resulted from the impo- sition of Turkic ones on the Islamic world. Turkish historians have ignored not only the Iranian nobility’s resistance but also that of the Abbasid caliphs. While pointing out Sultan Tuğrul’s investiture as ‘Ruler of the East and the West’, saying that it signified a henceforth permanent separation of powers in Islamdom, they ignore Sultan Malik-Shah’s demand that his grandson from the caliph’s marriage to his daughter was appointed heir apparent. Instead of suggesting that Sultan Malik-Shah may have been killed as a result, they have tended to blame his wife, Terken Khatun, going so far as to conjecture that she had Nizam al-Mulk killed first with the intention of gaining sole control. This ignores that the Ottomans finally absorbed the caliphate, as Sultan Malik- Shah seems to have intended. While on this point, it is noteworthy that although Köymen notes that Sultan Tuğrul consulted his wife, Altuncan Khatun, in affairs of state rather than his vizier Kunduri, Turkish historians have ignored the role of Seljuqid women in the Great Seljuq Empire. All the Seljuq women appear to have had their own commanders and diwans, and in some cases were responsible for whole regions. Instead, Turkish historians have claimed the Seljuqs led Sunni resistance against Shicism and then Christianity, marrying their women to the Abbasid caliphs as a mark of respect. At the end of this work the so-called Türkmen problem is discussed as resulting from the conquest of a largely agrarian environment by pastoralists. It is preceded by a discussion of Turkic migrations and followed by an analysis of the conquest of Anatolia. This approach extends itself naturally into a discussion of the basis of the Great Seljuq Empire, which encompasses governance, both military and civil, as well as legitimacy and succession. The power struggle that resulted from Sultan Malik-Shah’s attempt to impose Turkic ideology throughout the empire is discussed after this, as a resolution of tensions arising from the clash of cultures. 20 Introduction The review of Great Seljuq history is completed by separate discussions of common sovereignty and the role of Seljuqid women in state affairs.

Methodology Chapter 2, Overview, signposts events in Great Seljuq history with reference to the above-mentioned problems, in sections that correlate to Chapters 3–7. Chapter 3, Origins, then discusses Oguz, Türkmen and the Seljuqs’ origins after summariz- ing what is known of Eurasian steppes culture. In Chapters 4–7 the Turkish historians’ political histories of the Great Seljuqs are detailed. The narrative includes the Seljuqs in Anatolia, Kirman, and Syria and Palestine where their political histories overlapped with the Great Seljuqs’. Chapter 4, Conquest, starts with Seljuqid attempts to become established first in Transoxania and then Khurasan (983 or 985–1040), and ends with Sultan Tuğrul’s death (1040–63). Chapter 5, Empire, covers the reign of Sultan Alp-Arslan (1063–72) and that of Sultan Malik-Shah (1072–92). Chapter 6, Interregnum, covers the frac- tious reign of Great Sultan Berk-Yaruk (1092–1105). Chapter 7, Dissolution, covers the reigns of Sultan Tapar (1105–17) and Great Sultan Sancar (1121–57), and includes the history of the Seljuqs of Iraq until their demise in 1194. At the beginning of each chapter there is a brief discussion of the sources used and a reminder of the events covered. Köymen was the main historian on the Great Seljuqs. His works cover all but the reigns of Sultan Malik-Shah, Sultan Berk-Yaruk and Sultan Tapar, which nevertheless he discussed in his overview Selçuklu Devri Türk Tarihi (The Seljuqid Period of Turkish History). Kafesoğlu is the main source on Malik-Shah’s reign and Özaydın on Berk-Yaruk and Tapar’s reigns. For the political history of the Seljuqs of Kirman and of Syria and Palestine, Merçil and Sevim respectively are the main sources. Since Sümer remains the authority on the Oguz, his work is referred to throughout. Kafesoğlu’s booklet on the origins of the Seljuqs, his main article in the IA and his book on the Khwarazm-shahs have been used as supplementary sources, as have Turan’s books and articles. Chapter 8, Evaluation, first discusses the primary sources available with regard to Seljuq and Türkmen cultural expectations. It then identifies the modern historical and anthropological studies on nomadic pastoral organization on the Eurasian steppes. Great Seljuq history is discussed with reference to both of these. The Conclusion is presented in Chapter 9. Appendix 1 contains maps relevant to the text and Appendix 2 and 3 respectively chart the Oguz descent lines and the Seljuq rulers. 2 Overview

Turkish historians were not agreed on the periods that made up the history of the Great Seljuqs (1040–1194). Kafesoğlu (IA/10: 353–416) divided their reign into two: rise (1040–92) and fall (1092–1194). Turan (1993a: 236) added a period in between: rise (1040–92), stagnation (1092–1157) and fall (1157–94). In his overview of the Great Seljuqs, Köymen (1963) likewise divided Great Seljuq history into three periods: foundation (985/993–1040), empire (1040–1153) and fall (1153–94). However, in his main body of work he identified five periods: foundation (985/993–1040), first empire (1040–92), interregnum (1092–1114), second empire (1115–53) and dissolution (1153–94) (1954: 1–2, 1979: 24–5*). The Seljuqs did not cease their conquests until the interregnum following Sultan Malik-Shah’s death, but it is difficult to equate Tuğrul and Çağrı Beg’s reigns with empire. Even when Tuğrul Beg had been proclaimed sultan his hegemony was not secure. Besides his step-brother Ibrahim Yınal, there was his cousin Kutalmış who was still in full rebellion at the time of his death. Under Tuğrul and Çağrı Beg the Great Seljuq realms were ruled according to Turkic tradition, which was bi-partite. Other members of the dynasty were subordinated to their seniors in the East and the West. Despite the vizier Nizam al-Mulk’s efforts to put Great Seljuq hegemony on a traditional Irano-Islamic footing, under Sultan Alp-Arslan and Malik-Shah there was the emergence of the Seljuqs of Kirman and of Anatolia. While this does not prevent their reigns from being referred to as ones of empire, given the extent of the realms they ruled directly, after Sultan Malik-Shah Great Seljuq hegemony was once again put on a bi-partite basis. However, unlike under Tuğrul and Çağrı Beg, rather than help each other, Sultan Tapar and, after him, his uterine brother Great Sultan Sancar seem to have gone out of their way to undermine the Seljuqs of Syria and later the Seljuqs of Iraq. Moreover, as Köymen argued, having attempted to usurp power in Syria and Iraq during the interregnum, the commanders continued to struggle with the Seljuqs and each other after the interregnum as well, establishing their own dynasties first in Syria, then in Iraq al-Arab and finally throughout the Seljuq realms in the wake of the Oguz rebellion. Consequently, this work divides Great Seljuq history into four periods: conquest (985/993–1063), empire (1063–92), interregnum (1092–1105) and dissolution 22 Overview (1105–94), which are summarized herein with reference to themes that will be discussed in the penultimate chapter.

Origins Turkish historians refer to the Kök Türk, the Khazar and the Karakhanids in their attempts to demonstrate the Turkishness of Seljuqid rule. Consequently, the relationships between these polities and the Oguz gain considerable importance, particularly as there is still discussion as to whether Selçuk Beg was in the service of the Khazar Hakan-beh or the commander-in-chief of the Oguz Yabgu before he migrated to Jand, though there is no doubt he was of the Kınık. Situated between the Black and the Caspian seas, the Khazar kaganate (630–965) was established during the initial collapse of the Kök Türk Empire (552–740). The Oguz and the Karakhanids, on the other hand, emerged after the collapse of the Uygur Empire (740–840). Under the Kök Türk, the Uygurs were a member of the Tokuz Oguz. When their successor state collapsed, they migrated to north-western China where they founded city-states along the Silk Route. This is probably why the Khazar knew of them, through their trade caravans. As for the Tokuz Oguz, although archaeological evidence suggests some of them migrated westward, none of their lineages reappear among the Oguz situated between the Caspian and Aral seas. Consequently, it is not possible to compare known Uygur characteristics with those of the Oguz and hence the Great Seljuqs. There are also fundamental differences between the Khazar and the Karakhanids, the eastern neighbours of the Oguz. Although like the Karakhanids the Khazar kagan claimed descent from the Kök Türk imperial dynasty, he was considered sacred and governed through a regent. The Karakhanid kagans, on the other hand, remained politically and martially active, frequently at odds with each other over seniority, similar to the Kök Türk. While the Oguz were divided into two branches like the Kök Türk and the Karakhanids, their yabgu or ruler, who does not always seem to have come from the senior Boz-Ok branch, is thought to have been chosen, though there is some evidence the position may have become hereditary. It is, therefore, just as difficult to compare Seljuq characteristics with the Karakhanids. The question of Seljuq expectations is complicated further by the identity of the Türkmen. Recently it has been argued that the name applies to a much older tribal affiliation in present-day Turkmenistan. While this does not affect the consensus among Turkish historians that those Türkmen affiliated to the Seljuqs were mainly an amalgam of Oguz, Karluk and Kıpçak tribesmen who had converted to Islam, it does raise further questions with regard to the origins of Seljuq cultural expectations.

Conquest (985/993–1063) It is not known when Selçuk Beg migrated to Jand. However, unlike Yengi-Kent, where the Oguz yabgu wintered, and Huvara, both of which are thought to have Overview 23 been founded by Muslim colonists, Jand appears to have been much older. Also, it was situated on the marches between Üst-Yurt and Transoxania rather than within Oguz territory. Considered to be at the forefront of jihad, the war against non-believers, Transoxania had numerous fortifications built by local notables for gazis, warriors against infidels from their locality. This can be said to have been crucial to the Seljuqs’ fortunes. Claiming Muslims should not have to pay tax to the non-believer, Selçuk Beg is said to have rallied the gazis in the environs of Jand against the Oguz yabgu’s tax collector. Possibly as a direct result, sometime between 985 and 993, the Seljuqs are thought to have gained additional pasturage near the village of Nur, situated between Bokhara and Samarkand, which Selçuk Beg’s son Arslan appears to have taken up. In return, Selçuk Beg is thought to have undertaken to defend the Samanids’ northern marches. However, this could also indicate that Selçuk Beg had died and, having been ousted from Jand, the Seljuqs were forced to seek fresh pastures, or that the Seljuqs had grown in number. Whatever the cause, their presence in Jand and later Nur strongly suggests they may have become familiar with Irano-Islamic statecraft well before they arrived in Khurasan half a century later. Encouraged by dynastic struggles at the time, the Samanid realms were being dismembered by their Turkic slave soldiery, mamluks from their palace corps. One such commander, Sebük Tekin, had become established in Ghazna (977). Nevertheless, he seems to have remained loyal until his death (997). The governor of Khurasan, Ali Simcur, on the other hand, is said to have made a secret pact with the western Karakhanids who entered Transoxania and took the capital, Bokhara (992). However, when their leader fell ill they were forced to withdraw. Called upon to help, Sebük Tekin is said to have defeated Ali Simcur. Undeterred, Simcur seems to have allied with another commander, Faik, who then encouraged the eastern Karakhanids to enter Transoxania. On this occasion Sebük Tekin could only broker an agreement that left Bokhara to the Samanids (996). When he and the Samanid ruler died the following year, the eastern Karakhanids eventually re-entered Bokhara (999), taking the treasury and remaining dynasts to their capital Özkent. Having escaped, the Samanid Muntasir then embarked on a desperate struggle to regain his realms. It is not clear whether he met with the Oguz yabgu in Yengi-Kent or the Seljuqs’ Arslan, but Turkish historians have argued the latter on grounds that it must have been the Seljuqs who harassed and plundered the western Karakhanids when they withdrew from Bokhara in 992. Mention of an Oguz contingent under the command of a Hasan Tak, which defected to the Karakhanids during the final battle, sealing Muntasir’s fate (1004/5), puts this in some doubt. With the exception of Sümer, Turkish historians also argued that Selçuk Beg must have died soon after and the Seljuqs were forced out of Jand, possibly by the Oguz yabgu’s son Shah-Malik. These historians went on to argue that having failed to gain the support of the western Karakhanids, Çağrı Beg now left for eastern Anatolia, returning a few years later, both wiser and richer. What is certain is that little is known about the Seljuqs until they emerge as the allies of the 24 Overview western Karakhanid Ali Tekin in Bokhara (1020–1), a decade and a half after Muntasir’s demise. Having married Arslan Beg’s daughter, Ali Tekin seems to have backed the eastern Karakhanids’ claim to seniority rather than his elder brother Yusuf’s in order to remain in Bokhara and Samarkand. Unable to assert himself, Yusuf is said to have turned to the Ghaznavids for help. Sultan Mahmud, Sebük Tekin’s son, may have been more than ready to support him as it is thought Ali Tekin had led one of the eastern Karakhanid columns into Khurasan during a failed attempt to wrest the province from the Ghaznavids (1005/6). Sultan Mahmud and Yusuf’s meeting outside Samarkand (1025) resulted in the latter’s son being installed in Bokhara and the former tricking Arslan Beg into captivity. Imprisoned in India, Arslan Beg died in 1032. As for Ali Tekin, although Sultan Mahmud is said to have taken his wife and children prisoner, he does not appear to have been pursued. He returned to Bokhara and with the help of the Seljuqs ruled Transoxania until his death (1034). During this period the Seljuqs appear to have been led by the youngest of Selçuk Beg’s sons, Yusuf Yınal. During his leadership (1025–34), Çağrı and Tuğrul Beg and their uncle Musa may have wintered in Khwarazm. This does not seem to have prevented them from siding with Ali Tekin against its Ghaznavid governor. Although he was fatally wounded in a battle against them (1032), the governor’s son, the Khwarazmshah Harun, allied with Ali Tekin and the Seljuqs in an attempt to gain independence from the Ghaznavids. However, Ali Tekin died soon after and most likely Yusuf Yınal was killed by Ali Tekin’s commander-in- chief and regent. It is thought he may have been a Baran and possibly the last known Oguz yabgu’s son, Shah-Malik’s brother. Driven out of Transoxania, the Seljuqs are said to have been attacked and badly depleted by Shah-Malik en route to Khwarazm. Soon after Shah-Malik’s dawn raid, the Khwarazmshah Harun was assassinated on the orders of Mahmud’s son, Sultan Mascud. Having lost both their allies in quick succession and possibly under threat from the Oguz yabgu’s descendants, the Seljuqs crossed into Khurasan (1035), requesting pasturage from Sultan Mascud where his father Mahmud had tried to settle Arslan Beg’s affiliates, the so-called Türkmen of Iraq. Consisting of 4,000 households, they are said to have refused to affiliate with Çağrı and Tuğrul Beg after Arslan Beg’s imprisonment, citing cruelty. The number of households attributed show the Seljuqs were organized decimally, as known since the Hsiung-nu. When the Seljuqs arrived in Khurasan, the Türkmen of Iraq were in full-scale revolt. On Arslan Beg’s incarceration, in return for pasturage these Türkmen had offered to defend north-eastern Khurasan for Sultan Mahmud. Situated along the Atrek River, between the Caspian Sea and the Köpetdag Mountains, they are thought to have been joined by others from between the Caspian and Aral seas. Possibly due to overcrowding but more likely because the Ghaznavid governor of Khurasan had not wanted them in the first place, relations between the Türkmen and Ghaznavids seem to have deteriorated steadily so that Sultan Mahmud finally drove them out (1028). When he died soon after however, his son Mascud invited the Türkmen of Iraq’s Yağmur Beg back to gain the Ghaznavid throne from his brother. Having Overview 25 succeeded, Yağmur Beg persuaded him to allow back the remaining Türkmen of Iraq. Although said to have served Sultan Mascud well, they seem to have continued to disrupt sedentary life along the Atrek River, where they had been resituated. As a result Sultan Mascud appears to have ordered their commander while on campaign to press Yağmur Beg and his affiliates into permanent military service. When Yağmur Beg resisted and was executed (1033), the Türkmen of Iraq rebelled. This is thought to have encouraged yet more Türkmen to cross into Khurasan (1034), making it likely the Seljuqs were aware of the situation. Musa Yabgu and Çağrı and Tuğrul Beg each sent letters to the governor of Khurasan, citing the death of Yusuf, the dawn raid by Shah-Malik and the assassination of the Khwarazmshah Harun for their presence in Khurasan. In return for pasturage they offered to rid the province of the Türkmen, as well as protect it from further incursions from between the Caspian and Aral seas. Sultan Mascud is said to have refused. Having defeated the army sent against them, the Seljuqs sent an envoy. After protracted negotiations the Ghaznavids agreed to give each the title of dihkan or village elder, assigning the province of Dihistan to Çağrı Beg, the settlement of Nesa to Musa Yabgu and that of Ferave to Tuğrul Beg (27 August 1035). Contrary to their agreement, the Seljuqs are thought to have rallied the Türkmen of Iraq and those from Khwarazm to their banner, attracting still others from Balkhan on the Caspian Sea. A year later the Seljuqs requested three urban centres; Marv, Serakhs and Baverd. In their letter the Seljuqs now cited their increased number but refused responsibility for the depredations the Türkmen caused, though once again they promised to clear Khurasan and offered to support Sultan Mascud in his campaigns in Iraq al-Ajam. Interestingly, they also requested Sultan Mascud to appoint jurists and tax collectors (12 November 1036). What ensued can only be described as steppe warfare: an endless series of skirmishes punctuated by the occasional test of strength, which totally wore out the Ghaznavids. At first victorious, the Seljuqs appear to have been quick to divide up northern Khurasan at a kengeş or council, choosing Tuğrul Beg as leader. Thought to have been emboldened by the governor’s cruel and corrupt rule of Khurasan, Tuğrul Beg entered the capital Nishapur (1038). His acceptance by the city elders seems to have prompted a number of the Ghaznavids’ Turkic mamluk commanders to desert with their soldiery. Forced to return from India and take charge (7 October 1038), Sultan Mascud appears to have had some initial success (7 April and 13–22 June 1039), deciding to retrain his soldiery for desert warfare (1039/40). While he was thus occupied, the Seljuqs appear to have had time to help a branch of the Karakhanids defeat Ali Tekin’s sons and strengthen their ties to the Khwarazmshahs. The Oguz migration into Khurasan also seems to have intensified, possibly strengthening the Seljuqs before their final victory over Sultan Mascud at Dandankan the following spring (23 May 1040). Having sent a letter to the caliph requesting legitimization of their rule, the Seljuqs are said to have held another council at which they reapportioned Khurasan with a view to expanding their realm. Sistan and Kirman to the south (1040), Khwarazm and Iraq al-Ajam to the north and west (1042), Transoxania 26 Overview to the east (1047) and Azarbayjan to the north-west (1048) were incorporated directly or became vassals in the following decade. As for the Türkmen of Iraq, although before his death Kızıl Beg affiliated with Tuğrul Beg, giving his daughter in marriage, the remainder appear to have moved away to northern Mesopotamia during the Seljuqs’ westward expansion. When the Abbasid caliph was prompted to complain to Tuğrul Beg they moved further away still, into Anatolia. However, in the process most of them seem to have perished (1041–8). Not so the other Oguz; these were led into Anatolia first by Kutalmış (1045), and then by Yakuti and Musa’s son Hasan (1047). When Hasan lost his life, Tuğrul Beg appears to have assigned Ibrahim Yınal, at the time governor of Azarbayjan. Ibrahim Yınal is said to have defeated a Byzantine force near Erzurum, deep in eastern Anatolia, after which Seljuqid involvement seems to have ceased (1048). Six years later Tuğrul Beg appears to have taken charge personally and with Armenian help attempted to take the fortress of Malazgirt (Manzikert). Having failed, the remainder of his campaign is thought to have been cut short by the caliph’s invitation (1054). At the time of the battle of Dandankan (1040), Iraq al-Ajam (central and western Iran), Iraq al-Arab (Mesopotamia), northern Syria and south-eastern Anatolia were ruled by various Islamic dynasties. The most prominent of these was the Twelver Shicite Buyid state in Iraq al-Ajam. Although they continued to have the khutba, the sermon delivered at the noon congregation prayer on Friday, read in his name, Caliph al-Qa’im is considered to have been their prisoner. Consequently, the initial exchange of envoys between the caliph and the Seljuqs are thought to have been at the Buyids’ instigation, leading to an exchange of daughters in marriage between the Seljuqs and the Buyids, and recognition of the Seljuqs’ sovereignty in Khurasan by the caliph (1046). However, the Buyid ruler’s death the following year appears to have sparked internecine warfare, prompting their commander in Baghdad, the Shicite al-Basasiri, to strike out on his own. Within four years the khutba in Lower Mesopotamia was being read in the Fatimid caliph’s name (1050). This is thought to have brought the Abbasid caliph closer to the Seljuqs. When, having received an envoy from Tuğrul Beg, Caliph al-Qa’im declared the Fatimids heretics (1052/3), al-Basasiri is said to have accused him of secretly negotiating with the Seljuqs and imprisoned him with his vizier (1054). Nevertheless, Turan argued that the caliph’s invitation came only after Tuğrul Beg had promised to open the hajj or pilgrimage routes – on the condition he was proclaimed sultan. There can be no doubt that securing the hajj routes to Mecca and Medina would have necessitated the Seljuqs to declare war on the Fatimids. However, while this may explain the large army with which Tuğrul Beg is said to have come to Baghdad (1055), as Köymen argued, once there Tuğrul Beg’s actions show that he was more concerned with using the caliph’s standing to legitimize Seljuqid sovereignty. Blaming the widespread riots his troops caused in Baghdad on the Buyids, Tuğrul Beg seems to have used the events to trick the Buyid ruler into captivity Overview 27 and incorporate his realms. Once the riots had been quelled, Tuğrul Beg does not appear to have deployed his forces to open the hajj routes as would be expected if Turan’s interpretation of events was correct. Instead he assigned Ibrahim Yınal to deal with al-Basasiri after the latter had defeated Kutalmış (1056). When the caliph crowned Tuğrul Beg and in accordance with Turkic tradition proclaimed him Ruler of the East and the West (January 1058), al-Basasiri was still in north Syria. As a result, while Sultan Tuğrul dealt with the now rebellious Ibrahim Yınal, al-Basas- iri was able to take the caliph prisoner. Having defeated Ibrahim Yınal however, Sultan Tuğrul is said to have first secured the caliph’s wife and Çağrı Beg’s daughter Arslan Khatun’s release. When, on finally having been freed, the caliph asked for his wife, Sultan Tuğrul countered by asking to marry the caliph’s daughter first. Despite fatwas or legal pronouncements against the marriage, the caliph appears to have succumbed when Sultan Tuğrul confiscated his estates (1062). As for Ibrahim Yınal’s rebellion, after his victory over the Byzantines Ibrahim Yınal is considered to have become the most renowned Seljuqid outside Khurasan (1048). Earlier, when he had conquered Rayy, Tuğrul Beg had taken the city from him and made it his capital. When now he asked also for Hamadan and the fortifications between it and Rayy, Ibrahim Yınal refused and had his vizier tortured, holding him responsible for Tuğrul Beg’s request. Although defeated by Tuğrul Beg, Ibrahim Yınal appears to have been pardoned by him (1050). After Kutalmış’ defeat by al-Basasiri, Tuğrul Beg assigned him Mosul, asking him to deal with al-Basasiri (1056). Insulted by such a small appenage, Ibrahim Yınal is said to have used al-Basasiri to request money, titles, robes of honour and a banner from the Fatimid caliph, which appear to have been granted (August 1058). Although Ibrahim Yınal defeated Sultan Tuğrul several times, he lost the final battle outside Rayy (July 1059). This time Sultan Tuğrul had Ibrahim Yınal strangled with his own bowstring, in keeping with Turkic tradition. After Ibrahim Yınal, those Türkmen who had supported him joined Kutalmış and his brother in Girdkuh, near Damghan; Kutalmış appears to have taken refuge there after having rebelled separately. From here he and his brother repulsed Sultan Tuğrul’s forces with the help of the Türkmen (April–May and August 1061), mounting a bid for the throne on the sultan’s death (September 1063). These rebellions are rightly considered to be of some importance by Turkish historians; first, because Çağrı and Tuğrul Beg do not appear to have been willing to share sovereignty as customary and, second, because on rebelling the second time Ibrahim Yınal is said to have promised the Türkmen in Hamadan he would never make peace with the sultan, or force them to campaign in Iraq al-Arab, or appoint a vizier without consulting them, raising the fundamental question of what Seljuq, Iranian and Türkmen expectations were.

Empire (1063–92) In accordance with custom, on Çağrı Beg’s death Sultan Tuğrul had married his brother’s wife and since he was childless appointed her youngest son Süleyman heir apparent. In Köymen’s opinion this was his vizier al-Kunduri’s idea. Because 28 Overview Süleyman was underage, Köymen thought al-Kunduri may have hoped to become de facto ruler of the empire. This assumes Sultan Tuğrul was the sole ruler of the Great Seljuq Empire, an assumption contradicted by al-Kunduri’s letter to Alp- Arslan, which Köymen cites as evidence. In it al-Kunduri advises Alp-Arslan that by having the khutba read in Süleyman’s name he was fulfilling Sultan Tuğrul’s will and that Alp-Arslan should suffice with his father’s inheritance. It is highly unlikely the vizier was unaware that Alp-Arslan had moved to incorporate the sultan’s realms earlier on rumours of his death. Whether or not Alp-Arslan left Khurasan before Kutalmış had defeated al-Kunduri’s advance guard, Kutalmış’ approach appears to have changed the vizier’s mind. He sent an envoy, proclaiming his allegiance to Alp-Arslan and asking his help (16 November 1063). While Kutalmış and Alp-Arslan moved to claim Sultan Tuğrul’s realms, Çağrı Beg’s eldest son Kara-Arslan Kavurt is thought to have advanced from Kirman to Isfahan, where he seems to have waited. When Alp-Arslan defeated Kutalmış, who died in flight (7 December 1063), before retiring Kavurt is said to have had the khutba in Shiraz read in Süleyman’s name (25 December 1063). In view of their later confrontations, Kavurt’s decision to wait and see rather than help his brother seems to indicate he may also have moved to claim the childless sultan’s realms. As for Caliph al-Qa’im, having replaced the Seljuqs’ officials with his own and demanded they return their allocations of land or iqtacs, he is said to have invited the leaders of the local Arab and Kurdish dynasties to Baghdad to discuss the region’s future. On taking up residence in the Seljuq palace however, the Uqailid leader seems to have proceeded to plunder the caliph’s palace, leaving Baghdad only after news of Alp-Arslan’s victory. After receiving his envoy (6 March 1064), the caliph had the khutba read in Sultan Alp-Arslan’s name (11 May 1064), although the sultan never came to Baghdad. Having added Sultan Tuğrul’s realms to his, Sultan Alp-Arslan embarked on a campaign into the Caucasus and north-eastern Anatolia. This appears to have been aimed at clearing a path for the Türkmen situated on the Mukan steppe in present-day Azarbayjan. In order to raid north-eastern Anatolia, the Türkmen had to pass through the Armenian and Georgian kingdoms in the region, which made them doubly vulnerable when returning from Anatolia with loot. Having subdued the Bagragat dynasty in Georgia and cleared the Armenian fortifications along the Aras River, Sultan Alp-Arslan gained the allegiance of the Armenian Kingdom of Lori before attacking the Armenian Kingdom of Ani, south of present-day Kars, torching the city (August 1064). The Chinese graded nomadic pastoral raids as follows: looting border areas, robbing border areas, border encroachment, invasion, a great invasion, deep invasion and large-scale invasion. The key to escalation from the looting and robbing of border areas and border encroachments to an invasion of Anatolia appears to have been Ahlat. Situated on the western shores of Lake Van, the pastures in the surrounding countryside are comparatively lush, affording permanent settlement by pastoralists. It is likely that Ahlat was taken either during Sultan Alp-Arslan’s campaign or soon after, when possibly the fortification of Malazgirt also fell to Overview 29 the Seljuqs. With an unchallenged base in the heart of eastern Anatolia, it is not surprising that both the number and intensity of Türkmen raids increased steadily. The commander that Sultan Alp-Arslan is thought to have appointed to coordinate the raids was initially supported by the Türkmen Bekçioğlu Afşin Beg. Together they appear to have taken the fortifications between the Murat and Tigris rivers, enabling safe return from central Anatolia (1066). The following year the Türkmen sacked Kayseri (Cæsarea) and pillaged the environs of Konya (Iconium) (1067), signalling an invasion. Afşin Beg wintered in Aleppo not Ahlat. The reason for this appears to be that he had killed the sultan’s commander for killing his brother. In spring Afşin Beg besieged Antakya (Antioch), but withdrew on hearing of the new Byzantine emperor’s approach (April 1068). Romanos Diogenes’ advance on Aleppo seems to have surprised a couple of Türkmen groups deep in Anatolia. One had sacked Niksar, north-east of Kayseri, and another had been raiding along the Euphrates. At the time Aleppo still belonged to the Uqailids. On being besieged they paid off the emperor, but when Romanos Diogenes proceeded deeper into north Syria they joined forces with the Türkmen there and drove him off. The emperor’s campaign appears to have been a total failure; while Romanos Diogenes was thus occupied, Afşin Beg had raided as far west as the Sakarya River, destroying the city of Amorion. In spring Romanos Diogenes tried to retake the fortifications between the Murat and Tigris rivers and from there advance on Ahlat. On receiving news that Afşin Beg had attacked Malatya (Melitene) he turned back, but Afşin Beg withdrew into central Anatolia where he sacked Konya (1069). The following year the emperor stayed in Istanbul (Constantinople), assigning Manuel to the defence of Anatolia (1070). Instead of attempting to take the fight to the Türkmen, the Byzantines appear to have tried to counter their raids from a central base, Kayseri. When this also proved unsuccessful, Emperor Diogenes seems to have decided to take the fight to the Seljuqs. Having arrived at the shores of Lake Van from the north and taken back the fortress of Malazgirt, he was met by Sultan Alp-Arslan and, having been defeated, was captured (25 August 1071). In their effort to prevent an invasion, the Byzantines can be said to have precipitated deep invasions that, after their ignominious defeat at Malazgirt, turned into a large-scale invasion which proved final. Sultan Alp-Arslan’s apparent reluctance to reap the full reward of his victory at Malazgirt – he released Romanos Diogenes – does not weaken the Turkish historians’ argument that the Seljuqs wanted Anatolia. Sultan Tuğrul’s involvement in Anatolia was cut short by Caliph al-Qa’im and his ensuing struggles with al-Basasiri and then Ibrahim Yınal and Kutalmış. After his first campaign into the Caucasus, Sultan Alp-Arslan likewise had to turn his attention elsewhere. When the Kırgız ended the Uygur Empire (740–840), they did not establish a successor state. Within a century, the Mongol Khitay had ousted the Kırgız from present-day Mongolia (924), establishing the Liao dynasty (936–1122). However, the Liao were unable to expand south because the Chinese Sung dynasty was established soon after (960), becoming the dominant force in northern China (1004). Consequently the Liao appear to have attempted to expand westward, but failed to take the Issyk-Köl region north-west of the T’ien-shan Mountains 30 Overview or Kashgar immediately south-east at the westernmost point of the Tarim Basin (1017). Nevertheless, their attempt is thought to have displaced the remaining Kimek (Kuman/Kıpçak) westward along the steppes north of Transoxania and the Aral and Caspian seas. This must have impacted the Oguz because immediately after opening the routes into Anatolia (1064), Sultan Alp-Arslan is said to have campaigned in Üst-Yurt, between the Aral and Caspian seas. He appears to have been concerned with Kıpçak incursions into Khurasan (1065), returning to the Caucasus the following year to stop the Alan, Komuk, Sabir and other remnants of the Khazar from entering Azarbayjan (1066). In the meantime Kavurt had dropped Sultan Alp-Arslan’s name from the khutba. Kavurt had been trying to take Fars from the Shebankarids at the time of Sultan Tuğrul’s death. When the Shebankarids requested his help, Sultan Alp-Arslan appears to have accepted, giving them soldiery. Having been ordered to return to their realms, Kavurt seems to have persuaded the Shebankarids to join him instead. When Sultan Alp-Arslan finally advanced on Kavurt, he was forced to withdraw because of suspicion that some of his commanders had proclaimed allegiance to Kavurt (1069). Kavurt’s son Sultan-Shah apparently attempted to benefit and with Sultan Alp-Arslan’s support tried to depose his father, but lost (1070). Sultan Alp-Arslan was killed while on campaign in Transoxania (November 1072). Although Malik-Shah had been appointed heir apparent by his father Sultan Alp-Arslan (1066), on grounds that he was the eldest Seljuq Kavurt challenged Malik-Shah. Kavurt is thought to have hoped to gain the support of the Türkmen between Hamadan and Rayy, but Sultan Alp-Arslan’s vizier and Malik-Shah’s atabeg or guardian, Nizam al-Mulk, is said to have bought them off. Although the Türkmen do not appear to have sided with him, Malik-Shah won the day with the aid of the Seljuqs’ vassals, despite his father’s mamluk commanders who are said not only to have refused to fight, begging Kavurt’s mercy on the battlefield, but also pillaged the vassals’ camps afterwards. Their unruliness seems to have escalated when Kavurt was captured and imprisoned, threatening to turn into open rebellion against Sultan Malik-Shah. As a result Nizam al-Mulk is thought to have had Kavurt strangled with his own bowstring and his sons blinded in Malik- Shah’s presence (1073). The Great Seljuq Empire is considered to have reached its highest point under Sultan Malik-Shah, recovering from decades of warfare and migrations. Nevertheless, even if confined to the marches, warfare was not absent. Although Sultan Alp-Arslan had strengthened ties with the Karakhanids by marrying Malik-Shah to Terken Khatun, on his death the Karakhanid ruler of Samarkand took Tirmiz and Balkh. Concurrently the Ghaznavids attacked Çiğil-Kent and captured Malik-Shah’s uncle Osman (1073). Once Kavurt had been dealt with, unlike his father, Malik-Shah appears to have waited for news of the caliph’s endorsement before turning his attention to the Karakhanids and Ghaznavids, successfully dealing with both (1074–5). In the Caucasus, King Giorgi II of Georgia (1072–89) seems to have tried to benefit from Sultan Alp-Arslan’s death by expanding south. Having defeated him, Sultan Malik-Shah took Ganja (Gyandzha) from the Shaddadids and assigned it Overview 31 to his brother Tutuş (1076). However, after his departure the Georgians appear to have defeated the sultan’s commander Sav-Tekin several times, encouraging the former king of Ani to try and re-establish his kingdom. The struggle in the Caucasus lasted more than five years until, having been comprehensively beaten, Giorgi II was forced to travel to Isfahan and proclaim allegiance (1080–1). Although initially not instigated by him, the main territorial expansion during Sultan Malik-Shah’s reign was in Syria. Having entered the region as early as 1063, the Türkmen are thought to have been joined by others who had supported Kavurt against Sultan Alp-Arslan (1068–70). The Türkmen initially became established near Ramla in Palestine, from where they took Jerusalem and then Acre (1071, 1074). Although they besieged Damascus at regular intervals (1071, 1073, 1075), success did not come until Uvakoğlu Atsız took over. He took Tyre and Tripoli before gaining Damascus as well (1076). Next Atsız advanced on the Fatimids but seems to have hesitated outside Cairo, demanding money instead. Having given the Fatimids the chance to marshal their forces, he was badly beaten (1077). Thinking that Atsız had died, Sultan Malik-Shah is said to have assigned Syria to his brother Tutuş. On entering Damascus, Tutuş had Atsız strangled with his own bowstring (1078/9). After Malazgirt the Byzantines appear to have lost control in the Balkans as well. The Bulgar uprising was followed by the Nestor rebellions (1074, 1075), causing famine (1076). This encouraged ambitious generals to revolt in the hope of becoming emperor (1077). Having become established in Anatolia after their release upon Sultan Alp-Arslan’s death, Kutalmış’ sons Mansur and Süleyman- Shah appear to have accompanied one such general, Nicephorus Melissenos, who is said to have toured the walled cities and fortifications in central and western Anatolia in search of support, allowing the Türkmen to gain control. Mansur and Süleyman-Shah are thought to have entered Iznik (Nicea) in this manner, making it their capital. At this point the brothers appear to have switched their support to Botaniates, the Byzantines’ Anatolian commander. With help from the Türkmen Botaniates was able to defeat the main contender, Nicephorus Briennios of Thrace, and become emperor (3 April 1078). However, encouraged by what seems to have been a falling out between Mansur and Süleyman-Shah, Botaniates appears to have turned on the victor, Süleyman-Shah. It is not clear if Süleyman-Shah was helped by Sultan Malik-Shah who may have sent his commander Porsuk against Mansur. Whatever the case, the Byzantine attempt to regain Iznik failed and as a result Botaniates lost his crown to Alexius I Comnenus (1080). For a time, Süleyman-Shah’s domains seem to have stretched from the Bosporus to the Toros Mountains, but while he was busy with the Armenian principalities that had emerged in the wake of the Byzantine defeat at Malazgirt, Comnenus regained the western shores of the Bosporus, forcing Süleyman-Shah to make peace. The new boundary between the emergent Seljuqs of Anatolia and what remained of the once-mighty Byzantine Empire is thought to have been the stream of Dragos (Drakon), between Istanbul and Iznik (1082). Free to expand in south-eastern Anatolia, Süleyman-Shah appears to have entered Antakya on the invitation of its residents and consolidated his hold by 32 Overview taking the surrounding towns and fortifications (1085). However, this brought him into conflict with the Uqailids in Mosul who, having lost Aleppo, had recently retaken it as well as Harran (1080, 1081/2). Harran protected Aleppo from south-eastern Anatolia where, at the time, Sultan Malik-Shah’s Türkmen commander Artuk Beg was incorporating the Marwanids’ realms. Diyarbakır and Silvan (Mayyafariqin) fell to the Seljuqs in 1084 and 1085 respectively, while Sultan Malik-Shah was occupied in Balkh with his rebellious brother Tekiş (1084). When the Uqailid Muslim demanded Süleyman-Shah pay tax for Antakya, he was reminded of his vassal status. Muslim advanced but was defeated by Süleyman-Shah and died (23 June 1085). Immediately after this Süleyman-Shah besieged Aleppo but was unable to take it. He returned the following year (April 1086), but had to lift the siege on learning that Tutuş was advancing with Artuk Beg. They defeated Süleyman-Shah who died (5 June 1086). Having forgiven Tekiş, Sultan Malik-Shah came west and took Mosul, Harran, Aleppo, Antakya and Urfa in that order, appointing his commanders as governors general. Many of these were Türkmen begs rather than mamluks. Sultan Malik-Shah then left for Baghdad to marry his daughter Mah Melek to the caliph (13 March 1087). After a brief campaign against the Karakhanids in Transoxania (1088–9), Sultan Malik-Shah is said to have been getting ready to take the fight to the resurgent Fatimids who had gained much of Syria and Palestine from Tutuş. Possibly in anticipation of the demise of the Fatimid caliphate, Sultan Malik- Shah demanded that Caliph al-Muqtadi (1075–94) appoint his son Jacfar (b. 1088) from Mah Melek Khatun heir apparent. The caliph refused, sending Mah Melek Khatun home with her son Jacfar (1089). Sultan Malik-Shah came to Baghdad and ordered the caliph to leave (21 October 1092). During the ten days Caliph al- Muqtadi was granted to gather his belongings, the sultan fell ill while on a hunting trip and died (20 November 1092). Thought to have been poisoned, his death came hard on the heels of his vizier Nizam al-Mulk’s assassination by a batini, a member of the Shicite Nizari Ismacili sect (15 October 1092). Nizam al-Mulk had been in the Ghaznavids’ service before his appointment by Çağrı Beg as Alp-Arslan’s vizier. Not surprisingly, his policies are considered to be traceable to institutions prevalent not only under the Ghaznavids but their founder’s masters, the Samanids. In his famous treatise, the Siyasat-Nama, wherein Nizam al-Mulk advised the Seljuqs on how to govern their empire, he clearly attempts to avoid the pitfalls he thought inherent to the Ghaznavids’ relationship with their mamluk soldiery (Köymen 1999). Under the Buyids, mamluk commanders’ iqtacs or allocations of land allowed them to keep due taxes to fully train and equip a stipulated number of soldiery. Although the Buyids assigned family members to oversee such allocations, the system appears to have failed. After extorting what they could, commanders are said to have returned their allocations and demanded new ones. This is thought to have been largely due to their preference for residing in the capital so as to keep in touch with palace politics. By comparison, the Samanids seem to have trained their mamluk soldiery as a palace corps, paying them directly from the Treasury Overview 33 until they could be trusted with governorships. The Ghaznavids also paid their military directly. However, despite campaigns to India that provided their commanders and soldiery alike with plunder, the Ghaznavid sultans’ dependence on their soldiery for power appears to have encouraged their civilian governors to extort extraordinary taxes on their behalf. In order to prevent this, Nizam al-Mulk is thought to have attempted to establish a system whereby, as under the Samanids, a minimal élite force was paid from the Treasury, allocating lands commensurate with their financial responsibilities to the empire’s other military and administrative staff. Moreover, unlike under the Buyids, it seems he intended that military iqtacs be distributed throughout the Great Seljuq realms so that commanders could provide for their troops when on campaign there. It is not clear from what the Turkish historians have written however, if taxes were to be collected by administrative staff answerable to the diwan or council of ministers and then paid to the commanders. Whatever Nizam al-Mulk’s intention, with the exception of administrators who resided in the proximity of their allocations, under the Seljuqs commanders not only resided on their iqtacs but also passed them on to their heirs. Turan in particular argued that because commanders resided on their iqtacs and were able to inherit them, they had developed an interest in the land’s improvement and the well-being of its population. While this may have been so, the downside was that when central authority weakened, iqtac holders became established at the Seljuqs’ expense. Sultans were reduced to forcibly collecting back taxes from defiant commanders and administrators, mostly the latter. Commanders also refused to fight each other or go on campaigns when it did not suit their ambitions, which centred on establishing their own dynasties. It seems that in a further attempt to avoid the pitfalls he had witnessed under the Ghaznavids, Nizam al-Mulk also tried to dilute the pre-eminence of Turkic mamluk soldiery with recruits from other ethnic origins, but appears to have failed in this, too. When his Turkic commanders objected, Sultan Malik-Shah is said to have not only ousted the Armenian mamluks that appear to have been recruited under Sultan Alp-Arslan but also co-opted Türkmen begs with their affiliates. Nizam al-Mulk’s policy of establishing colleges of jurisprudence or madrasahs was by far his greatest success. Based on earlier examples in Khurasan, under him rent from the commercial buildings built around the madrasahs ensured not only their maintenance but also that appointed jurists were salaried and their students provided for. According to Turan, the colleges were primarily intended to counter Fatimid propaganda. There can be no doubt that the madrasahs provided a higher quality of administrative staff, which would have improved governance throughout the Great Seljuq Empire. While the Siyasat-Nama and the madrasahs explain the Turkish historians’ laudatory opinion of Nizam al-Mulk, it must not be forgotten that Sultan Malik- Shah had tried to groom at least one successor whom Nizam al-Mulk is said to have killed. The reasons for the young sultan’s disaffection seem to have stemmed from the power Nizam al-Mulk had come to think rightfully his. 34 Overview First, there was the size of his retinue, said to have numbered thousands and to include mamluk soldiery. Asked by Sultan Malik-Shah why he needed such a large force Nizam al-Mulk is said to have replied they were there to serve the sultan. Then there was the vizier’s nepotism. When one of his numerous sons appointed by him to high office took advantage of his position as governor and publicly humiliated the sultan’s appointed commander, the commander complained to the sultan. Apparently angered, Sultan Malik-Shah sent his wife’s vizier Taj al-Mulk to confront Nizam al-Mulk. Asked if he considered himself a partner to the sultan’s sovereignty and dominion, Nizam al-Mulk is said to have replied to Taj al-Mulk there would not be an empire without him. Nizam al-Mulk’s arrogance appears to have stemmed equally from his popularity with the local populaces. There can be little doubt that his championing of Irano-Islamic institutions must have been particularly welcome given the Seljuqs’ origins. They and their Turkic affiliates must have appeared undeniably foreign. Instead, Turkish historians have concentrated on the Seljuqs’ stated willingness from the start to rule in accordance with Irano-Islamic institutions. Welcomed to Nishapur by its notables, Tuğrul Beg is said to have told the senior jurist not to withhold his advice as he was a stranger to their ways (1038). By Sultan Malik- Shah’s reign however, the Seljuqs appear to have decided they would do better if both sultan and caliph were from the same dynasty. This was finally achieved under the Ottomans who, after Yavuz Sultan Selim (1512–20) had conquered Egypt, took the title of caliph and subordinated it to that of sultan.

Interregnum (1092–1105) Sultan Malik-Shah’s wife Terken Khatun is said to have kept her husband’s death secret while she negotiated with the caliph, offering to return his son Jacfar from Mah Melek Khatun if he had the khutba read in her son’s name. The caliph agreed on the condition that her mamluk commander Üner was appointed commander- in-chief and reported to Taj al-Mulk who had been appointed vizier on Nizam al-Mulk’s death. Having sent two commanders ahead with the royal seal and instructions to imprison the heir apparent Berk-Yaruk, Terken Khatun left for the capital Isfahan after Mahmud’s name had been read in the Friday khutba (26 November 1092). News of the sultan’s death and possibly the reading of the khutba in her son’s name seem to have reached Isfahan before Terken Khatun. On Berk-Yaruk’s imprisonment, Nizam al-Mulk’s mamluk soldiery, known as the Nizamiyya, are said to have raided the deceased vizier’s arsenal and released him from prison. They had the khutba in Isfahan read in Berk-Yaruk’s name before departing with him for Rayy. There they are said to have been joined by the deceased vizier’s remaining commanders and soldiery. According to Turan and Özaydın, the struggle that ensued was primarily due to Terken Khatun’s insatiable ambition. In justification they pointed out that her son Mahmud was only five. However, the heir apparent Berk-Yaruk was still only twelve. More importantly, Turan and Özaydın appear to have conveniently Overview 35 forgotten that when Sultan Alp-Arslan appointed Malik-Shah heir apparent, his eldest Ayaz was alive. Ayaz was also alive when Malik-Shah acceded to the throne. As Malik-Shah’s atabeg, Nizam al-Mulk was involved in both instances. Given that as the eldest of Sultan Malik-Shah’s remaining sons Berk-Yaruk had also been Nizam al-Mulk’s choice, as Kafesoğlu pointed out, it is more likely that Terken Khatun did not want the Great Seljuq Empire to continue to be administered by Nizam al-Mulk’s sons and sons-in-law. Consequently, at least initially, the internecine struggle that followed Sultan Malik-Shah’s death appears to have been between Nizam al-Mulk’s supporters and opponents. Once in Isfahan, Terken Khatun is said to have used the funds available in the Treasury to gain the allegiance of the royal corps. This does not seem to have prevented some of them from deserting Taj al-Mulk at the battle of Burujird (17 January 1093). Victorious, Berk-Yaruk besieged Isfahan and offered the captured Taj al-Mulk his vizierate. Although Taj al-Mulk is said to have gifted the Nizamiyya money, they lynched him. Berk-Yaruk lifted the siege in return for money, ceding Isfahan and Fars to Terken Khatun. Terken Khatun then offered to marry Yakuti’s son Ismacil. Ismacil advanced on Berk-Yaruk from Azarbayjan with the Türkmen he had gathered to his cause but was defeated when his mamluk soldiery defected. Ismacil took refuge in Isfahan (February 1093). Meanwhile the deceased sultan’s brother Tutuş had been preparing to make his bid. With the support of the governors general of Aleppo, Antakya and Urfa, first he advanced on the Uqailids, thought to have been reinstated by Terken Khatun as a buffer. Tutuş defeated the Uqailids and asked the caliph to have the khutba read in his name (February–April 1093). The caliph refused on grounds that he had neither the allegiance of the dynasty nor control of the Treasury. Tutuş then advanced on south-eastern Anatolia where the Marwanids likewise appear to have tried to re-establish themselves. Having defeated them, Tutuş appointed his Türkmen begs as the governors general of Diyarbakır, Bitlis and Hisn Kayfa. In Köymen’s opinion, Tutuş’ harsh treatment of the Uqailids and the Marwanids turned Aksungur and Bozan, the governor generals of Antakya and Urfa, against him. When Tutuş arrived in Azarbayjan, causing Berk-Yaruk to advance to Rayy, Aksungur is said to have persuaded Bozan to defect, forcing Tutuş to return to Antakya. Shunned by Terken Khatun after his defeat, Ismacil is said to have met further resistance from her commander-in-chief, Üner. However, when he defected to Berk-Yaruk, he was strangled or possibly even decapitated by Aksungur and Bozan (August–September 1093). If true, it seems ironic that Köymen thought Aksungur and Bozan left Tutuş because they considered him harsh. More likely their enmity stemmed from when Aksungur and Bozan had been assigned to Tutuş to help regain Syria and Palestine from the resurgent Fatimids. Aksungur had lifted the siege of Tripoli on grounds that the city had proclaimed its allegiance to Sultan Malik-Shah. Bozan seems to have followed Aksungur, returning to Urfa. Unable to enter Tripoli, Tutuş is said to have complained to Sultan Malik-Shah without success (1090/1). Consequently, Aksungur and Bozan may have thought 36 Overview their interests better served by the youthful Berk-Yaruk rather than the older, more forceful Tutuş. In so far as Tutuş’ treatment of the indigenous dynasties and their soldiery is concerned, Köymen may have been right. On taking Nisibin, which appears to have resisted, Tutuş is said to have killed its Arab commanders and thousands of residents (March 1093). While Berk-Yaruk journeyed to Baghdad where the caliph proclaimed him sultan, Tutuş prepared for the next stage by marrying his eldest son Ridwan with the daughter of the governor general of Antakya, Yağı-Sıyan. After this, Tutuş advanced on Aksungur in Aleppo. Although Sultan Berk-Yaruk is said to have ordered Bozan to help him, Aksungur was forced to attack when Bozan failed to cross an irrigation canal in time. The battle is noteworthy for two reasons: first, Tutuş seems to have been supported by the Arab Banu Kilab and Aksungur by additional Türkmen; second, Tutuş is said to have feared the Arabs would desert him. Ironically, it seems the Türkmen deserted Aksungur instead. Victorious, Tutuş had Aksungur decapitated along with most of his commanders (27 May 1094). On entering Aleppo where Bozan had sought refuge, Tutuş also had him decapitated. Tutuş gained control of Urfa by parading Bozan’s head outside the city walls (29 May 1094). Clearly emboldened by his success, Terken Khatun offered Tutuş marriage, which he is said to have accepted. However, their marriage was not to be; Terken Khatun fell ill and died en route from Isfahan. Nevertheless, her soldiery are thought to have joined Tutuş in Hamadan. Tutuş may also have acquired what remained of the treasury, because he is said to have sent the Türkmen Abıkoğlu Yakup Beg to Baghdad to request the caliph have his name read in the khutba. En route Yakup Beg came across Berk-Yaruk and defeated him. Berk-Yaruk withdrew to Isfahan but was unable to gain entrance. When he did negotiate entry, he was imprisoned despite having his trusted commanders and their soldiery with him. He is said to have escaped being blinded when coincidentally Terken Khatun’s son Mahmud died of smallpox (October 1094). Released, Berk-Yaruk also fell ill but seems to have recovered. Possibly unaware of his recovery, the caliph had the khutba read in Tutuş’ name (October–November 1094). With Terken Khatun and Mahmud dead, Berk-Yaruk seems to have finally gained the royal corps’ allegiance. He also appears to have rallied the remaining commanders with his vizier Nizam al-Mulk’s son Muayyad al-Mulk’s help. As a result, Tutuş was defeated outside Rayy where he is said to have been decapitated on the battlefield by one of Aksungur’s mamluks (26 February 1095). It would seem that Tutuş failed, not because of his harsh character, but because Sultan Malik-Shah’s former commanders thought the youthful Sultan Berk- Yaruk more malleable. Another factor appears to have been Tutuş’ low opinion of Nizam al-Mulk’s sons and sons-in-law. When he came across the dead vizier’s son Fakhr al-Mulk, Tutuş wanted him killed, but his commanders persuaded him instead to appoint Fakhr al-Mulk his vizier so as to gain the goodwill of the residents of Hamadan. This suggests that at least they had realized Nizam al-Mulk’s fame held greater sway with the Seljuqs’ subjects than endorsement by the caliph. Overview 37 However, unlike their father, Nizam al-Mulk’s sons do not appear to have been up to their appointed tasks. On defeating Tutuş, Sultan Berk-Yaruk had sent for his mother Zubaydah Khatun. Possibly afraid he would come under her influence because of his youth, his vizier Muayyad al-Mulk appears to have objected with the support of several commanders. Al-Balasani, a member of the diwan who accompanied Zubaydah Khatun on her journey appears to have made her aware of Muayyad al-Mulk’s objections. Learning of this, Fakhr al-Mulk, who had fallen out with Muayyad al-Mulk over their father’s inheritance, sent the sultan a considerable amount of money, requesting the post of vizier. Sultan Berk-Yaruk seems to have accepted, imprisoning Muayyad al-Mulk (1095). Muayyad al-Mulk may not have been released for a number of years, possibly until he and Fakhr al-Mulk had come to an agreement over their father’s jewellery. He reappears in Ganja, Azarbayjan, as Tapar’s vizier. Another of Sultan Malik-Shah’s sons, Tapar appears to have been brought to Isfahan by Terken Khatun, possibly with his uterine brother Sancar, but escaped and joined Berk-Yaruk who appointed him to Ganja with an atabeg (1093/4). When Muayyad al-Mulk approached him, Tapar had already killed his atabeg and was in the process of annexing Arran (1098). The assassination of one of Sultan Berk- Yaruk’s most influential and loyal commanders, Porsuk, appears to have given Muayyad al-Mulk the opportunity to spread rumours that al-Balasani had hired the responsible batini. Thought to have been influenced by Muayyad al-Mulk, Terken Khatun’s former commander Üner advanced on Sultan Berk-Yaruk from Isfahan, threatening to rebel if al-Balasani was not handed over to him. Although he was killed en route and his commanders joined Porsuk’s sons, Berk-Yaruk’s commanders then petitioned the young sultan to hand over al-Balasani. The sultan refused, but they killed al-Balasani nonetheless, forcing Berk-Yaruk to flee to Rayy where he was joined by his mother and Nizam al-Mulk’s remaining sons (August 1099). Tapar dropped Sultan Berk-Yaruk’s name from the Friday khutba and advanced from Ganja. Berk-Yaruk withdrew to Isfahan but once again was refused entry. Nonetheless, he appears to have left his mother there before retiring to southern Mesopotamia. Here he plundered Wasit before being joined by the ruler of Hilla, with whom he then advanced on Baghdad (December 1099). It seems the caliph had dropped his name from the khutba on Tapar’s request (4 November 1099). Hearing of Berk-Yaruk’s advance, Tapar’s emissary, the mamluk commander Gawhar-A’in is said to have sent word to the new sultan. However, when Sultan Tapar did not come to Baghdad, Gawhar-A’in and the other commanders there switched allegiances yet again, proclaiming fealty to Sultan Berk-Yaruk, who had had the khutba changed back (31 December 1099). When Sultan Berk-Yaruk finally arrived in Baghdad (2 January 1100), he had the caliph’s vizier arrested for back taxes owed from when his father had been in Diyarbakır. The caliph is said to have threatened to take action, which may suggest that while in Wasit Sultan Berk-Yaruk had plundered his estates. A confrontation appears to have been avoided by a promise of payment. Meanwhile Muayyad al-Mulk had journeyed to Isfahan and imprisoned 38 Overview Sultan Berk-Yaruk’s mother, forcing her to sign a promissory note before having her brought to Rayy and strangled. In spring Berk-Yaruk advanced on Tapar (17 April 1100), but was defeated (15 May 1100). Nonetheless, their struggle continued unabated with Sancar joining Tapar from Khurasan. During this time Sultan Berk-Yaruk gained the uncertain satisfaction of personally strangling Muayyad al-Mulk in revenge for his mother (1101). In the end, the commanders forced the young Seljuq princes to come to an agreement. After protracted negotiations conducted on the battlefield (2–31 January 1104), Tapar recognized Berk-Yaruk as Great Sultan. In return he was allowed to have the khutba read in his name outside Iraq – present-day Iraq and western Iran, excluding Azarbayjan. Having been wracked by illness throughout his short but turbulent life, Great Sultan Berk-Yaruk died that year (22 December 1104). While the Seljuqs and Nizam al-Mulk’s sons fought over their respective inheritances and each other, the Crusaders arrived (August/September 1096). Resistance was left first to the Seljuqs of Anatolia and then to the Seljuqs of Syria. Both were initially unsuccessful because they had not encountered mounted knights in battle before, but Tutuş’ sons, Ridwan and Dukak in Aleppo and Damascus respectively, also proved too young and suspicious of each other. They responded separately to calls for help by Antakya’s governor general, Yağı-Sıyan, and thus were also beaten separately (31 December 1097 and March 1098). Sultan Berk-Yaruk then assigned one of the deceased Aksungur’s mamluks, Kür-Boğa, to relieve Antakya. Kür-Boğa and his brother Altun-Taş had taken Harran south of Urfa and from there expanded into northern Mesopotamia, taking Nisibin, Mosul (October/November 1096) and Rahba from the Uqailids who had returned on Tutuş’ demise. Rather than immediately relieve Antakya, Kür-Boğa first tried to regain Urfa, which had just fallen to Godfrey of Bouillon (10 March 1098), but he failed (4–25 May 1098). His reasoning may have been to protect his rear but when he got to Antakya (7–10 June 1098) the city had fallen (2–3 June 1098). While besieging the city, Kür-Boğa seems to have spent his time scheming against Dukak and Ridwan. After Antakya, Kür-Boğa may have hoped to add Damascus and possibly also Aleppo to his recently won domains. Whatever his motives, sure of the outcome, he is said to have refused Peter the Hermit and Herluin’s offer to hand back Antakya in return for safe passage (27 June 1098). When the Crusaders sallied forth the next day, Dukak’s soldiery and Ridwan’s commanders-in-chief seem to have deserted Kür-Boğa (28 June 1098), forcing him to withdraw. Taking the coastal fortification and settlements en route, the Crusaders advanced on Jerusalem and took it as well (15 July 1098).

Dissolution (1105–94) Kafesoğlu was of the opinion that, with the exception of Great Sultan Sancar, Sultan Malik-Shah’s successors lacked the personal qualities to reign. His judgement is as misleading as it is harsh. Overview 39 When they arrived in Khurasan, not only was their guardian and uncle Musa still with them but Çağrı and Tuğrul Beg were in their forties. Having proven himself under his father’s tutelage, Alp-Arslan was thirty-two when he became sultan. Although Sultan Malik-Shah was only seventeen when he acceded, he had been on campaign with his father and had his atabeg Nizam al-Mulk as his vizier. By comparison, Berk-Yaruk was twelve and Tapar only ten when their father died. Thirty at the time, Tutuş was the only mature Seljuqid to challenge Berk- Yaruk and thus Nizam al-Mulk’s sons and supporters. Having been sent to conquer Syria and Palestine at the age of fifteen (1077), Tutuş was a seasoned military campaigner when he challenged Sultan Berk-Yaruk. Whatever his personal traits, in the end Tutuş was unable to overcome the commanders and administrators fighting over the spoils of empire because he lacked a personal powerbase, something Alp-Arslan had and Sancar was to have. Given how often Çağrı Beg’s sons came to their uncle’s rescue, and in view of Alp-Arslan and Sancar’s success, another reason for Tutuş’ failure may have been that the weight of Seljuqid martial power was in Khurasan, not Iraq al-Ajam or al-Arab. Perhaps not surprisingly, the Ismacili Nizari sect led by Hasan al-Sabbah seems to have been far better established in Iraq al-Ajam and northern Syria than in Khurasan. Like their leader in his mountain fortress in Alamut, the Nizaris appear to have taken care to ensconce themselves in relatively unassailable fortifications. One such citadel, Shahdiz, even threatened the main route from the capital Isfahan to Khurasan. Led by Ahmad Attash, the son of al-Sabbah’s mentor, the Nizaris are said to have taxed the local populace and raided the countryside surrounding Shahdiz. Rumours that Khurasan was in turmoil and the Anatolian Seljuqid Kılıç Arslan had entered Baghdad appear to have prevented Sultan Tapar from immediately besieging the citadel. Having taken Malatya (1106), Kılıç Arslan had laid siege to Urfa. Unable to take the city, from there he had marched on Mosul, entering it uncontested (22 March 1107). He was defeated soon after however, and drowned while fleeing (3 June 1107). As for Khurasan, unrest there appears to have been caused by Sultan Tapar’s cousin Mengü-Pars. Finding himself in financial difficulties, Mengü-Pars had journeyed to Nihavand intending to claim it but was captured and returned to Isfahan, where he was imprisoned with his cousins (1105–6). During the siege of Shahdiz, Sultan Tapar hanged his vizier Sacad al-Mulk and four of his ‘associates’ from the city gates of his capital for colluding with the Nizaris (5 July 1107). The sultan then replaced all his ministers from Iraq al- Ajam and al-Arab with administrators from Khurasan. When Shahdiz fell, Attash was flayed alive, having been paraded around the capital. Although the Nizaris continued to assassinate notables and foment discontent, Sultan Tapar seems to have left them alone except for a few token attempts to root out Hasan al-Sabbah. The reason for this may have been that he had far greater problems with the Shicite Sadaqa and the Crusaders. When the commander of Basra turned away his tax collector, Sultan Tapar asked his Mesopotamian vassal Sadaqa to deal with him. Sadaqa besieged the city and having sacked it installed his own commander (10 February 1106). He 40 Overview then took advantage of a similar situation in Tekrit, situated between Baghdad and Mosul. After Sultan Tapar’s commander Porsukoğlu Aksungur had failed to gain entry despite besieging the city for seven months, its Iranian mamluk commander invited Sadaqa to take charge (October 1106). Having gained Tekrit, Sadaqa appears to have prepared for the worst. Although the caliph interceded, he is said to have refused to disband his forces. When Sultan Tapar sent an envoy to reassure him, asking that he support his commander Çavlı Sakavu against the Crusaders, Sadaqa again refused. Sultan Tapar appears to have assigned Türkmen beg Muhammad, son of the deceased governor general Antakya Yağı-Sıyan, to take Wasit back from Sadaqa. But when the Türkmen and Sadaqa’s cavalry joined to plunder the city, the sultan took charge. Sadaqa was defeated and died on the battlefield (3–4 March 1108). Sadaqa’s sacking of Basra and that of Wasit with the Türkmen is similar to what appears to have happened in Tekrit. During the interregnum its commander is said to have taken turns to sack the city with the Türkmen beg Artukoğlu Sökmen (1098/9). It seems safe to conclude, therefore, that those in power – whether Turk, Kurd, Arab or Iranian – had little concern for the plight of their indigenous populations. Sultan Tapar does not appear to have led his soldiery into battle after Wasit. This may be because he did not have sufficient forces directly under his command. For instance, there is the case of Çavlı Sakavu, who seems to have gained control of Fars as well as Khuzistan during the interregnum. Having successfully resisted the Porsukid commander sent against him, Sakavu appears to have then relented, journeying to Isfahan to gain Sultan Tapar’s pardon, as a result of which he seems to have been assigned to fight the Crusaders. Sakavu was married to Porsukoğlu Porsuk’s daughter, which puts in question Sultan Tapar’s judgement in sending a Porsukid commander against him in the first place. Sakavu also seems to have been Kür-Boğa’s nephew. As noted, Kür-Boğa had been a mamluk of Sultan Malik- Shah’s governor general Aksungur. Given Ridwan was married to the deceased governor general Yağı-Sıyan’s daughter, it seems that under Sultan Malik- Shah the Great Seljuqs had established a hereditary Turkic warrior aristocracy supported by iqtacs. While Sultan Malik-Shah may have been able to control such an aristocracy, probably with the help of the royal mamluk corps, these not only became divided during the interregnum but also most probably died in the various battles immediately after his death, leaving Sultan Berk-Yaruk and his successors at the mercy of their regional commanders. As for the Crusaders, Sultan Tapar appears to have concentrated on regaining Urfa, though he had campaigns conducted in Syria too. The countryside on the eastern shores of the Euphrates is said to have been systematically laid to waste during the sieges of Urfa by various commanders and begs (1105, 1106, 1110, 1112 and 1114), so that during 1116–17 the Crusaders took Keysun, Ra’ban and Birecik to the west, and Gerger to the north from the Armenians to compensate for the population exodus. Confined to the western shores of the Euphrates, after 1118 the County of Urfa ceased to provide a serious threat to Aleppo and Mosul, except for the possibility of the odd raid. Overview 41 Sultan Tapar died in 1117. On the hour of his death, his wife Gevher Khatun is said to have been strangled on his orders. Two explanations are offered. First, that his long illness was the result of poisoning, presumably by her. More likely, however, he may have wanted to prevent her marriage to his brother as was customary. Gevher Khatun was Çağrı Beg’s son Yakuti’s granddaughter and seems to have inherited responsibility for Azarbayjan, which she ruled through her diwan. Sultan Tapar was succeeded by Mahmud, still only thirteen. Mahmud appears to have had four brothers: Mascud, nine; Tuğrul, eight; Süleyman and Selçuk- Shah. This does not seem to have prevented Mascud’s atabeg Ay-Aba from asking that Ganja be taken from Tuğrul and given to his charge. Before the rebellion could be resolved one way or another, Sancar declared himself sultan and advanced on Mahmud (14 June 1118), defeating and capturing him at Sawa (14 August 1119), which suggests he encountered resistance. Taking the title of Great Sultan, Sancar stripped Mahmud of Rayy, Mazandaran and Qomis in Iraq al-Ajam, as well as retaining taxes from Sawa and Huys. He assigned Tuğrul to north-east Jibal and Gilam, Selçuk-Shah to western Khuzistan and Fars, and left Mascud in Mosul as Sultan Mahmud’s subordinates. Great Sultan Sancar is said to have also assigned his own administrators to determine taxes due to Sultan Mahmud. This arrangement does not appear to have satisfied either Mascud’s atabeg Ay-Aba or his ally Dubays Sadaqa. Ay-Aba was defeated by Porsukoğlu Aksungur who persuaded Mascud to present himself at court, where Sultan Mahmud is thought to have retained him, but Dubays began to pillage the environs of Baghdad. When Caliph al-Mustarshid (1118–35) threatened to retaliate, Dubays advanced on Baghdad. The caliph appealed to Sultan Mahmud who occupied Hilla (September 1120). As a result Dubays appears to have sought refuge in Aleppo with his father-in-law, Artukoğlu Il-Gazi. Having assigned Mosul to Porsukoğlu Aksungur and increased Il-Gazi’s iqtacs, Sultan Mahmud sent an envoy to Great Sultan Sancar with Caliph al-Mustarshid (June 1121). A few months later, the caliph held a ceremony at which he presented Sultan Mahmud with two swords (29 October 1121), echoing Tuğrul Beg’s investiture. Possibly on the caliph’s instigation, Sultan Mahmud asked Il-Gazi to divorce his daughter from Dubays. Il-Gazi refused, though he is said to have taken care to reiterate his allegiance. Dubays is thought to have come to an agreement with the caliph before returning to Hilla, where he ousted Sultan Mahmud’s officials and their families, causing many to die. In response, Sultan Mahmud retook Hilla, but instead of ousting Dubays from Mesopotamia, as the caliph is said to have requested, he took Dubays’ son Mansur hostage and returned to Hamadan, assigning Aksungur to defend Baghdad. However, when Dubays defeated Aksungur and took Wasit while Sultan Mahmud was campaigning in the Caucasus, Caliph al-Mustarshid appears to have decided that the time had come to take matters into his own hands. Having assigned one of his chamberlains commander-in-chief and declared jihad on Dubays, something it must be noted neither he nor his predecessor had done against the Crusaders, the caliph took the field with Aksungur’s forces and some 42 Overview Türkmen from south-eastern Anatolia who answered his call. When he was sure of victory, the caliph is said to have also drawn his sword (1123). A pattern of events was thus established, which hastened the unravelling of Great Seljuq hegemony in Iraq al-Arab as well as Syria. The caliph took the field again when Dubays advanced with Sultan Mahmud’s brother Tuğrul (13 March 1125). After having withdrawn without giving battle, Dubays and Tuğrul are said to have journeyed to Khurasan for an audience with Great Sultan Sancar who imprisoned Dubays and retained Tuğrul at court. Emboldened, the caliph appears to have established a force of Arab and Kurdish cavalry and started to interfere in the affairs of the Seljuqs, coming to an agreement with Sultan Mahmud to depose Great Sultan Sancar. Having learnt of this, Great Sultan Sancar ordered Sultan Mahmud to march on Baghdad, imprison the caliph’s vizier and kill his Kurdish cavalry. Instead, Sultan Mahmud asked to come to Baghdad, but Caliph al-Mustarshid refused, setting up camp on the eastern banks of the Tigris. He moved his camp to the western banks on learning of Sultan Mahmud’s approach, sending a eunuch to take Wasit. Sultan Mahmud appears to have dispatched one of his commanders, Sultan Malik-Shah’s governor general of Aleppo, Aksungur’s son Zengi to retake it. When Sultan Mahmud finally arrived in Baghdad he seems to have had the caliph’s palace pillaged. The caliph is said to have enticed the populace against the Seljuqs and crossed the Tigris at the head of his forces (7 January 1126). While his and the sultan’s forces skirmished on the outskirts, Zengi took Baghdad by force. After a long illness Sultan Mahmud appears to have retired to Hamadan without effecting any of Great Sultan Sancar’s orders (12 April 1126). Great Sultan Sancar is thought to have called Sultan Mahmud to account in Rayy where he arrived at the head of a large army (1127/8). Tuğrul and Mascud appear to have joined their brother at court, where they were detained for over a year. At the end of that time Great Sultan Sancar is said to have appointed his daughter, Sultan Mahmud’s wife, as Dubays’ guardian, assigning Dubays to Mosul as governor of Jazira (northern Mesopotamia) and Syria. Sultan Mahmud seems to have journeyed to Baghdad to gain Dubays’ pardon (14 January 1129), but Caliph al-Mustarshid appears to have refused either to pardon him or approve of his appointment. Sultan Mahmud returned to Hamadan (25 May 1129), where his wife then died. After this Dubays appears to have taken matters into his own hands, leaving Hamadan for Hilla with Sultan Mahmud’s son; the sultan is said to have been seriously ill at the time. Although Dubays was captured, Sultan Mahmud’s son still seems to have been in captivity when the sultan died a few years later (11 September 1131). Neither the heir apparent Davud nor his uncles Mascud and Selçuk-Shah were able to persuade the caliph to have their name read in the khutba. He is said to have insisted they gain Great Sultan Sancar’s approval first. Having gone to Samarkand to deal with the Karakhanids (April 1130), Great Sultan Sancar once again arrived in Rayy at the head of his army (January–February 1132). Meanwhile Mascud and Selçuk-Shah seem to have come to an agreement with the caliph. In return for allowing him to rule Iraq al-Arab through an agent, the caliph appears to have Overview 43 promised to help them topple their uncle. The understanding between the brothers seems to have been that Mascud would be sultan and Selçuk-Shah heir apparent. However, when the caliph followed them out of Baghdad with his forces, he had included neither of their names in the khutba (27 March 1132). Nor was the caliph at their side when Great Sultan Sancar defeated his nephews near Dinavar (26 May 1132). The caliph seems to have been occupied with Zengi and Dubays outside Baghdad. Great Sultan Sancar made Tuğrul sultan, assigning Mascud to Azarbayjan after having several of his commanders decapitated. Said to have defeated Zengi and Dubays, due to the reluctance of their soldiery to fight him, Caliph al-Mustarshid now threatened to take Mosul by force if Zengi did not hand over Dubays. Although this did not transpire, Dubays does not appear to have been able to re-establish himself in Hilla where the caliph had installed his own commander. At this point the deceased Sultan Mahmud’s son Davud and his atabeg Ahmadili seem to have decided to challenge Sultan Tuğrul. Defeated, they fled to Baghdad where they were joined by Mascud. The caliph declared Mascud sultan and Davud heir apparent, giving them soldiery with which they are said to have defeated Tuğrul. The constant desertions by commanders from one Seljuqid to another appear to have encouraged the caliph who unsuccessfully besieged Mosul with the help of some Türkmen (July–October 1133). Tuğrul died the following year (24 October 1134). Although Great Sultan Sancar does not appear to have objected to his succession, Mascud seems to have been unable to gain his commanders’ allegiance. The commanders are said to have turned to the caliph for an alternate Seljuqid candidate. Caliph al-Mustarshid appears to have tried to take advantage by taking the field against Mascud, but was defeated and captured (24 June 1135). He was assassinated by a group of Ismacilis while in captivity (29 or 31 August 1135). Blaming Dubays for the caliph’s demise, Sultan Mascud had Dubays killed. Undeterred, the commanders and Türkmen begs seem to have continued to plot with the new caliph, al-Rashid (1135–6), but when the caliph attempted to lead an army against Sultan Mascud (27 May 1136) he was defeated and ended up a fugitive in Mosul (14 August 1136). Sultan Mascud took Baghdad by force and installed a new caliph, al-Muqtafi (1136–60). Al-Rashid was assassinated in Fars by Nizaris (8 June 1138). While Seljuq rule in Iraq al-Arab unravelled, Transoxania fell into turmoil. Pressure from the Mongolian Kara Khitay had displaced the Kanglı, causing some Karluk and the remaining Oguz to move south, a few of their Avşar, Salur and Yıva lineages continuing west into Iraq al-Ajam. Unable to cope with the unrest their displacements caused in Transoxania, the Karakhanids had asked the Kara Khitay to deal with the Kanglı and the Seljuqs to deal with the disaffected Karluk (1137). By the time Great Sultan Sancar marched into Transoxania, the disaffected Karluk appear to have been joined by the Kanglı who had gained the Kara Khitay’s protection. Despite being outnumbered, the Kara Khitay and their allies crushed the Great Sultan Sancar at Katvan (9 September 1141). Having escaped with his life, Great Sultan Sancar’s problems were only just beginning. Emboldened by his defeat, his vassal the Khwarazmshah Atsız took 44 Overview Serakhs and Marv (October 1141), after which his brother pillaged as far as Nishapur. Great Sultan Sancar appears to have waited for the Kara Khitay. When they withdrew having established their own administration in Transoxania, he marched on Atsız whom he is said to have forced to return his plunder (July 1143). Nevertheless Atsız appears to have continued to resist the officials sent to administer Khwarazm until Great Sultan Sancar marched on him again, forcing him to proclaim allegiance in person (2 June 1148). Forced out of Transoxania by the Karluk, the Oguz appear to have been situated in the region between Balkh and Ghazna. Having killed his tax collector, the Oguz then defeated and captured Great Sultan Sancar who had been persuaded to march against them (March–April 1153). The sultan was put in a cage and taken along by the Oguz while they repeatedly sacked Khurasan in the following years. Thought to have lost thousands of their commanders with their soldiery at Katvan, the Great Seljuqs seem to have been helpless to mount a counteroffensive. Köymen concluded that as much as a quarter of the urban population of Khurasan lost their lives during this period. The Oguz finally released Great Sultan Sancar (April 1156), who died the following year (22 or 29 April 1157). During his reign Sultan Mascud does not appear to have been able to gain control of his commanders, let alone re-establish Seljuq hegemony in either Iraq al-Arab or northern Syria. The fight against the Crusaders and for a time also the Byzantines seems to have been left to Zengi in Syria. Zengi was killed by a servant (14 September 1146) two years after he captured Urfa (24 December 1144). Having effectively handed over the day-to-day administration to an Oguz beg, Beg Arslan, Sultan Mascud died a year before Great Sultan Sancar’s capture (12 October 1152). After his death none of his brothers or sons were able to establish their sovereignty and with the death of the last known Great Seljuq Sultan Tuğrul on the battlefield (25 March 1194), the Great Seljuq Empire came to an end. 3 Origins

The first section in this chapter outlines what is known of the Eurasian steppe zone and its cultures. Map A1.1 in Appendix 1 shows known Turkic migrations in accordance with Turkish historiography. There follow sections on the origins of the Oguz, Türkmen and Seljuqs, in that order.

The Eurasian steppe zone The Eurasian steppes are interrupted by a complex of mountains, namely the Pamir, T’ien-Shan (Tengri or Ak-Tağ), Targabatay, Altay (Altun-Yış), Tannu-Ula (Kögmen), Hangay (Ötüken-Yış) and Sayan (Kögmen) mountains. The grasslands are higher and hence drier eastward toward Manchuria, while lush pastures abound westward into Eastern Europe. This is why to the east horses and sheep appear to have predominated, whereas to the west initially cattle were favoured alongside sheep, with horses being kept as draught animals. Since the steppes to the west also favoured agriculture, agrarian and urban concerns there remained to the fore (Kafesoğlu 1994: 165); a point confirmed by Turkic words incorporated into Hungarian, such as arpa (barley), buza from buğday (wheat), tarló from tarla (field), alma (apple), szölö from sidleg (grape), tyuk from tavuk (chicken), kos from koç (ram), bıka from buka or buğa (bull), ökör from öküz (cattle), etc. (Ràsonyi 1983: 10). Nevertheless, agriculture was practised in Inner Asia too (Ögel 1962: 88–90, 164–5), even if to a limited degree (Barfield 1996: 45–9). Nor must it be forgotten that migration routes were fixed, regardless of distances. Tombs and fortifications discovered in Inner Asia indicate pastoralists had seasonal residences (Esin 1980a: 4–5), most of which doubtless were winter quarters. Having begun to dry during the previous millennium, the Eurasian steppes are thought to have been at their driest around 800 BCE (Lamb 1995; also Khazanov 1994), which is when the Scythians entered the historical record. Although thereafter fully nomadic pastoralists seemed to irrupt periodically in western Eurasia, whether or not these were likewise due to drought must be a secondary consideration. Arguably it was the development and spread of agriculture throughout northern China that forced full-blown pastoralism to emerge in Inner Asia among cultures whose herders were driven to the margins (Ögel 1962; Esin 1980a; Eberhard 1987; Lattimore 1967). 46 Origins Having spread to forest people, pastoralism remained dependent on agriculture and manufacture (Lattimore 1967; Khazanov 1994), and hence trade (Jagchid and Symons 1989; also Kafesoğlu 1992). Although herders appear to stay healthier than farmers (Barth 1986; also Khazanov 1994), unlike agriculture, pastoralism has never provided the means of intensifying food production to absorb population growth – at least not to a comparable magnitude. Thus, for instance, it cannot be known at present whether the Huns, who evidently also practised agriculture (Maenchen-Helfen 1973: 174–8), were displaced by an expansion of predominantly agrarian culture(s) in the Pontic-Caspian region or by the collapse in Inner Asia of the Hsiung-nu Empire (210 BCE– CE 155). Even when nomadic pastoralists were united into empires, sedentary people outnumbered them by as much as 100:1. According to Tonyukuk, Bilge Kagan’s chief counsellor and commander-in-chief, the Kök Türk made up less than one per cent of the T’ang period Chinese (Divitçioğlu 1987: 210). Given that Tonyukuk was born in China and before helping to rejuvenate the Kök Türk Empire was in hereditary service in the Chinese administration (Tekin 1994: 2), this percentage has to be taken seriously. Based on a census conducted in CE 2, the Chinese are thought to have totalled 12.3 million hearths, or 58.5 million people (McNeill 1989: 119). Using the same multiplier for arithmetic consist- ency, by CE 703 the total number of hearths would have fallen to 6.9 million, thus 32.8 million people, rising by CE 742 to 8.6 million hearths or 40.9 million people (see Elvin 1973: 206). Consequently, less than one per cent would make the Kök Türk Empire’s population at the most 300,000, as by CE 716 Tonyukuk already was seventy years of age. For nearly 3,000 years people from Central and Inner Asia migrated across the Eurasian steppes zone, sowing fear wherever they appeared. Not that anyone could determine how many made up a ‘horde’ (Roux 1995: 82), or who they were. A person’s ancestry was secondary to the lineage to which he was politically affiliated at the time (Evans-Pritchard 1978; Sahlins 1968; Fox 1967), unless he was of noble birth (Sneath 2007). Affiliates, commoners as well as slaves, were organized decimally by the lineages or noble houses that led or owned them (Barfield 1993; Sneath 2007). Moreover, marriage alliances were as much between lineages of different ethnic and linguistic origins as between dynasties heading different polities (Türköne 1995; Togan 1981; Esin 1980a; Ögel 1962; also Fox 1967), further complicating the latter-day search for the ethnic origins (Togan 1981; Esin 1980a; see also Golden 1992). In short, nomadic pastoral peoples are currently thought to have been inclusive not exclusive (Barfield 1993; Sahlins 1968; Fox 1967; Leach 1977), their hierarchical organization making them indig- enously capable of statecraft (Sahlins 1968; Barfield 1993, 1996; Sneath 2007). This is what Turkish scholars have been arguing for some time, albeit for different reasons (Togan 1981; Kafesoğlu 1994; Köymen 1979). Recent research likewise concurs with Turkish scholarship that statecraft rather than a written record is the differentiating characteristic between primitive and advanced cultures (Sahlins 1968; Sahlins and Service 1973; Claessen and Skalník 1981). Arguably it is the organization and distribution of an agricultural surplus that Origins 47 necessitated writing, reading and arithmetic (Lattimore 1967; Wittfogel 1959). A nomadic pastoral way of life did not, except when the leaders of such polities needed to communicate with or administer sedentary empires. According to the Chinese, the Hsiung-nu did not have a script (Eberhard 1942: 76), which may explain why a few tantalising lines of Hsiung-nu poetry thought to be in Turkic were rendered in Chinese characters (Tekin 1993). Certainly, in their written communications with the Chinese, the Hsiung-nu used the latter’s language (Barfield 1996: 52–3), as did the Kök Türk, who also used Soghdian in Transoxania (Divitçioğlu 1987). While the Great Seljuqs and their military spoke Turkic among themselves, they administered their subjects in Persian (Kafesoğlu 1994 and IA/10: 353ff). Similarly, although the Mongols adopted the Uygur version of the Soghdian cuneiform script, they administered their empire(s) in local languages. Not surprisingly, the further one goes back, the less chance there is of finding records left by nomadic pastoralists. Nevertheless, the recent discovery of a tomb in Central Asia suggests that an archaic version of the runiform script used by the Kök Türk possibly existed as early as the fifth century BCE. Two of the letters were found in the Kök Türk alphabet, which is thought to have evolved from Türk clan seals (tamga). At present scholars appear to agree that at least one of the words inscribed on a silver cup found in the said tomb is Turkic, namely azuk (food prepared for a journey) (Esin 1980a: 1, 22–4; also Kafesoğlu 1994: 323).

The Oguz Oguz origins are of particular importance as the Seljuqs and the majority of their Türkmen affiliates stemmed from them. Also they finally destroyed the Great Seljuq Empire before going on to found three more: the Ottoman (1299–1923), Akkoyunlu (1378–1508) and Safavid (1501–1736). Uz, Guzz and Gur, also Guz and Gurr, come up repeatedly in the works of Muslim and other chroniclers and geographers with reference to the Oguz, the Ogur and the Uygur, Turkic polities that not only knew each other but also intermingled. Turkic nomenclature further complicates the problem of ascertaining their whereabouts at any specific time. The root for Oguz and Ogur appears to have been og– or ogu–, as in the Argu ogla for youth or hero (Atalay 1992/I: 129), or ogul for son (Atalay 1992/I: 51, etc.). Ogul initially meant human child (Arat IA/9: 376–8), so that urı oglan was a male child (Ibid.; also Atalay 1992/I: 88, 251) and kız oglan a female child (Ibid.). Appropriately, ogulçuk was womb (Atalay 1992/I: 149). Oguş meant family, or more correctly kin as in clan or relative (Atalay 1992/I: 61, 88 and II: 83, 103). Consequently, Og– or better still Ogu– and the collective suffix –z or –r are thought to mean the clans (Sümer 1992: 13–14; Kafesoğlu 1994: 141–2; Divitçioğlu 1994: 11). The collective suffix –z originated east of the Altay while –r prevailed westward (Kafesoğlu 1994: 51, 185–6; Yüce IA/12/2: 445ff; Bozkurt 1992). Interestingly enough, one meaning of ogur was to split a bone and separate its parts and another was felicity/statecraft (Atalay 1992/I: 53). Turko-Mongol 48 Origins differentiation between those descended from the White and Black Bone, Ak and Kara Kamag Budun, namely noble and commoners is well known (Sümer 1992; Kafesoğlu 1994), as is the connection between felicity and statecraft (Atalay 1992/I: 92, 301, 304, 320, 508 and II: 229). Possibly as a variation of og–, ok or arrow denoted a share of an inheritance (Atalay 1992/I: 37, 48). The Orkhun stone inscriptions refer to the Western Kök Türk as the On Ok (Tekin 1988: 13, etc.). Chinese annals mention that the Kök Türk kagan resident in the East presented the leader of each ruling lineage in the West with an arrow (Sümer 1992: 20), which is interpreted to denote that they were subordinate to him (Kafesoğlu IA/12/2: 219). Chinese annals refer to the Ogur, but not the Oguz. Initially mentioned between the Sayan and T’ien-shan mountains in the third century BCE, the Ogur seem to have spread westward as far as the Urals by the second century CE (Kafesoğlu 1994: 185–7). As for the Oguz, they and the Kıpçak are thought to have been the Wu-huan that Chinese annals mention immediately east of the Altay, between the Tannu-ola and Hangay mountains (Esin 1980a: 7). The earliest known reference to the Oguz, to be more precise the Altı (Six) Oguz under a certain Öz Yigen Alp Turan, is on the Yenisei stone inscriptions dated to the sixth and seventh centuries (Sümer IA/9: 378). The Yenisei River flows north from the Tannu-ola Mountains. The Oguz are next referred to on the eighth-century Orkhun stone inscriptions; the Tokuz Oguz are mentioned on the Tonyukuk inscription (Tekin 1994: 4, 5) and the Üç (Three) Oguz on the Bilge Kagan inscription (735) (Tekin 1988: 49). As is clear from these inscriptions (Tekin 1988: 13, 45 and 1994: 6), at the time the Tokuz Oguz were situated in present-day Mongolia. This is where Kutlug or Ilteriş Kagan (682–91) re-established the Kök Türk Empire by defeating their newly appointed Baz Kagan (Sümer 1992: 16; see Tekin 1994: 5). Although Baz can mean stranger (Atalay 1992/III: 148, 159), or subjugated (Divitçioğlu 1987: 184), in this context most likely it denoted the species of the goshawk, an Oguz avian emblem (Esin 1994: 202). Having said that, prior to this defeat the Tokuz Oguz were allied with the Otuz (Thirty) Tatar and the Chinese, going so far as to migrate to northern China (Tekin 1988: 51), where some are thought to have stayed. However, archaeological evidence suggests that others migrated westward from Inner Asia, reaching the region between Lakes Balkash and Issyk- Köl during the sixth to seventh centuries (Esin 1980a). The Oguz migration to Central Asia may explain the reference to the On Ok’s successors, the Türgiş (Tekin 1988: 13, 19, 23, 33, 43, 47), who, according to the Orkhun stone inscriptions, were represented at Bilge Kagan’s funeral (731) by a Makaraç tamgaçı and an Oguz Bilge tamgaçı (Tekin 1988: 23). Possibly Makaraç is a title derived from maharac (more commonly maharaja) (Tekin 1988: 153). Bilge means knowledgeable/wise, while tamga is seal (Atalay 1992/IV: 92, 567). On the other hand, of the six lineages that made up the Türgiş, who were originally on the Left Wing of the On Ok, four were headed by patriarchs with the title of Çor, which may indicate that they were of the Şul (from Çöl or desert, and hence Çor (Esin 1980a: 62–3)). The Şul are thought to have been in Jurjan, south- east of the Caspian Sea, from at least as early as the fifth century (Ibid.). As far Origins 49 as the Kök Türk’s great kagan was concerned, it seems that the Türgiş were the senior members of the Left Wing or eastern On Ok. The Right Wing or western On Ok, the Nu-she-pi, appears to have consisted of three lineages, namely the A-si-kie, the Ko-shu (the Ezkiş or Ezgiş) and the Pay- say-kan. These are thought to have become the Üç (Three) Ok of the Oguz, which included the Kınık. Once again, however, there is a link to Inner Asia. The regent of the tenth-century Oguz polity between the Caspian and Aral seas was called Kül Erkin (also Kuzerkin), the title of the senior rulers of the A-si-kie and Ko-shu. The Karluk, at the time situated from the northern slopes of the Pamir and central T’ien-shan Mountains toward the Irtish River, also called their regent Kül Erkin (Sümer 1992: 21, 33, 35–7; Togan 1982: 137; Kafesoğlu 1994: 137 and IA/12/2: 165; Divitçioğlu 1992: 130; Şeşen 1975: 37). The question of whether or not the Oguz were related to their namesakes in Inner Asia is confused further by the origins of the Uygur, who as the On (Ten) Uygur, namely nine Oguz and one Uygur were well known to the Khazar, most probably due to the Uygur merchants who plied the Silk Route. Uy– meant to go along, to ally (Atalay 1992/III: 146), and uya– relative/brother (Atalay 1992/I: 85). Accordingly, Uy+(o)gur (Uygur) would be allied/related/brotherly clans (Kafesoğlu 1994: 122; Bozkurt 1992: 66). The –r suffix suggests an origin to the west of the Eurasian mountain complex, which appears to be confirmed by Chinese sources that mention them in the Ili River valley north of Issyk-Köl (50–43 BCE) (Esin 1980a: 204). The Uygur re-emerge in present-day Mongolia during the initial collapse of the Kök Türk Empire. Having been recognized by the Chinese court (646), they seem to have reinforced their independence of the Kök Türk by helping defeat remnants of the On Ok (657), advancing as far as Tashkent. Around this time some of the Tokuz Oguz already appear affiliated to the Uygur (Kafesoğlu IA 12/2: 180). They are thought to have consisted of the Buku, Bayırku, Siker, K’i-pi, Ediz, Izgil (or Sse-ki (Divitçioğlu 1994: 11)), the Toñra, Kunı and one other (Divitçioğlu 1987; Sümer 1992: 33). The Izgil are sin- gled out in the Orkhun stone inscriptions as having been dispersed by Kül Tekin (Tekin 1988: 21). Thus the Esgil/Isgil mentioned in association with the Bulgar polity established on the Volga during the ninth and tenth centuries may possibly be of the Izgil and hence of the Karluk’s Çiğil (see ‘The Türkmen’ below). There is also the final collapse of the Kök Türk Empire and the ensuing struggle to consider (744–55). This may have given impetus to the westward migration mentioned earlier and explain an unconfirmed reference to the Oguz in conjunction with al-Muqanna’s (c. 776–83) revolt against the caliphate in Transoxania (Kafesoğlu 1994: 143; Divitçioğlu 1994: 13–14). Although during this struggle the Basmıl’s leader Iduk Kutu, a relation of the imperial Kök Türk A-shih-na clan (Tekin 1988: 47), declared himself kagan after defeating the Kök Türk Tengri Kagan’s successor, their allies, the Uygur and Karluk did not accept this. The Basmıl lineages were beaten, never to recover. The Uygur next defeated the Üç Karluk, consolidating their ascendancy. The Karluk (most likely the Çiğil and Yağma lineages) migrated south-west from between Lake Zaysan and Lake Ulungur, where the current borders of Kazakhstan, China, Mongolia and Russia 50 Origins meet in the Altay, displacing the Peçenek toward the Caspian Lowlands (Sümer 1992: 28ff; Divitçioğlu 1992: 131). Peçenek derives from Becenek (currently bacanak), meaning brother-in-law, but in the context of men who have married sisters. They are thought to have been initially of the Left Wing of the On Ok because two of the eight Peçenek lineages were called Çor and Çoban (Çopan), which were the titles of the Left Wing chiefs of the On Ok other than the Türgiş (Sümer 1992: 21, 44). Thus the Peçenek recorded as a lineage of the Üç Ok Oguz (Sümer 1992: 169–71; also Atalay 1992/I: 57), are thought to be a branch of the main body subsequently driven from between the Ural and Volga rivers by the Khazar and the Oguz in 898–902 (Sümer 1992: 45). Although there is an eighth-century reference to a Ba-cha-neg affiliation north-east of the Orkhun River, noted as having been at war with the Uygur, these may have been a branch that migrated eastward rather than westward. Consequently, as Sümer proposed (1992: 21, 33, 35–7), most likely the Oguz of the tenth century were Turkic people of western Eurasian origin with whom in time other Turkic people from Inner Asia became affiliated. The senior lineage of the Üç Ok Oguz, the Bayandır, for instance, whose name stems from bay, meaning rich (Atalay 1992/I: 349, III: 158, 239), thus from a rich place, are considered to have been originally of the Kimek. Lastly, the Eymür clans, also of the Üç Ok Oguz, are possibly the Aymur clans of the Uygur (Divitçioğlu 1994: 12–13).

The Türkmen Until recently ‘Türkmen’ has been used almost solely to identify Turkic tribesmen who had at least nominally accepted Islam. However, recently discovered Soghdian records appear to show that ‘Türkmen’ was used much earlier in both an ethnic and a geographical sense (Necef and Berdiyev 2003: 27). According to Russian Orientalists N. A. Aristov and later D. E. Yevemeyev (in 1869 and 1970 respectively), since in Turkic languages –men and –man mean human being, as in öğretmen or teacher, Türkmen denotes a Turkic human being (Necef and Berdiyev 2003: 31–2). The earliest version of Türkmen appears in the fifth-century Chinese annals, where the reference is to a Tö-kyo Ming. In Chinese annals of the eighth and ninth centuries there is also mention of the land of the Tö-kyo Ming. Tö-kyo is thought to be from Tu-kyu or Türk, and Ming from Mengi or eternal. The Kök Türk were referred to as Türk Mengi or Eternal Türk, because their rulers were considered sacred. Consequently, it seems that Türkmen is from Türk Mengi (Türkmeng) (Necef and Berdiyev 2003: 33–4, 36). The references seem to be confirmed by eighth-century Soghdian records wherein Trwkkm’n or Türkmen is likewise used to refer to a place rather than to the Kök Türk whose empire had collapsed in 740. There appear to be further references in Chinese annals of the sixth and seventh centuries to Turkic lands bordering those of the Soghdians. These references are made with respect to the K’ang-kuo (Kanglı) or Töles (T’ieh-le, originally Ti) north of Samarkand. Interestingly enough, references in Chinese annals to Origins 51 northern Transoxania, more specifically north of the Syr-Darya River and east to the Talas and Çu rivers, incorporating the region around Issyk-Köl, as Kangku (K’ang-kuo and hence as belonging to the Kanglı or Töles) begin from the second century BCE (Necef and Berdiyev 2003: 38–9). The Ho-shih, Ho-chie, Pu-ku (Buku), Pikan (Mukan or more correctly Mu-ğan and hence the steppe region in Azarbayjan named after them), Chü-hai, Ho-pi-his, Ho-ts’o-su, Pa-ya-wei and Ho-ta of the Töles are known to have been of the On Ok or western Kök Türk and later the Türgiş who were finally dislocated by the Karluk Çiğil (also Sikil or Izgil) (Necef and Berdiyev 2003: 41). This may explain why the Karakhanids, whose main affiliates were the Karluk, were also referred to as Türkmen (Pritsak IA/6: 253ff). In short, it seems that since a group of Karluk in the land of the Türkmen (present-day Turkmenistan) were the first to convert to Islam, all subsequent Turkic tribesmen who accepted Islam came to be referred to as Türkmen (Necef and Berdiyev 2003: 42).

The Seljuqs Togan maintained that Selçuk Beg (originally Salçuk, then Selçük) was of humble Oguz origin and that his father Dukak had converted to Islam. Basing his views on a version of the Oguznâme, according to Togan, the Seljuqs’ lineage was as follows: Togrul b. Mikâ’il b. Selçuk b. Dukak b. Ertogrul b. Lukman b. Toksurmuş İlçi b. Kerükiçi Hoca (1981: 182ff). Kerükiçi Hoca is thought to have been a tent pole sharpener (Kafesoğlu 1955). Togan also thought that Dukak had been in the service of the Khazar kagan (1981: 182ff); Divitçioğlu agreed (1994: 55ff). None of the above is accepted by the main historians on the Seljuqs (Kafesoğlu 1955: 22–5 and IA/10: 353ff; Turan 1993a: 31–4; Sümer 1992: 65; Köymen 1979: 7, 30 and 1963: 23). However, it is possible the Seljuqs were of humble origin. On the steppe a commoner could merit nobility through martial prowess and leadership (Mardin 1992b: 81). This is consistent with the Oguz yabgu taking Selçuk Beg under his wing when his father died. According to the Seljuqs’ legend of origin, at the time Selçuk Beg was between seventeen and eighteen years old. Later he is said to have been appointed commander-in-chief. Divitçioğlu also thinks that Selçuk Beg’s name stems from the Çuvaş for white; şüre, şarı or sarı and thus Sarçuk (the –çuk ending being an endearment denoting little), and through an orthographic mistake, finally Salçuk. Divitçioğlu argues that this interpretation is supported by the Seljuqs’ legend of origin. According to Divitçioğlu’s interpretation, Selçuk Beg’s father Dukak objected to a campaign against a group of Muslims with such vehemence that he came to blows with the Khazar kagan’s regent, the Hakan-beh, not the Oguz yabgu. Although the Hakan- beh ordered Dukak to be captured and killed, he was ignored. Instead the parties were persuaded to reconcile their differences. According to Divitçioğlu this refers to the reported refusal by Khazar soldiery to campaign against Muslims in the Caucasus (913). A year earlier they had attacked and killed Rùs merchants who had paid to be permitted to trade in Islamdom by the Khazar kagan (1994: 55ff). 52 Origins There is no doubt that Dukak was of the Kınık Oguz (Atalay 1992/I: 55). Dukak or Tukak (possibly from the original tugag or standard-bearer (Turan 1993a: 32), thus also Dokak or Tokak) was renowned as Temir Yalığ (Iron Bow). Temir is thought to refer to his strength (Turan 1993a: 31–4). Divitçioğlu thought Dukak’s name of distinction further proof that he was an alp or warrior in the service of the Khazars (1994: 55–8, 60–3). Selçuk Beg’s sons did have Biblical names: Israil (Israel), Mikail (Michael), Musa (Moses), Yusuf (Joseph) and/or Yunus (Jonah) (Sümer 1992: 69; Köymen 1979: 31). However, rather than confirming that the Seljuqs had been in the service of the Judaic Khazar kagans, this is thought to reflect Khazar cultural pre-eminence in the region where the Kınık were situated. Selçuk Beg’s sons also appear to have had Turkic names or titles: Israil’s was Arslan, Musa’s was Inanç and Yusuf’s was Yınal (Turan 1993a: 34–7, 49). Nevertheless, it is possible the Seljuqs were for a time in the service of the Khazars, although even at its peak their kaganate does not appear to have stretched further east than the Caspian Sea. The Khazar kaganate began to unravel in the middle of the ninth century, when the Peçenek intruded over the Caspian Lowlands (860–89) (Kurat IA/9: 537), causing, among others, the Magyar to leave. The Rùs took Kiev in 866 and a century later overran the kaganate (965–9) (Togan IA/5/1: 397ff; Kafesoğlu 1994: 157ff). But the Khazar are known to have sought help against the Peçenek from the Oguz with whom they allied to drive them from between the Ural and Volga rivers (898–902) (Sümer 1992: 45). Last but not least, whether or not Dukak had been in the service of the Khazars, legend has it that the Oguz yabgu’s wife became wary of Selçuk Beg’s growing popularity as commander-in-chief and of his flagrant ambition, insisting he be killed. When Selçuk Beg heard this he left with his family and affiliates (Köymen 1979: 7–11, 1963: 23–4; Sümer 1992: 65; Kafesoğlu IA/10: 353ff; Turan 1993a: 40–2; Şeşen 1975: 37). 4 Conquest (985/993–1063)

The events covered in this chapter relate the Seljuqs’ emergence in Transoxania and the problems they faced before being forced to migrate to Khurasan, where they struggled with the Ghaznavids. After their crushing victory at Dandankan, the Seljuqs divided up Khurasan, assigning their chosen leader, Tuğrul Beg, to conquer Iraq with his step-brother Ibrahim Yınal and his cousin Kutalmış, while his uncle Musa and elder brother Çağrı Beg and their sons secured Khurasan and Sistan to the south. Tuğrul Beg was highly successful, but this led to contention with Ibrahim Yınal and Kutalmış. When Tuğrul Beg died, having been declared sultan, Kutalmış and his brother Resul Tekin were still in rebellion with the support of the Türkmen who had backed Ibrahim Yınal. After defeating him for the second time, Sultan Tuğrul had had Ibrahim Yınal and his brothers strangled with their bowstrings. The chapter follows Köymen’s Büyük Selçuklu I˙mparatorluğu – Kuruluş Devri (1979) for the Seljuqs’ early years in Transoxania and Khurasan in conjunction with Sümer’s Oğuzlar (first published in 1965) and Pritsak’s article in the IA on the Karakhanids. Köymen’s Tuğrul Bey ve Zamanı (1976) is the main source for the years between Dandankan and Sultan Tuğrul’s death. Köymen’s Selçuklu Devri Türk Tarihi (1963) has been referred to alongside other works for supplementary material pertinent to the events being reviewed, such as Kafesoğlu’s article in the IA. The regions and cities mentioned in the remainder of the text are identified on Maps A1.2 and A1.3 in Appendix 1.

Transoxania The decision to migrate to Jand, which was on the left bank of the Syr-Darya near the Aral Sea, south-east of Yengi-Kent where the Oguz yabgu wintered, appears to have been a judicious one (Köymen 1979: 18–19 and 21–3; also Kafesoğlu IA/10: 353ff and Turan 1993a: 40–5). Current archaeological work near Perowsk suggests Jand existed well before the spread of Islam into Central Asia (Roux 2001: 251). Furthermore, Yengi-Kent and Huvara, which had been founded by Muslim colonists, were inside Oguz territory whereas Jand was on the marches (Köymen 1979: 17 and 1963: 24). 54 Conquest (985/993–1063) The border lands between Islamdom and the Oguz polity stretched from Khwarazm, south of the Aral Sea, north-east to Isfijab (Şeşen 1987: 210), the region around present-day Çimkent. As such the region had nearly 1,700 ribats, namely fortifications built by notables or a city for the gazi, those from their locality warring against the infidel (Barthold 1990: 190). Transoxania as a whole was considered at the forefront of jihad, the war against non-believers (Köymen 1979: 18; see also Şeşen 1987: 210). Selçuk Beg appears to have taken advantage of this. Having arrived in Jand, he is said to have rallied the local gazis against the Oguz yabgu’s tax collector on grounds that Muslims should not pay tax to the non-believer (Köymen 1979: 24; also Kafesoğlu IA/10: 353ff and Turan 1993a: 40–5). This and similar gazi activity seems to have earned Selçuk Beg the title of al-Malik al-Gazi (Kafesoğlu IA/10: 353ff; Turan 1993a: 40–5), increasing his already substantial wealth and number of affiliates (Köymen 1979: 26, 30). Possibly as result Selçuk Beg appears to have come to an agreement with the Samanids sometime between 985 and 993. According to this the Seljuqs would defend their borders in exchange for pasturage around the village of Nur, between Bokhara and Samarkand. However, it is thought the Türkmen who took up these pastures were under the command of Selçuk Beg’s son Arslan (Israil) (Köymen 1979: 34–5; also Kafesoğlu IA/10: 353ff and Turan 1993a: 40–5). The Samanids had emerged at the beginning of the ninth century (819). Soon after they had established themselves in Transoxania, they had taken Isfijab from the Karakhanid Bilge Kül Kadir Khan (840) (Pritsak IA/6: 253ff). By the end of the century the Samanids’ hegemony extended south-west into Khurasan. After this their power had waned due to internecine warfare and the jealousies that had arisen between their Turkic slave soldiery, the palace corps’ mamluk commanders who had started to vie for power (Ibid.; see also Büchner IA/10: 140–3 and Barthold 1990: 225ff). Alp Tekin (from Tigin, then Tegin) was first to try and establish his own dynasty in Ghazna but he failed (962) (Ibid.). His mamluk Sebük Tekin, however, was successful (977), although he remained loyal to his masters until his death (997) (Ibid.). Sebük Tekin was rewarded with the governorship of Khurasan after conducting two successful campaigns against its governor, Ali Simcur, another of the Samanids’ Turkic mamluk commanders. The first campaign came after Ali Simcur had invited the Karakhanids to Transoxania on the condition that he would retain Khurasan (Dames IA/4: 742ff). Although Buğra Khan Harun of the eastern Karakhanids took Isfijab, with the consent of the local notables, and then entered Bokhara, the Samanid capital, he was forced to withdraw due to illness and died en route (992) (Pritsak IA/6: 253ff; also Köymen 1963: 27–8 and Barthold 1990: 276–9, 289). The second campaign came within a few years, when Ilek Khan Nasr of the western Karakhanids re-entered Transoxania with the encouragement of another of the Samanids’ Turkic commanders, Faik, who had allied with the defeated Simcur. Sebük Tekin once again defeated the latter (995), but this time the Karakhanids appear to have stayed. Sebük Tekin is said to have negotiated Conquest (985/993–1063) 55 a settlement according to which Faik became governor of Samarkand and the Karakhanids retained the region between the Syr-Darya River and the Katvan steppe, leaving only Bokhara to the Samanids (996). The following year Sebük Tekin and the Samanid ruler both died (997). Ilek Khan Nasr entered Bokhara and took the Samanid treasury to his capital Özkent with the remaining members of the dynasty (999) (Ibid.). The following year the Samanids’ Ismacil Muntasir appears to have retaken Bokhara but lost it again (1000). Having gone to Khurasan, Muntasir was defeated by Mahmud of Ghazna, Sebük Tekin’s son, and returned to Transoxania to seek help. One view is that he went to Yengi-Kent, where he sought assistance from the Oguz Yabgu Ali, the Seljuqs having decided to side with the Karakhanids (1003) (Sümer 1992: 68–71; see also Pritsak IA/6: 255). However, in the main Turkish historians are of the opinion that the Oguz said to have plundered Harun’s forces during his retreat from Bokhara were the Seljuqs. According to them, Muntasir travelled to Jand not Yengi-Kent, where he regained the Seljuqs’ support (Köymen 1979: 48 and 1963: 28; Sümer 1992: 68–71; Turan 1993a: 47–9; Kafesoğlu IA/10: 353ff). The basis for this seems to be the reference to the Oguz chieftain with whom Muntasir had his meeting as Peygu, which Barthold thought should be read as Yabgu (Peygu/Beygu/Yabgu) (1990: 289). Although Arslan is known to have appropriated the title of yabgu, like Çağrı and Tuğrul (originally Togrul), respectively a blue-eyed merlin, the avian emblem of the Kınık, and a crested goshawk (Esin 1994: 202), also Arslan (the lion was a Buddhist heraldic emblem that became frequent as a name or title after it was used by the Buddhist Kök Türk Taspar Kagan (572–81) (Esin 1980a: 68 and 101)), paygu was a totemic name prevalent among the Karakhanid (Pritsak IA/6: 253ff). Consequently, Peygu may be paygu rather than Yabgu. Nevertheless, it is also possible that having visited the Oguz yabgu in Yengi-Kent and failed to secure his help, Muntasir was successful with the Seljuqs. Thus, according to Turkish historians, having agreed to help Muntasir, the Seljuqs defeated the Karakhanids near Samarkand, surprising Ilek Khan Nasr and capturing eighteen of his commanders. The Seljuqs are then said to have hesitated, undecided what to do with the commanders, so that Muntasir withdrew fearing they would come to an agreement with the Karakhanids who are thought to have pursued Muntasir across the frozen Amu-Darya River into Khurasan (Köymen 1979: 49–50 and 1963: 29; Sümer 1992: 68–71; Turan 1993a: 49–52; Kafesoğlu IA/10: 353ff). Muntasir is known to have returned and been defeated again, withdrawing to Nur where he not only rallied the people of Samarkand but with Arslan’s help also surprised and defeated Ilek Khan Nasr (1004). Although there do not seem to be any further references to the Seljuqs, Turkish historians have explained this by the Seljuqs’ possible satisfaction with their plunder rather than a decision that Muntasir’s was a lost cause. Given the events, this latter would seem the more plausible. Although supported by 5,000 Oguz under Hasan Tak, Muntasir was defeated one last time when he defected to Nasr on the battlefield (1004/5) (Köymen 1979: 59–61, 100 and 1963: 29; also Sümer 1992: 68–71). 56 Conquest (985/993–1063) According to Köymen, soon after this Selçuk Beg died, and Çağrı and Tuğrul Beg were forced to leave Jand with their uncle Musa, their father Mikail having died earlier (1007–8) (1979: 97ff). Sümer (1992: 71) and Turan (1993a: 47) disagreed. They were of the opinion that Musa and his nephews’ departure came as a result of pressure by the Baran clan. Although Sümer did not consider the Baran to be of the Oguz, Turan thought them affiliated to the Yazır. Moreover, according to Turan, Baran was not only the name of the ruler of Yengi-Kent, but also the father of Ali Khan and hence Shah-Malik’s grandfather (1993a: 47–52). The latter is known to have reigned in Jand by at least 1030, eventually forcing the Seljuqs from Transoxania. For his part, Kafesoğlu did not offer any comment as to when Selçuk Beg may have died or why the Seljuqs left Jand (IA/103: 53ff). With the exception of Sümer (1992: 76), Turkish historians are also agreed that after finally defeating Muntasir, Ilek Khan Nasr gathered his forces with the intention of driving the Seljuqs out of Transoxania. To counter this, Çağrı and Tuğrul Beg are said to have approached the Karakhanid Buğra Khan near Talas. However, Tuğrul Beg is thought to have been taken hostage by Buğra Khan. According to this version of events, in a counter move Çağrı Beg surprised Buğra Khan’s soldiery and having captured his commanders traded them for Tuğrul Beg. Next Çağrı Beg is said to have taken 3,000 mounted archers with him to Anatolia, leaving Tuğrul Beg behind (Köymen 1979: 96–104 and 1963: 32; see also Turan 1993a: 49–52 and Kafesoğlu IA/10: 353ff). Originally proposed by Yınanç and accepted by all Turkish scholars except Sümer (1992: 76), Çağrı Beg’s expedition to Anatolia is considered to have lasted five years (1016–21). Once there Çağrı Beg is said to have enlisted the aid of the Türkmen in the region and attacked the Armenian kingdom of Vaspurakan (1018– 19). After this he and the Türkmen are thought to have attacked the Shaddadid dynasty. Having sacked Georgia, they were denied by the Armenians in Ani (near present-day Kars), finally separating to go their own ways (1021). With his pres- tige enhanced, on his return Çağrı Beg is said to have rallied the Türkmen around Bokhara. When Tuğrul Beg rejoined him, Arslan is said to have advised his nephews to disperse, in order not to appear to threaten the Karakhanids in Transoxania (Yınanç 1944: 35–6; see also Köymen 1979: 104–14 and 1963: 33, Kafesoğlu IA/10: 353ff; Turan 1993a: 52–3). What appears more likely is that the Seljuqs had allied with the western branch of the Karakhanids after initially having helped Muntasir against the eastern Karakhanid Ilek Khan Nasr. Buğra Khan was the head of the western Karakhanids. Rather than having been taken hostage, therefore, Tuğrul Beg may have established a lasting friendship with their leader Yusuf, who later became Great Kagan (1026/7–1032) (Pritsak IA/6: 251ff). This interpretation also helps to put into perspective the Seljuqs’ subsequent relations with Yusuf’s youngest brother Ali Tekin. First, however, it is necessary to review relationships between the Karakhanids and the Ghaznavids after the demise of the Samanids. Having defeated Muntasir for the last time, Ilek Khan Nasr is known to have sent two columns into Khurasan (1005–6). Sultan Mahmud of Ghazna and his brother Nasr defeated these forces, Conquest (985/993–1063) 57 one of which might have been commanded by Yusuf’s brother Ali Tekin. As a result, Ilek Khan Nasr is said to have requested help from his cousin Yusuf but Sultan Mahmud crushed their combined forces outside Balkh (1008). Their defeat appears to have thrown the Karakhanids into disarray, prompting Nasr to declare independence in Transoxania (1008–10) (Pritsak IA/6: 251ff). On Ilek Khan Nasr’s death, his brother Mansur is thought to have replaced him (1012/13). Soon after this Mansur appears to have occupied the capital Özkent and declared himself Great Kagan in place of his ailing brother Ahmed (1015/16). It is at this time that Ali Tekin, who had become the ruler of Bokhara (1014/15), is thought to have been imprisoned by Mansur. However, it seems that Ahmed was not done; he fought an indecisive battle with Mansur outside the capital (1016/17). Subsequently, still referred to as Great Kagan, Ahmed is thought to have pursued and crushed the folk displaced by Sultan Mahmud’s conquest of Khwarazm (1017/18). On Ahmed’s death that same year, the western Karakhanid Yusuf claimed the title of Great Kagan, which is thought to have been rightfully his. Although Yusuf and Mansur appear to have reached a compromise, Yusuf’s brother Ahmed objected, occupying Özkent and then Ahsikash (1019/20) (Ibid.). It is not clear whether Ali Tekin escaped captivity or was freed, but he appears to have taken back Bokhara with the Seljuqs’ help (1020/1), marrying Arslan’s daughter. He also seems to have come to an understanding with Mansur, recognizing him as Great Kagan rather than his brother Yusuf. This appears to have enabled him to acquire Samarkand as well. When Mansur withdrew to become a dervish (1024/5), Yusuf’s brothers Ahmed and Ali are said to have rebelled. Ahmed declared himself Great Kagan and Yusuf sought Sultan Mahmud’s help. This is thought to have suited Mahmud because of Ali Tekin’s presence in Bokhara and Samarkand. Sultan Mahmud crossed the Amu-Darya River and met Yusuf outside Samarkand where they agreed that Yusuf’s second son Muhammad should replace Ali Tekin and that the Karakhanids and the Ghaznavids should exchange daughters in marriage. Immediately afterwards Sultan Mahmud appears to have tricked Arslan into captivity, not pursuing Ali Tekin with whom Arslan had withdrawn into the desert (Ibid. see also Köymen 1979: 68–78). Sultan Mahmud is said to have sent an envoy to Arslan, inviting him to discuss the defence of Khurasan. When Arslan duly arrived at court, Mahmud asked how strong a force he thought he could muster. Reportedly Arslan claimed his followers totalled 30,000 but that he could call also on his uncles Musa and Yusuf who commanded another 10,000. When Sultan Mahmud pressed him for more, Arslan is said to have bragged that he could call on 100,000 horsemen from Balkhan, a mountain situated between the Kara-Kum Desert and the Caspian Sea, below Üst- Yurt, and if need be on a further 200,000 from Turkmenistan. Whatever the truth of Arslan’s claims, Mahmud is known to have imprisoned him in the fortress of Kalinjar, in India, where he died (1032) (Köymen 1979: 78–89 and 1963: 31; see also Kafesoğlu IA/10: 353ff; Sümer 1992: 70–1; Turan 1993a: 52; Barthold 1990: 292–5, 299–304). However, relations between Yusuf and Mahmud did not continue smoothly. Sultan Mahmud refused to honour the marriage agreement, claiming Yusuf 58 Conquest (985/993–1063) should first put his house in order. Yusuf’s brother Ahmed finally capitulated after a series of defeats (1026–9), but Sultan Mahmud died, as a result of which Ali Tekin appears to have regained full control of Transoxania. Having acceded to his father’s throne in Ghazna with the help of Arslan Beg’s affiliates, Sultan Mascud now decided to implement his father’s agreement with the Karakhanid Yusuf and sent his governor of Khwarazm, Altun-Taş, against Ali Tekin. Altun- Taş was defeated and mortally wounded at Dabusiya by Ali Tekin and the Seljuqs (1032). Altun-Taş’ son Harun then sought and gained Ali Tekin’s support against Sultan Mascud, but Ali Tekin died (1034) (Pritsak IA/6: 251ff). After Arslan’s imprisonment Ali Tekin is thought to have given Yusuf the Seljuqs’ leadership. According to Turkish historians, Çağrı Beg had to restrain Tuğrul Beg from attacking their uncle as Musa was the eldest. In this interpretation of events, Ali Tekin then had Yusuf killed and attacked the Seljuqs who suffered heavy losses (1029). Kafesoğlu and Turan thought that as a result the Seljuqs withdrew to Khwarazm, but Sümer and Köymen were of the opinion that Çağrı and Tuğrul Beg had been wintering there for some time. Sümer was also of the opinion that Yusuf had gained the leadership of the Seljuqs, but on merit. He dismissed Çağrı and Tuğrul Beg’s alleged conflict with Ali Tekin. Instead he thought that after Ali Tekin’s death the Seljuqs had fallen out with his commander-in-chief and regent Alp Kara and that he had Yusuf killed. In support he pointed out that the Seljuqs are thought to have fought with Ali Tekin against Altun-Taş. This would appear to be the more likely. Turan thought Alp Kara was of the Baran, possibly Ali Khan’s brother, which he thought explained the sudden enmity of the Karakhanids toward the Seljuqs (Köymen 1979: 121–8, 131–60 and 1963: 34–5, 37 and Sümer 1992: 72 and 76–7; see also Kafesoğlu IA/10: 353ff, Turan 1993a: 52, 56–7 and Barthold 1990: 315–17). Upon Yusuf’s death, it is thought that the Karakhanid Yusuf may have helped the Seljuqs withdraw from Transoxania to Khwarazm (Pritsak IA/6: 251ff), but without success. En route the Seljuqs were surprised by Shah-Malik and are thought to have lost 7,000–8,000 men, countless women and children being taken captive. They are said to have been so badly depleted that the gazi of the Nemek ribat on the northern banks of the Amu-Darya River considered killing them. When Sultan Mascud had their new ally, the Khwarazmshah Harun assassinated, the Seljuqs were forced to cross from Khwarazm into Khurasan (1035) (Ibid.). In order to better understand the ensuing developments it is necessary first to follow Arslan Beg’s affiliates’ (the so-called Türkmen of Iraq) fortunes after their master’s imprisonment. They are known as the Türkmen of Iraq in Turkish historiography, because of their adventures in western Iran, Iraq al-Ajam. They are also referred to as the Türkmen of Balkhan, the Yabgulu or the Navekkiye (men of the quiver).

The Türkmen of Iraq On Arslan’s incarceration some if not all of his affiliates, reportedly 4,000 families, are said to have asked Sultan Mahmud for pasturage in Khurasan in return Conquest (985/993–1063) 59 for their martial services. As will be remembered this is the offer Sultan Mahmud is thought to have made Arslan. The leaders of these Türkmen, namely Yağmur, Buka, Gök-Taş and Kızıl are said to have claimed persecution by Çağrı and Tuğrul Beg. Sultan Mahmud appears to have granted their request despite objections from Arslan Jadhib, the governor of Tus (1025–6). The Türkmen of Iraq appear to have been situated in the vicinity of Nesa, Baverd and Ferave along the Atrek River, which flows from the Köpetdag Mountains into the Caspian Sea, south of the Kara-Kum Desert and the Balkhan Mountain. During the following two years other Türkmen from the region around the Balkhan Mountain appear to have joined them, straining relationships with the local administration and inhabitants. The Türkmen are said to have allowed their flocks to graze in the farmers’ fields and gone off with their animals. On the other hand, when they were unable to pay their taxes, allegedly collectors confiscated their livestock and sold their children into slavery. The worsening situation appears to have led Sultan Mahmud to command Arslan Jadhib to restore public order. Despite several attempts, Jadhib seems to have failed and asked Sultan Mahmud to take the field. Sultan Mahmud defeated the Türkmen near the Ferave ribat and is said to have killed 4,000 men, women and children (1028) (Köymen 1979: 165–74 and 1963: 39–40; see also Sümer 1992: 73; Kafesoğlu IA/10: 353ff; Turan 1993a). Having cleared Khurasan of the Türkmen, Sultan Mahmud then took Rayy, outside present-day Tehran, appointing his son Mascud governor (1029). However, when Sultan Mahmud died the next year (1030), Mascud revolted against his brother Muhammad who had acceded to the throne. Having invited Yağmur back to Khurasan, Mascud was able to gain the throne of Ghazna from his brother. On Yağmur’s instigation, Sultan Mascud now allowed Kızıl, Buka and Gök-Taş back as well. Although the Türkmen of Iraq are said to have served Sultan Mascud well on campaign, their conduct appears to have remained a cause for complaint around Serakhs and Baverd, where they were situated. When they resisted a Ghaznavid force sent to restore order, Sultan Mascud is said to have ordered his commander-in-chief in Iraq al-Ajam, Taş-Faraş, to press them into service under Humar-Taş, their campaign commander. On his way back from Iraq, Yağmur was executed with fifty other Türkmen leaders, which suggests they had resisted (1033). Despite Sultan Mascud’s order that they too were executed, Buka and Gök- Taş, who were in Rayy at the time, appear to have survived along with Yağmur’s son and Kızıl, both of whom were in Balkhan. The Türkmen of Iraq now took up arms against the Ghaznavids. When other Türkmen from Transoxania joined them, Sultan Mascud was forced to take the field (1034) (Köymen 1979: 174–97 and 1963: 40–1; also Sümer 1992: 74–5 and Kafesoğlu IA/10: 353ff). It is at this juncture that the Seljuqs entered Khurasan with Ibrahim Yınal, their step-brother. According to custom, Çağrı and Tuğrul Beg’s paternal uncle Yusuf had married their mother upon Mikail’s death. The Seljuqs are said to have crossed into Khurasan with only 900 mounted archers but soon attracted more (reportedly 10,000). Three envoys, representing Musa Beygu, Çağrı and Tuğrul Beg presented a letter to the Ghaznavid governor of Khurasan, requesting pasturage around Nesa, Baverd and Ferave. As will be remembered, this is 60 Conquest (985/993–1063) where Sultan Mahmud and also Mascud had situated Arslan’s followers. The Seljuqs’ letter clearly stated their reasons for coming to Khurasan – Ali Tekin’s and the Khwarazmshah Harun’s deaths. They offered to send members of their dynasty in rotation to the sultan’s palace as hostage for good behaviour. When Sultan Mascud’s minister received the letter, he is reported to have exclaimed that Khurasan was lost as until then they had been dealing with the Türkmen, not their masters (Köymen 1979: 197–201 and 1963: 41; also Sümer 1992: 77 and Kafesoğlu IA/10: 353ff). Sultan Mascud responded to the Seljuqs’ request by sending an army equipped with war elephants. Although his commander Beydoğdu objected that the chain of command Sultan Mascud set up was too complicated, he appears to have been disregarded. As the Ghaznavid advance units came upon the Seljuqs’ camp, despite Beydoğdu’s orders to push on they appear to have started to kill the shepherds, and women and children were left behind, taking their animals and tents. As a result, Beydoğdu is said to have been forced to lead the Ghaznavid centre through the mêlée to the village of Nesa, where the Seljuqs were lying in wait. During the battle that followed, some of the more inexperienced Ghaznavid commanders seem to have tried to water their horses, thirsty from the hot desert sand, causing their soldiery to panic, thinking they were being abandoned. The Seljuqs routed the Ghaznavids, plundering their baggage train (29 June 1035). Sultan Mascud capitulated without further action, giving Dihistan to Çağrı Beg, Nesa to Tuğrul Beg and Ferave to Musa Yabgu. The Seljuqs did not send their members to the palace in rotation and are said to have laughed at the presents sent by the sultan. Their success seems to have attracted fresh groups to Khurasan, including a branch of the Türkmen of Iraq – though which is not specified. Within four months complaints that livestock were being grazed in the fields and animals were being taken reached Ghazna. In response Sultan Mascud appears to have deployed a force of 15,000 men from Nishapur. The Seljuqs now requested three urban centres, namely Marv, Serakhs and Baverd, claiming their numbers had grown (1036). In return, they offered to clear Khurasan of marauding bands. When their request was not met, having instead been appointed dihkans or notable land owners who acted as tax collectors, the Seljuqs embarked on a series of raids (Köymen 1979: 201–37 and 1963: 42–5; also Sümer 1992: 78–9 and Kafesoğlu IA/10: 353ff). Apparently encouraged by the Seljuqs’ success, Buka and Gök-Taş Beg attacked and routed the Ghaznavid forces under Taş-Faraş, whom they dismembered, going on to sack and plunder Rayy (1036). Buka, Gök-Taş and two others, Oğuzoğlu Mansur and Dana, then migrated to Azarbayjan, possibly at the invitation of Vahsudan of the Rawwadid dynasty, known to have been at odds during this period with al-Khaija of the Hezebaniyye clan. In view of their relentless plundering, however, Vahsudan, his adversary al-Khaija and the local populace united against the Türkmen, who withdrew to Iraq al-Ajam (1037). Having married Vahsudan’s sister or daughter, however, Dana appears to have stayed behind. Next, Gök-Taş and Mansur besieged Hamadan, while Buka joined Kızıl outside Rayy. Having sacked Rayy again (1038), Buka and Kızıl joined Gök-Taş with whom they also sacked Hamadan (1038). Kızıl appears to have Conquest (985/993–1063) 61 married Tuğrul and Çağrı Beg’s sister at some point thereafter and settled in Rayy, where he died (1041) (Sümer 1992: 80–2; also Yınanç 1944: 37–8).

Victory at Dandankan Although strongly advised to do so by his council of ministers, Sultan Mascud decided not to lead his forces against the Seljuqs. Arguing he had taken the requisite defensive measures throughout Khurasan, the sultan instead embarked on a campaign into India (1037), where he learned of the sacking of Rayy by the Türkmen of Iraq and of the Seljuqs’ raids on urban centres in the environs of Jurjan. Insisting on the wisdom of his decision, he ordered the force based in Nishapur to engage the Seljuqs. They did so, but as his ministers had feared lost the battle (1038). The Seljuqs now divided up northern Khurasan; the regions around Serakhs, Marv and Nishapur were apportioned respectively between Musa Yabgu, Çağrı and Tuğrul Beg. More importantly, Tuğrul Beg was chosen as leader, or more correctly appointed primus inter pares. He is said to have entered Nishapur with 3,000 mounted archers, a drawn bow in his hand and three arrows in his belt. Since the Seljuqs had sent only two envoys with their request for Marv, Serakhs and Baverd, the suggestion is that Musa Yabgu had lost any authority he had over Tuğrul and Çağrı Beg (Köymen 1979: 238–78 and 1963: 46–8; also Sümer 1992: 82 and Kafesoğlu IA/10: 353ff). Upon receiving news of his army’s defeat, Sultan Mascud appears to have finally decided to take the field against the Seljuqs. However, once in Balkh he is said to have prevaricated, undecided whether to deal with the Karakhanid Böri Tekin who had been raiding Khurasan (Barthold 1990: 20). Having decided to go ahead, he was prevented from gaining an advantage by the severe winter conditions and forced back. In Balkh he received news that Çağrı Beg had skirmished with his soldiery near Pul-i Karvan. Sultan Mascud advanced and met Çağrı Beg there, but the Seljuqs did not commit their forces. They sent out a company of archers and when these tired replaced them with another. Frustrated by these tactics, Sultan Mascud took command and advanced on the Seljuqs, whereupon they are said to have melted away into the desert (April 1039). Çağrı Beg is considered to have been assessing the Ghaznavids’ strength and commitment under Sultan Mascud. The outcome seems to have proved crucial at a council meeting where Tuğrul Beg and Ibrahim Yınal suggested they evacuate Khurasan for the environs of Rayy and Jurjan, from where it would be possible to raid Azarbayjan and Anatolia at will. Apparently Çağrı Beg objected, arguing that Sultan Mascud would follow them there, too. According to Çağrı Beg the Ghaznavids were superior in number, equipment and supplies, but their dependence on the latter and their reliance on war elephants and shock troops on foot slowed them down, making them vulnerable to the hit-and-run tactics of steppe warfare. Tuğrul Beg and Musa Yabgu are said to have also expressed concern for the loyalty of the Turkic mamluks who had defected, but once again Çağrı Beg appears to have prevailed. He is said to have argued that they had deserted because their commanders had been executed on Sultan Mascud’s orders and proposed they should be put at the 62 Conquest (985/993–1063) vanguard to prove their allegiance (Köymen 1979: 278–90 and 1963: 49; also Sümer 1992: 83 and Kafesoğlu IA/10: 353ff). The Seljuqs held their ground when Sultan Mascud advanced on Serakhs. They are said to have harried and harassed the Ghaznavid forces, preventing their cavalry from procuring sufficient water and pasturage. When he tried to engage in battle at the end of Ramadan, the Seljuqs withdrew (14–21 June 1039). On his war council’s advice, Sultan Mascud offered the Seljuqs peace on the condition they returned to Nesa, Baverd and Ferave. Although they are said to have accepted, the Seljuqs did not evacuate Serakhs, Marv and Nishapur, or send an envoy to Herat as agreed. During this period the Seljuqs are thought to have strengthened their alliances with the Khwarazmshah and Böri Tekin whom they helped to defeat Ali Tekin’s sons. Meanwhile the Oguz migration into Khurasan seems to have also intensified. When an old woman with one eye, one hand and one leg, who was wielding an axe, was asked why she had come, she is said to have replied that she wanted a piece of the treasures she had heard were being unearthed (Köymen 1979: 290–315 and 1963: 49–51; also Sümer 1992: 83–4 and Kafesoğlu IA/10: 353ff). At the end of autumn Sultan Mascud took command of a lightly equipped force to counter the Seljuqs’ superior manoeuvring ability and the second phase of their war commenced (November 1039). Despite countless skirmishes, however, Sultan Mascud appears once again to have failed to engage the Seljuqs who withdrew to Balkhan Mountain. Although there was famine in Khurasan, due to drought and the constant warfare, Sultan Mascud retired to Nishapur where he had his troops trained in desert tactics (January 1040). In the spring he advanced on the Seljuqs via Tus (March 1040). The Seljuqs again considered whether to leave Khurasan, but once more Çağrı Beg prevailed. Sultan Mascud was not able to stay in Tus, due to the desperate shortage of supplies, and attacked Serakhs. Although he took the city, the Ghaznavids appear to have lost countless animals and foot soldiers en route (May 1040). The sultan was advised to return to Herat, where there was ample food, water and pasturage, but instead decided to push on to Marv in pursuit (May 1040). This proved disastrous as the drought had dried all the wells. The Seljuqs continued to harry the now abject Ghaznavid forces. Although Sultan Mascud reached the fortification of Dandankan, due to a lack of sufficient water there, too, he ordered his forces on to a nearby reservoir. Despite the sultan’s and his son Mawdud’s personal attempts to marshal their forces against the Seljuqs’ raids, when a group joined their former comrades, their army lost its discipline and they were routed (23 May 1040). During this last encounter, which is said to have never progressed into a set piece battle, the Seljuq leaders remained behind the lines until informed of Sultan Mascud’s flight. Thus it was the former Ghaznavid mamluks who claimed victory and took most of the plunder. In open appreciation of their success, Tuğrul Beg is thought to have promised them land in the environs of Rayy and Isfahan (Köymen 1979: 312–45 and 1963: 52–3; also Sümer 1992: 84–6, Kafesoğlu IA/10: 353ff and Turan 1993a: 63–6). Conquest (985/993–1063) 63 Seljuqid sovereignty According to Köymen the nature and purpose of the newly founded Seljuq polity can be deduced from the words attributed to Tuğrul Beg at the council held in Marv later that month (May 1040). These are that world dominion cannot be achieved without unity. Once a letter had been sent to the caliphate requesting the legitimization of Seljuqid rule, in keeping with Turkic ideology, the Seljuqs apportioned among the dynasty their newly gained realm and those to be conquered. Çağrı Beg took Marv, thus the major part of Khurasan. Musa Yabgu was given the task of conquering Herat and Sistan. Later that year Ibrahim Yınal’s brother Er-Taş was assigned to Musa Yabgu with this purpose. Çağrı Beg’s eldest son Kara-Arslan Kavurt was given the separate task of conquering Kirman. In the West, Tuğrul Beg took upon himself the conquest of Iraq al-Ajam and al-Arab (western Iran and Mesopotamia) from Nishapur, with his subordinates Ibrahim Yınal, Çağrı Beg’s son Alp-Sungur (more commonly referred to as Yakuti) and Arslan’s eldest son Kutalmış. Ibrahim Yınal was assigned the conquest of Kuhistan (Jibal), Kutalmış that of Jurjan north-east of Tabaristan. The khutba referred to Tuğrul and Çağrı Beg respectively as Sultan and Malik; theirs was now senior branch of the Seljuq dynasty (Köymen 1979: 345–66 and 1963: 53–5, 100; also Kafesoğlu IA/10: 353ff). Selçuk Beg’s only living son, Musa, now referred to as Yabgu Kalan (more correctly Kelan from Yabgu-i kelan, literally the remaining yabgu (Turan 1993a: 49)), took Herat and from there Er-Taş incorporated Sistan, whose ruler al-Fadl proclaimed allegiance (November 1040). Er-Taş then fell out with Yabgu Kalan and returned to Khurasan because Musa wanted to appropriate Sistan from him. Meanwhile, when he had decided to take the field against the Seljuqs, Sultan Mascud had offered Khwarazm to Shah-Malik on the condition he proclaimed allegiance (1038). Shah-Malik, however, had not moved until after Sultan Mascud’s demise at Dandankan. Now Shah-Malik defeated Altun-Taş’ son Ismacil, who withdrew to Gurganj (February 1041). However, on learning that a Ghaznavid army was on its way, the Khwarazmshah Ismacil left Gurganj and took refuge with Çağrı Beg (March 1041). Shah-Malik entered Gurganj unaware that Sultan Mascud had been killed (January 1041) and had the khutba read in the latter’s name (April 1041). Shah-Malik at first defeated Çağrı Beg and the Khwarazmshah Ismacil, but when Tuğrul Beg joined them he was routed and Gurganj fell (1042). The Seljuqs did not restore Khwarazm to Ismacil and instead a governor was appointed. Er-Taş had returned to Sistan upon news from al-Fadl that Sultan Mawdud of Ghazna had sent an army to Khwarazm. He defeated this army (July 1042) and later captured Shah-Malik on his way to Ghazna with his family and treasury (1042/3). Er-Taş handed him over to Çağrı Beg who had Shah-Malik imprisoned and killed. (Köymen 1963: 221; also Sümer 1992: 88, Kafesoğlu IA/10: 353ff, Barthold 1990: 322–3 and Turan 1993a: 69–71). When Çağrı Beg became ill the following year (1043/4), his son Alp-Arslan (thought to have been born on 20 January 1029) is said to have defeated another army sent by Sultan Mawdud. Alp-Arslan conquered Tirmiz and its environs, 64 Conquest (985/993–1063) immediately north of Balkh and the Amu-Darya, successfully defending these against the Karakhanid Arslan Khan who was forced to recognize the Seljuqs’ sovereignty along the Amu-Darya from Tirmiz to Bokhara (1047). Alp-Arslan then helped Çağrı Beg in his unsuccessful attempts to conquer Ghazna. When Ibrahim acceded to the throne of Ghazna, however, Çağrı Beg made peace (1059). As the Hindu Kush became the agreed border between them, the Ghaznavids lost communication with Transoxania and the Karakhanids. Çağrı Beg died the following year at the age of seventy (1060) (Köymen 1972: 3–7 and 1963: 222; also Kafesoğlu IA/10: 364 and Turan 1993a: 69–71). Before being called back by Çağrı Beg to help fight Shah-Malik, Tuğrul Beg had incorporated Jurjan and Tabaristan. When Kızıl Beg died, Ibrahim Yınal took Rayy and Tuğrul Beg moved his capital from Nishapur to Rayy (1042). The Seljuqs’ success is thought to have attracted more Türkmen. Possibly it was their depredations that caused Caliph al-Qa’im (1031–75) to send his chief judge, al-Mawardi. Tuğrul Beg is said to have countered that the Seljuqs’ realms had grown too small for their number, referring to the Türkmen as ‘my soldiers’ (1043/4). It is possible, however, that these were the Türkmen of Iraq. On Kızıl’s death Tuğrul Beg had sent an envoy to Gök-Taş, Buka, Oğuzoğlu Mansur and Anasıoğlu who had refused his invitation, stating their suspicion that he wanted to trick them into captivity. Presumably unable to return to Khurasan because Ibrahim Yınal had cut off their retreat, or to Azarbayjan where Vahsudan had had thirty Türkmen leaders killed at a feast in their honour (1041), the Türkmen of Iraq are said to have moved toward Lake Urmia, immediately west of Tabriz. Here others already in Azarbayjan appear to have joined them and together they entered eastern Anatolia. When the Byzantine governor general refused them passage, they seem to have headed south. With Mansur possibly busy elsewhere, Gök-Taş, Buka and Anasıoğlu defeated a force gathered by the Uqailid and Marwanid dynasties. Nevertheless, the head of the latter dynasty, Nasr al-Dawla, appears to have tricked Mansur into captivity. Although released to Gök-Taş, Buka and Anasıoğlu, with many goods in exchange for their departure, the Türkmen apparently reneged. While Buka and Anasıoğlu raided the environs of Diyarbakır (Amid), Gök-Taş and Mansur took Mosul, having the khutba read in Tuğrul Beg’s name (1043). Their rough treatment of the inhabitants caused the Buyid ruler, who was in Baghdad at the time, to complain with the Marwanid Nasr al-Dawla. Tuğrul Beg seems to have promised to bring them under control, but before he could do so the local Arab and Kurdish rulers rallied to the Uqailids and together they defeated Gök- Taş and Mansur who returned to the environs of Diyarbakır (April 1044). There they received an envoy from Tuğrul Beg who asked that instead of raiding within Islamdom they should move to Azarbayjan and join in the jihad against the Byzantines. En route they defeated a Byzantine force near Lake Van and captured its commander, the governor who had refused them passage. He was sold on the slave market in Tabriz (1045). Although there is no further record of Gök-Taş and Mansur, Buka and Anasıoğlu are said to have fought when drunk outside Diyarbakır, which they had been assigned to conquer, and died Conquest (985/993–1063) 65 from their wounds (Köymen 1963: 160–1 and 241–2; also Sümer 1992: 88–9, Kafesoğlu IA/10: 364, Yınanç 1944: 40–4 and Sevim 1990b: 2–3). The Türkmen’s Anatolian campaigns appear to have begun by joint raids with the Shaddadid dynasty on the Armenian Kingdom of Ani and Georgia. The Byzantines, who had pursued a policy of annexation in eastern Anatolia since the beginning of the eleventh century, sent an army to Ani and another against the Shaddadids (1045). Tuğrul Beg is said to have responded by dispatching Kutalmış with Yabgu Kalan’s son Hasan and Çağrı Beg’s son Yakuti. Kutalmış defeated a Byzantine army near Erzurum, deep in eastern Anatolia, under the command of the Georgian Prince Liparit (1046). The assignment of dynastic members to lead the Türkmen from the Mukan steppe in Azarbayjan appears to have been supported by a redirection there of other Türkmen. Apart from those who had threatened to sack Nishapur (1047), there appears to have been also a large group from Transoxania who offered Ibrahim Yınal their services (1048). When Hasan and Yakuti tried to expand the Türkmen raids southward, they were met by a Byzantine army commanded by Katakalon and Aaron, resulting in the death of Hasan (1047/8). This appears to have prompted Tuğrul Beg to appoint Ibrahim Yınal governor of Azarbayjan and assign Kutalmış to him (1048). By now the regions immediately south of the Caspian Sea, Dailam and Jibal had been incorporated fully into the Seljuq realms and Ibrahim Yınal’s name was being read after Tuğrul Beg’s in the khutba. Ibrahim Yınal defeated both Prince Liparit and another Byzantine army under Katakalon and Aaron, which had been reinforced by Prince Liparit, at Hasan-Kale, on the plain of Pasin near Erzurum (September 1048). Prince Liparit is said to have been among the prisoners Ibrahim Yınal sent Tuğrul Beg in Rayy. It is not clear whether Emperor Constantine Monomachus or Tuğrul Beg then sued for peace. Nonetheless, this seems to have fallen through when Tuğrul Beg demanded an annual tribute once paid to the Abbasids (1050). However, Constantine Monomachus appears to have agreed to repair the mosque in Istanbul (Constantinople) and have the khutba read in Tuğrul Beg’s name (Köymen 1963: 163, 232, 245–8; also Sümer 1992: 88–9, 95, Kafesoğlu IA/10: 364–5, Yınanç 1944: 45–8, Sevim 1990b: 3–4 and Turan 1993a: 79–85). After this Ibrahim Yınal became the most renowned commander in Iraq al-Ajam. Although he had conquered Rayy, Tuğrul Beg had taken the city for his capital. When Tuğrul Beg asked also for Hamadan and the fortresses he had conquered in Iraq al-Ajam as well, Ibrahim Yınal refused. Said to have held his vizier responsible for the request, he left to gather his forces after he had him tortured. Although Tuğrul Beg defeated Ibrahim Yınal, who retired to the fortress of Sarmaj, he appears to have pardoned him (Köymen 1963: 59). According to Yınanç (1944: 49–50), when Ibrahim Yınal was defeated, Kutalmış fled to eastern Anatolia and laid siege to Kars. If true, not only does Kutalmış appear to have joined Ibrahim Yınal but was likewise forgiven. After dealing with Ibrahim Yınal and possibly Kutalmış, Tuğrul Beg is thought to have spent the next four years consolidating his conquests in Iraq al-Ajam. Toward the end of 1054 he appears to have once again taken command of the raids into Anatolia, incorporating into his realms the Rawwadid, Shaddadid and 66 Conquest (985/993–1063) Marwanid dynasties as vassals. Having failed to take the fortress of Malazgirt (Manzikert), near Lake Van, despite help from the Armenians, he is said to have advanced deep into eastern Anatolia where he annihilated a Byzantine army commanded by Gagik. On the invitation of Caliph al-Qa’im, Tuğrul Beg then left for Baghdad (1055), assigning Yakuti to co-ordinate the raids into Anatolia from Azarbayjan (Köymen 1963: 248–51; also Kafesoğlu IA/10: 366, Yınanç 1944: 49–50 and Sevim 1990b: 14).

Ruler of the East and the West It is thought that by sending his chief judge al-Mawardi as his envoy to Tuğrul Beg, Caliph al-Qa’im’s real aim had been the establishment of peaceful relations between the Seljuqs and the Buyids (1043/4). On receiving Tuğrul Beg’s envoy in return, Caliph al-Qa’im had given him a diploma finally recognizing the Seljuqs’ sovereignty over Khurasan (1046/7). That year the Buyid ruler Abu Kalijar married his daughter to Tuğrul Beg and took one of Çağrı Beg’s daughters for his son Mansur. However, when he died the following year the Buyids began to war. One of their members, Khusraw, took refuge with Tuğrul Beg and married a Seljuq princess. It is at this juncture that the Buyids’ Turkic commander responsible for Baghdad and its environs, Arslan al-Basasiri, began to gain influence, such that the khutba came to be read in his name throughout Lower Mesopotamia (1050). After receiving Tuğrul Beg’s envoy bearing gifts and money (1052), Caliph al- Qa’im appears to have declared the Fatimids heretics (1052/3) (Köymen 1963: 168–9, 172; see also Kafesoğlu IA/10: 366 and Kitapçı 1994: 21) Matters with al-Basasiri appear to have come to a head when the Uqailids took the city of Enbar, which was under his command, and secretly sent two envoys to Baghdad (November/December 1054). When the caliph’s vizier prevented their arrest, al-Basasiri is said to have imprisoned both him and the caliph accusing them not only of keeping in touch with the Seljuqs but encouraging them to come to Baghdad. When al-Basasiri then took back Enbar, his soldiery in Baghdad was attacked and personal wealth plundered (November 1055). Although Caliph al- Qa’im is said to have requested al-Basasiri to be withdrawn from Baghdad, his departure appears to have more to do with the Buyid ruler al-Malik al-Rahim’s proclamation of allegiance to the Seljuqs. Ironically this seems to have prompted most of al-Malik al-Rahim’s Turkic soldiery to defect to al-Basasiri (Köymen 1963: 169–73; also Kafesoğlu IA/10: 366 and Kitapçı 1994: 21). Tuğrul Beg entered Baghdad at the head of a 60,000-strong force that included eight war elephants (25 December 1055). The size of his force is taken to indicate that he did not expect a warm welcome. Events appear to bear this out. When the following day a soldier tried to buy some straw for his horse he is thought to have been misunderstood because he did not speak Arabic. Thinking instead that he wanted to appropriate the bales free of charge, a crowd is said to have formed and attacked the Seljuq troops billeted in their houses and the Buyids’ Turkic garrison. Although initially Shicite members of the public are said to have tried to protect the Seljuq forces, some of the Buyids’ Turkic garrison joined the rioting. When the Seljuqs regained Conquest (985/993–1063) 67 control many inhabitants were enslaved and part of Baghdad sacked. Tuğrul Beg appears to have taken the opportunity to blame the rioting on the Buyids and used Caliph al-Qa’im to trick al-Malik al-Rahim into captivity (Köymen 1963: 175–7; also Kafesoğlu IA/10: 366, Sümer 1992: 89–90 and Turan 1993a: 92–3). Now fully in charge, Tuğrul Beg seized the treasury, minted money and appointed his own commander to ensure public order in Baghdad, increasing the caliph and his council’s annual allowances. Caliph al-Qa’im also married Çağrı Beg’s daughter, Khadija Arslan Khatun. Finally, Tuğrul Beg had a palace built on the outskirts of Baghdad, with villas for his commanders and barracks for their soldiery, hoping to prevent further disturbances. In the meantime al-Basasiri seems to have added to his soldiery with the help of the Fatimids. When he defeated Kutalmış at the head of a Türkmen force near Mosul (December 1056), Tuğrul Beg advanced with Ibrahim Yınal and Yakuti. Al-Basasiri withdrew to Syria. After a number of minor conquests and proclamations of allegiance, Tuğrul Beg appointed Ibrahim Yınal to Mosul and its environs, giving him the requisite robes of honour and 20,000 dinars. On his return to Baghdad, Tuğrul Beg was crowned and proclaimed Ruler of the East and the West by Caliph al-Qa’im, a title not decreed before and one in accord with Turkic ideology (January 1058). Turkish historians take this as confir- mation that Sultan Tuğrul now had sole sovereignty over worldly matters in Sunni Islamdom (Köymen 1963: 59, 177–80 and 1976: 102; also Kafesoğlu IA/10: 366–7, Sümer 1992: 90 and Turan 1993a: 93–5).

Seljuqid discord While Tuğrul Beg was thus occupied, Kutalmış’ brother Resul-Tekin is said to have rebelled (1057). More importantly, a rumour spread that Ibrahim Yınal was preparing to revolt. Having defeated Resul-Tekin, Sultan Tuğrul sent Ibrahim Yınal gifts, inviting him to Baghdad (April 1058). While Ibrahim Yınal was in Baghdad, the caliph’s vizier seems to have approached Sultan Tuğrul seeking assurance that forgiven once, Ibrahim Yınal would not be forgiven again if he rebelled. Having been reassured, the vizier appears to have called Ibrahim Yınal and informed him of Caliph al-Qa’im’s goodwill and satisfaction concerning his allegiance. Although no reason is given by Turkish historians for the caliphate’s involvement, there is the caliph’s reported reluctance to allow Sultan Tuğrul to take the field against al-Basasiri. On the caliph’s suggestion, therefore, Sultan Tuğrul is said to have entrusted Ibrahim Yınal to deal with al-Basasiri. Once back in Mosul, however, under the guise of inviting al-Basasiri to proclaim his allegiance, Ibrahim Yınal is said to have secretly requested money, robes of honour, titles and a banner from the Fatimid caliphate, which were granted (August 1058). In exchange Ibrahim Yınal promised to have the Shicite khutba proclaimed throughout his realms once he had dethroned Sultan Tuğrul. Alerted that Sultan Tuğrul was advancing on Mosul, Ibrahim Yınal left for Hamadan. Sultan Tuğrul sent his wife and vizier back to Baghdad, hurrying to head off Ibrahim Yınal (Köymen 1963: 60–1; also Sümer 1992: 90, Kafesoğlu IA/10: 367, Yınanç 1944: 52 and Turan 1993a: 95–6). 68 Conquest (985/993–1063) The sultan is said to have reached Hamadan first (23 November 1058), preventing the money, weapons and supplies stocked there from falling into Ibrahim Yınal’s hands. Although the region was his stronghold, Ibrahim Yınal had to promise the Türkmen he would never make peace with the sultan, or force them to campaign in Iraq al-Arab, or appoint a vizier without consulting them. Initially defeated, Sultan Tuğrul returned to Hamadan, from where he sent word to his wife and his vizier Kunduri (10 December 1058). Despite the caliph’s objection, Altuncan Khatun took the treasury and Seljuq soldiery in Baghdad with her, leaving the vizier who is said to have attempted to crown Altuncan Khatun’s son in place of Sultan Tuğrul during the following three months. When Ibrahim Yınal learnt of Altuncan Khatun’s approach he seems to have sent a force to have her intercepted. In the hope of gaining plunder, other Türkmen are said to have joined them, such that the sultan was able to make a sortie, defeat Ibrahim Yınal’s forces and retire to Rayy, where his wife joined him, having managed to disengage from the Türkmen sent to intercept her by leaving behind her personal baggage train. After this setback Ibrahim Yınal is said to have requested help from his nephews, Muhammad and Ahmad, who reportedly turned up with 30,000 men, their father Er-Taş having died in 1048/9. The sultan advanced with the reinforcements his wife had brought. Ibrahim Yınal defeated him, but when Çağrı Beg’s sons Kavurt, Yakuti and Alp-Arslan joined Sultan Tuğrul, Ibrahim Yınal lost the battle outside Rayy and was captured (July 1059). In accordance with custom, the sultan had Ibrahim Yınal strangled with his own bowstring. Muhammad and Ahmad were also killed. Meanwhile, al-Basasiri took Mosul and then Baghdad, sacked the caliph’s palace and placed him in captivity elsewhere while he embarked on the conquest of Basra. However, he withdrew on learning that Sultan Tuğrul was approaching Baghdad. The caliph was released voluntarily by his captor and the sultan set off after al-Basasiri, whom he defeated and had killed (January 1060) (Köymen 1976: 62–3 and 67–70 and 1963: 61–2; also Sümer 1992: 90–1, Kafesoğlu IA/10: 367 and Turan 1993a: 97–8). It appears that Kutalmış and his brother Resul-Tekin had sided with Ibrahim Yınal. When Sultan Tuğrul dispatched an army against them, they withdrew to the fortress of Girdkuh, near Damghan, from where they are said to have repulsed Sultan Tuğrul’s forces with the help of the deceased Ibrahim Yınal’s Türkmen (April–May and August 1061) (Köymen 1963: 62–4; also Sümer 1992: 92, Kafesoğlu IA/10: 367–8 and Turan 1993a: 98–100). Sultan Tuğrul is said to have secured the release of the caliph’s wife, Arslan Khatun, before attempting to have the caliph freed. When the caliph’s wife sent an envoy to request that Arslan Khatun return to Baghdad, Sultan Tuğrul dispatched the chief jurist for Rayy, requesting to marry the caliph’s daughter. Although Caliph al-Qa’im refused, eventually he seems to have accepted on the condition the sultan resided in Baghdad, having handed over Wasit, 300,000 dinars in compensation and the deceased Altuncan Khatun’s real estate. However, by the time Arslan Khatun returned to Baghdad with Kunduri to finalise the arrangements (April 1061), the caliph appears to have had a change of mind. In subsequent discussions, first with Arslan Khatun and then Kunduri, Caliph al-Qa’im is said Conquest (985/993–1063) 69 to have threatened to leave Baghdad if the sultan insisted. As a result Kunduri is thought to have made preparations to depart, prompting Baghdad’s chief jurist and a member of the caliph’s retinue to warn the caliph of the dangers of refusing the sultan’s request. Although apparently persuaded by them to go through with the marriage, the Hanafi and Shafici jurists in Baghdad now issued a fatwa against the marriage. The caliph once again changed his mind and Kunduri left (17 July 1061). When Sultan Tuğrul instructed the governor of Iraq to confiscate his iqtacs, however, the caliph changed his mind one last time and sent an envoy to Rayy, agreeing to marry his daughter to the sultan (22 August 1062). The marriage ceremony took place in Baghdad (17 February 1063). Despite her protests, Sultan Tuğrul is said to have taken his new wife with him on leaving Baghdad with the caliph’s permission (13 April 1063). Kunduri once again laid siege to Girdkuh (May 1063), but the sultan died soon afterwards (September 1063) (Köymen 1963: 190–3; also Kafesoğlu IA/10: 367–8 and Turan 1993a: 101–2). 5 Empire (1063–92)

The events in this chapter cover the reigns of Sultan Alp-Arslan and his son Malik-Shah. During his short reign Sultan Alp-Arslan first secured the Caucasus and north-eastern Anatolia for the Türkmen raiding Anatolia from the Mukan steppe in Azarbayjan. Soon after the Türkmen appear to have established a base in Ahlat, near Lake Van in eastern Anatolia, precipitating the Byzantine response that led to Emperor Romanos Diogenes’ defeat and capture by Sultan Alp-Arslan at nearby Malazgirt (1071). At the beginning of Sultan Malik-Shah’s reign (1072), Kutalmış’ sons Mansur, Süleyman-Shah, Devlet, Alp-Ilek and another appear in Anatolia, where they take up leadership of the Türkmen. Using the divisions that followed the crushing defeat at Malazgirt, Mansur and Süleyman-Shah appear to have quickly wrested Anatolia from the Byzantines. However, in the process only Süleyman-Shah was left alive. Having made peace with the resurgent Byzantines (1081), Süleyman- Shah was killed in battle with Sultan Malik-Shah’s son Tutuş while trying to expand from Antakya into north Syria (1084). By the end of Sultan Malik-Shah’s reign, south-eastern Anatolia, Syria, Bahrain, Mecca and Medina had all been incorporated into the Great Seljuq Empire, marking its highpoint. Both Sultan Alp-Arslan and Malik-Shah were challenged by Çağrı Beg’s eldest Kara-Arslan Kavurt. Although he appears to have had the support of the Great Seljuqs’ mamluk commanders, Sultan Malik-Shah finally defeated Kavurt with the help of his vizier Nizam al-Mulk. Having been captured, Kavurt was strangled with his own bowstring on Nizam al-Mulk’s orders (1092). The last years of Sultan Malik-Shah’s reign were dominated by his struggle with Nizam al-Mulk, which was acted out between his wife Terken Khatun and his commanders on the one side and the vizier and his sons on the other. Matters came to a head when Sultan Malik-Shah died soon after Nizam al-Mulk was assassinated by an adherent of Hasan al-Sabbah’s Ismacili Nizari sect. Turkish historians in particular have blamed the interregnum that followed the sultan’s death on his wife Terken Khatun, going so far as to accuse her of having the sultan poisoned. In support of their view they point to her negotiations with the caliph, as a result of which in return for his son Jacfar from Sultan Malik-Shah’s daughter Mah Melek Khatun, the caliph agreed to proclaim Terken Khatun’s infant son Mahmud sultan despite the existence of an heir apparent, Berk-Yaruk, Nizam al-Mulk’s original Empire (1063–92) 71 choice. Since Nizam al-Mulk had been replaced by her former vizier Taj al-Mulk and the royal corps assigned to Üner, likewise her commander-in-chief, Terken Khatun is depicted as a meddlesome woman of unbridled ambition. What is more likely, however, is that the caliph had Sultan Malik-Shah poisoned after having refused to name Jacfar his heir apparent. Angered by the refusal, the sultan had given him a fortnight to vacate Baghdad. The chapter follows Köymen’s Alp-Arslan ve Zamanı (1972) for Sultan Alp-Arslan’s reign and Kafesoğlu’s Sultan Melikşah Devrinde Büyük Selçuklu I˙mparatorluğu (1953) for Sultan Malik-Shah’s reign. Although Köymen mentions that Kafesoğlu’s work needed to be updated in light of new material that had emerged since its publication (Köymen 1979: 12*), this has not happened. Kafesoğlu’s work is augmented with Merçil’s Kirman Selçukluları (1980), Turan’s Selçuklular Zamanında Türkiye (1993c) and Sevim’s Suriye Selçukluları I – Fetihten Tutuş’un Ölümüne Kadar (1965) for Kirman, Anatolia, and Syria and Palestine respectively. As in the previous chapter, references to other works are for supplementary material pertinent to the events being reviewed.

Unification of the empire A fortnight after learning of Sultan Tuğrul’s death, Caliph al-Qa’im is said to have invited the Seljuqs’ Arab and Kurdish vassals to discuss the future of the region. The Seljuq representative in Baghdad, Sacid Qa’ini, seems to have protested but then apologized on learning that both Kutalmış and Alp-Arslan had refused to recognize Süleyman. Having dropped Sultan Tuğrul’s name from the khutba (17 September 1063), the caliph is said to have sent an envoy to Qa’ini demanding he return his iqtacs and depart. Concurrently the caliph appears to have replaced officials appointed to his diwan by Sultan Tuğrul with others of his choice and lifted non-Shari taxes. Presumably to legitimize his actions, the caliph also asked the jurists in Baghdad for a fatwa concerning conduct in the face of disobedience and rebellion. They appear to have complied by ruling the caliph should wage war against those culpable, presumably Kutalmış and Alp-Arslan (Köymen 1972: 11–12, 96–100). However, contrary to the caliph’s expectations, which can be said to have centred on driving out the Seljuqs with help from the local Arab and Kurdish rulers, these latter appear to have tried to improve their positions at the caliph’s expense. The Uqailid Muslim is said to have pillaged the caliph’s palace, withdrawing only after receiving additional lands from him (25 December 1063). More likely, however, given that the date coincides with news of Alp-Arslan’s victory over Kutalmış, Muslim may have thought it best to vacate the Seljuqs’ palace where he had taken up residence. During this period Seljuq dominion in Iraq al-Ajam also appears to have been challenged. In Hamadan, for instance, the inhabitants are said to have attacked officials, killing them and 700 of their men (Köymen 1972: 12, 100–1). Süleyman was the youngest of Çağrı Beg’s sons, but perhaps more importantly Sultan Tuğrul had married his mother on Çağrı Beg’s death. Childless, Sultan 72 Empire (1063–92) Tuğrul is thought to have appointed Süleyman heir apparent on Çağrı Beg’s wife’s instigation and with his vizier Kunduri’s support. According to Köymen, Kunduri may have wanted to become the de facto ruler of the Great Seljuq Empire. If so, he appears to have failed at the first hurdle. On Süleyman having been declared heir apparent, one of his mamluk commanders, Erdem, is said to have left to join Alp-Arslan. Another, Yağı-Sıyan, also appears to have left. Once in Qazvin, they are said to have had the khutba read in Alp-Arslan’s name. Nevertheless, having distributed money and suits of armour to secure the remaining commanders’ allegiance to Süleyman, Kunduri is said to have sent Alp-Arslan a letter in which he stated that he was merely executing the sultan’s will and would not hesitate to take up arms if challenged, advising Alp-Arslan to make do with his father’s inheritance (Köymen 1972: 14–15). When Kutalmış arrived in Sawa, however, some of the mamluk commanders are said to have accepted his invitation. Encouraged, Kutalmış is said to have advanced at the head of a large force of Türkmen and irregulars who raped and pillaged the surrounding countryside. Arriving outside Rayy (16 November 1063), Kutalmış appears to have defeated Kunduri’s advance guard under Inanç Beg, capturing him with his soldiery. Besieged, the vizier appears to have sent an envoy to Alp-Arslan, proclaiming his allegiance and seeking help. Receiving news that Alp-Arslan’s advance units had arrived in Damghan, under the command of Erdem (24–25 November 1063), Kutalmış is said to have attacked and forced him to seek refuge in a nearby village. Despite having flooded the valley below the village to hinder Alp-Arslan, Kutalmış appears to have lost the ensuing battle and died in flight when he fell off his horse (7 December 1063). On receiving news that Alp-Arslan was on his way to Rayy (25 December 1063), Süleyman seems to have gone to Shiraz where Kara-Arslan Kavurt had the khutba read in his name (Köymen 1972: 16–17, 82–5). Alp-Arslan does not appear to have been in any hurry to gain an audience with the caliph, embarking instead on a campaign in the Caucasus (22 February 1064) and sending an envoy to request that the khutba be read in his name (6 March 1064), which appears to have been effected sometime later (11 May 1064). Nonetheless, the caliph seems to have continued to resist the Seljuqs. Prior to the arrival of the sultan’s envoy, the governor of Iraq, Nihavandi, is said to have sent a deputy to Baghdad who appears to have been unsuccessful in re-establishing Seljuq authority. Sometime after, when Nihavandi took up residence in Baghdad with his shihna (commander responsible for enforcing official decisions such as those by a jurist or tax collector), the caliph’s vizier is said to have objected that Seljuq officials were interfering with the administration and taxation of his and his officials’ iqtacs. Having complained to the caliph, Nihavandi appears to have confiscated the iqtacs in question when al-Qa’im failed to support him. The caliph is said to have appealed to Nizam al-Mulk in Khurasan. Although reprimanded, Nihavandi does not seem to have taken notice as at the time Kunduri was still the sultan’s vizier. Nihavandi does not appear to have returned the iqtacs until after news of Sultan Alp-Arslan’s conquest of Ani (September–October 1064), but to no avail. Soon afterwards Nizam al-Mulk is known to have replaced Kunduri Empire (1063–92) 73 (29 December 1064), installing Sacid Fasi instead of Nihavandi. Sacid Fasi appears to have taken up his three-year tenure in exchange for money. The caliph is said to have dismissed his vizier on grounds he received robes of honour from the sultan without his permission, taking care to blame him for inviting the Uqailid Muslim to Baghdad. When Sultan Alp-Arslan insisted the caliph appoint al-Ala as his vizier, however, al-Qa’im appears to have reinstated him. Nevertheless, the sultan seems to have sent al-Ala to Baghdad as his deputy, decreeing half of the caliph’s vizier’s iqtacs to him (December 1071) (Köymen 1972: 19, 35, 103–6).

The Caucasus and Transoxania During the years prior to Sultan Tuğrul’s death, Türkmen begs such as Dinar, Kapar, Cemcem, Tuğ-Tekin and Salar-i Khurasan are thought to have continued raiding eastern Anatolia under Yakuti, who is said to have defeated a Byzantine force under the command of Briennios with the help of Sabuk Beg (1057). This appears to have enabled the Türkmen to besiege Kars and Ani before joining with others to raid as far as Erzincan, while yet more Türkmen under Dinar Beg sacked Malatya (Melitene) (1058). The following year, on Sultan Tuğrul’s orders, the Türkmen are also said to have once again entered Anatolia under Yakuti’s direction. Salar-i Khurasan seems to have failed to take Urfa (Edessa), but Sabuk Beg was able to sack Sivas (Sebasteia) the year after that (1060). Following this a Byzantine army under Pankaras is said to have been defeated (1061). Before his death Sultan Tuğrul is thought to have visited Azarbayjan to review Yakuti’s conduct of operations (1062) (Sevim 1990b: 5–6; Turan 1993a: 109). Yakuti appears to have died the same year as Sultan Tuğrul, as having arrived in Azarbayjan Sultan Alp-Arslan is said to have divided his forces into two on Tuğ-Tekin’s, not Yakuti’s advice (February 1064). As a result Sultan Alp-Arslan entered Georgia and having conquered the region south-west of Tbilisi turned back toward Kars, in north-eastern Anatolia (July 1064). The force commanded by Sultan Alp-Arslan’s nine-year-old son Malik-Shah, doubtless with his atabeg Nizam al-Mulk’s help, appears to have taken the fortresses along the Aras River south of Kars. Having gained the Armenian kingdom of Lori’s allegiance, Sultan Alp-Arslan was joined by Malik-Shah. They continued to sack the fortified towns in their path, devastating the countryside before finally laying siege to Ani. When the city fell it too is said to have been pillaged without remorse and torched. Lastly, the sultan is thought to have assigned the conquered realms to his commanders (August 1064) (Köymen 1972: 24–34; see also Kafesoğlu IA/10: 368 and 1953: 9–12; Turan 1993a: 110–13; Sevim 1990b: 7–8; Yınanç 1944: 57–9). Having thus secured Azarbayjan and Armenia, Sultan Alp-Arslan appears to have returned to Khurasan to quell the ageing Yabgu Kalan’s or his son’s revolt in Herat. There is also mention of a campaign elsewhere in Khurasan against two vassals who likewise had rebelled. Both their lands are said to have been incorporated. In Marv, Sultan Alp-Arslan married his son Malik-Shah with a Karakhanid princess, the niece of the ruler of Samarkand, and apportioned the now united Seljuq realms among remaining members of the dynasty. He is said to 74 Empire (1063–92) have assigned Balkh, Khwarazm, and Tukharistan and Chaganiyan to his brothers Süleyman, Arslan-Argun and Ilyas respectively. His sons Arslan-Shah and Togan-Shah were assigned Marv and Herat. Er-Taş’ remaining sons Mawdud and Mascud were assigned Bagshur and Isfizar. Finally, Inaç or Inanç Yabgu (possibly Musa’s son or his regent) was assigned Mazandaran. The absence of Yakuti’s name seems to confirm his death. Also, Sultan Alp-Arslan’s sons Ayaz and Ismacil are not mentioned. From Marv, Sultan Alp-Arslan seems to have gone to Khwarazm, where the Türkmen are said to have been raiding passing caravans with the Kıpçak. When these took refuge on an island, the sultan is said to have taken their women and children prisoner, confiscated their livestock and subordinated the Kıpçak chieftain. Moving on to Jand, he is said to have visited his grandfather’s tomb, receiving the allegiance of its Kıpçak ruler (April 1066). On his return to Marv, the sultan is said to have appointed Malik-Shah heir apparent (May 1066) (Köymen 1972: 75–8, 83–4 and 90; also Kafesoğlu IA/10: 368 and 1953: 12–15 and Turan 1993a: 113–16). While the sultan was thus occupied in Khurasan, said to have been encouraged by his vizier, Kavurt appears to have dropped Alp-Arslan’s name from the khutba and minted coins. This is taken to suggest his intentions were limited to establishing an independent polity rather than challenging his brother for the throne, but this was not their first encounter. Immediately prior to Sultan Tuğrul’s death, Kavurt had taken Fars in south Iran from the Buyids’ former vassal, the Shebankarid dynasty (1062). When the Shebankarids complained to Sultan Alp- Arslan, Kavurt was ordered to return Fars. The consensus among historians is that the sultan wanted to reduce Kavurt’s realm because he had advanced only as far as Isfahan on Sultan Tuğrul’s death, withdrawing when Sultan Alp-Arslan had defeated Kutalmış. Encouraged by Sultan Alp-Arslan’s support, the Shebankarids had attacked Shiraz but were repulsed (1064). On being informed, Sultan Alp- Arslan had advanced on Shiraz and Kavurt had withdrawn to a fortress by the sea, sending his treasury and family back to Kirman (January 1065). Although their forces skirmished, Sultan Alp-Arslan had not pursued the issue and Kavurt had withdrawn from Fars. This time, however, the sultan marched on Kirman via Isfahan (June/July 1067). When his advance units were defeated and fled the fortified city of Bardasir where he had taken up position, Kavurt sent an envoy proclaiming allegiance and asking to be forgiven. Sultan Alp-Arslan received him at court and reinstated him. Matters did not end there, however, for Kavurt now changed tactics. Having allied with the Shebankarids, he ensconced himself in one of their fortresses. Sultan Alp-Arslan again advanced from Isfahan (October 1068). When he arrived in Shiraz the Shebankarid ruler’s brother asked to be entrusted with the conquest of the fortresses in Fars. Instead, Sultan Alp- Arslan appointed his vizier Nizam al-Mulk to the task and marched on Kirman to besiege Bardasir, where Kavurt had withdrawn after the falling out among the Shebankarids. On learning that Nizam al-Mulk had captured his ally, once again Kavurt sent an envoy to proclaim allegiance and ask forgiveness. As it turned out, however, while besieged, Kavurt had gained the allegiance of the sultan’s commanders. They were killed, but leaving Malik-Shah in Bardasir with a small force, Empire (1063–92) 75 the sultan hurriedly evacuated Kirman (October 1069) (Merçil 1980: 32–43; Köymen 1963: 65–8 and 1972: 85–7; Kafesoğlu IA/10: 353ff; Turan 1993a: 116). After his withdrawal, Sultan Alp-Arslan is said to have received a letter of allegiance from Kavurt’s son Sultan-Shah who asked for an audience. Sultan-Shah appears to have offered to take Kirman from his father. Although entrusted with a force comprising of mamluk commanders and Türkmen irregulars, Sultan-Shah was defeated by his father (1070) (Merçil 1980: 43–4; Köymen 1963: 68 and 1972: 88–9).

The war for Anatolia The Byzantine Empire had recovered from earlier losses of territory under the Macedonian dynasty founded by Basil I (867–86). At the end of Basil II’s reign (976–1025) its domains stretched from the Alps to the Caucasus and included southern Italy. However, central and eastern Anatolia had been devastated by warfare with the Iranians and then Arabs, its depleted regions being resettled from time to time with Slavs, Bulgars, and Armenians. For instance, during the incorporation of Armenia (1020–2), it appears 40,000 people were forcibly resettled in Sivas and Kayseri (Sevim 1983: 7–10). Quite apart from any cultural disjunction such resettlements may have introduced, Basil II’s reliance on a new group of families loyal only to him seems to have encouraged the concentration of power in fewer hands, aided by the replacement of the old aristocracy with civilian administrators. This process coincided with attacks on Byzantine realms by the Uz and the Peçenek in the Balkans, the Normans in Italy and the Seljuqs in Anatolia, coming to a head under Constantine IX Monomachus (1042–55). During his reign, the new nobility who are all said to have established private armies, rebelled (1041–7). Although the rebellions were suppressed, it is thought to have further weakened the empire. There were the Ducas in Bythnia, the Comnenus in Paphlagonia, the Dalassenos in Armeniakon, the Diogenes in Harisanda, the Botaniates in Anatolik and the Phocas in Cappadocia. Of these families Romanos Diogenes had been successful against the Peçeneks in the Balkans. As a result he became emperor through marriage to the Empress Eudoxia (January 1068), but appears to have been unable to undertake any fundamental changes due to the empress’ firm grip on the administration. In particular, forces stationed in Anatolia seem to have been reduced to plundering in order to clothe and feed themselves. Romanos IV Diogenes crossed into Anatolia now under pressure from the north, east and south under these circumstances (Sevim 1990b: 13–15, 21; Yınanç 1944: 62; Kafesoğlu IA/7: 242ff and 1953: 77; see also Cahen 1984: 79–80 and Karolidis 1993: 17–19, 37–41). After his campaign in the Caucasus, Sultan Alp-Arslan had appointed Gümüş- Tekin to lead the raids into Anatolia (1066). He and the Türkmen beg Bekçioğlu Afşin set off from Ahlat on the north-western shore of Lake Van, taking many small fortresses situated between the Murat and Tigris rivers. Afşin seems to have extended his raids towards the Euphrates where he is said to have defeated and captured the Byzantine general Aruandanos, whom he ransomed. When they 76 Empire (1063–92) returned to Ahlat Afşin appears to have killed Gümüş-Tekin for killing his brother. Possibly afraid of Sultan Alp-Arslan’s wrath, Afşin returned to Anatolia (August 1067), sacking Kayseri (Cæsarea) and raiding as far as Konya (Iconium) before withdrawing to Aleppo. A Byzantine force sent under the command of Botaniates dispersed without engaging him. The following year, while Romanos Diogenes prepared his army Afşin laid siege to Antakya (April 1068), but on receiving the sultan’s pardon, raised the siege in exchange for money and war materials. At the same time Sultan Alp-Arslan entered Georgia with Nizam al-Mulk and his senior commander, the eunuch Sav-Tekin. Under pressure from the Kimek (Kıpçak/ Kuman) migration, the Alan, Komuk, Sabir and various remnants of the Khazars had been pushed toward the Caucasus. The sultan’s campaign may have proved only partially successful as the Karakhanid ruler’s death and Kavurt’s rebellion forced him to turn back (Köymen 1972: 40–2; also Kafesoğlu IA/7: 242ff, Turan 1993a: 116–21, Sevim 1990b: 19–22 and Yınanç 1944: 62–5). Possibly encouraged by Afşin’s siege of Antakya, a group of Türkmen thought to have been able to field between 500 and 1,000 men affiliated to a certain Harun took two Byzantine fortifications toward Antakya with the support of Arab contingents (July 1068). Referred to as Hanoğlu, literally the son of a khan, Harun is thought to have migrated from Central Asia or further east still, having fallen out with his father who may have been from a branch of the Karakhanid dynasty. He appears to have been active between Aleppo and Diyarbakır since 1063, selling his services to the Marwanids in Diyarbakır and the Mirdasids in Aleppo, as well as the Byzantines. Harun seems to have helped the Mirdasid Mahmud take Aleppo from his nephew (1065) (Sevim 1965: 19–28). The army that Romanos Diogenes had gathered is said to have consisted of Greeks – mainly from his home region of Cappadocia – Macedonians, Bulgars, and those Uz and Peçenek settled in the Balkans, and mercenaries recruited from the Franks, the Germans, the Varank and the Normans. In view of Afşin’s devastating raids the previous year, Romanos Diogenes first moved toward Syria (March 1068). Before he arrived in Kayseri, however, he learnt that Niksar, north- east of Kayseri toward the Black Sea, had been sacked. Leaving his army in Sivas under the command of Ducas, Romanos Diogenes advanced on the Türkmen with his best troops. Surprised, they are said to have left behind most of their captives and plunder. The emperor then marched back south. Has Inal, who had been raiding along the Euphrates, seems to have withdrawn so that the emperor pressed on to Aleppo, where he is said to have ransacked the countryside. When the Uqailid Mahmud paid him off Romanos Diogenes laid siege to Manbij (Ierapolis), one of the most important fortifications in northern Syria. Mahmud appears to have retaliated, harrying and harassing the Byzantine reserves around Aleppo with Harun’s help. When the forces sent to rescue them were routed, Romanos Diogenes was forced to return to the environs of Aleppo. Mahmud and Harun seem to have encircled him but he broke out, returning to Manbij (20 November 1068). Said to have been unable to supply his forces because of the devastation he had inflicted earlier and fearful he would be cut off, Romanos Diogenes appears to have withdrawn toward the Toros Mountains, retaking the fortifications that had fallen to Empire (1063–92) 77 Harun earlier that year. Here the emperor is said to have been informed that Afşin had penetrated as far as the Sakarya River, destroying the city of Amorion en route. Although he tried to cut off Afşin’s retreat the emperor seems to have been unsuccessful. When he returned to Istanbul, he learnt that the Italian Crispino was extorting money on the grounds that local Byzantine administrators were disrespectful. Crispino defeated the forces sent against him under the Bulgarian Alufasianos. When Romanos Diogenes marched on him Crispino surrendered but was pardoned in view of the parlous circumstances in Anatolia (Köymen 1972: 42–3; also Sevim 1965: 24–7 and 1990b: 23–4, Yınanç 1944: 65–6, Kafesoğlu IA/7: 242ff, Turan 1993a: 123ff and Karolidis 1993: 44–9). Throughout that winter the Türkmen begs, thought to have been Bekçioğlu Afşin, Sabuk, Ahmad-Shah, Uvakoğlu Atsız and his brother Çavlı, Türkman, Demleçoğlu Mehmed, Duduoğlu, Serhenkoğlu, Tavtavoğlu, Tarankoğlu and Arslan-Taş are said to have continued to raid south-eastern Anatolia from Aleppo. After Easter, Romanos Diogenes divided his forces into three. Manuel and Philaretos Brachamios were sent to Sivas and Malatya respectively while the emperor marched to Kayseri. There he found the Türkmen from Ahlat positioned in the surrounding hills. In the engagement that followed many of the Türkmen begs were taken prisoner and killed despite offers of ransom. When Romanos Diogenes advanced to take back the fortresses between the Euphrates and Tigris rivers, however, Afşin attacked Malatya. Defeated, Philaretos joined the emperor still in Harput, immediately to the north-east. Afşin and the Türkmen begs did not lay siege to Malatya. Instead, they raided deep into central Anatolia, sacking Konya. Romanos Diogenes returned to Kayseri to cut off the Türkmen retreat but Afşin crossed the Toros Mountains and retired to Aleppo (Köymen 1972: 43–5; also Kafesoğlu IA/7: 242ff, Turan 1993a: 123ff, Sevim 1990b: 24–5, Yınanç 1944: 66–7 and Karolidis 1993: 49–52). The following year Romanos Diogenes remained in Istanbul (1070). He is thought to have been pressured to assign Manuel to the defence of Anatolia. Having situated himself in Kayseri, Manuel was forced to move his headquarters to Sivas given the Türkmen raids along the Pontic Mountains parallel to the Black Sea. When the Türkmen based in Aleppo entered south-eastern Anatolia, however, Manuel was ordered by Romanos Diogenes to send an army against them. Manuel did not realise these Türkmen were under the command of Sultan Alp-Arslan’s brother-in-law Er-Basgan, who had rebelled with Kavurt the previous autumn. Manuel engaged them on the banks of the Kızılırmak River between Sivas and Kayseri. Er-Basgan feigned retreat and the Byzantine forces separated in pursuit, severing their chain of command. Er-Basgan captured Manuel with some of his generals. Afraid of Sultan Alp-Arslan’s wrath, Er-Basgan continued westward, raiding along the Toros Mountains. On learning that the sultan had sent Afşin in pursuit, Manuel is thought to have persuaded Er-Basgan to seek refuge in Istanbul. Afşin is said to have set up camp in the hills overlooking the Bosporus and sent an envoy to Romanos Diogenes demanding he hand over Er-Basgan and his Türkmen. Whether he went as far as the Bosporus or not, Afşin appears to have returned to Azarbayjan empty-handed (Köymen 1972: 45–6; also Kafesoğlu 78 Empire (1063–92) IA/7: 242ff, Turan 1993a: 128–31, Sevim 1990b: 22–4, Yınanç 1944: 68 and Karolidis 1993: 52–4).

Victory at Malazgirt Romanos IV Diogenes embarked on his last campaign, as Sultan Mascud had done, with the intention of taking the fight to the Seljuqs (13 March 1071). In addition there were now also Kıpçak, Khazar, Goth, Bulgar, Georgian, Armenian and Rùs mercenaries. The army is thought to have numbered close to 100,000 with the inclusion of some 30,000 infantry, 3,000 supply wagons and a siege machine that required 1,200 people to operate. While still on the banks of the Sakarya River, however, he appears to have ordered back those generals he did not trust, such as the experienced Botaniates. Nonetheless, at a war council held in Sivas, Briennios and those who remained, such as the Georgian Trachaniotes, are said to have tried to change Romanos Diogenes’ mind, arguing it was better to defend Anatolia from Sivas, but if the emperor was set on engaging the Seljuqs this would be accomplished best on the Pasin Plain east of Erzurum. With the support of the younger generals, Romanos Diogenes is said to have insisted the Seljuqs and hence the Türkmen would be driven back into Transoxania, Byzantine governors general appointed to Khurasan, Iraq al-Ajam and al-Arab, and mosques replaced with churches (Kafesoğlu IA/7: 242ff; Yınanç 1944: 71; Süer and Çakın 1981: 16; Turan 1993c: 21–32 and 1969: 132–4; Sevim 1993: 219ff). On the way Romanos Diogenes is thought to have put down a rebellion by the Franks. While procuring supplies in Erzurum, he is said to have sent back Er-Basgan and his son to Istanbul, lest they desert, and deployed a small force to assist the Bagratid dynasty regain Georgia from the Seljuqs. In order not to be outflanked while attacking the fortress of Malazgirt, he is thought to have assigned the Norman commander Roussel (referred to also as Urselio and/or Ruselio) to his rearguard with Trachaniotes, ordering them to advance on Ahlat. Although the commander of Malazgirt surrendered, most of the Seljuq soldiery seem to have been put to the sword (Sevim 1993: 219ff; Dirimtekin 1993: 230ff; Kafesoğlu IA/7: 242ff; Yınanç 1944: 71–2; Turan 1993c: 21–32 and 1969: 134ff; see also Karolidis 1993: 56–7). Romanos Diogenes may have sent a letter to the sultan, offering Manbij in return for Ahlat and Malazgirt, or simply demanding these and another fortification. Whatever the content, Sultan Alp-Arslan is said to have received the Byzantine envoys outside Aleppo, en route to Egypt at the invitation of the Fatimid vizier. This meeting is thought to have occurred on the day the sultan learnt of the Byzantine advance. Although what passed between the Byzantine envoys and the sultan is not known, it is further assumed Romanos Diogenes gained the impression that Sultan Alp-Arslan had withdrawn toward Iraq al-Ajam. This is then taken to explain the deployment of his forces prior to the fall of Malazgirt, which are considered to have been militarily unsound (Sevim 1993: 219ff; Dirimtekin 1993: 230ff; Kaymaz 1993: 261; Kafesoğlu IA/7: 242ff; Yınanç 1944: 71; Turan 1993a: 134ff). Empire (1063–92) 79 In view of the difficulty of procuring supplies en route, or because many of their mounts drowned, Sultan Alp-Arslan is thought to have crossed the Euphrates with only 15,000–20,000 horsemen, including 4,000 mamluks (27 April 1071). Before arriving in Mosul, however, the sultan is assumed to have doubled back toward Urfa, proceeding north to Ahlat over Diyarbakır. Whatever the route, clearly the Byzantines were unaware of the sultan’s approach. When Roussel and Trachaniotes were engaged outside Ahlat by its commander Sabuk Beg and withdrew, Romanos Diogenes sent a reconnaissance unit under the Armenian Vasilakios. Sabuk Beg feigned retreat, surrounding and capturing him. Sent to help, Briennios appears to have managed to return wounded. Vasilakios is said to have been taken to Sultan Alp-Arslan’s headquarters, thought to have been established 15–20 km east of Malazgirt, in the hills overlooking the Zaho (also Zehve, Zahva or Rahve) Plain (23 August 1071). In view of Romanos Diogenes’ subsequent advance on Ahlat, Sabuk Beg’s successful skirmishes are thought to have cloaked Sultan Alp-Arslan’s manoeuvre around Malazgirt. When he learnt of Sultan Alp-Arslan’s true position, however, the emperor is said to have returned, sending word for Roussel and Trachaniotes to join him. If so, they appear to have decided it was safer to put as much distance as possible between themselves and the Seljuqs (Sevim 1993: 219ff; Dirimtekin 1993: 230ff; Kafesoğlu IA/7: 242ff; Süer and Çakın 1981: 17, 20–1; Turan 1993c: 21–32; Yınanç 1944: 72–3; Köymen 1972: 48–9; see also Karolidis 1993: 57–60). On the way, Kutalmış’ sons Süleyman and Mansur are thought to have joined Sultan Alp-Arslan along with the Türkmen begs who had been raiding Anatolia. Although probable, it seems unlikely that as a result the Seljuqs’ forces more than doubled to 40,000–55,000, including 10,000 Kurdish auxiliaries from vassals in south-eastern Anatolia and those Uz and Peçenek who deserted the Byzantines before the battle. Despite these desertions, those forces ordered back to Istanbul, the one sent to Georgia and those with Trachaniotes and Roussel, the Byzantine army is thought to have still numbered nearly 100,000, which no doubt is equally exaggerated (Sevim 1993: 219ff; Dirimtekin 1993: 230ff; Kafesoğlu IA/7: 242ff; Yınanç 1944: 73–4; Süer and Çakın 1981: 17, 21; Turan 1993c: 21–32 and 1969: 134ff; Togan 1981: 192; Köymen 1972: 49). Most likely in view of the disparity between their forces, Sultan Alp-Arslan is thought to have approved an envoy from the caliph, the jurist al-Muhalban (or al-Muhallaban), to sue for peace. The emperor is said to have responded with the retort that talks could begin only in Rayy, asking whether it would be better to winter in Isfahan or Hamadan, adding that he had been informed Hamadan would be cold. Turkish historians note that according to al-Azrak the envoy agreed Hamadan would be best for the emperor’s horses but that he could not profess to know where Romanos Diogenes would be spending the winter. Once the jurist had returned (24 August 1071), the Seljuqs appear to have started to harry and harass the Byzantine camp (Sevim 1993: 219ff; Dirimtekin 1993: 230ff; Kafesoğlu IA/7: 242ff; Süer and Çakın 1981: 22; Yınanç 1944: 74; Köymen 1972: 52–3; Turan 1993a: 136–7; see also Karolidis 1993: 60–1). 80 Empire (1063–92) The following morning (25 August 1071), Romanos Diogenes took command of the Byzantine centre, the Cappadocian Attaliates of the wing consisting of heavy cavalry from Anatolia and the remaining Peçenek, while Briennios took command of the other wing, consisting of heavy cavalry from the Balkans and the remaining Uz. It must be pointed out, however, that Turkish historians are disagreed as to who was in charge of which wing (see Köymen 1972: 62). The rearguard, which is thought to have included Armenian nobility, was under the command of Andronicos Ducas. According to Turkish military historians (Süer and Çakın 1981: 25–9), Sultan Alp-Arslan is said to have divided his forces into three, hiding the main part in the hills to his left and rear. Thus, instead of a rearguard, it is thought a small reserve was assigned to each wing, presumably the mamluks under his command being in the centre, with the exception of Sav-Tekin. Köymen disagrees with this, accepting only that Sultan Alp-Arslan set a trap by hiding the forces under the command of Taranges (Köymen 1972: 62–3), which seems to be a misspelling of Sarhang, Sav-Tekin’s rank (Sümer and Sevim 1971: xx). As pious Muslim legend would have it, having put aside his bow and quiver, the sultan took up only his sword and mace like the common soldiery under his command. Dressed all in white, equally unlikely, he is said to have tied his horse’s tail into a knot. Finally, after Friday prayers, and therefore possibly at one-thirty, Sultan Alp-Arslan is thought to have finally launched the attacks aimed at dispersing the cavalry shielding the Byzantine centre and drew out Romanos Diogenes into his trap. The Byzantine cavalry appear to have broken under what is thought to have been a relentless shower of arrows. Whether he decided the Seljuqs’ main force was also at their centre, or in order to bolster the gaps opened in the cavalry lines, Romanos Diogenes is known to have advanced and Sultan Alp- Arslan feigned retreat. Thus the Byzantines would have entered his trap fighting around four in the afternoon. It is at this juncture the Uz and Peçenek said to have deserted in battle would have broken ranks. If so, this would have coincided with the fresh onslaught unleashed by the main body of Seljuq soldiery and hence possibly the reported break in the Byzantine right wing, which exposed their centre to bloody fighting. While Romanos Diogenes tried to reorder and bolster his right flank to stabilize the situation, the left wing is said to have also collapsed – possibly under attack from Sav-Tekin, originally hidden to the sultan’s rear. Afraid that his camp would fall, it is at this juncture that Romanos Diogenes seems to have ordered a retreat. In view of the distance his camp now would have been from the battle, however, and possibly under the impression the main force had broken, Andronicos Ducas is known to have taken flight with the Armenian nobility following suit. Completely encircled as a result, by seven in the evening the Byzantines’ defeat would have been final. Romanos Diogenes was wounded and captured by a soldier as night fell, allegedly remaining tied to the man’s tent pole until next morning (Köymen 1972: 62–4; Süer and Çakın 1981: 25–9; Kafesoğlu IA/7: 242ff; Sevim 1993: 219ff; Dirimtekin 1993: 230ff; Yınanç 1944: 75–6; Turan 1993c: 21–32 and 1969: 137ff). When his captor brought Romanos Diogenes to Sultan Alp-Arslan’s tent, apparently Vasilakios fell to his knees in tears. The jurist al-Muhalban appears to Empire (1063–92) 81 have confirmed his identity. Sultan Alp-Arslan is said to have criticized Romanos Diogenes for having refused his peace offer and ticked off his errors on the battlefield. When asked by the sultan what kind of treatment he expected, Romanos Diogenes is said to have replied that most likely he would be killed, or put in chains to be paraded throughout Islamdom, or possibly that he could be sent back to Istanbul as an ally (Kafesoğlu IA/7: 242ff; Yınanç 1944: 76–82; Turan 1993c: 21–32). Sultan Alp-Arslan allowed Romanos Diogenes to return, but the conditions of his release remain unclear. Whatever these may have been, upon news of his capture his enemies declared Empress Eudoxia’s son Michael Ducas emperor. Romanos Diogenes was defeated at Sivas and later in Adana, where he gave himself up on the condition he was allowed to live out his life as a monk. Having been blinded, he died soon after in a monastery (Köymen 1972: 69–73; see also Karolidis 1993: 65–72). Lauded and celebrated throughout Islamdom, his victory at Malazgirt having been equated with those previously won at Yarmük (636) and Qadisiyya (637) against the Byzantines and the Sasanians respectively, Sultan Alp-Arslan embarked on his final campaign. The sultan’s brother Ilyas, who was then in Khwarazm, had been at war with the western Karakhanid ruler, Shams al-Mulk Nasr Khan, who resided in Samarkand. While in Transoxania, however, Sultan Alp-Arslan was stabbed by a fortress commander he had captured and died (November 1072) (Kafesoğlu IA/10: 370).

Contention for the realm Although Malik-Shah was his favourite, Ayaz was the eldest of Sultan Alp- Arslan’s sons, of whom Togan-Shah, Tutuş and Tekiş were all younger than Malik-Shah (Kafesoğlu 1953: 14 and IA/10: 353ff). The sultan appears to have been quite specific in his will, which is said to have been communicated verbally on his deathbed, assigning Nizam al-Mulk as executor. According to this, Çağrı Beg’s real estate and 500,000 dinars in Balkh were to be given to Ayaz on the condition he supported Malik-Shah. Otherwise Malik-Shah was free to use his grandfather’s fortune to fight Ayaz. Furthermore, Fars and Kirman were to remain Kavurt’s, on the condition he married the sultan’s widow and settled in Shiraz (Kafesoğlu 1953: 16–17; also Merçil 1980: 45). Malik-Shah acceded at the age of eighteen (25 November 1072). His first action is said to have been to increase the pay of the soldiery (accession money) by 700,000 dinars. The new sultan then buried his father in Marv, next to Çağrı Beg, and moved to Nishapur where he is said to have distributed the contents of the treasury kept in the inner castle among his retinue and commanders, and informed his father’s governors and vassals he had acceded to the throne. Encouraged by some Turkic mamluk commanders who secretly offered their services, Kavurt refused to include Malik-Shah’s name in the khutba on grounds that he was the eldest. Malik-Shah is said to have replied that while there is a son, a brother cannot inherit. Kavurt appears to have hoped to gain the support of the Türkmen between Hamadan and Rayy, departing for the latter city with 2,000 cavalry and 82 Empire (1063–92) 4,000 foot soldiers. Arriving in the environs of Hamadan two days before Kavurt, Sultan Malik-Shah is thought to have given 500,000 dinars to the Türkmen, with 5,000 suits of armour and weapons (Kafesoğlu 1953: 18–21; Merçil 1980: 46–7; Köymen 1963: 69). Sultan Malik-Shah’s forces are thought to have included not only his mamluks but the Türkmen (though they may have remained impartial (Köymen 1963: 163)), as well as Arab and Kurdish contingents, which appears certain. The latter were the forces of two Seljuqid vassals who had been on their way to Sultan Alp-Arslan and included an envoy from the caliph. The two armies met outside Hamadan (1073). When Kavurt attacked the sultan’s right wing, the mamluk soldiery under Sav-Tekin begged for mercy. The critical blow appears to have been dealt by the Seljuq’s vassals on the left wing, under the command of Temirel (or Temrek). These are said to have disrupted Kavurt’s right wing, enabling Sultan Malik-Shah and Nizam al-Mulk at the centre to win the day. Having been thwarted on the battlefield, the mamluks are said to have pillaged both the vassals’ and the caliph’s tents (Merçil 1980: 47–51; Köymen 1963: 69; Kafesoğlu 1953: 22). Although Kavurt escaped, he was reported and captured. Kavurt appears to have insisted he acted upon the invitation of the sultan’s commanders, possibly providing letters as proof, which Nizam al-Mulk is said to have burned in the sultan’s presence without opening. Kavurt seems to have been moved to Hamadan to prevent further provocation, but when the commanders gathered outside Nizam al-Mulk’s tent, asking for more money and iqtacs and shouting Kavurt’s name, appearing to acquiesce, the vizier is said to have had Kavurt killed. Given the sultan’s reported reluctance, consensus is that Nizam al-Mulk acted with alacrity. It is not clear if Kavurt was strangled according to Turkic tradition with his own bowstring by a one-eyed Armenian, or by a Kıpçak with whom he fought for an hour before suc- cumbing, or if he took poison. Kavurt is said to have been buried next to Ibrahim Yınal (Merçil 1980: 52–9; Köymen 1963: 70–1; Kafesoğlu 1953: 23, 26). The confusion surrounding Kavurt’s demise extends to his eight sons. Current opinion is that his two eldest, Sultan-Shah and Emiran-Shah, and possibly also Turan-Shah, were with him on the battlefield with Ibrahim Yınal’s son. All are said to have been blinded in Sultan Malik-Shah’s presence after their father’s execution, Emiran-Shah dying in the process. Upon hearing of these events, some of the Turkic mamluk soldiery may have dispersed and begun pillaging the environs of Hamadan, cursing Nizam al-Mulk. Whatever the case, Sultan-Shah and possibly also Turan-Shah were not disabled completely by their ordeal and escaped. Certainly, Sultan-Shah (1074–85) and Turan-Shah (1085–98) lived to rule Kirman, Fars and Oman. Sultan Malik-Shah is known to have mounted campaigns against Sultan-Shah, but clearly these did not alter the status quo (1075–6 or 1079–80) (Merçil 1980: 59–63, 69–74; also Kafesoğlu 1953: 25–6).

Transoxania Upon hearing of Sultan Alp-Arslan’s death and Sultan Malik-Shah’s return to Khurasan, the Karakhanid Shams al-Mulk Nasr Khan entered Tirmiz (December Empire (1063–92) 83 1072). Taking advantage of Ayaz’s absence, he also entered Balkh. Ayaz appears to have taken back this city (January 1073). However, when he attempted to take back Tirmiz as well, he was defeated and lost most of his force retreating across the Syr-Darya (March 1073). Concurrent with the Karakhanid advance, the Ghaznavids attacked Çiğil-Kent in Tukharistan and captured the sultan’s uncle Osman with his treasury and took him to Ghazna (January 1073) (Kafesoğlu 1953: 19–20). Unlike his father, Sultan Malik-Shah did not move to secure his eastern realms until he had the caliph’s endorsement (October 1073). Ayaz appears to have died soon after in Balkh (1073–4). The young sultan is said to have advanced on Tirmiz via Herat, sending Sav-Tekin ahead to sever connection with Samarkand. The sultan appears to have laid siege to Tirmiz once Sav-Tekin had defeated a Karakhanid army by the Syr-Darya. Isolated, Shams al-Mulk Nasr Khan’s brother capitulated. The sultan is said to have released him and having garrisoned Sav-Tekin in Tirmiz advanced on Samarkand. When Shams al-Mulk Nasr Khan evacuated the city and asked to be forgiven, the sultan retired to Rayy. Here he received the Ghaznavid envoy, sent with a considerable amount of money and gifts to secure the sultan’s sister’s hand for Sultan Ibrahim (1058–99), which was granted. It is presumed, therefore, that his uncle Osman had been released earlier, during Sultan Malik-Shah’s advance through Herat. Certainly, the sultan appears to have journeyed to Serakhs next, where he is said to have received his uncle and assigned him the governance of a city, the regions of Gur and Herat having been assigned to Böri-Pars. Similarly, those of Balkh and Tukharistan were assigned to the sultan’s brother Tekiş (Kafesoğlu 1953: 28–30; also Öngül 2000: 325).

The Caucasus Next, Sultan Malik-Shah is said to have campaigned against the Georgian King Giorgi II (1072–89). This appears to have transpired upon the instigation of the commander Ioannes Liparit. Having put down a rebellion he seems to have then sold the castle concerned to the Shaddadid dynasty in Ganja (Gyandzha). At the end of this campaign, which both Liparit and his son appear to have joined, the sultan took Ganja from the Shaddadids and assigned it to his brother Tutuş, leaving Sav-Tekin in Yerevan as commander (1076). King Giorgi II defeated Sav-Tekin twice, gaining Yerevan as well as Kars. This seems to have prompted the former king of Ani, Gagik, to try to re-establish his kingdom. Liparit now began to resist Seljuq authority from the castle he had been given. The sultan returned, laid siege to Liparit and took him prisoner, leaving Sav-Tekin additional forces (1078–9). However, King Giorgi II again defeated Sav-Tekin, so that the sultan sent an army under Ahmad who appears to have finally established Seljuqid authority throughout the Caucasus, inclusive of Trabzon (Trebizond) on the Black Sea (1080–1). As a result King Giorgi II was forced to travel to Isfahan to proclaim allegiance. Subsequently, the Caucasus region was included in the province of Azarbayjan under Yakuti’s son Ismacil (1084), but the sultan appears to have had to return yet again. Although in the meantime Ganja and Ani seem to 84 Empire (1063–92) have been sold to the Shaddadids, these were taken back by force. At the end of this campaign, Ganja and the surrounding region were incorporated directly into the Great Seljuq Empire (1086) (Kafesoğlu 1953: 113–17; also Sevim 1965 and Ross IA/11: 381–2).

Anatolia It is at this conjuncture that Kutalmış’ sons, Mansur, Süleyman-Shah, Alp-Ilek (or Alp-Ilig) and Devlet (or Donat) enter the historical record and the Türkmen colo- nization of Anatolia gains coherence. Although most Turkish historians agreed that Kutalmış’ sons were assigned to the conquest of Anatolia by Sultan Malik- Shah, possibly on the instigation of the caliph (Yınanç 1944: 57–8, 86, 88; thus also Kafesoğlu 1953: 62 and IA/10: 370, Köymen 1963: 102–3, Sevim 1990b: 70), more likely they benefited from Sultan Alp-Arslan’s death, making their own ways to Anatolia in the ensuing confusion (Kaymaz 1993: 261). After Malazgirt, the new emperor, Michael VII Ducas, had appointed Isaac Comnenus with his brother Alexius and the Norman Roussel to clear Anatolia of the Türkmen. However, when they arrived in Kayseri, which had been badly dam- aged by an earthquake, Roussel fell out with Isaac and departed with the Norman mercenaries in his company. Whether for this reason or the task he had been set was now beyond the means of the Byzantine Empire, Isaac was defeated and taken prisoner by a Türkmen force that appeared in the environs. Having bought his freedom, he rejoined his brother in Ankara, from where they appear to have had to evade yet another group of Türkmen who then followed them west across the Sakarya River (Kafesoğlu 1953: 63; also Turan 1993c: 51). Said to have augmented his force from the region he was pillaging between Sivas, Kayseri and Ankara, Roussel defeated a Byzantine force sent out under the emperor’s uncle, Caesar Ioannes Ducas, whom he took prisoner with his son and Nicephorus Botaniates (1073). The battle is said to have taken place near the Sakarya River, which is taken to indicate Roussel had been retreating ahead of the advancing Türkmen under Artuk Beg. Certainly, the emperor is said to have made an agreement with Artuk Beg who defeated and captured Roussel with Ioannes Ducas near Izmit (Nicomedia), only 100 km from Istanbul. However, Artuk Beg does not appear to have handed over his captives, ransoming them instead to their families. At this point Artuk Beg is said to have been recalled by Sultan Malik- Shah. Roussel seems to have re-established himself in Amasya. Although Alexius was assigned to deal with him, at first he failed because the local populace is said to have preferred Roussel. However, when a large Türkmen force under Tutak Beg entered central Anatolia, Alexius appears to have been able to persuade them to capture Roussel. Since Roussel had also approached Tutak Beg, Alexius is said to have had to reward Tutak Beg handsomely. On his way back to Istanbul with Roussel, Alexius was engaged by another group of Türkmen who had penetrated north-western Anatolia up to the Black Sea (Kafesoğlu 1953: 64–8; also Turan 1993c: 51–2). Empire (1063–92) 85 Syria and Palestine Türkmen presence in Syria can be traced to 1063, with the arrival of Hanoğlu Harun in the environs of Diyarbakır in south-eastern Anatolia. After Ibrahim Yınal’s brother and Sultan Alp-Arslan’s brother-in-law Er-Basgan’s unsuccessful involvement with Kavurt and flight (1068–9), an affiliate of his, Kurlu, is thought to have broken away with Uvakoğlu Atsız and his brother Çavlı (1069–70). They appear to have been joined by a group of Harun’s Türkmen with whom they came to the environs of Ramla, north-west of Jerusalem, some of the Türkmen moving on to the Jordan Valley. Kurlu Beg is said to have revived the town by improving the olive harvest and attracting farmers back. Otherwise he appears to have sold his services to the highest bidder, namely the Fatimid governor of Acre (Akka) and the Mirdasid ruler of Aleppo, pillaging and confiscating when not paid. Kurlu Beg laid siege to Damascus (Dimashq) and then Acre where he died (1071) (Sevim 1965: 30–7 and 1990b: 33; also Turan 1993a: 45–50, Kafesoğlu IA/10: 353ff and Cahen 1984: 89). On his death, Uvakoğlu Atsız of the Kınık Oguz became the leader of the Türkmen in Palestine. His first action appears to have been to besiege Jerusalem. Its Fatimid governor, who was of Turkic origin, is said to have handed over the city on the condition it was not sacked (1071). Next Atsız Beg besieged Damascus (1071 and 1073), while his lieutenants mounted expeditions along the coast toward Acre. When Acre fell to the Türkmen in the same manner as Jerusalem had done (October/November 1074), their beg, Şöklü, refused Atsız Beg’s demand of half the spoils. In order to strengthen his hand against his former leader, Şöklü Beg seems to have married his sister to the Fatimid governor of Damascus and exchanged hostages with branches of Arab tribes in the region. Before his new allies could be of use, however, Atsız Beg is said to have lifted the siege of Damascus long enough to attack and defeat him (April/May 1075). Undaunted, Şöklü Beg is thought to have invited one of Kutalmış’ sons, said to have arrived from south-eastern Anatolia with a younger brother and nephew. These are thought to have been Alp-Ilek and Devlet (Sevim 1990a: 24), but Resul-Tekin’s son is not known (Kafesoğlu IA/10: 353ff). On Şöklü Beg’s instigation they are said to have proclaimed allegiance to the Fatimid caliph to attract help from this quarter. However, Atsız Beg appears to have pre-empted them by advancing from Jerusalem and defeating them. Having executed Şöklü and his son, Atsız Beg informed Sultan Malik-Shah of the three captive Seljuqid princes (autumn 1075) (Sevim 1965: 43–7 and 1990b: 35–7; also Kafesoğlu 1953: 33). Meanwhile another of Kutalmış’ sons, thought to be Süleyman-Shah (Turan 1993c: 48), had laid siege to Aleppo. Persuaded by a combination of military force and money to lift the siege, Süleyman-Shah then asked Atsız Beg to return his brothers and nephew. The request appears to have been refused and the captive princes sent to Isfahan as requested (December 1075). Next Süleyman-Shah is said to have besieged Antakya, forcing its Byzantine governor to offer annual tribute. Upon hearing that the 3,000-strong Türkmen force Atsız Beg had requested from Sultan Malik-Shah was approaching Aleppo, Süleyman-Shah is said to have 86 Empire (1063–92) attacked and killed some, plundering their baggage train before returning to the environs of Antakya (Sevim 1965: 47–8 and 1990b: 37–8).

The rise of Süleyman-Shah An absence of princes from the reigning branch of the Seljuqid dynasty in Anatolia appears to have been critical in enabling Kutalmış’ sons to take up leadership of the Türkmen there. In particular those who had rebelled against the Seljuqid sultans at one time or another would have been more than amenable to Kutalmış’ sons. That said, the dire straits the Byzantine Empire found itself in after Malazgirt seems to have more than contributed to the power vacuum, helping Kutalmış’ sons to establish a polity of their own. The gravity of the situation in Anatolia had not escaped the attention of Pope Gregory VII. He wrote to the emperor that the time had come to unite the Orthodox and Catholic churches (9 July 1073). Although Michael VII Ducas acquiesced, the Pope’s call for help (2 February and 1 March 1074) was not answered for another quarter of a century. In the meantime the emperor is known to have asked the caliph to mediate a peace agreement with Sultan Malik-Shah (June 1074). It is not certain, however, if the sultan received the emperor’s envoys as a result (1076), because a Byzantine embassy is also dated during his last campaign to the Caucasus (1086). This latter date seems the more likely given the problems discussed below. In any case, as with the reputed agreement between their predecessors, the outcome of such contacts is also not known (Kafesoğlu 1953: 74, 116; Turan 1993c: 52–3, 57). Since those unable to take refuge in the towns and cities were more vulnerable to the Türkmen, Michael VII Ducas is thought to have relocated large numbers of people from the countryside to the Balkans. However, although there may have been some migrations westward, it is doubtful this was on a large scale. There was the Bulgar uprising (1074) and the Nestor rebellion (1075), as well as plague and famine in the Balkans (1076). Amidst all this, the Byzantine army under Nicephorus Briennios in Thrace also revolted (1075). While still in Edirne (Adrianople), Briennios declared himself emperor and advanced on Istanbul (1077). The commander of the Byzantine army in Anatolia, Botaniates, who had been busy lobbying senators, clergy and the populace of Istanbul, also rebelled. In response, the emperor is said to have released Roussel from prison and sent him against Briennios with Alexius Comnenus, arranging to capture Botaniates with Kutalmış’ sons, thought to have been in Konya (Iconium) at the time. Süleyman-Shah is said to have situated his men at key crossroads and passes in order to intercept Botaniates who was travelling at night and hiding by day. When his advance units finally engaged Botaniates near Iznik (Nicea), it seems that at Er-Basgan’s instigation Süleyman-Shah switched sides and Botaniates was declared emperor in Iznik (1078). Until then the Türkmen seem to have been supporting Nicephorus Melissenos. On deciding to become emperor, Melissenos is said to have toured walled cities and fortifications in central and western Anatolia not yet under Türkmen control, hoping to gather support for his claim to the throne. Empire (1063–92) 87 He appears to have done so under the armed protection of the Türkmen, enabling them to enter these, too. With the addition of 2,000 horsemen from Süleyman- Shah, Botaniates is said to have continued to Istanbul where his accession was ratified (3 April 1078). Interestingly enough, this Türkmen force is deemed also to have enabled Alexius Comnenus to defeat Briennios, who seems to have been as dependent on the Peçenek and Uz (Turan 1993c: 53–5, 60–2 and Kafesoğlu 1953: 70–3, 76 and IA/10: 385ff; also Cahen 1984: 90). Süleyman-Shah is thought to have fallen out with Mansur, but it is not clear if Mansur died or was killed in battle. According to the latter view, Sultan Malik- Shah sent his commander Porsuk to help Süleyman-Shah. In this context, the report that the caliph then addressed Süleyman-Shah as sultan is interpreted as divisive (Yınanç 1944: 105–7; Köymen 1963: 102–6; Kafesoğlu 1953: 75; Sevim 1990b: 29). Turan disagreed, interpreting the report concerning the caliph as con- firmation that Sultan Malik-Shah had sent his commander against Kutalmış’ sons, Mansur dying in the ensuing battle. Porsuk’s campaign is coincident with the sultan’s brother Tutuş being assigned Syria and Palestine and the Türkmen begs returning from raids to Anatolia being ordered to join him (1078). Turan argued that Sultan Malik-Shah and Emperor Michael VII Ducas may have come to an agreement earlier (1074). In his view this explained Mansur’s and Süleyman- Shah’s support for Botaniates as in return for his emperorship he would have been amenable to their sovereignty over part of Anatolia (hopefully as his vassals) (Turan 1993c: 56–60; also Kaymaz 1993: 263 and Cahen 1984: 90–1). Whatever the case, unable to deal with the empire’s problems, Emperor Botaniates was forced to abdicate in favour of Alexius Comnenus who was declared emperor by the Byzantine forces in Thrace (1081) (Kafesoğlu 1953: 76–9; also Turan 1993c: 60–2). Having learnt he could not engage the Türkmen in large-scale combat, at least not until they saw they had the advantage, Alexius I Comnenus resorted to raiding. He is said to have disembarked eight to ten soldiers on the Asian shores of the Bosporus at odd hours of the night, having them return as soon as they achieved their goal. In this way he was able to regain a foothold, eventually disembarking as many as fifty cavalry at a time, even during the day. When the Türkmen had withdrawn toward Izmit, possibly because at the time Süleyman-Shah was engaged in the conquest of Tarsus in the south-east, the emperor offered peace. The acceptance of a stream, which is thought to be that of present-day Dragos (Drakon), as the boundary between the Byzantine Empire and the newly founded Seljuq polity clearly confirms the conquest of Anatolia (1082) (Kafesoğlu 1953: 80–1; also Turan 1993c: 60–2). The agreement not only left Emperor Alexius free to grapple with the Normans, but enabled Süleyman-Shah also to consolidate his hold on Tarsus and its environs (1083). This brought Süleyman-Shah in direct contact with the Armenian Philaretos Brachamios, a former commander of Romanos Diogenes and thus also Botaniates’ comrade, who until then had enjoyed a free hand in much of south-eastern Anatolia. Prior to the Seljuqs, the region south of the Caucasus Mountains between the Caspian Sea and Lake Van had been mainly under the control of members of the Armenian Bagratid dynasty, a branch of which had 88 Empire (1063–92) also resided in north-eastern Anatolia. This latter region and that of the Armenian Ardzruni dynasty, namely the region around Lake Van had been incorporated into the Byzantine Empire during the reign of Basil II (976–1025). As a result many Armenians had been resettled in Sivas and Kayseri. Subsequently, members of these and other leading Armenian families were all but eradicated by Constantine IX Monomachus (1042–55). In the years after the battle of Malazgirt, many of these appear to have re-emerged in eastern and south-eastern Anatolia. Although Philaretos Brachamios had gained control of these fledgling princedoms as well as Antakya (1078), his cruel and wanton approach is said to have made the local Christian population resentful as they were already under pressure from the Türkmen. In order to maintain his position, therefore, Philaretos Brachamios appears to have proclaimed his allegiance to Emperor Botaniates while paying tribute and sending gifts to both Sultan Malik-Shah and his brother Tutuş, as well as the rulers of Aleppo and Mosul (Sevim 1983: 7–10, 21–2; also Kafesoğlu 1953: 82–3 and Turan 1993c: 68–9). Possibly on the invitation of Brachamios’ son, Süleyman-Shah appears to have advanced on Antakya in secret while the Türkmen laid siege to Malatya and various other urban centres. Süleyman-Shah is said to have taken only a few hundred men with him to avoid alerting either Tutuş or Muslim, the Uqailid ruler of Mosul. Having been let in at night (12 December 1084), Süleyman-Shah took the inner city despite some resistance (11 January 1085). He consolidated his hold by also taking the towns and fortifications in the environs of Antakya (Sevim 1965: 80–4; Kafesoğlu 1953: 83; Turan 1993c: 69–73).

Incorporation of Syria As noted, two of Kutalmış’ sons, thought to have been Alp-Ilek and Devlet, had been captured by Uvakoğlu Atsız and sent to Sultan Malik-Shah (1075). Although at the time Atsız Beg requested help from the sultan in the conquest of Syria and Palestine, when the city of Tyre (Sur) capitulated on the condition that the Fatimid khutba would not be changed, he agreed. Atsız Beg next took Tripoli (Tarabulus) and then Damascus. In the latter case he is said to have been aided by a revolt against the city’s Fatimid governor (July/August 1075). Although at first the city’s commander continued to resist Atsız Beg, the inhabitants are said to have become restless, forcing him to capitulate. Atsız Beg appears to have rewarded him with two coastal cities (March 1076). Unlike Tyre, however, in Damascus the Fatimid khutba was replaced (Sevim 1965: 48–52 and 1990b: 38–40; also Kafesoğlu 1953: 31–2). In the previous three years the former governor of Damascus and Acre, Badr al- Jamali, had put Fatimid affairs in order. Of Armenian origin, while in Damascus Badr al-Jamali had formed a personal guard of his countrymen. Invited to Cairo by the caliph, he was able to surprise and kill the Turkic commanders with their help, ending the commanders’ anarchy. The son of one of the commanders appears to have escaped with his father’s valuables and some of his men, and taken refuge with Atsız Beg. Encouraged by him, Uvakoğlu Atsız marched on the Fatimids at Empire (1063–92) 89 the head of a mixed force of Türkmen, Arabs and Kurds (1076). Having taken Rif (Gaza), Atsız Beg demanded money from Badr al-Jamali. Although promised 150,000 dinars, he advanced on Cairo after a two-month wait. Badr al-Jamali defeated Atsız Beg outside Cairo (January/February 1077). On the way back Atsız Beg appears to have been refused entry to Rif and also Ramla, and is said to have arrived in Damascus with only ten or fifteen horsemen (Sevim 1965: 52–6 and 1990b: 41–2; Kafesoğlu 1953: 34–6). In the aftermath of his disastrous campaign Atsız Beg seems to have lost control of many if not all the urban centres of Syria and Palestine which are said to have reverted to the Fatimid khutba. Aided by the arrival of a fresh group of Türkmen from Anatolia, Atsız Beg embarked on a bloody re-conquest, which he started by saving his wife and family from enslavement in Jerusalem where they had been left en route to Egypt. However, upon returning to Damascus he is said to have found the city’s population reduced to a mere 3,000. Badr al-Jamali now sent out a force under Nasr al-Dawla al-Juyushi, a commander from his personal guard, to besiege Damascus (1077/8). Although this was unsuccessful, the following year al-Juyushi returned with an army that invaded Palestine before once again laying siege to Damascus. This appears to have prompted Atsız Beg to ask the sultan’s brother Tutuş for help. On entering Damascus, Tutuş had Uvakoğlu Atsız and his brother Çavlı strangled with their bowstrings (1078/9) (Sevim 1965: 57–60 and 1990: 43–5; Kafesoğlu 1953: 37–8). Sultan Malik-Shah is said to have assigned his brother Tutuş to Syria and Palestine thinking that Atsız Beg had been killed in Egypt. According to this version of events, on hearing of this Atsız Beg sent gifts and money to the sultan, earning a reprieve (1077/8) (Sevim 1965: 59). While this may be true, at the time Sultan Malik-Shah was conducting a campaign against the resurgent Georgians. Also his commander Porsuk may have been involved in Anatolia against Kutalmış’ son Mansur, so that it is also possible that Sultan Malik-Shah left Atsız Beg in place until he was better prepared. It is at this conjuncture that Türkmen begs such as Afşin, Sabuk, Dilmaçoğlu and Davdavoğlu are thought to have been ordered by the sultan to join Tutuş. However, it is not known if these Türkmen were with Tutuş at the siege of Aleppo. Tutuş is said to have failed to take the city because the Uqailid Muslim had instructed his soldiery from Mosul to join in the city’s defence (Sevim 1965: 66–7; Turan 1993c: 57). It was when Tutuş returned to Aleppo the following spring that he was called to Damascus by Atsız Beg. With Tutuş busy in Damascus and its environs, the Uqailid Muslim is said to have requested and gained permission from Sultan Malik-Shah to take Aleppo on the invitation of its inhabitants. Having done so (1080), Muslim then took all the fortifications in the region where the Türkmen were situated, killing or imprisoning them. Next he extended Uqailid power into south-eastern Anatolia by taking Harran (1081–2). Consequently, the Byzantine governor of Urfa, and subsequently also Philaretos Brachamios in Antakya, are said to have proclaimed their allegiances (1082–3). When Tutuş advanced on Aleppo with Artuk Beg, Sultan Malik-Shah ordered his brother back to Damascus and recalled Artuk Beg. 90 Empire (1063–92) However, Muslim now laid siege to Damascus, withdrawing only when Harran revolted (May/June 1083) (Kafesoğlu 1953: 40–4; Sevim 1965: 66–79). At this point Caliph al-Qa’im’s vizier, Fakhr al-Dawla, who was for a time his successor Caliph al-Muqtadi’s vizier, retired to the capital Isfahan (June 1083). Originally from Mosul, Fakhr al-Dawla had started his career as the Marwanids’ vizier. Consequently, he seems to have been more than familiar with their fortunes, said to have taken a downturn with the appointment of a new administration. Despite his advanced years he appears to have persuaded Sultan Malik-Shah that the Marwanids’ domains, which included numerous cities and fortifications in south-eastern and eastern Anatolia, should be incorporated directly into the empire (Kafesoğlu 1953: 46–8). The commander of Iraq al-Ajam, Türkmen begs such as Artuk, Sabuk and Dilmaçoğlu, and two neighbouring vassals were assigned to Fakhr al- Dawla and his son Muhammad. Alerted to the approaching Seljuq forces, the Marwanid ruler Mansur asked the Uqailid Muslim for help. Muslim is said to have gathered his forces and arrived in Diyarbakır, which Mansur had promised him along with other towns and fortifications. Faced by his fellow Arabs and townsmen, Fakhr al-Dawla appears to have prevaricated. When the Marwanid and Uqailid leaders opened negotiations, Muhammad ordered Artuk Beg to retreat in order to allow the Uqailid forces to withdraw. Artuk Beg refused and during the night the Türkmen attacked under Çubuk Beg, apparently without the knowledge of either Artuk or Fakhr al-Dawla (31 July 1084). Defeated, the Marwanid and Uqailid forces retired to Diyarbakır, which was put under siege. Fakhr al-Dawla then ordered Artuk Beg to collect the plunder as well as the money gained by the Türkmen and send it to Isfahan. Artuk Beg appears to have refused. Furthermore, he seems to have come to terms with Muslim and withdrawn from Diyarbakır (August 1084) (Kafesoğlu 1953: 49–55; also Sevim 1965: 85). Informed of Muslim’s support of the Marwanids, Sultan Malik-Shah is said to have assigned the Uqailids’ realms to Fakhr al-Dawla Muhammad’s son, ordering the Türkmen begs ostensibly under his father’s command and his commander Aksungur Beg to join him. Although Mosul appears to have capitulated without resistance and the sultan set off from Isfahan, the Uqailid Muslim seems to have gained an audience and been reinstated (November 1084). Possibly the sultan was more concerned with his brother Tekiş who had again rebelled. On a prior occasion, Turkic soldiery dismissed for lack of discipline in Rayy had joined Tekiş in Balkh and incited him to rebel. Tekiş had taken Tirmiz with their help, but when the sultan arrived in Nishapur ahead of him, he had asked to be forgiven (1080–1). On this occasion, Tekiş appears to have been unable to take Serakhs, withdrawing to a fortification near Tirmiz having been tricked into believing Nizam al-Mulk was on his way to relieve the city. When instead the sultan arrived from Mosul he had Tekiş caught and blinded, appointing his son Ahmad in his stead (1084–5). Tekiş was finally strangled with his own bowstring on Sultan Berk-Yaruk’s orders (March–April 1094) (Kafesoğlu 1953: 56–9; also Öngül 2000: 325–6, 329–31, Sevim 1965: 85 and Turan 1993c: 73). Empire (1063–92) 91 Meanwhile Fakhr al-Dawla appears to have laid siege to Silvan (Mayyafariqin), where the Marwanid Mansur had withdrawn after Muslim had paid Artuk Beg to lift the siege of Diyarbakır. When the Seljuq forces began to take the other Marwanid cities, towns and fortifications, Mansur is said to have journeyed to Isfahan but without success. While he was thus occupied the inhabitants of Diyarbakır appear to have opened the city gates (4 May 1085). Soon thereafter Silvan also fell (31 August 1085). Fakhr al-Dawla and his son Muhammad took charge of Silvan and Diyarbakır respectively, the remainder of the Marwanid realms being apportioned between the Türkmen begs ostensibly under their command (Kafesoğlu 1953: 49–55; also Sevim 1965: 85). At the beginning of that year Süleyman-Shah had taken Antakya on the invitation of its residents (11 January 1085). When the Uqailid Muslim asked him to pay the annual tax Philaretos Brachamios had promised for Antakya, Süleyman- Shah is said to have replied that not only was he a Muslim but likewise a vassal of Sultan Malik-Shah. After each had raided the other’s environs, Muslim appears to have set off to besiege Antakya. Süleyman-Shah met him en route where the Türkmen under Çubuk Beg appear to have deserted Muslim who was defeated and killed (23 June 1085). Although Süleyman-Shah seems to have besieged Aleppo immediately after, he was unable to take it. When he returned the following spring (April 1086), its ruler is said to have played for time, sending Süleyman-Shah money and saying he was waiting for Sultan Malik-Shah’s permission. Tutuş advanced on Aleppo with Artuk Beg. Having failed to regain the sultan’s favour after Diyarbakır, Artuk Beg had taken refuge with Tutuş and been rewarded with Jerusalem. The two armies are thought to have met at Ain Salm, where Süleyman- Shah’s all-out attack on the detachments from Damascus failed and the Türkmen under Artuk Beg routed his forces. Süleyman-Shah is said to have been deserted not only by Çubuk Beg but also some of his closest comrades-in-arms, possibly because Tutuş had succeeded in winning them over prior to the engagement. Süleyman-Shah is thought to have committed suicide on the battlefield (5 June 1086) (Kafesoğlu 1953: 86–90; Köymen 1963: 106–8; Sevim 1965: 89–92; Turan 1993c: 73–6). Unlike Süleyman-Shah, Tutuş took no notice of the Uqailid ruler of Aleppo who appears to have tried the same ruse with him too. Although able to breach the outer walls through a defending commander’s defection, Tutuş is said to have abandoned his siege of the inner city and retired to Damascus on learning that Sultan Malik-Shah was on his way (September 1086). He is thought to have been fearful of the sultan’s wrath after Süleyman-Shah’s death, but the fact of the matter may have been the siege of Damascus by the Fatimids. In any event, the sultan is said to have arrived in Aleppo via Mosul and Harran, taking any town or fortification that resisted. Having appointed Aksungur to Aleppo, Sultan Malik- Shah sent Bozan to take Urfa from Brachamios’ son and advanced on Antakya. Süleyman-Shah’s vizier handed over the city and Süleyman-Shah’s son Kılıç Arslan. Sultan Malik-Shah assigned Antakya to Yağı-Sıyan. With Urfa assigned to Bozan who had taken it (28 February 1087), the sultan returned to Aleppo. From there he went on to Baghdad for the first time (13 March 1087). Here he 92 Empire (1063–92) attended his daughter’s wedding to the caliph (Kafesoğlu 1953: 91–4; also Sevim 1965: 94–8 and Cahen 1979: 92–3). On learning of Tutuş’ return the Fatimids appear to have lifted the siege of Damascus and retired (1087/8). While Sultan Malik-Shah was preoccupied in Baghdad, Tutuş seems to have tried to consolidate his hold on northern Syria while expanding the Seljuq realms back along the Mediterranean coast into Palestine. The following year Badr al-Jamali sent out al-Juyushi who was able to re-conquer most of the Mediterranean coast because Tutuş is thought to have left it undefended, thus putting pressure on Damascus (1089). Caught unprepared, Tutuş asked the sultan for help, being assigned the commanders Aksungur, Bozan and Yağı-Sıyan. However, after some initial success Tutuş and Aksungur appear to have fallen out during the siege of Tripoli. Aksungur is said to have refused to continue after secretly receiving money and claiming the city had proclaimed allegiance. When Bozan also left for undisclosed reasons, Tutuş was forced to withdraw to Damascus. Although he is said to have sent his son to Sultan Malik- Shah to complain, Aksungur does not appear to have been punished (1090/1) (Kafesoğlu 1953: 98–101; Sevim 1965: 99–103).

Anatolia after Süleyman-Shah On Süleyman-Shah’s death al-Kasim is thought to have revoked the agreement with Emperor Alexius, raiding toward the Bosporus (1085). When on this occasion the emperor’s tactics of launching night raids from small boats proved unsuccessful, Alexius Comnenus is said to have suggested an alliance against Sultan Malik-Shah, flattering al-Kasim’s military prowess. Instead al-Kasim appears to have taken a fortification near Gemlik on the Sea of Marmara and started building a navy. The emperor sent Butumites to burn the boats while a large force advanced under Tatichios of Peçenek origin. Al-Kasim seems to have attacked Tatichios first but was unable to defeat him, allowing Butumites to burn his fledgling navy. However, Tatichios appears to have withdrawn on receiving news of Porsuk’s advance. Al-Kasim set off in pursuit and took Izmit. Porsuk is thought to have been after al-Kasim as the latter now accepted the emperor’s peace offer, journeying to Istanbul. The emperor seems to have taken this opportunity to dispatch his commander Eustathios to Izmit by sea. Eustathios appears to have had little difficulty in persuading them that al-Kasim had agreed to a joint defence of the city. Once the Byzantine force had been allowed into Izmit, however, Eustathios took control and had the walls strengthened. Al-Kasim is said to have only become aware of the emperor’s duplicity while returning to Iznik on Porsuk’s withdrawal (probably 1086/7) (Turan 1993c: 84–5; Kafesoğlu 1953: 101–4; also Cahen 1984: 94). While still on campaign in south-eastern Anatolia, the sultan is thought to have sent an envoy to Emperor Alexius, requesting his daughter in marriage for one of his sons and offering some seaports on the Anatolian coast. This was how the emperor is thought to have gained control of Sinop on the Black Sea from Kara- Tekin, thought to have been an affiliate of Erbasgan. It is more likely, however, that Empire (1063–92) 93 Kara-Tekin withdrew due to Sultan Malik-Shah’s commander Bozan’s advance and subsequent siege of Iznik. Bozan seems to have been supported by Porsuk, thought to have stopped en route in Konya and Aksaray. Al-Kasim is thought to have defended Iznik against Bozan with help from the Byzantines, forcing Bozan to lift the siege after a number of unsuccessful attempts to storm the walls. Under pressure from the sultan and the emperor’s opportunism, al-Kasim seems to have journeyed to Isfahan to proclaim allegiance. Unable to gain an audience with Sultan Malik-Shah, he is said to have been apprehended by Bozan and strangled with his own bowstring. After this the military governors general of Antakya and Aleppo, Yağı-Sıyan and Aksungur, are thought to have joined Bozan on a campaign against al-Kasim’s son, al-Gazi, who once again thwarted them at Iznik. The commanders are said to have returned to their cities on Sultan Malik-Shah’s death (1092), having considered and abandoned the idea of going on to take Istanbul instead (Turan 1993c: 86–7; Kafesoğlu 1953: 105–7). While this seems highly fanciful, it is interesting that in attempting to build a fleet al-Kasim appears to have been following the example of Çaka (or Çakan) Beg of the Çavuldur Oguz in Izmir. Çaka Beg is thought to have been captured by the Byzantines after Malazgirt and educated at the palace in Istanbul, being given the aristocratic title of protonobilissimos when Botaniates became emperor with Süleyman-Shah’s help (1078). His position is said to have deteriorated when Alexius Comnenus took the throne (1081), such that he escaped, eventually to emerge in Izmir and its environs (1086). Unlike the other Türkmen begs, however, Çaka’s power was based on a naval force he created with the aid of Greek shipbuilders from Izmir and the islands. This appears to have gained him some degree of control over the northern Aegean and the Dardanelles, despite repeated attempts by the Byzantines to stop him (Turan 1993c: 87–9; Kafesoğlu 1953: 108–11). Çaka Beg’s emergence coincides with the Peçenek onslaught in the Balkans. Having defeated them, Alexius Comnenus had refused the Peçeneks’ peace offer (1086), intending to deal them a final blow (1087). Although the Peçeneks had requested help from the Kumans, they inflicted a terrible defeat on the Byzantines before their arrival and forced the emperor to retire to Istanbul (1087–9). When the Kumans asked for a share of the spoils, however, the Peçeneks appear to have fallen out with them and allied with Çaka Beg. The emperor once again reverted to diplomacy, inviting the Kuman begs Manyak and Tongurdak to Istanbul where Byzantine generals of Kuman origin, Karaca and Uzan, hosted them. While the Peçeneks were waiting for Çaka Beg, the Byzantines and Kumans attacked and massacred them (29 April 1091). Those able to save their lives were christened and settled in Macedonia, east of the Vardar River, serving in the Byzantine army as Vardoroit (Turan 1993c: 90–2; Kafesoğlu 1953: 111–12). According to Turan, Çaka Beg was the first Türkmen beg to have appreci- ated that Istanbul could not be conquered from land alone. In his opinion Çaka Beg’s intentions can be deduced from his self-ascribed imperial title of Basileus (1993c: 93–5). 94 Empire (1063–92) Central Asia The Karakhanid ruler of Samarkand is said to have been treating his subjects badly, confiscating the possessions of wealthier inhabitants, too, so that the residents complained to Sultan Malik-Shah. The sultan appears to have responded by taking Bokhara and Samarkand before advancing on Kashgar. En route he incorporated the regions around Talas and Issyk-Köl, so that the Karakhanids became his vassals (1088–9). Sultan Malik-Shah’s commander in Samarkand fell out with the local Çiğil beg. Afraid of the sultan’s wrath, the Çiğil beg is said to have invited Yakup Tekin, the brother of the Karakhanid ruler of Kashgar who is thought to have been Harun Buğra Khan. While the sultan re-conquered Samarkand, the ruler of Kushan, Yınaloğlu Tuğrul, captured Harun Buğra Khan. Deciding against pursuit, Sultan Malik-Shah appointed Yakup Tekin to Kashgar (Kafesoğlu 1953: 119–23).

Arabian peninsula and the Gulf When Artuk Beg was recalled from Anatolia (1076–7), he was assigned to deal with the Shicite Carmathians in Ahsa and Bahrain (Kafesoğlu 1953: 38). Artuk Beg is said to have soon realized he could not fulfil his mission without replacing his horses with camels and returned to Basra. Properly equipped, he took Bahrain, then Ahsa (1076–7) (Kafesoğlu 1953: 38–9). Although the khutba in Mecca was now being read in the Abbasid caliph’s name (1068), after the death of al-Qa’im it had reverted to the Fatimids (1073). While Artuk Beg advanced down the eastern coast of the Arabian Peninsula, the commander for Khurasan arrived in Mecca with some jurists and the khutba once again was read in the Abbasid name there and in Medina (1076). However, there appears to have been a brief setback in Medina (1077). After the Abbasid khutba had been re-imposed (1079–80), the Fatimid caliphs’ names were erased from the Qa’bah (1086–7). However, following the Fatimid recovery in Palestine Sultan Malik-Shah is said to have called his brother Tutuş and the commanders Bozan, Aksungur and Çubuk to Baghdad along with others (December 1091). Before returning to Isfahan (March 1092), he assigned Tutuş to the conquest of Palestine and Egypt with Gawhar-A’in, Bozan and Aksungur, and the caliph’s former vizier to that of Yemen and Aden with Çubuk. Certainly, the latter was achieved that year by Yarınkuş, who took over the campaign when the commander Turşek (probably Çubuk) died of natural causes (Kafesoğlu 1953: 123–7).

Coming of the assassins The leader of the Nizari Ismacili sect Hasan al-Sabbah’s family is thought to have been Arabic. Although they were of the Twelver Shicite sect, al-Sabbah is thought to have come under the influence of the Ismacilis while being educated in Rayy. Impressed by his intelligence and energy, the daci or religious official for Iraq al-Ajam, Ibn Attash, suggested he journey to Cairo (1071). Al-Sabbah does not Empire (1063–92) 95 appear to have left Rayy until the Seljuqs intensified their pursuit of Shicites, going first to Isfahan (1077). From there he is thought to have travelled to Azarbayjan before arriving in Cairo (1079). When the Fatimid Caliph Mustansir appointed his younger son instead of the heir apparent Nizar, possibly because he was the vizier Badr al-Jamali’s son-in-law, al-Sabbah is thought to have been unable to remain in Cairo because he was a vociferous supporter of Nizar. He returned secretly to Isfahan (May 1081) (Kafesoğlu 1953: 128–30). Nizam al-Mulk is known to have assigned the governor of Rayy with the capture of al-Sabbah who appears to have escaped first to south-eastern Iran, where Shicites were in the majority. After putting an end to the local dynasty (1077), Sultan Malik-Shah had left this region in the control of various commanders lodged in redoubtable mountain fortresses. Al-Sabbah is thought to have gained converts from among these as well as the local populace and begun to send out missionaries. He appears next in Alamut castle (5 September 1090). Ensconced in his ‘eagle’s nest’, which he did not leave until his death, from there al-Sabbah launched a violent campaign of assassinations for which the batinis are remembered to this day (Köymen 1963: 211; also Kafesoğlu 1953: 130–2). Although Alamut was besieged, it was lifted when the commander Yorun- Taş died of natural causes. After this Sultan Malik-Shah is said to have assigned Arslan-Taş to the capture of al-Sabbah and Kızıl-Sarıg to the pursuit and capture of his missionary in Kuhistan, Hussein Qa’ini. However, al-Sabbah was able to call on his daci in Kazvin; he surprised Arslan-Taş with an armed force gathered from the fortifications the Nizaris had established near Alamut (September 1092). Routed, Arslan-Taş withdrew. As for Kızıl-Sarıg, he is said to have pursued al-Sabbah’s followers in towns and villages throughout Kuhistan, but news of Sultan Malik-Shah’s death caused him to withdraw (Köymen 1963: 212; also Kafesoğlu 1953: 132–4 and Turan 1993a: 246). The sultan’s death came within weeks of Nizam al-Mulk’s assassination, which is commonly attributed to the Nizaris. The vizier had made numerous enemies during some thirty years in office. Quite apart from his grip on the diwan (public financial register, thus government council) as chief officer, he had placed his numerous sons and sons-in-law in key positions. The vizier’s retinue is said to have numbered nearly 20,000, including his mamluk soldiery, the Nizamiyya (Kafesoğlu 1953: 197, 203). However, a vizier’s power ultimately depended on his relationship with the sultan. For his part Sultan Malik-Shah does not seem to have been beyond showing his displeasure, or punishing transgressions of his authority by Nizam al-Mulk’s sons, regardless of the fact that he had once been his guardian. The sultan’s court jester provides an example. Reports of his mimicry of the renowned vizier are said to have coincided with rumours at court that the former vizier of the military governor of Fars was being groomed to replace Nizam al-Mulk. Angered by this, Nizam al-Mulk’s son Jamal al-Mulk is said to have journeyed to Isfahan from Balkh where he was governor. Sultan Malik-Shah was none too pleased when Jamal al-Mulk not only scolded the jester in his presence but later had him killed by having his tongue pulled out through his neck. Not satisfied with this, Jamal al-Mulk is said 96 Empire (1063–92) to have had his father’s supposed contender blinded as well. Jamal al-Mulk died subsequently, having been poisoned (November 1082) (Kafesoğlu 1953: 198–9). Nizam al-Mulk’s successor, Taj al-Mulk, was likewise recommended to court by a governor general, Sav-Tekin. He became Sultan Malik-Shah’s wife Terken Khatun’s vizier coincident with the death of the heir apparent Ahmad (1088). After Ahmad’s death Terken Khatun wanted her newborn son Mahmud to become heir apparent, but Nizam al-Mulk appears to have insisted that the sultan’s eldest living son Berk-Yaruk, from the sultan’s marriage to his uncle Yakuti’s daughter, was made heir apparent. Soon after, one of Nizam al-Mulk’s sons, the governor of Marv appears to have fallen out with its commander. Publicly insulted, the commander is said to have complained in person to the sultan who sent Taj al-Mulk to confront Nizam al-Mulk. The vizier’s response seems to have been quite haughty. When asked whether he thought himself partner to Sultan Malik- Shah’s sovereignty and dominion, Nizam al-Mulk is said to have replied that there would not be an empire without him. Threatened by Sultan Malik-Shah with dismissal if he continued to abuse his power, the vizier is considered to have become vulnerable enough for Terken Khatun and Taj al-Mulk to use a batini assassin. Nizam al-Mulk was killed while en route to Baghdad with the sultan, Terken Khatun and her vizier Taj al-Mulk who was appointed in his place (15 October 1092) (Kafesoğlu 1953: 200–6). Sultan Malik-Shah appears to have continued on his journey despite his vizier’s demise. Although initially Caliph al-Muqtadi had ignored the sultan’s request he marry his daughter from Terken Khatun, he appears to have had a change of heart and subsequently asked for Mah Melek Khatun through the good offices of Nizam al-Mulk (1081). The marriage had taken place after protracted negotiations (1087), producing a son Jacfar (1088). The marriage did not go well; Mah Melek Khatun and her Turkic retinue returned home with Jacfar the following year (28 June 1089), shortly after which she died (December 1089). After he had a caliphal building erected outside Isfahan (spring 1092), Sultan Malik-Shah journeyed to Baghdad, demanding Caliph al-Muqtadi appoint Jacfar heir apparent. The caliph refused. Arriving in Baghdad, Sultan Malik-Shah ordered the caliph to leave (29 October 1092). The caliph appears to have gained ten days grace through the good office of Taj al-Mulk. Before the ten days had elapsed however, Sultan Malik-Shah fell ill while on a hunt and died at the age of thirty-eight (20 November 1092) (Kafesoğlu 1953: 94–8, 206–8; Kitapçı 1994: 109–56; Ocak 2002: 379–82). 6 Interregnum (1092–1105)

The internecine warfare that followed Sultan Malik-Shah’s death was carried out between his and Nizam al-Mulk’s sons, a situation that appears to have only benefited the commanders. They finally brokered a deal between Sultan Berk-Yaruk and his brother Tapar, as a result of which Seljuqid hegemony became bi-partite once more (1104). After Great Sultan Berk-Yaruk’s death Sultan Tapar did not declare his brother Sancar sultan, but he never interfered in Khurasan. Soon after his death (1118), Sancar declared himself Great Sultan and Tapar’s eldest son sultan (1121). Nizam al-Mulk’s eldest son Muayyad al-Mulk played a key role in Berk- Yaruk’s struggles with Terken Khatun, Sultan Malik-Shah’s son Tutuş, and finally Tapar and his uterine brother Sancar. At first he and Berk-Yaruk were supported by Nizam al-Mulk’s soldiery, the Nizamiyya. However, these and the royal corps in Isfahan appear to have been quickly decimated in the bloody battles that initially took place. This left the commanders and their soldiery in the regions, supported by incomes from lands allocated to them under Sultan Malik-Shah. These were without exception either mamluks of Turkic origin or Türkmen begs. They appear to have not only avoided taking the field against each other but also refused to leave their iqtacs at harvest time. Having been replaced by his brother Fakhr al-Mulk (1095), with whom he is said to have fallen out over their father’s jewellery, Muayyad al-Mulk encouraged Tapar to challenge Sultan Berk-Yaruk at a time when he had finally achieved a position from which he may have been able to put the affairs of empire into order and threw the Great Seljuq Empire into further disarray (1098). In the process Muayyad al-Mulk went so far as to have the sultan’s mother strangled, causing him finally to lose his life in a similar fashion in the sultan’s hands. While Sultan Berk-Yaruk and Tapar struggled, the Crusaders arrived and took Urfa and Antakya (1096), then Jerusalem (1099). After this they and the Seljuqs avoided open conflict unless the field of battle gave their diametrically opposite military strengths an advantage. In the process the Great Seljuqs lost control of south-eastern Anatolia and Syria to various commanders. The chapter follows Özaydın’s Sultan Muhammed Tapar Devri Selçuklu Tarihi (498–511/1105–1118) (1990) and Sultan Berkyaruk Devri Selçuklu Tarihi (485– 498/1092–1104) (2001), as well as ‘Büyük Selçuklu Emiri Kürboğa’ (2000). In 98 Interregnum (1092–1105) the case of the Seljuqs of Syria, the chapter follows Sevim’s Suriye Selçukluları I – Fetihten Tutuş’un Ölümüne Kadar (1965) and Suriye-Filistin Selçuklu Devleti Tarihi (1989). Turan’s Selçuklular Zamanında Türkiye (1993c) has been referred to for Anatolia. As in the previous chapters, references to other works are for supplementary material pertinent to the events being reviewed.

Contention for the realm Terken Khatun kept the sultan’s death secret, offering to send back the caliph’s son Jacfar by Mah Melek Khatun on the condition that the caliph accepted to have the Friday khutba read in her son Mahmud’s name. Mah Melek Khatun had died the previous year. At first the caliph objected that Mahmud was too young, but finally agreed. He is said to have stipulated that Terken Khatun’s commander Üner was appointed commander-in-chief and reported directly to the vizier Taj al-Mulk. After the khutba had been read as agreed (26 November 1092), Terken Khatun set off for Isfahan, having dispatched the commanders Kür-Boğa and Kumac ahead with the royal seal and orders to imprison Berk-Yaruk (Özaydın 2001: 16–18; also Köymen 1963: 72–3 and Kitapçı 1994: 163–71). Although the commanders succeeded, on news of the sultan’s death Nizam al-Mulk’s mamluks, the Nizamiyya, are said to have raided the deceased vizier’s arsenals and released Berk-Yaruk, having the khutba read in his name, before retiring to Rayy where Nizam al-Mulk’s commander Erkuş joined them with his soldiery. Once in Isfahan, Terken Khatun is said to have spent vast sums to ensure the allegiance of her soldiery and that of the royal corps. Despite this some of the commanders appear to have deserted Taj al-Mulk for Berk-Yaruk at the battle of Burujird (17 January 1093). Victorious, Berk-Yaruk pursued Terken Khatun’s forces and besieged Isfahan. On the advice of her commander- in-chief, Terken Khatun is said to have offered Berk-Yaruk money to lift the siege. Apparently Berk-Yaruk accepted, ceding Isfahan and Fars to her. During the siege Taj al-Mulk appears to have been captured and brought to Berk-Yaruk who offered him the vizierate. Although Taj al-Mulk is said to have gifted the Nizamiyya money and valuables, they lynched him on Nizam al-Mulk’s regent Osman’s instigation (February 1093). After the siege Terken Khatun approached Yakuti’s son Ismacil in Azarbayjan with an offer of marriage. Having accepted, Ismacil gathered his Türkmen forces and advanced with Sav-Tekin and Kür- Boğa, whom Terken Khatun had sent in support. Berk-Yaruk defeated Ismacil when some of his commanders deserted him on the battlefield. Ismacil withdrew to Isfahan (February 1093) (Özaydın 2001: 18–22 and 2000: 405; also Köymen 1963: 73 and 1966: 25, Kitapçı 1994: 172–4, Turan 1993a: 177 and Zetterstéen IA/2: 556ff). Tutuş was en route to Baghdad to refresh his allegiance to Sultan Malik-Shah when he learnt of his brother’s death. After the death of Süleyman-Shah, Tutuş is thought to have been reluctant to appear at court in case he was punished. Having hurried back to Damascus, he is said to have sent word to the governors of Aleppo, Antakya and Urfa, namely Aksungur, Yağı-Sıyan and Bozan. Thus Interregnum (1092–1105) 99 strengthened, Tutuş took Rahba (February 1093), then Raqqa and Habur. When he advanced on Nisibin, however, he met with resistance. On taking the city, Tutuş is said to have had twenty Arab commanders and 2,000 people killed, including those who had sought refuge in the mosques (March 1093). From Nisibin Tutuş advanced on Mosul where the Uqailid Ibn Quraysh had been reinstated. On her return to Isfahan, Terken Khatun is thought to have released him to gain an ally against Tutuş. Ibn Quraysh and his uncle were killed in a bloody battle outside the city even though Tutuş was outnumbered three to one (2 April 1093). Tutuş asked Caliph al-Muqtadi to have the khutba read in his name but was refused. The caliph is said to have asked to gain the allegiance of the remaining Seljuqids and control of the treasury in Isfahan. Tutuş advanced on south-eastern Anatolia, preventing the Marwanids from regaining control of their former domains. He assigned Diyarbakır and its environs to Yınaloğlu Ibrahim, Bitlis and its environs to Dilmaçoğlu Doğan-Arslan, Hisn Kayfa and its environs to Kızıl-Arslan. From here Tutuş is thought to have continued north into Azarbayjan where Aksungur is said to have persuaded Bozan to defect, forcing Tutuş to withdraw (Sevim 1965: 104–12 and Özaydın 2001: 30–6 and 2000: 406; also Köymen 1963: 74–5, Kitapçı 1994: 167 and Sümer 1992: 20–1). Although his name was being read after Mahmud’s in the Friday khutba and coins had been struck with both their names, Ismacil is thought to have felt threatened by Terken Khatun’s commander-in-chief. But when he defected to Berk-Yaruk who was in Rayy at the time, Ismacil was strangled or possibly even decapitated by Aksungur and Bozan (August–September 1093) (Özaydın 2001: 22; Köymen 1963: 74; Kitapçı 1994: 174–5). The commanders appear to have accompanied Berk-Yaruk to Baghdad (October 1093), where he was declared sultan (3 February 1094). Possibly while in Baghdad, Sultan Berk-Yaruk had his uncle Tekiş strangled with his bowstring for plotting with Tutuş (March 1094) (Özaydın 2001: 51). Having returned to Damascus (December 1093), Tutuş married his eldest son Ridwan to Yağı-Sıyan’s daughter before advancing on Aksungur with his aid (March–April 1094). Aksungur appears to have asked Sultan Berk-Yaruk for help. Having ordered Bozan to join Aksungur, Sultan Berk-Yaruk also sent Kür-Boğa, originally one of Aksungur’s mamluks (Özaydın 2001: 36 and 2000: 422; Kafesoğlu IA/6: 1084ff), as well as a force of Türkmen under Abukoğlu Yusuf, the military governor of Rahba. Aksungur is said to have been forced to attack Tutuş without Bozan and Kür-Boğa who were unable to cross the irrigation canal near Tel Sultan in time. Abukoğlu Yusuf and the Türkmen appear to have defected and Aksungur was captured (27 May 1094), forcing Bozan and Kür- Boğa to seek refuge in Aleppo. Tutuş is said to have had Aksungur and fourteen others decapitated. Before Sultan Berk-Yaruk could relieve Bozan and Kür-Boğa, some of their soldiery opened the gates (29 May 1094). Tutuş executed Bozan but on Üner’s intervention imprisoned his son-in-law Kür-Boğa and his brother Altun-Taş. From Aleppo Tutuş advanced on Urfa where he is said to have gained access by parading Bozan’s head on a spear (Sevim 1965: 112–16; Özaydın 2001: 37–40 and 2000: 406–7). 100 Interregnum (1092–1105) Terken Khatun now approached Tutuş with an offer of marriage, which he seems to have accepted. Although Terken Khatun set off from Isfahan once Tutuş had re-asserted his authority in Azarbayjan and began his advance on Hamadan, she was taken ill (possibly poisoned) and had to return. Terken Khatun died soon after (September 1094), but most of her forces appear to have joined Tutuş who is said to have dispatched Abukoğlu Yusuf to Baghdad with orders to have the khutba read in his name. In the meantime Abukoğlu Yusuf’s brother Yakup appears to have come upon Sultan Berk-Yaruk and defeated him. On receiving news of this Caliph al-Mustazhir (r. 1094–1118) had the khutba read in Tutuş’ name (October–November 1094). Although Berk-Yaruk was able to withdraw toward Isfahan with his atabeg Gümüş-Tekin, Porsuk and a few other loyal commanders, he was not allowed in. When he had negotiated entry he was tricked by Üner and imprisoned. However, Mahmud caught smallpox and died before Berk- Yaruk could be blinded (October 1094). Released, Berk-Yaruk is said to have also fallen ill but recovered (Sevim 1965: 116–19; Özaydın 2001: 40–1). Tutuş had been unable to take Hamadan. However, when its commander Inanç Yabgu gave pursuit and began to plunder his baggage train, Tutuş defeated him and took the city. In Hamadan Tutuş seems to have come across Nizam al-Mulk’s son Fakhr al-Mulk who had been ousted from Balkh by the sultan’s great-uncle, Arslan- Argun. Fakhr al-Mulk had been en route to Isfahan to see his elder brother Muayyad al-Mulk when he had been attacked and forced to flee. At first Tutuş is said to have wanted to execute him, but on Yağı-Sıyan’s instigation appointed him vizier to ensure the resident’s good will. While in Hamadan, on hearing that Berk-Yaruk was ill with smallpox, Tutuş seems to have asked the commanders in Isfahan to join him. Having stalled, they are said to have refused when Berk-Yaruk recovered. Berk-Yaruk now appointed his deceased vizier Izz al-Mulk’s brother Muayyad al-Mulk. Muayyad al-Mulk appears to have been able to rally other commanders to join Berk-Yaruk (December 1094). As a result, Sultan Berk-Yaruk appears to have gathered a large force that included the commander of Hamadan, Inanç Yabgu, who is thought to have defected while on a supply mission. Tutuş now withdrew from Hamadan, taking Rayy (January–February 1095). Possibly equally mistrustful of its inhabitants, Tutuş seems to have accepted battle outside the city (26 February 1095). Seeing Malik-Shah’s banners on the battlefield, many of his army are said to have either deserted or refrained from fighting. Captured on the battlefield, one of Aksungur’s commanders, Sungurca decapitated him (Sevim 1965: 119–21; Özaydın 2001: 41–4; Köymen 1963: 76, 78).

Coming of the Crusaders Süleyman-Shah’s son Kılıç Arslan and his younger brother Davud are said to have left for Anatolia on Sultan Malik-Shah’s death, with a large group of Türkmen referred to as the Yabgulu. On their arrival in Iznik, the deceased al-Kasim’s brother al-Gazi handed over the city, which was under pressure from the Byzantines (1092). As will be remembered, at the time Çaka Beg held the Aegean coast while the Danişmends held Ankara, Kayseri and Sivas, eastern Interregnum (1092–1105) 101 Anatolia being divided among various Türkmen begs, some of whom had been appointed by Tutuş (Turan 1993a: 95–7; also Cahen 1984: 97 and Sümer 1990: 2). In view of Byzantine advances along the Sea of Marmara, Kılıç Arslan’s first move appears to have been the establishment of good relations with Çaka Beg, whose daughter he married. As noted, although Çaka Beg’s attempts to ally with the Peçenek in Thrace and thus take Istanbul had fallen through, he does not appear to have given up. Possibly he hoped to renew his onslaught with Kılıç Arslan’s help. However, Kılıç Arslan’s forces, which had been sent to halt a fresh Byzantine advance, were defeated and their commander captured. Whether or not aware of this, when Çaka Beg advanced north along the Dardanelles and besieged Abydos, Emperor Alexius sent an offer of peace to Kılıç Arslan, warning that his father-in-law meant to usurp his domains, not Byzantium. Although Kılıç Arslan is thought to have taken heed, rather than being killed by his son-in-law, Çaka Beg most likely withdrew when Kılıç Arslan advanced (Turan 1993a: 93–8; also Kafesoğlu 1953: 112). Kılıç Arslan appears to have turned his attention on Malatya, ruled by the Armenian Gabriel. He is thought to have laid siege to the city to pre-empt the Danişmends (1095/6). While thus occupied Kılıç Arslan learnt of the arrival of the Crusaders (August/September 1096). Possibly because his brother Davud attacked and killed a large number of pilgrims outside Izmit, Kılıç Arslan seems to have taken his time. However, on arrival he found Iznik besieged by the Byzantines and the main body of Crusader knights (May 1097). He was forced to retire with heavy losses and the city surrendered to the Byzantines under the Peçenek Tadik (26 June 1097). Having asked for help from the Danişmend Gümüş-Tekin, Hasan Beg (the ruler of Kayseri) and other Türkmen begs, Kılıç Arslan rejoined battle with the Crusader knights outside Eskişehir but once again withdrew with heavy losses (4 July 1097). Kılıç Arslan was defeated one last time outside Konya (September 1097). After this he seems to have avoided shock combat against knights with body armour and long lances. Hasan Beg, however, was manoeuvred against the mountain now bearing his name near Kayseri and annihilated. Thus with the Crusaders’ help the Byzantines re-established their authority in the towns and cities along the Aegean, Mediterranean and Black Sea (1097), pushing the Türkmen back into the Anatolian steppe (Turan 1993a: 98–104; also Sevim 1989: 94–8 and Köymen 1963: 288–9). Meanwhile Tutuş’ sons were trying to establish themselves in Syria. The elder Ridwan was en route to Rayy with his atabeg, local Arab commanders and the deceased Aksungur’s soldiery when he learnt of his father’s demise. Presumed to have feared for his life, Ridwan retired to Aleppo where he appears to have negotiated the release of Kür-Boğa and his brother Altun-Taş in return for his younger brother Dukak. They were joined by Dukak’s atabeg, Yağı- Sıyan and Abukoğlu Yusuf (April/May and August 1095). Ridwan took Urfa after Artukoğlu Sökmen proclaimed allegiance in Suruç. Having been given the inner citadel of Urfa, Yağı-Sıyan began to plot against Ridwan with Tutuş’ former regent al-Kasim who appears to have been imprisoned by Ridwan’s atabeg, vizier and commander-in-chief Hussein. Alerted, Hussein returned to 102 Interregnum (1092–1105) Aleppo, so that Yağı-Sıyan and al-Kasim withdrew to Antakya with Abukoğlu Yusuf (1095/6) (Sevim 1989: 83–6; Özaydın 2000: 407–8). On their release Kür-Boğa and Altun-Taş are said to have been joined by soldiery in the region, enabling them to take Harran south-east of Urfa. The brothers then took Nisibin further to the east, which with Mosul to the south had reverted to the Uqailids. When they besieged Mosul, the Uqailid Ali asked help of Çökürmüş. Although Altun-Taş defeated Çökürmüş, Kür-Boğa seems to have won him over and Çökürmüş joined the siege. After nine months Mosul also fell (October–November 1096), its ruler fleeing to Hilla. Prevented from sacking the city, when Altun-Taş confiscated the notables’ wealth, Kür-Boğa is said to have executed his brother. After this Kür-Boğa also took Rahba to the south-west and had the khutba read in Sultan Berk-Yaruk’s name (Özaydın 2000: 408–10). While Kür-Boğa was busy appropriating the Uqailids’ realms, Dukak seems to have been secretly invited to Damascus by its commander Sav-Tekin. Ridwan besieged Damascus while Dukak was on campaign but appears to have been unsuccessful because of Artukoğlu Sökmen’s departure for Jerusalem, which had been granted to him and his brother Il-Gazi by Tutuş. Although joined by Il-Gazi, who had been released by Dukak, Sökmen was unable to defend Jerusalem against the Fatimids (August 1096). Undaunted by Sultan Berk-Yaruk’s acceptance of Dukak as his vassal, once again Ridwan seems to have advanced on Damascus only to find that Dukak had been joined by Yağı-Sıyan. Presumably on Sökmen’s instigation, Ridwan appears to have changed plans and moved on Jerusalem. Dukak is thought to have taken advantage and advanced on Aleppo, raiding the countryside (January/February 1097). He was met by Ridwan and Sökmen who defeated him and Yağı-Sıyan (22 March 1097). As a result, Dukak accepted to have the khutba read in his brother’s name (Sevim 1989: 86–90; also Kafesoğlu IA/6: 1084ff). Said to have been disturbed by the influence the Shicites had over Ridwan, his second-in-command Hussein appears to have left for his iqtac in Hims with his wife, Ridwan’s mother, and soldiery. Approached by the Fatimids with a promise of financial and military aid, Ridwan agreed to have the khutba read in Caliph Mustacli’s name. However, this brought him under pressure not only from Sultan Berk-Yaruk and the Abbasid caliph, but also from Yağı-Sıyan, who had traken over Hussein’s functions, and Sökmen. Ridwan retracted, sending letters of apology (September 1097). As he was preparing for another campaign against Dukak and Hussein with the help of Yağı-Sıyan and Sökmen, news arrived of the Crusaders’ approach (Sevim 1989: 90–3). Returning to Antakya before the Crusaders’ arrival (20 October 1097), Yağı-Sıyan evacuated most of its Christian inhabitants, sending his son Shams al-Dawla to Ridwan, Dukak, Hussein and the Banu Kilab for help. His younger son Muhammad was dispatched to ask help of Kür-Boğa, Artukoğlu Sökmen, Arslan-Taş, Il-Gazi Süleyman and other Türkmen begs in south-eastern Anatolia. Dukak, his atabeg Tuğ-Tekin and Hussein appear to have responded immediately. They came across Bohemond and Robert of Flanders en route to Antakya with Shams al-Dawla (31 December 1097). Badly defeated, the Seljuqs withdrew to Interregnum (1092–1105) 103 Hamah near Hims from where Shams al-Dawla appears to have journeyed north to Aleppo. Ridwan put a local contingent with Sökmen and his Türkmen under his command (February 1098), but before they could reach Antakya they were attacked by the Crusaders and withdrew to Harim, where Ridwan’s soldiery is said to have returned to (March 1098). When Shams al-Dawla and Artukoğlu Sökmen also departed to join Kür-Boğa, apparently the Armenians massacred the town’s Muslim inhabitants and took control (Özaydın 2000: 411–12). That same month Godfrey of Bouillon took Urfa from the Armenians (10 March 1098). Appointed by Sultan Berk-Yaruk to deal with the Crusaders, this is thought to have diverted Kür-Boğa because possibly it threatened his right flank. After a futile three-week siege (4–25 May 1098), he is said to have given up and finally taken command of the forces gathered at Marj Dabik, north of Aleppo, under Dukak, his atabeg Tuğ-Tekin, Ridwan’s atabeg Hussein, Arslan- Taş and Artukoğlu Sökmen. When the pilgrims outside Antakya heard of this, many are said to have panicked and left. Bohemond and the Crusader knights, now supported by Godfrey of Bouillon, continued the siege (29 May 1098). Possibly Bohemond was encouraged by his secret negotiations with an Armenian armourer, Firuz, whose stocks had been confiscated by Yağı-Sıyan as punishment for hoarding. Having come to an agreement with Bohemond through his son, Firuz is said to have allowed the Crusaders to scale the tower he was charged with defending (2–3 June 1098). Caught totally unawares, Yağı-Sıyan seems to have thought the citadel had fallen and left the city with some horsemen. When he learnt his son Shams al-Dawla still held the citadel, however, he is said to have become so distraught he fell off his horse repeatedly and was left behind. An Armenian woodcutter decapitated him and received a bounty from Bohemond (Özaydın 2000: 413–15; also Köymen 1963: 290, Kafesoğlu IA/6: 1084ff, Sevim 1989: 100–3 and Turan 1993a: 179). On learning that Antakya had fallen, Kür-Boğa dispatched advance units to Demir Köprü (al-Hadid) on the Âsi River (Orontes) and also to the fortified town of Artah. The Crusaders defending Demir Köprü were routed (5 June 1098). From captives taken there, Kür-Boğa learnt that the citadel had not as yet fallen. His units camped outside Antakya and within a few days had the city totally surrounded (7–10 June 1098). Although Shams al-Dawla proclaimed his allegiance, Kür-Boğa replaced him with his own commander whom he charged with infiltrating the city from the citadel. Aware of the danger, Bohemond and Raymond are said to have hurriedly constructed a wall separating it from the city. This appears to have enabled Raymond of Flanders, Raymond of Normandy and Hugue of Vermandois to withstand a fierce attack launched from the citadel. As a result Kür-Boğa is thought to have decided to starve the Crusaders into submission and decamped on to the plain from the mountains south-east of Antakya. Growing desperate, the Crusaders seem to have sent envoys to Dukak and Kür-Boğa while trying to persuade the Fatimids to open a second front. Having escaped, when William informed Alexius Comnenus that the Crusaders had been defeated, the Byzantines preparing to relieve them are said to have returned to Istanbul, concluding that they could no longer regain parts of Syria as agreed (Özaydın 2000: 415–16). 104 Interregnum (1092–1105) Kür-Boğa appears to have been busy with his own affairs. For reasons best known to him, he is said to have received Ridwan’s envoy Abukoğlu Yakup, whose brother Yusuf had been killed on Hussein’s orders. This is thought to have made Hussein and Dukak suspicious, weakening their allegiance to Kür-Boğa. Abukoğlu Yakup seems to have sown dissent among the Banu Mirdas Wassab’s soldiery, too, so that some of the Türkmen under his command left. Confident they would surrender from hunger, Kür-Boğa is said to have refused an offer by Peter the Hermit and Herluin to hand back Antakya in return for safe conduct home (27 June 1098). The next day he allowed the Crusaders to leave the city in small parties, refusing to pick them off so as to annihilate them. When the Crusaders attacked, Dukak, who is said to have received news that the Fatimids were advancing on Damascus, seems to have refused battle and departed. His departure is thought to have forced Kür-Boğa and Artukoğlu Sökmen and Hussein, both of whom had been waiting in ambush, to withdraw. The Muslim irregulars who had volunteered were thus left behind to face the Crusaders who initially thought Kür-Boğa was feigning retreat (28 June 1098). Kür-Boğa’s commander in the citadel surrendered and having converted to Christianity joined Bohemond (Özaydın 2000: 416–18; also Sevim 1989: 88, 103–6, Köymen 1963: 291, Kafesoğlu IA/6: 1084ff and Turan 1993a: 179).

The rise of Muhammad Tapar Sultan Berk-Yaruk’s great uncle Arslan-Argun, assigned Khwarazm by Sultan Alp-Arslan, had tried to occupy Nishapur, but the inhabitants resisted. Nevertheless, after having taken Balkh from Nizam al-Mulk’s son Fakhr al-Mulk with help from the commanders for Marv, Kavdan (or Kodan) and Yaruk-Taş, Arslan-Argun gained not only Nishapur but also Tirmiz. Now effectively in control of Khurasan as well as Khwarazm, Arslan-Argun is thought to have made it clear to Muayyad al-Mulk that he did not want to challenge Sultan Berk-Yaruk. For his part Muayyad al-Mulk seems to have persuaded the young sultan to send his uncle Böri-Pars against Arslan-Argun. Having been defeated by Böri-Pars, who was supported by his vizier ‘Imad al-Mulk, Muayyad al-Mulk’s brother, and the commanders Mascud and Altun-Taş (early 1095), Arslan-Argun seems to have withdrawn to Balkh rather than Khwarazm. Böri-Pars is thought to have returned to Herat, which he had been assigned by Sultan Malik-Shah (Özaydın 2001: 47–51; also Köymen 1963: 77–8 and Turan 1993a: 178). On defeating Tutuş (26 February 1095), Sultan Berk-Yaruk had sent for his mother Zubaydah Khatun despite Muayyad al-Mulk who, after gaining the support of a group of commanders, advised him not to. Majd al-Mulk Abu’l-Fadl al-Balasani, a member of the diwan sent to accompany Zubaydah Khatun is thought to have told her of this. Having fallen out with his brother over their father’s jewellery, this appears to have encouraged Fakhr al-Mulk to send Sultan Berk-Yaruk a considerable sum of money, requesting his brother’s office. As a result Muayyad al-Mulk was imprisoned and Fakhr al-Mulk appointed vizier (1095). Nevertheless, Majd al-Mulk is said to have been able to use his position Interregnum (1092–1105) 105 with Zubaydah Khatun to undermine Fakhr al-Mulk’s authority to such an extent as to render his appointment meaningless (Özaydın 2001: 12–13). Possibly encouraged by Muayyad al-Mulk’s demise, Arslan-Argun retook Marv with a Türkmen force. After his commander Ahur had the sultan’s commander Mascud and his son killed by pretending to desert him, Arslan-Argun captured Böri-Pars and imprisoned him in Tirmiz where he was strangled with his own bowstring. Böri-Pars’ vizier ‘Imad al-Mulk was also killed and his wealth confiscated. Sultan Berk-Yaruk is known to have taken his brother Sancar with him to deal with their great-uncle, but Arslan-Argun seems to have been strangled by one of his slaves before they arrived (3 February 1097). Sultan Berk-Yaruk pardoned his son Alp-Argun and appointed him to an undisclosed place. Having appointed Sancar to Khurasan, Sultan Berk-Yaruk is said to have ordered the commander for Khwarazm Ekinci to join him. Instead Ekinci appears to have left his soldiery and gone to Marv, where he was killed by the deceased Arslan-Argun’s commanders, Kavdan and Yaruk-Taş, while inebriated. After they appear to have gone to Khwarazm and claimed it, saying they had been sent by Sultan Berk-Yaruk. When the sultan sent the commander Habashi to Khwarazm, Yaruk-Taş attacked without waiting for Kavdan and captured him. On hearing of this Kavdan’s soldiery are said to have rebelled and plundered his treasury. Kavdan fled to Bokhara where for a time he was imprisoned. Having joined Sancar in Balkh, he died soon after. At the time Sancar appears to have been busy putting down a rebellion by Çağrı b. Süleyman’s son Muhammad, who was married to one of Arslan-Argun’s daughters. Blinded on being captured, Muhammad appears to have been supported by the Ghaznavid Sultan Ibrahim (1059–99) (Özaydın 2001: 51–3). After his victory over Tutuş, Sultan Berk-Yaruk appears to have appointed Terken Khatun’s former commander Üner governor of Fars. Terken Khatun had dispatched Üner to take the province from the Seljuqs of Kirman (June/July 1094), but Üner had been routed by Turan-Shah. Now the Shebankarids, who appear to have to become re-established in Fars after Turan-Shah’s death, asked for help from his successor Iran-Shah (1097–1101). Once again Üner was defeated and withdrew to Isfahan (1098–9). Here he seems to have come under the influence of Muayyad al-Mulk. On having gained his freedom, Muayyad al-Mulk had sought refuge with Sancar’s uterine brother Tapar in Ganja. Tapar had been in Baghdad when Sultan Malik-Shah died. From there he had been taken to Isfahan by Terken Khatun, but had escaped during its siege by Berk-Yaruk who took him to Baghdad. On having been declared sultan, Berk-Yaruk had assigned him Ganja with an atabeg (November 1093–February 1094). Having had his atabeg killed, Tapar seems to have been in the process of trying to include Arran when Muayyad al-Mulk joined him as vizier (Özaydın 2001: 11–13; also Köymen 1963: 79). Whatever Tapar’s ultimate ambitions, it is Muayyad al-Mulk who is credited with fanning the flames of contention with Sultan Berk-Yaruk. The sultan’s problems seem to have started with his commander Porsuk’s assassination by batinis, which his sons Zengi and Ak-Böri are said to have ascribed to al-Balasani. As Tapar advanced from Ganja after omitting Sultan 106 Interregnum (1092–1105) Berk-Yaruk’s name from the khutba, Muayyad al-Mulk’s men fanned the rumours against al-Balasani. At the same time Üner advanced from Isfahan, threatening that unless Sultan Berk-Yaruk handed over al-Balasani he would rebel. Although he was killed en route by his soldiers, his commanders Inanç Yabgu, Bilge Beg and Toga-Yürek asked Porsuk’s commanders to support them in return for their allegiance. Sultan Berk-Yaruk is said to have received their ultimatum near Hamadan. Apparently the sultan refused, but when al-Balasani was killed nevertheless he left for Rayy with a few horsemen. The rest of his army then joined Tapar who had arrived nearby. In Rayy, Nizam al-Mulk and Anuş-Tekin’s sons, Mansur and Yınal, joined the sultan along with his mother Zubaydah Khatun (Özaydın 1990: 14, 23, 54–5 and 2001: 57–8; also Köymen 1963: 80). Pursued by Tapar, Sultan Berk-Yaruk left for Isfahan where the inhabitants are said to have refused him entry. Nevertheless, he seems to have left his mother there. With Tapar in Rayy (20 September 1099), Muayyad al-Mulk appears to have gone on to Isfahan where he had Zubaydah Khatun imprisoned, forcing her to sign a promissory note. Although counselled that her death might provoke the soldiery, he had her brought to Rayy and strangled. The commander of Baghdad, Gawhar-A’in, and still others such as Çökürmüş in Jazira now joined Tapar in Qumm. Tapar then sent Gawhar-A’in to Baghdad, requesting the khutba was read in his name; the caliph complied (4 November 1099). Meanwhile Berk-Yaruk appears to have plundered Wasit with Anuş-Tekin’s son Yınal before advancing on Baghdad with the ruler of Hilla, Sayf al-Dawla Sadaqa. En route he is said to have survived an assassination attempt by the commander for Isfahan (Özaydın 1990: 15–16 and 2001: 58–60; also Köymen 1963: 80–1). Hearing of Berk-Yaruk’s approach, Gawhar-A’in, Artukoğlu Il-Gazi and others sent word for Sultan Tapar and Muayyad al-Mulk to join them in Baghdad. Instead the sultan appears to have dispatched Kür-Boğa and Çökürmüş. The latter is said to have excused himself, claiming problems at home in Jazira. The commanders then sent word to Berk-Yaruk that they would not oppose him. Possibly isolated, Gawhar-A’in and Kür-Boğa also seem to have proclaimed allegiance to Berk-Yaruk, returning to Baghdad with him (2 January 1100). The city’s inhabitants are said to have welcomed Berk-Yaruk who had succeeded in having the khutba read in his name a few days earlier (31 December 1099). When the caliph sent his vizier to greet him, however, he was arrested in lieu of taxes owed by his father from Diyarbakır and Jazira. On hearing of his predicament, the caliph is said to have sent a letter threatening that the sultan should not be fooled by his inaction to date given his previous mistakes. The confrontation was resolved when the vizier promised to pay up (Özaydın 1990: 16–18 and 2001: 60–1; also Köymen 1963: 82). Since the caliph appears to have had estates near Wasit, the ‘mistakes’ referred to may have been their plunder. After having wintered in Baghdad, where he replaced his vizier Fakhr al-Mulk with al-Mahasin Dihistani, Sultan Berk-Yaruk advanced on Tapar with the support of numerous Türkmen (17 April 1100). They met near Hamadan (15 May 1100), but the sultan was defeated, having to flee. Captured on the battlefield, his Interregnum (1092–1105) 107 vizier was released and sent to Baghdad to arrange for the khutba to be read in Tapar’s name, which was effected (25 May 1100). Sultan Tapar also asked for the caliph’s vizier to be dismissed, which seems to have been accepted. The vizier and his brothers were imprisoned and a sum of money confiscated (10 July–8 August 1100) (Özaydın 1990: 18–20 and 2001: 62–4; also Köymen 1963: 82–3). Meanwhile Berk-Yaruk is said to have approached the commander for Tabaristan and Jurjan, Habashi, who had fallen foul of Sancar. Although willing, apparently Habashi was unable to join Berk-Yaruk who came to his aid instead. Sancar defeated Berk-Yaruk and Habashi who was handed over by the Türkmen in a nearby village. Habashi’s cavalry are said to have been supported by Ismacili foot soldiers. Having escaped to Damghan, Berk-Yaruk was joined by Çavlı Sakavu and some other commanders. With them he advanced on Isfahan, asking Porsuk’s sons Zengi and Il-Begi for help, but Sultan Tapar appears to have arrived before him. After a couple of months Berk-Yaruk appears in Hamadan, where Ayaz, who had become commander upon Inanç Yabgu’s assassination, also joined him. Ayaz seems to have been of the opinion that Inanç Yabgu had been poisoned by Muayyad al-Mulk. Seeing the balance of power shift to Berk-Yaruk, Sultan Tapar advanced on Hamadan. Berk-Yaruk is said to have received twelve consignments of weaponry on the morning before the battle, eight of which were shields for his foot soldiers. Outnumbered, Sultan Tapar lost and Muayyad al-Mulk was taken prisoner by Majd al-Mulk (5 April 1101). Berk-Yaruk is said to have personally killed Muayyad al-Mulk despite an offer of money for the vizierate (Özaydın 1990: 20–3 and 2001: 64–7; also Köymen 1963: 83–4). Sultan Tapar appears to have withdrawn to Damghan where his brother Sancar now joined him. Apparently Sancar’s soldiery pillaged the city, forcing the inhabitants to take refuge in the nearby citadel of Girdkuh. Sultan Tapar and Sancar then journeyed to Rayy where the Nizamiyya joined them. For his part Berk-Yaruk is said to have attracted a massive following after his victory, which could not be supplied. Whether or not for this reason, many of the commanders left Berk-Yaruk under various pretexts while Kür-Boğa was dispatched to put down Ismacil b. Yakuti’s son Mawdud’s rebellion (Özaydın 1990: 23–6 and 2001: 67–8; also Köymen 1963: 85–8). When Sultan Tapar and Sancar advanced on Hamadan, Porsuk’s sons are said to have refused to help Berk-Yaruk and forced him to return to Baghdad. There the caliph once again had the khutba read in his name. Sultan Berk- Yaruk appears to have been joined by the commander for Hamadan, Ayaz, who had been relieved of his iqtac by Tapar despite a proclamation of allegiance. Regardless of his title, Sultan Berk-Yaruk appears to have been so short of money as to have asked the caliph for funds. Despite an infusion from him, however, Sultan Berk-Yaruk’s soldiery plundered Baghdad. When the sultan asked Sadaqa for the money left from the Public Treasury and back taxes, Sadaqa added Kufa to his dominion and had the khutba read in Tapar’s name. As Tapar and Sancar approached Baghdad with the inclusion of Artukoğlu Il-Gazi, Sultan Berk-Yaruk became ill. He was evacuated on a litter but fell into a coma in Ramla, on the western shore of the Tigris, where Tapar’s advance 108 Interregnum (1092–1105) units caught up with him. They are said to have shouted across the lines that he and his soldiery were batinis, a rumour that had been circulating in Baghdad. After withdrawing from Ramla, the sultan’s soldiery seems to have continued pillaging all the way to Wasit (Ibid. and Özaydın 2001: 68–71). Although Tapar and Sancar arrived in Baghdad and the khutba was changed yet again, they left soon afterwards due to Karakhanid incursions into Khurasan (23 October–11 November 1101). Köymen was of the opinion that the sultan did not as yet have commanders whose allegiance he could trust in the same way Sancar could. Having recovered from his illness, Berk-Yaruk seems to have occupied some of the caliph’s estates around Wasit and begun to speak ill of him, so that the caliph recalled Sultan Tapar and offered to join in the fight against Berk- Yaruk, which was turned down. Coming to Baghdad long enough to appoint a new mayor and Il-Gazi as commander, the sultan is said to have rejoined his troops. When Sultan Tapar and Berk-Yaruk met, however, the severe cold prevented battle. Next day their soldiery embraced on the field and the sultan’s commanders brokered a treaty with Berk-Yaruk’s vizier (27 December 1101). According to this, Tapar accepted Berk-Yaruk as sultan and as malik of Azarbayjan, Ganja, Diyarbakır, Jazira and Mosul, also accepted to pay an annual tax. In return, Sultan Berk-Yaruk agreed to help Tapar whenever necessary. Sultan Berk-Yaruk also accepted Sancar as malik in Khurasan (Özaydın 1990: 27–8 and 2001: 70–3; also Köymen 1963: 88–90). Angered at being forced to sign, Tapar had his commanders Besmel assassinated and Ay-Tekin blinded. When the sultan’s commander Yınal, who had joined the Ismacilis, defected along with another, Tapar declared himself sultan in Rayy. Nevertheless, when he took the field against Sultan Berk-Yaruk, his soldiery dispersed without bloodshed and his treasury was plundered (February/ March 1102). Despite this setback, Tapar does not appear to have had any problem entering Isfahan, where he was besieged by Sultan Berk-Yaruk. Tapar seems to have escaped (25 September 1102), but was caught up by Ayaz who is said to have confiscated his banners. Tapar appears to have been en route to Hamadan where he was joined by fresh troops from Azarbayjan under Kızoğlu, Nizam al-Mulk’s son Mansur and Muayyad al-Mulk’s son Muhammad. Having lifted the siege of Isfahan (6 October 1102), Sultan Berk-Yaruk now advanced on Hamadan. Tapar is said to have split his forces (15 October–13 November 1102), sending Yınal and his brother Ali to Rayy. Their purpose seems to have been to raise money, which they appear to have done by force before leaving (December 1102). From Hamadan Tapar seems to have journeyed to Ardabil on the Caspian Sea, where his brother-in-law Mawdud, the deceased Ismacil’s son, joined him. Although Mawdud died unexpectedly (January 1103), his commanders proclaimed allegiance to Tapar who headed back into Azarbayjan where Sultan Berk-Yaruk caught up with him, inflicting another bloodless defeat (19 February 1103). Apparently from there Tapar fled to Ahlat. However, he seems to have returned to Tabriz with fresh troops. It is at this point Sultan Berk-Yaruk finally offered peace though ostensibly victorious (Özaydın 1990: 28–33 and 2001: 73–8; also Köymen 1963: 90–2). Interregnum (1092–1105) 109 In Özaydın’s view Berk-Yaruk was not only ill but also most likely realized the commanders’ endless demands could not be satisfied. After some negotiation (2–31 January 1104), the half-brothers effectively split the Great Seljuq Empire between them on the condition the commanders and military governors could change allegiance whenever they wanted. Berk-Yaruk’s name was to be excluded from the khutba in towns and cities in Tapar’s dominion and correspondence between the two conducted through their viziers. With the exception of Jazira and Sadaqa’s dominions in Lower Mesopotamia, Iraq al-Arab and al-Ajam were to be Berk-Yaruk’s while the remainder of the Great Seljuq Empire was to be Tapar’s. Finally, Tapar was to be Great Sultan after Berk-Yaruk (Özaydın 1990: 34–5 and 2001: 78–80; also Köymen 1963: 92). Although Sultan Tapar’s commanders handed over Isfahan to Great Sultan Berk-Yaruk, Çökürmüş refused to hand over Mosul to Sultan Tapar, claiming Great Sultan Berk-Yaruk had ordered him not to. Sultan Tapar advanced on Mosul and besieged it (23 October–20 November 1104). On learning of Great Sultan Berk-Yaruk’s death Çökürmüş proclaimed allegiance to Sultan Tapar (28 January 1105) (Özaydın 1990: 40–1). Great Sultan Berk-Yaruk is said to have died of tuberculosis and haemor- rhoids en route to Baghdad (22 December 1104). In accordance with his wishes, the khutba was read in his five-year-old son Malik-Shah’s name (6 January 1105). When Sultan Tapar arrived in Baghdad (10 February 1105), Great Sultan Malik-Shah II’s atabeg, Ayaz prevaricated. Against the wishes of some of his commanders, he finally proclaimed allegiance to Tapar, ostensibly reuniting the Great Seljuq Empire (13 February 1105) (Özaydın 1990: 39, 41–2).

Anatolia and Syria While Berk-Yaruk and Tapar fought over the Great Seljuq Empire the First Crusade progressed. After Kür-Boğa’s failures in Urfa and Antakya, Raymond of Toulouse took the citadel of Ruj and subsequently Godfrey of Bouillon besieged Bara, which also belonged to Ridwan. When the commander of Azaz, Omar, declared his independence, Ridwan besieged the citadel (September 1098). Through his French wife, Omar asked Godfrey for help. Faced by superior numbers, Ridwan is said to have lifted the siege but Godfrey seems to have been equally reluctant to engage in battle and appears to have likewise withdrawn. When Ridwan returned Omar capitulated. However, the Crusaders appear to have taken Bara and Ma’arrat al-Nu’man (November and 12 December 1098). Although Ridwan regained Qalla, which had also been taken by the Crusaders, Bohemond of Antakya defeated him (July 1101). Bohemond and Tancred are thought to have been preparing to move on Aleppo when Malatya came under pressure from Danişmend Gümüş-Tekin. Taking advantage of Bohemond’s hurried departure Ridwan is said to have raided the Crusaders supply dumps while Hussein recovered some of the lesser citadels (Sevim 1989: 106–9). Apart from Hasan Beg’s defeat and death, the Danişmends appear to have been unaffected by the First Crusade. Having captured Bohemond (1101), the 110 Interregnum (1092–1105) Danişmendid Gümüş-Tekin annihilated a relief force that had already been badly mauled by Sultan Kılıç Arslan. Malatya fell to Gümüş-Tekin despite a further attempt to relieve it by Baldwin of Urfa (1102). The strengthening of the Danişmends is thought to have prompted Sultan Kılıç Arslan to make peace with Emperor Alexius I Comnenus, ceding his territories along the Aegean and Mediterranean seas (Turan 1993c: 105). In Syria Ridwan is said to have settled his differences with his former atabeg, Hussein, in Hims. Nevertheless, he is credited with Hussein’s assassination by three Ismacilis during Friday prayers (May 1103). Hussein’s death seems to have prompted Raymond to attack Hims. Learning of this, Hussein’s wife who was also Ridwan’s mother is said to have summoned her son. Fearful they might suf- fer because of Ridwan’s enmity toward the deceased Hussein, the city leaders invited Dukak instead. Despite Dukak’s absence at the time his regent Ay-Tekin appears to have responded immediately, forcing both Ridwan and Raymond to retire. Ridwan now came under pressure from Tancred who began to systematically raid the environs of Aleppo. Reluctant to take the field, apparently Ridwan tried to pay him off but was unable to stem the pillaging. It is at this juncture that Ridwan appears to have received a request from Sultan Kılıç Arslan for supplies. The sultan is said to have been planning a campaign on Antakya. Regardless of whether or not Ridwan was in a position to comply, this plan seems to have collapsed when Gümüş-Tekin released Bohemond on payment of his ransom and started negotiations with the Byzantines for Richard’s ransom (1103). Angered, Sultan Kılıç Arslan marched on Gümüş-Tekin instead, defeating him (Sevim 1989: 109–11; also Turan 1993c: 107 and Özaydın 1990: 59, 76). Evidently growing more desperate, Ridwan appears to have approached Artukoğlu Il-Gazi in Mardin and other Türkmen begs in south-eastern Anatolia. Il-Gazi is credited with persuading Ridwan that they could not succeed against the Crusaders without Çökürmüş’ soldiery. When as a result they besieged Nisibin near Mosul in order to wrest it from Çökürmüş (presumably because he refused to help), Arslantaşoğlu was wounded and retired to his iqtac in Sincar. Although incapacitated by a stroke, Çökürmüş seems to have returned in time to subvert some of Ridwan’s commanders, promising Ridwan funds and soldiery if he imprisoned Il-Gazi. Instead Ridwan appears to have asked Il-Gazi to allow Çökürmüş to join them, which he is said to have refused. More inclined to fight Sultan Berk-Yaruk and hence possibly his suggestion that Ridwan move against Çökürmüş in the first place, Il-Gazi was captured by Ridwan and sent to Nisibin. When Il-Gazi’s Türkmen began to pillage the immediate environs, however, Ridwan is said to have withdrawn (Sevim 1989: 113–14). Once the struggle between Sultan Berk-Yaruk and Tapar was resolved, Çökürmüş appears to have acted without hesitation. Combining with Artukoğlu Sökmen he defeated the Crusaders under Baudouin of Bourg, Joscelin of Courtenay, Bohemond and Tancred in Harran, south of Urfa, capturing Baudouin and Joscelin (7 May 1104). Although Ridwan regained many of the citadels in the environs of Aleppo and Hims as a result, he lost all his gains when he took the field against Tancred and was defeated (April 1105). In the meantime Dukak Interregnum (1092–1105) 111 had died of tuberculosis (1104). Although at first the khutba was read in his one- year-old son Tutuş’ name, Dukak’s atabeg, counsellor and commander-in-chief Tuğ-Tekin released Dukak’s twelve-year-old brother Er-Taş and had the khutba read in his name (September 1104). Soon after both Tutuş and Er-Taş appear to have died, so that Tuğ-Tekin took control of Damascus and its environs on behalf of Sultan Tapar (April 1105). Around this time Danişmend Gümüş-Tekin also died (1104/5), Tapar acceded to the Seljuq throne on Sultan Berk-Yaruk’s death (1105), and Sultan Kılıç Arslan took Malatya from the Danişmends (2 September 1106) (Sevim 1989: 111–12, 162–6; Özaydın 1990: 59, 90, 99; Turan 1993c: 107). 7 Dissolution (1105–94)

Sultan Tapar and Great Sultan Sancar’s reigns oversaw the century-long dissolution of the Great Seljuq Empire. The end came after the crushing defeat at Katvan (1141), inflicted by the Turko-Mongol forces of the Kara-Khitay. It left the Great Seljuq Empire unable to deal with Great Sultan Sancar’s defeat and capture by none other than the Oguz (1153–7). The process can be said to have started in the previous decade, during the interregnum, but Sultan Tapar and his sons do not seem to have been able to regain control of their commanders as Great Sultan Sancar did leading them by example. It is possible Sultan Tapar was reluctant to lose face in the case of defeat by the Crusaders, but he never took charge against rebellious commanders either. Having conducted the siege of Shahdiz and defeated the rebellious Sadaqa, he appears to have been satisfied to assign various commanders to besiege Urfa or to campaign in Syria. Urfa gave access to Aleppo. Having suffered a crushing defeat at Harran immediately south of Urfa (1104), the Crusaders were forced to try and take Aleppo or Damascus from Antakya or Jerusalem, failing repeatedly. Their heavily armoured mounted knights and poorly equipped foot soldiers never properly overcame the desert conditions. Although Urfa did not fall until 1144, the various Seljuq commanders besieging it appear to have systematically decimated the countryside, forcing the Crusaders to vacate the eastern banks of the Tigris and thus making it almost impossible to mount an attack on Aleppo from Urfa. In the few months after Sultan Tapar’s death the heir apparent Mahmud and his brothers re-enacted the state of affairs that followed Sultan Malik-Shah’s death. Mahmud was thirteen and his eldest brothers Mascud and Tuğrul only nine and eight. Sancar put an end to the fighting started by the commanders, which involved Sadaqa’s son Dubays. Sancar defeated and captured Sultan Mahmud at Sawa (1119). However, given that a year elapsed between Sancar’s arrival in Rayy and the victory at Sawa, it would seem Sultan Mahmud did not give up without a struggle. Great Sultan Sancar then permanently weakened the Seljuqs of Iraq as they have come to be referred to by stripping Sultan Mahmud of various provinces and openly favouring his brother Tuğrul. Sultan Mahmud countered, involving Caliph al-Mustarshid against Great Sultan Sancar. This started a series of events during which the caliph and his successor al-Rashid took the field against the Seljuqs and lost their lives, ostensibly to Nizari assassins. Dissolution (1105–94) 113 After Great Sultan Sancar had lost the majority of his commanders and their soldiery at Katvan, the reigning sultan in Iraq was unable to gather a force strong enough to counter the Oguz. Having defeated, captured and then caged Great Sultan Sancar, the Oguz took him along while they sacked Khurasan repeatedly. Khurasan was finally taken over by one of Great Sultan Sancar’s vassals, the Khwarazmshahs. The last Great Seljuq, Sultan Tuğrul, died on the battlefield fighting the Khwarazmshah Tekiş (1194). This chapter continues to follow Özaydın’s Sultan Muhammed Tapar Devri Selçuklu Tarihi (498–511/1105–1118) (1990) for the reign of Sultan Tapar and Sevim’s Suriye-Filistin Selçuklu Devleti Tarihi (1989) for the Seljuqs of Syria. For the reign of Great Sultan Sancar and the Seljuqs of Iraq, the chapter follows Köymen’s Büyük Selçuklu I˙mparatorluğu – I˙kinci I˙mparatorluk Devri (1954). As in previous chapters, in this chapter Kafesoğlu’s Harezmşahlar Devleti Tarihi (first published in 1956), Sümer’s Oğuzlar (first published in 1965) and Turan’s Selçuklular Zamanında Türkiye (first published in 1971) have been referenced alongside others for supplementary material pertinent to the events being reviewed.

Unruly servants and vassals One event that possibly captures the atmosphere at the beginning of Sultan Tapar’s reign is the demise of the commander Ayaz. A fortnight after the khutba had been read in Sultan Tapar’s name, Ayaz gave a banquet in his honour (25 February 1105). The Seljuq vassal in Hilla, Sayf al-Dawla Sadaqa also seems to have attended. After Ayaz presented valuables that had belonged to Muayyad al-Mulk, he ordered his mamluks to fetch weaponry from the armoury as a further mark of respect for the sultan. Presumably in a festive mood, the mamluks are said to have forced a strangely attired scribe they came upon to put on chain mail under his woollen clothing. The scribe, however, seems to have escaped his tormentors and sought refuge with the sultan’s entourage. Becoming alarmed by the scribe’s sweaty, frightened demeanour, the sultan is said to have departed without notice when informed of the chain mail. Invited to discuss Kılıç Arslan’s advance in south-eastern Anatolia, Ayaz was decapitated (1 March 1105). Ayaz’s vizier was also caught and killed (17 May–5 June 1105) (Özaydın 1990: 43–4). When Böri-Pars’ son Mengü-Pars found himself in financial difficulties, he journeyed from Isfahan to Nihavand where he rebelled with the aid of the local commanders. Although Mengü-Pars approached the Porsuks, they had received a letter from their brother, Porsukoğlu Zengi, informing them he would be killed by Sultan Tapar if they helped. As a result, the Porsuks are said to have tricked Mengü-Pars into captivity and taken him back to Isfahan where he was imprisoned with Tekiş’ sons (1105–6). One of these, Böri-Tekin seems to have subsequently escaped. Unable to gain refuge in either Aleppo or Damascus, he continued to Cairo where the Fatimids put him on a retainer (Özaydın 1990: 44–5). The commander Ismacil, responsible for the policing of Rayy during Sultan Malik-Shah’s reign, had been assigned to Basra Sultan Berk-Yaruk. When Sultan 114 Dissolution (1105–94) Tapar sent a tax collector, Ismacil refused him entry. Rather than sending a force of his own, the sultan appears to have re-assigned the city to his long-time supporter, Sayf al-Dawla Sadaqa, ordering him to dispossess Ismacil. Although on Sadaqa’s instigation at first Ismacil appears to have allowed some taxes to be collected, he seems to have then arrested the man and confiscated the money. As a result, despite Ismacil’s proclamation of allegiance Sadaqa besieged Basra and having taken the city by force pillaged it ‘from end to end’ before appointing his own commander (10 February 1106). Soon afterwards Sadaqa added Tekrit to his realms in a similar manner. Situated near the Tigris between Mosul and Baghdad, Tekrit had been assigned to a certain Kaykubad from Dailam by Majd al-Mulk. Kaykubad appears to have been quite remorseless toward the inhabitants, on one occasion taking turns with Artukoğlu Sökmen to pillage the city (1098/9). Although Sultan Tapar assigned Tekrit to Porsukoğlu Aksungur, when he failed to take possession after a seven-month siege Kaykubad invited Sadaqa who duly took charge (October 1106) (Özaydın 1990: 45–7). Sadaqa was a Shicite, which appears to have enabled those jealous of his growing power, such as the deceased governor of Antakya Yağı-Sıyan’s son Muhammad to accuse him of being an Ismacili sympathizer. Sadaqa seems to have refused to return a local Shicite iqtac holder, Kayhusraw, who was fearful of the sultan’s wrath because accused of Ismacili sympathies. Learning the sultan was en route from Isfahan because of his acquisition of Tekrit, Sadaqa appears to have prepared his forces. When the caliph sent an envoy counselling Sadaqa not to rebel and offering to mediate on his behalf, Sadaqa is said to have expressed concern for his life even in the event of reconciliation. Presumably at the caliph’s instigation, the sultan seems to have sent his own envoy, ordering Sadaqa to assist Çavlı Sakavu against the Crusaders with Çökürmüş. Both Sadaqa and Çökürmüş appear to have declined. Sakavu appears to have had a reputation for cruelty and injustice, but had successfully prevented another Porsukid commander, Kür- Boğa’s nephew Mawdud b. Anuş-Tekin, from dispossessing him of Khuzistan and Fars on the sultan’s orders. After being besieged for eight months he is said to have journeyed to Isfahan and begged mercy. Sultan Tapar assigned him to Rahba and ordered him to mount a campaign against the Crusaders (Özaydın 1990: 47–8, 51–3; also Kafesoğlu IA/10: 375 and Turan 1993a: 183–4). Sultan Tapar appears to have been more interested in the citadel of Shahdiz (February–March 1107), which threatened the capital Isfahan. As well as having imposed taxation on the local populace, the leader of the Ismacilis in Isfahan and its environs, Hasan al-Sabbah’s mentor Ibn Attash’s son Ahmad, had been pillaging the surrounding countryside from Shahdiz. At first Sultan Tapar seems have been delayed by various rumours, namely that Baghdad had fallen to Kılıç Arslan and Khurasan was in turmoil. After these were proven to be unfounded, the citadel was completely surrounded (April 1107). Said to have been unable to procure supplies, Ahmad Attash seems to have asked for a fatwa, a proclamation legalizing the Ismacili Nizari creed in accordance with Shari law. Although many jurists are said to have been favourably disposed, Sultan Tapar does not appear to have been so inclined despite sending Sacid al-Hanafi and others to discuss the Dissolution (1105–94) 115 issue. He tightened the siege such that Ahmad Attash offered Shahdiz in return for safe conduct to the fortress of Khalinjan, further from Isfahan. When assassins sent to kill the commanders more active in the siege wounded one, the sultan ordered Khalinjan to be razed to the ground (Özaydın 1990: 79–80). On learning of this Ahmad Attash is said to have asked for safe conduct to Alamut instead, on the condition that one group from Shahdiz was escorted to the fortress of Nazir and another to that of Tabas. Although the sultan complied Ahmad Attash reneged, giving rise to rumours he was being aided by Sacd al-Mulk, the sultan’s vizier. Whether or not actually an Ismacili, Sultan Tapar seems to have decided he was plotting to have him assassinated. The vizier and four of his ‘associates’ were hanged from the city gates (15 July 1107). Shahdiz fell soon after. Ahmad Attash’s wife is said to have committed suicide by leaping from the walls, but he was captured and skinned alive after having been paraded around Isfahan (Özaydın 1990: 80–2). After Shahdiz Sultan Tapar seems to have been content to contain the Ismacilis, replacing ministers in his diwan from Iraq al-Ajam and al-Arab who were sus- pected of Nizari sympathies with others from Khurasan. After he had dealt with Sadaqa the sultan seems to have dispatched his vizier Ahmad al-Mulk and Çavlı Sakavu to besiege Alamut and a nearby fortress, but the sieges appear to have been lifted after the Ismacilis in their vicinity were hunted down and killed (1109). A few years later, Anuş-Tekin Şirgir is said to have taken the fortress of Bira and having given the inhabitants safe conduct to Alamut systematically destroyed the crops throughout the region (1111). Şirgir appears to have continued this war of attrition in the environs of Alamut until assigned several commanders for a final assault on Alamut (July 1117). But this likewise appears to have come to nothing, even if due to Sultan Tapar’s death (8 April 1118) (Özaydın 1990: 78, 82–4). After the fall of Shahdiz, the sultan is said to have dispatched Yağı-Sıyan’s son Muhammad and his Türkmen to take Wasit from Sadaqa (8 December 1107). Although the city was spared, the Türkmen began to systematically pillage Sadaqa’s domains. Sadaqa sent his cousin, but was defeated. While the Türkmen pursued his soldiery, killing those they could not capture, another group apparently ransacked Wasit with Sadaqa’s men. Although Muhammad is said to have stopped them, Wasit was assigned to Porsukoğlu Aksungur, also commander of Baghdad. Sultan Tapar now advanced on Sadaqa, but appears to have halted en route while the caliph tried one last time to mediate (18 January 1108). Having finally accepted the caliph’s order to obey the sultan, Sadaqa is said to have assigned his son to head an embassy (Özaydın 1990: 48). Seeing reconciliation in the offing, Muhammad appears to have started to pillage Sadaqa’s realms. Apparently Muhammad lost his life, but Sadaqa now recalled his son. The caliph is said to have sent yet another envoy with documents signed by the sultan, pardoning Sadaqa. Possibly interpreting this as a sign of weakness, Sadaqa demanded the reinstatement of Kayhusraw as well as rec- ompense for the Türkmen plunder. When the two armies met between Hilla and Wasit (3–4 March 1108), Sadaqa lost his life. Sultan Tapar released Sadaqa’s son Dubays after he proclaimed allegiance (Özaydın 1990: 48–51). 116 Dissolution (1105–94) The Crusaders Immediately prior to Muhammad Tapar’s accession, Bohemond had returned to Europe for recruits, putting Tancred in charge of Antakya. In turn, Tancred seems to have entrusted Urfa to Richard of Salerne, pending Baudouin’s release (September 1104). Baudouin had been captured by Çökürmüş (7 May 1104). Said to have been influenced by his good standing with the inhabitants, Sultan Tapar assigned Mosul to Çökürmüş (28 January 1105), who then moved on Urfa. Richard seems to have been inexperienced if not martially inept. A defensive sortie is said to have ended in disaster when the gates were shut behind him to prevent Çökürmüş from following his panicked soldiery into the city. However, Çökürmüş appears to have withdrawn after destroying the harvest and killing as many field hands as possible (Özaydın 1990: 91–3). When the following year Çökürmüş refused to support Çavlı Sakavu against the Crusaders, Sultan Tapar re-assigned Mosul to Sakavu (31 October 1106). Although Çökürmüş is said to have gained the help of al-Khwaja al-Kürdi in Irbil, Sakavu defeated and later killed him. However, the city dignitaries do not appear to have wanted Sakavu and asked Kılıç Arslan, Sadaqa and Porsukoğlu Aksungur to help Çökürmüş’ son Zengi to defend Mosul (Özaydın 1990: 51–3, 93–4). Having taken Malatya, Kılıç Arslan appears to have been besieging Urfa when he received their invitation. Although a good part of his soldiery is thought to have been in the Balkans, helping Alexius Comnenus against Bohemond, when Kılıç Arslan arrived in Nisibin Sakavu withdrew and Porsukoğlu Aksungur returned to Baghdad. Sadaqa is thought to have refused the invitation. With Sultan Tapar busy in Shahdiz, Kılıç Arslan entered Mosul uncontested and had the khutba read in his name (22 March 1107). Sakavu appears to have gone to Sincar and persuaded Artukoğlu Il-Gazi to help him take Rahba (26 February–19 May 1107). While thus occupied, he received an invitation from Ridwan to join against the Crusaders. Sakavu asked for help to relieve Mosul first, which Ridwan seems to have found acceptable, joining Sakavu as soon as he had made peace with Tancred. Although Kılıç Arslan advanced on Sakavu, the Türkmen begs are said to have left him en route because he was outnumbered. Kılıç Arslan was defeated and drowned in the Habur River while fleeing (3 June 1107). His son Malik- Shah was sent to Isfahan, but his youngest son, Tuğrul Arslan, was released with his mother, Aisha Khatun, who returned to Malatya (Özaydın 1990: 59–63; also Sevim 1989: 115–16 and Turan 1993a: 108–9). Having taken charge of Mosul, Sakavu imprisoned Il-Gazi and reneged on his agreement with Ridwan who quickly withdrew. Although Sultan Tapar had sanctioned his subsequent conquests in the environs of Mosul, Sakavu is said to have disobeyed repeated orders to support the sultan’s campaign against Sadaqa. As a result, after his victory over Sadaqa, the sultan ordered the Porsukid commanders Sökmen and Mawdud to dispossess Sakavu of Mosul. Leaving his wife, Porsukoğlu Porsuk’s daughter, to defend the city, Sakavu began to pillage its environs (14 April–13 May 1108). Sakavu’s wife seems to have stripped the inhabitants of whatever she could before they let in the sultan’s forces (September Dissolution (1105–94) 117 1108). Mawdud took command of Mosul and after imprisoning her in a tower for eight days is said to have given her safe conduct (Özaydın 1990: 53–4). Sakavu appears to have been joined by Sadaqa’s sons Badran and Mansur, as well as the new commander for Rahba. They are thought to have persuaded him to try and establish himself in Syria on grounds that the region had insufficient Muslim soldiery to counter the Crusaders. Learning of Sakavu’s intentions, Ridwan intercepted Baudouin’s ransom and took countermeasures. Establishing an alliance with the Arab Numayr tribe, on learning that he had lost the citadel of Balis on the Euphrates to Sakavu, Ridwan also sent word to Tancred (27 September 1108). Tancred is said to have been angry with Sakavu for releasing Baudouin because this had forced him to relinquish Urfa when its Patriarch intervened on Baudouin’s behalf (18 September 1108). Thus Tancred came to face Baudouin, Joscelin and Sakavu at Tell-Bashir, with Ridwan’s cavalry. Initially defeated Tancred fled, but when Sakavu’s men rode off with Baudouin and Joscelin’s spare horses, he returned and won the day. For his part Ridwan was able to retake Balis (November 1108) (Özaydın 1990: 99–102; also Sevim 1989: 116–17). Sakavu seems to have regained the sultan’s favour by capturing Tekiş’ son Bektaş and handing him over, in return being assigned Fars as atabeg to the sultan’s two-year-old son Çağrı (1108–9). During his seven-year tenure until his death, Sakavu is considered to have secured the region from the resurgent Shebankarids for Sultan Tapar (Özaydın 1990: 54–8; also Merçil 1980: 92–4). Meanwhile the Crusaders appear to have increased their efforts to take Tripoli (16 March and 13 April 1108). When they first besieged Tripoli (1104), its ruler, the jurist Fakhr al-Mulk Ammar, had asked Artukoğlu Sökmen for help, promising him financial and military support. Although Sökmen also received a similar request from Tuğ-Tekin in Damascus, which he is said to have favoured, he died en route (October 1104). Ammar approached Sultan Tapar next (1106), but this also came to nothing when Çökürmüş and Sadaqa failed to support Sakavu. After Sultan Tapar’s victory over Sadaqa, Ammar is said to have decided to ask for help personally, journeying to Baghdad. While in Damascus he learnt his cousin had rebelled in collusion with the Fatimids. He sent word for him to be captured and imprisoned, which seems to have been done. On his return from Baghdad, however, where he and Tuğ-Tekin’s son Böri were well received by the sultan, Ammar found the Fatimids in Tripoli (25 August 1108). Soon after Sultan Tapar announced he intended to march on the Crusaders (26 November–25 December 1108), but this did not transpire (Özaydın 1990: 95–8). Eighteen months later, Sökmen al-Kutbi and Artukoğlu Il-Gazi joined Mawdud in Mosul from Ahlat and Sincar respectively, and decided to besiege Urfa (2–11 May 1110). King Baudouin I of Jerusalem learnt of this during his siege of Beirut. After its fall (27 May 1110), he left for Urfa. En route he was joined by a force under Bertrand of St Gilles, another under Tancred and smaller groups of Crusaders, including the Armenians Gog-Vasil and Apilgarip. Baudouin II, who was in Urfa, informed the Crusaders the Seljuqs had withdrawn towards Harran where they appear to have waited for Tuğ-Tekin who had arrived in Raqqa. Having strengthened the city defences and soldiery, the Crusaders retreated along 118 Dissolution (1105–94) the Euphrates. The Seljuqs gave pursuit without Tuğ-Tekin. Although Crusader leaders crossed safely, almost a third of the soldiery are said to have been killed as they watched helplessly. The Seljuqs did not attempt a crossing and returned to Urfa, but were unable to take it (Özaydın 1990: 103–8). Although Tuğ-Tekin was not involved in the fighting, having returned to Damascus from Raqqa, Baudouin seems to have forced him to give up a third of his crop (19 August–16 September 1110). He also took Sidon and imposed a ransom on its inhabitants (4 December 1110). Ridwan seems to have suffered likewise for having taken the opportunity to regain as many fortresses as possible and raid the environs of Antakya. Tancred is said to have retaliated by pillaging the environs of Aleppo. Despite several offers of money by Ridwan, Tancred is said to have taken half a dozen fortresses around Aleppo (15 December 1110–12 January 1111). Tancred’s activities seem to have provoked a mass exodus that was worsened by the conditions of peace he imposed. Ridwan appears to have given up his annual crop as well as money and horses. In order to compensate those afflicted he seems to have sold treasury land at rock-bottom prices (Özaydın 1990: 108–10; also Sevim 1989: 118–20). The Crusaders’ response to the siege of Urfa caused a group of jurists, Sufis and merchants to journey from Aleppo to Baghdad. On 17 February 1111, they railed against both sultan and caliph at the sultan’s mosque, going so far as to smash the pulpit and preventing Friday prayers from being held. Sultan Tapar’s administrators and commanders are said to have finally quietened the crowd with promises of help against the Crusaders. The group and their supporters repeated their demands the following Friday at the caliph’s mosque. Sultan Tapar had to intervene personally and stop the caliph from having the perpetrators punished (Özaydın 1990: 110–11, 146; also Sevim 1989: 120–1). The Seljuqs’ campaign was as indecisive as their last. After numerous commanders once again joined Mawdud in Harran, they besieged the new Crusader fortress in Tell-Bashir (28 July 1111). After Joscelin bribed one of them and he withdrew, the main body journeyed to Aleppo and camped outside the city walls. Thought to have been unable to supply them, Ridwan appears to have shut the city gates. The soldiers are said to have requisitioned their needs by force from the local populace. After they had been joined by Tuğ-Tekin, they journeyed to Ma’arrat al-Nu’man (6 September 1111). The environs were once again pillaged. Having become suspicious of Tuğ-Tekin’s involvement – possibly another reason why he withheld support for the Seljuq forces ostensibly sent to help him – Ridwan is thought to have tried to have him killed by some of the lesser commanders. Possibly due to the approaching harvest, Mawdud and Tuğ- Tekin appear to have been left by the other commanders and withdrew to nearby Jalali on the Âsi (Orontes) River. Seeing the Seljuq forces disperse, Tancred alerted Baudouin and Bertrand, but the Crusaders were defeated near Ma’arrat al-Nu’man (6 September–6 October 1111) (Özaydın 1990: 111–16; also Sevim 1989: 121–3). The following year Mawdud once again besieged Urfa (30 April–4 May 1112). As in the campaign the year before, the Seljuq forces destroyed the Dissolution (1105–94) 119 crops, leaving the inhabitants to face yet another year of deprivation given the bad harvest in the intervening year. While he was pillaging further afield in the environs of Seruj (28 June–27 July 1112), Joscelin seems to have made a successful sortie and driven off most of the Seljuqs’ horses. Although Mawdud returned and was able to get a few men on to a tower with the help of the Armenians in Urfa, Joscelin appears to have caught them by surprise and saved the city (Özaydın 1990: 116–18). Under pressure from Baudouin I in Jerusalem, the next year Tuğ-Tekin enlisted Mawdud’s help. Joscelin had left Baudouin II in Urfa and been given Taberiyye (Tiberias). Despite Joscelin’s attempts to negotiate peace, Tuğ-Tekin joined Mawdud. They were supported by the Artukid Türkmen from Sincar. Advancing on Jerusalem, they came upon the Crusaders as they were crossing the River Jordan and severely defeated them (28 June 1113). Baudouin was released after having been stripped of his sword because his captor did not recognize him. The Crusaders withdrew to Taberiyye while the Seljuq forces raided as far as Jerusalem and Jaffa (August 1113). Deciding to winter in Damascus on Tuğ- Tekin’s invitation (6 September 1113), Mawdud was wounded by an assassin after Friday prayers (10 October 1113), and subsequently died (9–10 March 1114) (Özaydın 1990: 118–26; also Turan 1993a: 182–3). Next year the new military governor for Mosul, Porsukoğlu Aksungur, undertook the Seljuq campaign on Urfa (15 May 1114), supported by Sultan Tapar’s son Mascud and Artukoğlu Il-Gazi’s son Ayaz. During the month-long siege, said to have been quite bloody, the Seljuqs once again systematically pillaged the fields and orchards in the countryside. While thus occupied Aksungur seems to have arrested Ayaz on grounds that his father had not joined them as ordered, going on to pillage the environs of Mardin as well. Given Mascud’s presence, possibly this was effected on the sultan’s orders. Even if this was the case, it proved disastrous. While negotiating with the Armenian ruler of Maraş Gog-Vasil’s widow, Aksungur was attacked and defeated by Ayaz’s father Il-Gazi (June 1114). This enabled Baudouin rather than Aksungur to gain dominion over Gog-Vasil’s domains, strengthening the Crusaders’ position in the region (Özaydın 1990: 126–8). Although Artukoğlu Il-Gazi immediately released Mascud, whom he captured during his victory over Aksungur, Sultan Tapar threatened to dispossess him of his iqtac. Il-Gazi seems to have turned to Tuğ-Tekin for help. Under suspicion because of Mawdud’s assassination, Tuğ-Tekin is said to have suggested that they approach Roger of Antakya. Having agreed, Il-Gazi appears to have started drinking heavily en route home and was captured by the commander of Hims, Kirhan (January 1115). When Sultan Tapar’s soldiery was delayed, Kirhan and Il-Gazi came to terms. Il-Gazi left his son Ayaz hostage against an attack on Hims by Tuğ-Tekin. However, on being released he seems to have gathered some Türkmen from the environs of Aleppo and besieged Hims until the arrival of the sultan’s forces (Özaydın 1990: 128–30). In the meantime Ridwan died of tuberculosis (10 December 1113). As his father is said to have done in the case of his brothers Talib and Bahram-Shah, Alp- Arslan appears to have had his brothers Mubarak-Shah and Malik-Shah killed. 120 Dissolution (1105–94) On Sultan Tapar’s request, Alp-Arslan also had the Ismacili leader Tahir and his associates decapitated, confiscating their property and disposable wealth with those of their followers. Having established good relations with Tuğ-Tekin, said to have proclaimed his allegiance, Alp-Arslan invited him to Aleppo to sort out its governance. Presumably due to his youth, however, he seems to have been unable to shake off his atabeg, the eunuch Lü’lü’s influence, killing many of his father’s loyal commanders before Tuğ-Tekin’s arrival. As a result Ridwan’s mother left Aleppo for Damascus with Tuğ-Tekin. Having gained the remaining commanders’ consent, Lü’lü had Alp-Arslan killed (September 1114). This ostensibly made the six-year-old Sultan-Shah ruler of Aleppo. Devastated by more than a decade of war, Aleppo also suffered an earthquake (29 November 1114), which collapsed some of the city walls and a gate tower. Unable to cope, Lü’lü offered the city and its environs to Sultan Tapar (Sevim 1989: 124–36). Sultan Tapar seems to have assigned the governor of Hamadan Porsukoğlu Porsuk with the tasks of taking over Aleppo, dealing with the rebellious Il-Gazi and Tuğ-Tekin and organizing the annual campaign against the Crusaders. Having started preparations in winter (29 January–27 February 1115), Porsuk crossed the Euphrates near Raqqa (7 May 1115). When he informed Lü’lü and his commander-in-chief to prepare for the handover, sending the necessary documen- tation with the sultan’s seal, they sent word to Tuğ-Tekin and Il-Gazi, requesting their aid. Apparently Tuğ-Tekin got to Aleppo while Porsuk was still in Balis. On learning that Lü’lü had joined the rebellion (14 June 1115), Porsuk besieged and took Hama where Tuğ-Tekin had left his baggage train. After being plundered for three days and nights, the city was handed over to Kirhan in Hims (Özaydın 1990: 130–1; also Sevim 1989: 136–7). Unaware of the fate of Hama, Tuğ-Tekin and Lü’lü’s commander-in-chief appear to have journeyed to Efamiya where they joined up with Il-Gazi and Roger. Informed of the size of the force under Porsuk’s command, they are said to have decided there was little to be done but wait until they dispersed to their iqtacs for the winter. At the beginning of September Porsuk appears to have feigned withdrawal so that his foes dispersed instead. Attacking a Crusader fortress between Aleppo and Ma’arrat al-Nu’man, Porsuk took it and killed everyone within. After this show of force, Porsuk turned south toward Aleppo. Presumably confident his foes had been duly cowed, he seems to have thrown caution to the wind, allowing his baggage train, which was at the head of his column, to be ambushed by Roger. His main force is said to have followed into the trap, being forced to withdraw with heavy losses before dispersing (14 September 1115) (Özaydın 1990: 130–5; also Sevim 1989: 137–8). In Aleppo, Lü’lü tried to escape but was killed (1116). After a brief period under his former commander-in-chief, Yaruk-Taş, Sultan Tapar is said to have asked Il-Gazi to take charge of the city and its environs, which he did (1117/8) (Sevim 1989: 139–40). As for Tuğ-Tekin in Damascus and Aksungur, who had been demoted to Rahba, the former journeyed to Baghdad to refresh his allegiance (9 April–31 July 1116). Assigned with the task of fighting the Crusaders, Tuğ- Tekin was joined by Aksungur. On learning that Pons, the ruler of Tripoli, was in the Bekaa Valley, Tuğ-Tekin and Aksungur surprised the Crusaders and routed Dissolution (1105–94) 121 them (Özaydın 1990: 137–8). After this, however, Tuğ-Tekin seems to have been content to maintain his position in Damascus, helping out his ally Il-Gazi in Aleppo as and when necessary (Sevim IA/12/2: 44–6). Arguably, therefore, Porsuk’s rout ended Great Seljuq dominion in Syria and south-eastern Anatolia, allowing first the emergence of vassals and then of successor states (Köymen 1963: 295).

Transoxania and the Ghaznavids Sancar appears to have been highly active during his uterine brother’s reign. Having killed the eastern Karakhanid ruler for rebelling against the Seljuqs during the interregnum, Sancar is said to have reorganized Transoxania under the Karakhanid Muhammad II, whom he instated in Samarkand under the name of Arslan Khan. However, Arslan Khan does not appear to have been popular. When the Karakhanid Omar Khan displaced Arslan Khan, Sancar pursued him personally and, having defeated him in Khwarazm, killed him. Nevertheless, Sagun Beg, a descendant of Ali Tekin, now challenged Arslan Khan who defeated him with Sancar’s soldiery (1109–10). Finally secure, Arslan Khan appears to have pillaged the urban centres in his realm, confiscating the local inhabitants’ wealth. Sancar advanced against him, but pardoned him after he is said to have dismounted and kissed the ground across the Amu-derya River (1113–14) (Özaydın 1990: 139–40). By comparison, relations between the Seljuqs and Ghaznavids appear to have remained peaceful during Sultan Mascud III’s reign (1099–1115). However, when Shirzad acceded to the Ghaznavid throne, his brother Arslan-Shah rebelled and defeated him in a battle near the capital (17 February–16 March 1116). Shirzad was killed soon after in a subsequent battle. Although Arslan-Shah imprisoned his remaining brothers, Bahram-Shah escaped. Unable to hold his own against Arslan-Shah, Bahram-Shah sought refuge in Kirman from where he was sent to Sancar. Although Sancar sent an envoy to Ghazna entreating Arslan-Shah to settle his differences with Bahram-Shah, he does not seem to have received a favourable response. Hearing that Sancar was mobilizing, Arslan-Shah approached Sultan Tapar. The sultan’s envoy is thought to have been too late to stop Sancar who defeated the forces deployed against him. As a result Arslan-Shah is said to have sent his mother, Sancar’s sister, but without avail. Sancar defeated Arslan-Shah and having entered Ghazna instated Bahram-Shah (25 February 1117). Although Arslan-Shah returned on Sancar’s departure and regained control, a Seljuq army from Balkh pursued him with Sultan Bahram-Shah, pillaging and destroying any urban centres that offered Arslan-Shah refuge. As a result, given up to the Seljuqs, Arslan-Shah was strangled on Sultan Bahram-Shah’s orders (30 September 1118) (Özaydın 1990: 140–4).

The Seljuqs of Iraq On Sultan Tapar’s death (April 1118), his heir apparent, the eldest Mahmud (b. 1105) acceded to the throne. The cause of Sultan Tapar’s illness is not known, 122 Dissolution (1105–94) but the report that his wife Gevher Khatun was strangled in her quarters on the hour of the sultan’s death may explain the rumour that she had poisoned him and he knew of this. According to another rumour the sultan was afraid that if Sancar married her as was customary then his sons would never rule (Özaydın 1990: 149–51). Gevher Khatun was Yakuti’s granddaughter and as such ultimately responsible for Azarbayjan, which she seems to have ruled through her diwan (Turan 1993a: 184). Although no challenges are reported from his four brothers, namely Mascud (b. 1109), Tuğrul (b. 1110), Süleyman and Selçuk-Shah, the reassignment of Dubays al-Sadaqa to Hilla and that of Mengü-Pars as commander to Baghdad in place of Mascud’s former atabeg Porsukoğlu Aksungur seems to have created a power vacuum in Iraq al-Arab. While Dubays gathered a force of Arabs and Kurds, Aksungur refused to hand over Baghdad and killed Mengü-Pars’ brother who had come to negotiate the handover. This conjuncture of events appears to have prompted Mascud’s current atabeg Ay-Aba (also known as Çavuş Beg) to journey from Mosul to Baghdad and propose Aksungur join him in a raid on Dubays. Although this seems to have failed due to Mengü-Pars siding with Dubays, Sultan Mahmud is thought to have regarded Ay-Aba’s action as incit- ing Mascud to rebellion and he gathered his forces. Undaunted, Ay-Aba is said to have asked for Mascud’s domains to be expanded, gaining Azarbayjan (Sümer IA/8: 135; Köymen IA/12/2: 14–15 and 1954: 30–2). Since Tuğrul is reported in Ganja, this suggests his atabeg Gündoğdu was unsuccessful against the resurgent Georgians. Around this time the Bagratid David II married the Kıpçak Kara-Khan’s son Atrak’s daughter and allowed many of his affiliates to settle in Georgia, apparently taking a number of youths into service in his royal corps. Thus strengthened, David II is said to have refused to pay his annual tribute and forbidden seasonal Türkmen migrations, which caused many to migrate to Anatolia (Turan 1993a: 183; also Minorsky IA/12/1: 267). In the meantime Sancar declared himself sultan and marched on his nephew (14 June 1118). He defeated and captured him near Sawa (14 August 1119). Although he allowed Mahmud to keep the title of sultan, Sancar stripped him of Rayy, Mazandaran and Qomis, as well as retaining the revenues from Sawa and Huy. Not satisfied with this he assigned Gilan and north-eastern Jibal to Tuğrul and Fars and western Khuzistan to Selçuk-Shah, creating two intermediate Seljuqid dominions ostensibly subordinate to Sultan Mahmud. Last but not least, he is said to have assigned his own tax collectors to determine the revenues due to Sultan Mahmud (Köymen 1954: 13–18; also Turan 1993a: 185–7). In short, Great Sultan Sancar is credited with creating besides the Seljuqid dominions already established in Kirman and Anatolia that of Iraq, which included Azarbayjan and south-eastern Anatolia.

Sultan Mahmud Possibly in the knowledge that Dubays al-Sadaqa would support him in Hilla, Ay-Aba appears to have moved suddenly on the capital Hamadan. Despite this Dissolution (1105–94) 123 Sultan Mahmud was able to gather sufficient forces and defeat him (14 June 1120). The battle seems to have enabled Porsukoğlu Aksungur to redeem himself. Apart from fighting at the forefront, he is credited with persuading Mascud to present himself at court. After his vizier, the renowned jurist Fakhr al-Mulk Ammar, was killed on having been captured, Mascud seems to have been attempting to return to Mosul. Although he forgave his brother, Sultan Mahmud is thought to have retained Mascud at court (Köymen 1984: 32–5; also Sümer IA/8: 135–6). On learning of Ay-Aba’s defeat, Dubays is said to have burnt the crops in the environs of Baghdad, where he had been pillaging. Caliph al-Mustarshid (1118–35) threatened to retaliate, but Dubays marched on Baghdad and the caliph appealed to Sultan Mahmud (September 1120). Having threatened Caliph al-Mustarshid that if he did not withdraw his appeal to Sultan Mahmud he would consider himself free to act as he pleased, Dubays seems to have sought reconciliation on the sultan’s arrival (September–October 1120). Notwithstanding, the sultan occupied Hilla and Dubays was forced to seek refuge in Aleppo with his father-in-law Artukoğlu Il-Gazi. Caliph al-Mustarshid seems to have persuaded Sultan Mahmud to remain in Baghdad, offering to meet his military expenses for the next four months (Özaydın 1990: 37–40). While there Sultan Mahmud is said to have rewarded Porsukoğlu Aksungur with Mosul and Jazira, and increased Artukoğlu Il-Gazi’s iqtacs. He seems to have also re-imposed non-Shari taxes that had been suspended by his father. The most interesting development, however, is considered to be the embassy he sent to Great Sultan Sancar with the caliph (June 1121). At a ceremony held later that year (29 October 1121), Caliph al-Mustarshid presented Sultan Mahmud with two swords. While these are taken to signify a reiteration of Seljuqid sovereignty over worldly affairs, since they were not presented to Great Sultan Sancar, the purpose of the ceremony is thought to have been divisive. The caliph and Sultan Mahmud also asked Il-Gazi to divorce his daughter from Dubays. When Il-Gazi declined, though he is said to have taken care to reiterate his allegiance, Dubays is thought to have secretly approached the caliph. However, having returned to Hilla Dubays then expelled the Seljuqs’ commanders and tax collectors. In retaliation Sultan Mahmud re-occupied Hilla. Although the caliph now asked for Dubays to be ousted from the region, the sultan accepted Dubays’ brother Mansur as hostage and departed for Hamadan (10 May 1122) (Özaydın 1990: 43–55). Aksungur advanced on Dubays but was defeated (9 June 1122). Dubays took Wasit, once again forcibly removing the Seljuq officials, and demanding Caliph al-Mustarshid remove his vizier Ibn Sadaqa, which he appears to have done. Sultan Mahmud imprisoned Mansur and asked the caliph to appoint Ahmad al-Mulk in Ibn Sadaqa’s place. Dubays confiscated the caliph’s livestock. When the caliph objected, he is said to have given him five days to have Ibn Sadaqa killed, Ahmad al-Mulk removed, Aksungur sent back to Mosul and Mansur released. The caliph declared jihad, arming those inhabitants of Baghdad who responded. More importantly, he appointed one of his chamberlains commander-in-chief despite the fact that on his request Aksungur returned from the Caucasus where he was on campaign with Sultan Mahmud. The remainder of the caliphal army seems to 124 Dissolution (1105–94) have consisted of local Arab and Türkmen commanders. The latter are thought to have been Togan-Arslan, the commander of Bitlis in south-eastern Anatolia, Kıpçak b. Arslan-Taş, the commander of Shehrizor, the Saltukids from Erzurum and the Buka. Notwithstanding all the pomp and circumstance, Aksungur’s forces are said to have made up the majority and it was he who appears to have prepared the army for the battle at which Dubays was defeated. Satisfied with this victory, toward the end of which he drew his sword and joined the fighting, Caliph al-Mustarshid returned to Baghdad without displacing Dubays (January– February 1123) (Özaydın 1990: 62–8; also Sümer 1992: 113, 118). Sultan Mahmud’s campaign in the Caucasus appears to have been as unsuccessful as Tuğrul’s two years earlier with Il-Gazi and Dubays (1121), which allowed the Bagratids to take Tbilisi. Turan was of the opinion that this was due to Great Sultan Sancar’s vizier al-Kumumi, regarded to have been an enemy of Nizam al-Mulk’s sons. Great Sultan Sancar is said to have ordered Sultan Mahmud to dismiss his vizier Osman al-Mulk while still in the Caucasus, forcing him to return. According to Köymen, Sultan Mahmud may have taken the opportunity to blame the caliph’s unilateral actions on his vizier Osman al-Mulk’s brother, Ahmad al-Mulk, and asked the caliph to dismiss him. Having complied with the sultan’s request, Caliph al-Mustarshid seems to have asked for Aksungur to be removed from Baghdad. At this juncture Dubays appears to have enticed Tuğrul with the promise of his own dominion in Mesopotamia. Presumably after his defeat Sultan Mahmud had removed Tuğrul from Ganja, which he had been assigned by Great Sultan Sancar. On learning of Tuğrul’s arrival, Caliph al-Mustarshid is said to have instructed his vizier and the Seljuqs’ new commander Barankuş to mobilize (Özaydın 1990: 69–77 and IA/12/2: 14ff; also Turan 1993a: 197–8). The following year, during which Tuğrul and Dubays appear to have been inactive, Caliph al-Mustarshid asked for and married Great Sultan Sancar’s daughter. When they finally advanced, the caliph took the field and they withdrew (13 March 1125). Dubays is said to have secretly tried to come to an accom- modation with the caliph and failed, journeying to Khurasan with Tuğrul for an audience with Great Sultan Sancar. Retaining Tuğrul at court, Great Sultan Sancar imprisoned Dubays (Özaydın 1990: 75–81 and IA/12/2: 14ff). Having established a military force consisting of Arab and Kurdish cavalry, Caliph al-Mustarshid now began to interfere in the Seljuqs’ affairs, threatening Barankuş when he objected. As a result, Barankuş journeyed to Hamadan to warn Sultan Mahmud (11 August 1125). The sultan, however, sent a letter thanking the caliph for his recent action against Tuğrul, which he interpreted as having been conducted on his behalf. After a further exchange of envoys Caliph al-Mustarshid agreed to help Sultan Mahmud depose Great Sultan Sancar. When the latter learnt of this he sent his nephew a letter. Having warned him of the caliph’s intentions, he reminded him that despite having defeated him in battle he had appointed him his heir apparent and given him a second daughter in marriage after the first had died. He now ordered Sultan Mahmud to march on Baghdad where he was to imprison the caliph’s vizier Ibn Sadaqa, kill his Kurdish cavalry and destroy Dissolution (1105–94) 125 the caliphal garrison. Finally, his envoy Dergüzini was to be appointed vizier to oversee his instructions (Özaydın 1990: 81–7). Caliph al-Mustarshid refused Sultan Mahmud’s proposal to come to Baghdad, saying that having just emerged from famine the region could not support two armies and asking that he should come next year. The caliph then ordered the inhabitants of Baghdad to vacate the suburbs on the eastern banks of the Tigris where he set up camp. Sultan Mahmud appears to have been surprised by this development and sent an envoy who confirmed the caliph’s determination to fight. Although the caliph withdrew to the western banks of the Tigris when Sultan Mahmud advanced on Baghdad, he sent the eunuch Afif to take Wasit. The sultan countered by sending Zengi, Sultan Malik-Shah’s governor Aksungur’s son (Alptekin IA/13: 526ff), who defeated Afif’s forces. On arriving in Baghdad (4 January 1126), Sultan Mahmud sent another envoy. It seems the caliph refused to back down and the sultan had his palace pillaged. Since the caliph’s concubines were still in residence there, al-Mustarshid is said to have been able to rouse the populace against the Seljuqs. Having crossed the river, his soldiery attacked and killed the Seljuq soldiers in the palace while the populace pillaged the homes of Seljuq soldiery normally resident in Baghdad. The caliph then crossed the river with his army (7 January 1126). After some inconclusive skirmishes the sultan called on Zengi who ferried his soldiery into Baghdad on crafts he procured from Wasit and Basra, outflanking the caliph (Köymen 1984: 88–101; also Alptekin IA/13: 526ff). Sultan Mahmud appears to have been reluctant to push home his advantage, preferring to negotiate his entry into Baghdad. Once there the sultan requested from the caliph its inhabitants return their plunder, which they are said to have done. Having minted coinage in his name, Sultan Mahmud appointed Zengi to Baghdad, returning to Hamadan after a long illness without having fulfilled Great Sultan Sancar’s demands (12 April 1126) (Özaydın 1990: 101–11). Three months later Sultan Mahmud dismissed and imprisoned his vizier Dergüzini for conspiring with the caliph’s vizier Ibn Sadaqa while in Baghdad (July 1127). The accusation appears to have been made by Barankuş (Özaydın 1990: 111). However, Great Sultan Sancar arrived in Rayy at the head of his army and invited Sultan Mahmud there. It is not clear when Sultan Mahmud journeyed there or how long he remained (Özaydın 1990: 118–19). As a result Tuğrul’s atabeg Şirgir was replaced by one of Great Sultan Sancar’s commanders, Karasungur. Although Arran was re-assigned to him, Tuğrul appears to have stayed with Great Sultan Sancar (Özaydın 1990: 116, 127). Great Sultan Sancar also took charge of Mascud to whom he assigned Jurjan (Özaydın 1990: 127–8). Having ordered Sultan Mahmud to release Dergüzini, Great Sultan Sancar made him vizier to his daughter, the sultan’s wife. He then appointed her Dubays’ protector, instructing the sultan to ensure Dubays became governor of Jazira and Syria. Apart from the caliph’s enmity toward Dubays, a further problem with this was that Sultan Mahmud had assigned the said region to Zengi, who had not wanted to stay on as commander in Baghdad (Özaydın 1990: 117–24). When Porsukoğlu Aksungur was killed by Ismacili assassins (26 November 1126), Sultan 126 Dissolution (1105–94) Mahmud left Mosul and its environs in Aksungur’s son Mascud’s hands. On his death the following year (July 1127), Aksungur’s commander Sakavu in Mosul sent an envoy requesting to be confirmed in Mascud’s son’s name. However, the envoy seems to have suggested Zengi instead and Sultan Mahmud is said to have agreed, appointing Zengi atabeg to his son Alp-Arslan (Alptekin IA/13: 526ff). Sultan Mahmud arrived in Baghdad without Dubays (14 January 1129). The caliph not only refused to pardon Dubays but also protested his appointment to Mosul as governor of Jazira and Syria in place of Zengi, offering the sultan financial compensation. At first the sultan seems to have ignored the caliph’s offer and ordered Zengi to hand over Mosul to Dubays. When Zengi likewise offered the sultan money to retain Mosul, the sultan is said to have relented. Having received Zengi in Baghdad and allowed Dubays to enter the city, Sultan Mahmud left for Hamadan with Dubays (25 May 1129) (Köymen 1984: 134–9). When Sultan Mahmud’s wife died and the sultan became ill, Dubays appears to have left Hamadan for Hilla with the sultan’s youngest son. The commander for Hilla, Bihruz, does not seem to have offered any resistance, returning to Hamadan (August–September 1129). On learning Sultan Mahmud had dispatched the commander Ahmadili, Dubays once again approached the caliph. It is thought that the open hostility of the inhabitants of Baghdad to Dubays prevented the caliph from accepting reconciliation. After Ahmadili (5 October 1129), Sultan Mahmud also arrived in Baghdad (18 October 1129). While Ahmadili advanced on Hilla, Dubays sent an envoy to the sultan threatening to withdraw to the desert unless the caliph agreed to a pardon. Sultan Mahmud refused and Dubays took the sultan’s son with him. On learning Dubays had occupied Basra and pillaged the sultan and caliph’s estates, Sultan Mahmud dispatched the commander Kızıl as well. Sultan Mahmud remained in Baghdad while Ahmadili and Kızıl pursued Dubays without success (1130) (Özaydın 1990: 140–8). Having left Sultan Mahmud’s son in a fortress, Dubays is said to have sought refuge with the Crusaders. Apparently unsuccessful, he was later caught by the deceased Tuğ-Tekin’s son Böri (6 July 1131). On learning of his capture, the caliph sent a letter to Böri in Damascus asking him to surrender Dubays. Although Böri accepted, while he was waiting for the caliph’s soldiery Zengi offered to release his son Sevinç in exchange for Dubays (2 October 1131). Zengi had imprisoned Sevinç when instead Böri sent him on being requested to help Zengi against the Crusaders (24 September 1130). Zengi had also taken Hama. Now he intercepted the caliph’s envoy and plundered his forces’ baggage train before sending them back, releasing Dubays (Alptekin IA/13: 526ff). While still in Baghdad Sultan Mahmud learnt his brother Mascud was advancing (February 1131). Having declared his son Davud heir apparent, Sultan Mahmud prepared to meet him despite the caliph’s protestations that Dubays would return to Hilla if he departed. The caliph also refused to include Davud’s name in the khutba on grounds that such a request would have to come from Great Sultan Sancar. Once back in Hamadan, however, Sultan Mahmud and Mascud are said to have met and been reconciled. Having snubbed Great Sultan Sancar’s envoy, Sultan Mahmud arrested some of his commanders and confiscated the Dissolution (1105–94) 127 disposable wealth of the inhabitants of Hamadan before omitting the Sancar’s name from the khutba. He is thought to have been preparing to march on Baghdad with the intention of replacing Caliph al-Mustarshid when he fell ill and died (11 September 1131) (Köymen 1984: 164–9, 173).

Dissension Although some of the commanders supported the heir apparent Davud’s accession and succeeded in having the khutba read in his name outside Iraq al-Arab, the caliph refused their application. Sultan Mahmud’s vizier Dergüzini appears to have journeyed to Rayy with such commanders as Porsukoğlu Porsuk, Kızıl, Karasungur and Karadoğan to await Great Sultan Sancar’s orders. At the time Great Sultan Sancar is known to have been in Samarkand. It appears after the Karakhanid Arslan Khan suffered a stroke leaving him partially incapacitated his son Nasr Khan had taken over. Nasr Khan seems to have been unpopular, particularly among the Karluk, the majority of the Karakhanids’ fighting force. When Nasr Khan was killed by an Alawi who conspired with the head of the town militia or ra’is, Arslan Khan sent word to his other son (possibly Mahmud) in Turkistan and Great Sultan Sancar. On arrival his son appears to have killed the Alawi and imprisoned the ra’is. Arslan Khan is said to have informed Great Sultan Sancar of this and withdrawn his invitation, sending a detachment of soldiery whose attempt to assassinate him during a hunt en route was thwarted by the Khwarazmshah Atsız. The sultan took Samarkand by force and allowed his soldiery to pillage it (April 1130), sending Arslan Khan to Marv where he died. During the year or more he is thought to have spent in Samarkand, Great Sultan Sancar appears to have appointed a commander Hasan Tekin (Kılıç Tamgaç Khan) as his regent. Hasan Tekin died after Great Sultan Sancar returned to Khurasan, as a result of which Arslan Kahn’s son appears to have reinstated (Özaydın 1990: 158–63, 174–5; also Kafesoğlu 1992: 45–6, 50). After Sultan Mahmud’s death Mascud appears to have journeyed to Shehrizor and gained the support of its commander Kıpçak and taken Tabriz, where his nephew Davud besieged him (October 1131). Davud is thought to have withdrawn when his commanders decided to support Mascud instead. Mascud is next reported in Hamadan, from where he is said to have requested the caliph to have the khutba read in his name. The caliph refused him as well on grounds that such a request had to come from Great Sultan Sancar. In the meantime Zengi is said to have agreed to support Mascud. Selçuk-Shah seems to have arrived in Baghdad soon after at the head of forces from Fars and Khuzistan. Although the caliph allowed Selçuk-Shah to take up residence in the Seljuq palace he does not appear to have included his name in the khutba. However, Selçuk-Shah’s atabeg Karaca is said to have attacked and defeated Zengi near Samarra. Mascud now agreed with his brother Selçuk-Shah and Caliph al-Mustarshid to topple Great Sultan Sancar who had arrived in Rayy at the head of his army (January–February 1132). Mascud was to be sultan and Selçuk-Shah his heir apparent, but Iraq al-Arab was to be ruled by the caliph’s regent. However, when the caliph followed Mascud and Selçuk-Shah 128 Dissolution (1105–94) from Baghdad he had not included their names in the khutba even though he had omitted Great Sultan Sancar’s (27 March 1132) (Özaydın 1990: 176–81, 185, 190; Sümer IA/8: 136 and 1992: 113; Alptekin IA/13: 526ff). Caliph al-Mustarshid did not join Mascud and Selçuk-Shah, advancing to Hanikin near Halvan (21 May 1132). Great Sultan Sancar seems to have ordered Zengi and Dubays to attack Iraq al-Arab while he advanced from Hamadan with Tuğrul. Without Zengi or the caliph, Mascud and Selçuk-Shah seem to have decided to retreat to Azarbayjan, but Great Sultan Sancar cut them off and forced them to give battle (26 May 1132). Defeated, Mascud is said to have fled, but appears to have returned on Great Sultan Sancar’s inducement, being assigned Azarbayjan. There is no mention of Selçuk-Shah. Rather than Davud, Tuğrul was appointed sultan with Dergüzini as his vizier. Great Sultan Sancar returned to Nishapur (5 August 1132) (Özaydın 1990: 191–200; also Sümer IA/8: 136). Caliph al-Mustarshid appears to have returned to Baghdad on learning of Zengi and Dubays’ approach. In the ensuing battle near Baghdad Zengi and Dubays are thought to have been defeated due to the reluctance of their soldiery to fight the caliph who is said to have donned the Prophet’s vest. Zengi retreated to Tekrit and Dubays returned to Hilla. When Zengi offered peace, Caliph al-Mustarshid demanded he hand over Dubays, threatening to take Mosul by force. Although Zengi did not, Dubays seems to have been unable to re-establish himself in Hilla, which the caliph had assigned to his commander Iqbal (Özaydın 1990: 200–3; also Alptekin IA/13: 526ff). Davud appears to have remained in Tabriz throughout this period. Now he advanced on Hamadan but was defeated by Sultan Tuğrul. Davud fled to Baghdad with his atabeg Ahmadili (July–August 1132), where Mascud joined him, taking up residence in the Seljuq palace (December 1132). After the caliph had the khutba read in Mascud’s and his heir apparent Davud’s names, Mascud and Davud departed for Azarbayjan with the caliph’s commander Nazar and his soldiery (27 January 1133). In Azarbayjan Mascud defeated Sultan Tuğrul’s regent Karasungur. Mascud advanced on Hamadan and having defeated Sultan Tuğrul occupied the capital where he is said to have had Davud’s atabeg Ahmadili assassinated by Ismacilis (25 May 1133). Sultan Tuğrul fled to Rayy. When he journeyed to Isfahan, however, he met with resistance and was forced to return. Pursued by Mascud’s forces he may have been defeated en route. Having had his vizier Dergüzini killed while in Rayy Sultan Tuğrul is thought to have spent the winter in Tabaristan where he appears to have gained the confidence of Mascud’s commanders. When Sultan Tuğrul returned the following year with Dubays, these commanders deserted Mascud on the battlefield (June 1134). Mascud is said to have fled to Baghdad dressed as a Türkmen (Özaydın 1990: 203–15, 237–41; Sümer IA/8: 137). After Mascud’s initial victory, the caliph had advanced on Zengi with Mascud’s permission (June 1133). Having arrived outside Mosul (26 July 1133), the caliph is said to have besieged the city for eighty days, but without success. After he returned to Baghdad, Zengi sought to appease the caliph with gifts, attacking and defeating the commander of Hisn Kayfa for aiding the caliph (26 April 1134) (Özaydın 1990: 215–18; Alptekin IA/13: 526ff). Dissolution (1105–94) 129 Selçuk-Shah seems to have been in Baghdad when Mascud arrived. Despite pressure from the caliph to advance on Sultan Tuğrul in Hamadan, Mascud appears to have remained. When the caliph discovered the commanders who had supported him against Zengi had got in touch with Sultan Tuğrul he is said to have arrested one, causing the remainder to seek Mascud’s protection. Finally asked to leave (15 October 1134), Mascud seems to have camped outside Baghdad despite a gift of money from the caliph (Özaydın 1990: 242–7; Sümer IA/8: 137). Mascud journeyed to Hamadan only after receiving news of Sultan Tuğrul’s death (24 October 1134). His arrival appears to have prevented commanders such as Kızıl, Sungur and Barankuş Bazdar from deciding on their own Seljuqid candidate. These and others are said to have been opposed to Mascud due to his wife, Sultan Berk-Yaruk’s daughter, Zubaydah Khatun’s patronage of Karasungur, Sultan Tuğrul’s former regent in Azarbayjan. Great Sultan Sancar accepted Mascud’s accession, but ordered him to behead the commanders Porsukoğlu Porsuk, Kızıl, Barankuş Bazdar and others he is said to have held responsible for Sultan Tuğrul’s problems. Apparently Sultan Mascud showed the letter from Great Sultan Sancar in the hope of gaining the commanders’ confidence. Instead the commanders asked the caliph to send them a Seljuqid prince of his choice (possibly Selçuk-Shah) to lead them against Sultan Mascud from Porsukoğlu Porsuk’s iqtac in Khuzistan (Özaydın 1990: 248–53; Sümer IA/8: 137). Presumably because Dubays was with them, at first the caliph appears to have refused. The commanders tried to capture Dubays, but he escaped with the intention of rejoining Sultan Mascud. On learning of this Sultan Mascud is said to have advanced on Khuzistan, which caused the commanders to flee to Baghdad. Sultan Mascud confiscated Porsukoğlu Porsuk’s iqtac, pillaging it. Zengi sent his son to the caliph with the keys of Mosul (20 November 1134), siding against Sultan Mascud. Dubays also sent an envoy to the caliph requesting to be pardoned, but was refused (21 November 1134). Soon after the commander for Basra, Beg- Aba, seems to have proclaimed allegiance to Caliph al-Mustarshid. When Sultan Mascud’s advance units appeared near Hulvan under Dubays’ command, the caliph asked Zengi and Beg-Aba to join him and sent out his own advance units. Dubays defeated these (4 May 1135). After replacing Great Sultan Sancar and Sultan Mascud’s names with Davud’s in the khutba, the caliph asked for and gained a fatwa from jurists in Baghdad for war on Sultan Mascud. Despite his vizier’s advice to the contrary, Caliph al-Mustarshid then gave the order to advance (25 May 1135). En route Beg-Aba deserted to Sultan Mascud, but Porsukoğlu Porsuk joined with his soldiery. Nevertheless, the caliph seems to have made slow progress, providing Sultan Mascud with the opportunity to entice the Turkic commanders leading his forces with the promise of new iqtacs (Özaydın 1990: 255–60; Sümer IA/8: 137–8). Caliph al-Mustarshid held a war council in Kirmanshah (22 June 1135). Advised that it was too late to turn back, the caliph accepted battle and was captured (24 June 1135). Having brought the caliph to Hamadan, Sultan Mascud gave him an ultimatum. Caliph al-Mustarshid was to pay an undisclosed sum in war reparations, disband his remaining soldiery and never campaign again. 130 Dissolution (1105–94) Sultan Mascud also appointed Beg-Aba commander to Baghdad where his arrival seems to have caused riots. Beg-Aba had the city walls partially destroyed and took over the caliph’s fledgling administration in Iraq al-Arab. After a brief campaign to Azarbayjan (July 1135), Sultan Mascud is said to have received a letter from Great Sultan Sancar directing him to return the caliph to Baghdad after surrendering Dubays to him (13 July 1135). Instead Sultan Mascud persuaded the caliph to pardon Dubays. Before the caliph left for Baghdad a military detachment sent by Great Sultan Sancar arrived. During the welcoming ceremony Caliph al-Mustarshid is said to have been killed by Ismacilis of whom a number were caught (29 or 31 August 1135). Soon after Sultan Mascud appears to have had Dubays killed for the caliph’s murder (Özaydın 1990: 270–83; Sümer IA/8: 138; Turan 1993a: 199). Possibly because the amount was excessive, Caliph al-Rashid rejected to pay his father’s promissory note (29 October 1135). When Sultan Mascud’s envoy Barankuş insisted, he and Beg-Aba were forcibly ejected from Baghdad and the Seljuqs’ palace pillaged. Beg-Aba retired to Wasit and Barankuş returned to Hamadan. Davud arrived in Baghdad soon after (14 November 1135). Zengi followed with his charge Alp-Arslan, whom he wanted declared sultan. Those commanders opposed to Sultan Mascud joined them. Davud’s name replaced Great Sultan Sancar and Sultan Mascud’s in the khutba (22 November 1135). However, when Caliph al-Rashid arrested his father’s trusted commander Iqbal and confiscated his wealth, Zengi demanded his release. After he had secured this, Zengi also demanded and received the money the caliph had confiscated. Zengi then consolidated his influence by appointing Davud’s vizier. The caliph left Baghdad at the head of his forces (27 May 1136), but returned on learning from Zengi that Sultan Mascud was planning a surprise raid (7 June 1136). Zengi defeated Sultan Mascud’s advance units. On realizing that the sultan was in negotiation with the caliph’s commanders Zengi is said to have decided to leave for Mosul, offering the caliph refuge. Possibly afraid he would be killed like his father, Caliph al-Rashid accepted (14 August 1136) (Özaydın 1990: 285–96; Alptekin IA/13: 526ff; Sümer IA/8: 138; Kafesoğlu 1992: 26). Sultan Mascud entered Baghdad having besieged it for fifty days (15 August 1136). Once order was established, he appointed al-Muqtafi caliph (18 August 1136). Having recognized Caliph al-Muqtafi, Great Sultan Sancar ordered Sultan Mascud to ask Zengi to expel al-Rashid from Mosul. Ousted, al-Rashid is said to have unsuccessfully sought help from Sultan Mascud (1116–56) in Anatolia and the Danişmendid Muhammad (1134–42), while in Sincar. However, Davud and the governors of Fars and Khuzistan, Mengü-Pars and Boz-Aba, offered him protection. Sultan Mascud sent Karasungur to Azarbayjan, but when he was repelled by Davud the sultan advanced and defeated him, having Mengü-Pars killed. While the sultan’s forces were busy pillaging, however, Boz-Aba counter- attacked and captured most of his commanders, killing them. Sultan Mascud fled while Davud quickly occupied Hamadan. Evidently in charge, rather than pursue Sultan Mascud or occupy Baghdad, Boz-Aba appears to have decided to occupy Mengü-Pars’ iqtac in Fars first. While preparing to march on Isfahan, allegedly Dissolution (1105–94) 131 al-Rashid was assassinated by the Ismacilis (8 June 1138) (Özaydın 1990: 298– 305; also Sümer IA/8: 138, Turan 1993a: 199 and Kafesoğlu 1992: 26).

Defeat at Katvan While Sultan Mascud was thus occupied, Great Sultan Sancar was fighting with the Ghaznavids and the Khwarazmshah. First he appears to have set out against Sultan Bahram-Shah for withholding the annual tribute (August–September, 1135). When he arrived outside Ghazna, Sultan Bahram-Shah asked his pardon. Nonetheless, he seems to have balked at proclaiming allegiance in person and withdrew to India. Great Sultan Sancar entered Ghazna uncontested and having collected the taxes due, confiscated all the removable wealth therein. He returned to Khurasan after receiving another envoy from Sultan Bahram-Shah and pardoning him (July 1136) (Özaydın 1990: 306–11). After this, the Khwarazmshah Atsız, who is said to have accompanied the sultan on all of his campaigns and is credited with discovering the plot by the Karakhanid Arslan-Khan to have Great Sultan Sancar killed (1130), appears to have started a many-faceted campaign. Extending his dominion in bloody battles against the Türkmen on the Mankishlag peninsula on the north-eastern coast of the Caspian Sea and toward Jand, Atsız also tried to affiliate the Türkmen in the environs of Marv, the Seljuq capital. When the sultan accused him of spilling Muslim blood without his permission, Atsız arrested the Seljuq officials in Khwarazm and confiscated their disposable wealth (Özaydın 1990: 311–14; Kafesoğlu 1992: 47; Sümer 1992: 102–3). Great Sultan Sancar advanced on Khwarazm from Balkh (September 1138). Atsız is said to have opened the irrigation canals and flooded the surrounding countryside along the Amu-Darya before withdrawing to the fortress of Hezaresb, en route to his capital Gurganj. At the battle that took place (16 November 1138), Atsız was forced to flee. The sultan had Atsız’s son Atlig killed. As with the death of Caliph al-Mustarshid, this remained a cause for contention between the Khwarazmshahs and the Seljuqs. Although Great Sultan Sancar re-occupied Khwarazm and assigned it to Süleyman-Shah (February 1139), his youngest nephew, Atsız, returned and forced him to flee with his commanders and other officials. Atsız raided Bokhara soon after, capturing and killing its governor and partially destroying the city walls (Özaydın 1990: 318–20; Kafesoğlu 1992: 48–9). Meanwhile the Jurchen had superseded the Khitay, who had founded the Liao dynasty in northern China (936–1122). Although many of the Khitay accepted Jurchen rule, a small group led by Yeh-lü Tah-shi, a member of the Khitay dynasty, fled westward, into the Tarim Basin (1123). Having subjugated the Uygurs in Turfan, Kuça and Beşbalig, the Khitay attempted to incorporate Kashgar, which they had tried and failed to take once before (1017). They were defeated again (1128). As a result they appear to have turned north (1130), only to be forced westward again by the Kırgız (Kafesoğlu 1992: 51–2). Having established the city of Imil east of Lake Balkash, Yeh-lü Tah-shi appears to have affiliated some lineages of the Kanglı and Karluk and become 132 Dissolution (1105–94) known as Kür Khan, or the Khan of Khans, establishing the Kara-Khitay Empire (1124–1211). According to Sümer, the Kara-Khitay displaced the Kanglı, pushing the Oguz and Karluk into Transoxania. Some of the Oguz, the Avşar, Salur and Yıva lineages continued into Iraq al-Ajam, namely Khuzistan and Shehrizor. The Oguz that remained in Transoxania established good relations with the Karakhanids, but the Karluk rebelled against Nasr Khan who requested Great Sultan Sancar’s help (1130). The Kara-Khitay next occupied Balasagun and took Khoten on the southern edge of the Tarım Basin and Kashgar in the west (1137) (Kafesoğlu 1992: 52–3; Sümer 1992: 100; Turan 1993a: 189). When the Karluk rebelled against Mahmud Khan in Samarkand, they appear to have had help from the Kara-Khitay. Defeated, Mahmud Khan asked Great Sultan Sancar for help. After receiving Atsız’s proclamation of allegiance (25 May 1141), the sultan advanced on Transoxania with forces from Sistan, Gur, Ghazna and Mazandaran (July 1141). Once in Samarkand he appears to have sent his commanders against the Karluks. Although they are said to have tried and failed to appease him with gifts of livestock, it appears that it was Kür Khan’s letter asking for them to be pardoned that set the Great Seljuqs and Kara-Khitay against each other. Despite his vizier, Nizam al-Mulk’s grandson Nasr al-Din Tahir’s counsel to the contrary, Great Sultan Sancar is said to have threatened Kür Khan that unless he accepted Islam he would march against him. At the battle that took place in Katvan, near Samarkand, the Kara-Khitay may have applied cavalry tactics not yet known in Central Asia. In any case, they surrounded the Seljuqs and drove them against the Dergam River (9 September 1141). Great Sultan Sancar is said to have broken through and fled to Tirmiz when the malik for Sistan, al-Fazl Nasr, stood under the royal canopy in his stead (Köymen 1984: 321–3, 328–32; Turan 1993a: 190–1; Barthold IA/6: 274). While in Tirmiz, Great Sultan Sancar returned Mazandaran to the Seljuqs of Iraq and asked Sultan Mascud to await his orders in Rayy. He also asked for soldiery from Sistan, whose malik had been captured along with the sultan’s daughter Terken Khatun (the Karakhanid Arslan Khan’s widow) and his commander Kamac and his son. Although al-Fazl Nasr was released, Terken Khatun and Kamac had to pay a ransom, apparently remaining in captivity for nearly a year. With Great Sultan Sancar thus occupied, Atsız took Serakhs in Khurasan (October 1141). From there he advanced on the capital Marv, which he entered after some resistance (21 October 1141). He took Great Sultan Sancar’s treasury with him to Khwarazm where it seems the Kara-Khitay had begun to raid. Said to have agreed an annual tribute, Atsız returned to Khurasan and marched on Nishapur, threatening to pillage the city unless the khutba was read in his name. Although this seems to have been accepted (29 May 1142), possibly on Great Sultan Sancar’s return from Tirmiz, it once again reverted (July 1142). During this time Atsız’s brother Yınal-Tekin was raiding in Khurasan (Köymen 1984: 336–42; Kafesoğlu 1992: 54–6; Turan 1993a: 192). Once the Kara-Khitay had established an administration in Transoxania and withdrawn, Great Sultan Sancar moved on Khwarazm where he besieged Atsız in the capital Gurganj (July 1143). The Seljuqs entered the city but were repulsed. Dissolution (1105–94) 133 Nevertheless, Atsız sent an envoy bearing gifts and begged pardon. This seems to have been granted on the condition he returned all he had plundered while in Khurasan and proclaimed allegiance. Although Atsız is said to have complied with these conditions, he continued to resist Seljuq authority. When the sultan’s envoy learnt that Atsız had sent two Ismacilis to assassinate Great Sultan Sancar and alerted him, the assassins being found and killed in Marv, Atsız is said to have drowned the envoy in the Amu-Darya. When Great Sultan Sancar advanced on Khwarazm (October 1147), Atsız once again withdrew to the fortress of Hezaresb and flooded its environs. After a two-month siege the sultan took Hezaresb by force and Atsız fled to Gurganj where he begged pardon. Great Sultan Sancar demanded he proclaim allegiance in public, by the banks of the Amu-Darya River. Contrary to Turan, Atsız appears to have refused to kiss the ground as demanded, being forgiven nevertheless (2 June 1148) (Köymen 1984: 345–53; Kafesoğlu 1992: 58–9; Turan 1993a: 192).

Sultan Mascud When, instead of pursuing Sultan Mascud, Boz-Aba took Davud and al-Rashid with him to Fars in order to claim Mengü-Pars’ iqtac, the sultan appears to have taken the opportunity and re-established his authority. Although Selçuk-Shah was assigned Ahlat with his atabeg Oğuzoğlu after proclaiming allegiance, Sümer (IA/8: 139) and Köymen (1984: 392) were of the opinion that during this period Zengi continued to undermine the sultan not only by provoking the commanders against him but by his growing renown. Zengi was besieging Hims when he learnt that the Byzantine Emperor Ioannes Comnenus II had entered Syria (May 1137). Zengi lifted the siege and instead besieged the Crusader citadel of Bacrin (Montferrandus), which controlled the route between Aleppo and Hims. The ruler of Tripoli, Count Raymond, asked King Fulk in Jerusalem to help, but Zengi surprised them. King Fulk managed to flee to Bacrin but Raymond was captured. Although Joscelin II set out from Urfa with the intention of relieving Bacrin, the Crusaders capitulated before he could arrive. During the siege of Bacrin Ioannes Comnenus II took Antakya. The emperor is said to have sent Zengi an envoy, however, he seems to have come to an agreement with the Crusaders that in the following year Aleppo, Sheyzer, Hama and Hims would be attacked jointly and if taken left to Raymond. Before the onset of winter Zengi also took Ma’arrat al-Nu’man and Kafr Tab from the Crusaders (Alptekin IA/13: 529–30). In the spring a joint Byzantine and Crusader force besieged and took Büza’a north-east of Aleppo (9 April 1138). This seems to have caught Zengi by surprise. He sent one of his commanders, Ali Küçük, to Aleppo to strengthen the city defences and asked for help from Sultan Mascud in Baghdad. Seeing Aleppo was well prepared, Ioannes Comnenus II struck camp after three days (20 April 1138) and having taken al-Asarib (21 April 1138) besieged Sheyzer. Although his forces breached the outer walls, the inner citadel is said to have resisted for twenty days, so that the emperor raised the siege in return for a fee. Despite 134 Dissolution (1105–94) attempts to draw him into battle, Zengi appears to have successfully eluded the emperor who withdrew to Antakya. Having thus thwarted the joint campaign by the Byzantines and the Crusaders, Zengi re-entered Syria, where he gained Hims in exchange for Baalbek by marrying Zümürrüd Khatun, the mother of the ruler of Damascus, Mahmud, who in turn married Zengi’s daughter. Zengi then attacked and captured Araka from the Crusaders, as well as retaking Büza’a (16 September 1138), al-Asarib and Kafr Tab (7 October 1138) (Alptekin IA/13: 530). For his part Sultan Mascud seems to have done little other than promise Zengi help, at least until Zengi’s envoy the jurist Kamal al-Din succeeded in provoking the populace. Kamal al-Din is said to have given money for people to rise and demand action during Friday prayers in both the caliph’s and the sultan’s mosques. Joined by soldiery, apparently the populace continued their demonstra- tions in front of Sultan Mascud’s palace. Although as a result it seems the sultan gave orders for an army to be prepared, when the Byzantine emperor lifted the siege of Aleppo, Zengi is said to have instructed his envoy to stop Sultan Mascud as he would take the opportunity to occupy his lands. It seems Kamal al-Din was successful in this, too, as the sultan’s army is said to have decamped and returned home (Köymen 1984: 390–1). After al-Rashid’s assassination Sultan Mascud had dismissed his vizier and then had him killed. Appointed on Great Sultan Sancar’s instigation, the vizier is said to have been responsible for Seljuqid policy against al-Rashid. Nevertheless, Sultan Mascud’s new vizier al-Hazin was also one of Great Sultan Sancar’s administrators, having been responsible for the Treasury before being appointed governor to Rayy. During his short tenure the vizier appears to have been highly successful in reforming the administration and increasing tax revenue. As a result the commanders are said to have united behind Karasungur, Davud’s atabeg in Azarbayjan. Karasungur sent an ultimatum to Sultan Mascud, threatening to serve another sultan unless they were sent the vizier’s head. To back his demand, Karasungur is said to have brought Davud with a large army to the outskirts of Hamadan. When Sultan Mascud capitulated, Karasungur imposed his own vizier ‘Izz al-Mulk in al-Hazin’s stead (June 1139). Karasungur next journeyed to Fars and dispossessed Boz-Aba, assigning it to Selçuk-Shah, but Boz-Aba appears to have returned and imprisoned the prince in a fortress where he subsequently died or was killed (Köymen 1984: 383–4; Sümer IA/8: 138–9). While Sultan Mascud was dealing with Karasungur’s ultimatum, the ruler of Damascus, Mahmud, was killed by one of his men (23 June 1139), who succeeded in instating Mahmud’s brother Muhammad (1139/40). Mahmud’s other brother Bahram-Shah seems to have sought refuge with Zengi. Having been invited by Zümürrüd Khatun to attack Damascus, Zengi first took Baalbek (20 October 1139). Next, he cut off all routes to Damascus (6 December 1139), but on learning that the inhabitants were prepared to resist he is said to have lifted the blockade. When Muhammad died soon after (29 March 1140), however, and his son Abak was instated, Zengi marched on Damascus. This seems to have prompted the deceased Muhammad’s vizier Üner to make a pact with King Fulk in Jerusalem. Accordingly, Abak would pay the king tribute each month as well as jointly attack Dissolution (1105–94) 135 Zengi’s citadel Banyas, which if taken would remain with the Crusaders. As a result Zengi withdrew to Havran where he seems to have waited for a month in case of a Crusader advance against him. When this did not materialize, Zengi returned to Guta, near Damascus. While Zengi was in Guta, his commander in Banyas, Ibrahim Turgut, attacked a Crusader force en route to Jerusalem, but lost his life. This time Zengi seems to have withdrawn to Hama while a combined force from Damascus and Jerusalem besieged Banyas, which surrendered. Undeterred by the Crusaders’ occupation of Banyas, Zengi is said to have assigned Baalbek to a group of Yıva Türkmen from Arbil and Shehrizor, and in a surprise move besieged Damascus (22 June 1140). Apparently Abak capitulated and accepted to have the khutba read in Zengi’s name. On his way back to Mosul, Zengi also took the fortification of Bahmard by force from the commander Davud of Hisn Kayfa (Alptekin IA/13: 530–1; Sümer 1992: 117). Apart from the commander for Mardin in south-eastern Anatolia proclaiming his allegiance to Zengi (1141), there appears to have been little further activity in Iraq until Sultan Mascud gathered an army and announced that he intended to march on Zengi (1144). While the sultan was preparing his forces in Baghdad, Zengi is said to have sent money and his eldest son Sayf al-Din Gazi as hostage. His claim that he was too busy fighting the Crusaders to proclaim allegiance in person appears to have been true. On learning that Joscelin II had left Urfa, Zengi sent an advance unit under the command of Yağı-Sıyan whom he joined with a force of Türkmen. Zengi breached the walls and took Urfa, assigning it to his Türkmen commander Ali Küçük (24 December 1144). Two years afterwards, however, Zengi was killed by one of his servants while besieging the fortress of Jaber, which was in the hands of the Uqailids (14 September 1146) (Alptekin IA/13: 531–2; also Köymen 1984: 393 and Sümer IA/8: 139 and 1992: 116). The following year, the commanders Boz-Aba and Abbas in Fars and Rayy respectively appear to have set out from Isfahan with an undisclosed number of Seljuqid princes on the pretext of proclaiming allegiance to Sultan Mascud (1145). Realizing they were going to attempt to depose him when the commander Çavlı Candar failed to join him from Arran, the sultan appears to have withdrawn to Baghdad with the commanders Togan Yürekoğlu and Beg Arslan (Hasbey). However, three months later Sultan Mascud met Çavlı in Azarbayjan and marched on the rebellious commanders who are said to have retired to their iqtacs. When the following year Togan Yürekoğlu joined the conspirators, the sultan sent him to the Caucasus with Beg Arslan who killed him en route. The sultan also appears to have had Abbas killed (1146), prompting Boz-Aba now to march on Hamadan. However, Beg Arslan appears to have returned from Ardabil in time to defeat and kill Boz-Aba (1147) (Sümer IA/8: 139). After this Sultan Mascud is said to have handed over the running of his domains to Beg Arslan, the son of an Oguz beg who had migrated to Azarbayjan. Beg Arslan appears to have killed some of the commanders who opposed him. As a result those afraid or jealous of his growing influence over Sultan Mascud rallied around his nephew Muhammad (1148). When the caliph refused to support them, the commanders besieged Baghdad. Apparently displeased with Sultan Mascud’s 136 Dissolution (1105–94) dependence on Beg Arslan, Great Sultan Sancar once again journeyed to Rayy where Sultan Mascud seems to have persuaded him of Beg Arslan’s virtues. As a result several of the commanders are said to have journeyed to Rayy to proclaim their allegiance to Sultan Mascud. Others appear to have proclaimed allegiance on the sultan’s arrival in Baghdad (April 1150). Having toured his domains, Sultan Mascud died from typhoid at the height of his power while in Hamadan (2 October 1152) (Sümer IA/8: 140).

The Oguz rebellion It is not clear when the Oguz migrated out of Transoxania into Tocharistan (Kafesoğlu 1992: 47, 60–1, 67; Sümer 1992: 100–4; Köymen 1984: 406), however, they seem to have been involved on the Gurids’ side in their conflicts with the Ghaznavids, switching to the Seljuqs’ side at the end. After Katvan the Gurids appear to have taken Herat on the invitation of its inhabitants and successfully resisted Great Sultan Sancar’s commander Kamac’s efforts to retake it. As this endangered neighbouring Sistan, its malik al-Fazl Nasr is known to have offered military assistance. However, this does not appear to have been required as the Gurids accepted to pay an annual tribute (March 1147) (Köymen 1984: 353–60). Subsequently, Muhammad, a member of the Gurid ruling dynasty, appears to have fallen out with his half-brothers Mascud and Suri, the senior by blood. As a result Muhammad is said to have sought refuge in Ghazna with Sultan Bahram-Shah who had him poisoned because he was persuaded Muhammad was trying to overthrow him. On learning of his brother’s death, Suri marched on Ghazna and Sultan Bahram-Shah withdrew to Lahore. Possibly lulled into a false sense of security by the support he received from the Ghaznavid administration, Suri is said to have allowed most of his soldiery to return. Sultan Bahram-Shah is thought to have learnt of this and advanced from Lahore at the head of his army (13 March 1149). After some skirmishes between their advance units near Kabul (11 May 1149), Sultan Bahram-Shah defeated Suri (12 May 1149). Suri fled but was caught and hanged in Ghazna (19 May 1149) (Köymen 1984: 360–7). Suri’s brother Sam is said to have died of ‘grief’ en route to Ghazna. The task of avenging the deaths of his brothers thus fell on Hussein. This time Sultan Bahram-Shah withdrew into Sistan, but Hussein pursued and defeated him twice in succession. Once again the sultan fled to Lahore. Hussein took Ghazna by force and allowed his soldiery to burn and pillage the city. He is also said to have dug up the bones of Sultan Bahram-Shah’s ancestors and burnt them while exhuming his brothers’ remains for proper burial in Gur. These actions seem to have earned him the distinction Jihansuz or burner of the world. Hussein also refused to pay the Gurids’ annual tribute to Great Sultan Sancar. After his visit to Rayy (1150), Great Sultan Sancar moved against the Gurids with support from al-Fazl Nasr in Sistan and defeated him (probably 24 June 1152). Hussein was captured and sent to Marv (Köymen 1984: 368–82, 396). Dissolution (1105–94) 137 As Great Sultan Sancar’s affiliates it appears the Oguz were expected to supply a certain number of sheep each year for use in the palace kitchens in Marv (Sümer 1992: 104). On this occasion, however, the Oguz are said to have killed the tax collector because he insulted them when they refused to bribe him. Although the chamberlain responsible for the sultan’s kitchens seems to have met the shortfall from his personal income, Kamac, the governor of Balkh, is said to have brought the issue to Great Sultan Sancar’s attention and persuaded him to appoint him shihna to the Oguz. According to Köymen, a shihna was responsible for the establishment of migratien routes, the protection of peasants and townsfolk along these and the timely collection of tribal taxes. When the Oguz refused to recognize him, Kamac marched on them with his son but they were defeated and killed (Köymen 1984: 400–1, 406–9). The Oguz are said to have offered Great Sultan Sancar money and numerous Turkic slaves for killing Kamac and his son, but his commanders entreated him not to let the Oguz go unpunished. When the sultan arrived in Balkh the Oguz are said to have begged his pardon with their women and children, increasing their offer. Apparently Kamac’s grandson, now the commander for Balkh, and others insisted on reprisal, but when they attacked they were defeated (March– April 1153) (Köymen 1984: 409–12). Sancar is thought to have fled toward Balkh where he suffered another defeat after which he fled to Marv. He seems to have withdrawn from there as well, as the Oguz are known to have sacked the city (July 1153). The Oguz captured Great Sultan Sancar soon after and brought him back (Sümer 1992: 105). Sancar is said to have refused the Oguz their request to become their sultan. When he also refused the Oguz Bahtiyar’s demand for a part of Marv, the Oguz sacked the city again (August–September 1153). After returning to Balkh, the Oguz appear to have caged Sancar and taken him with them wherever they went (Köymen 1984: 414–17; Turan 1993a: 195). Great Sultan Sancar’s commanders appear to have gathered in Tabaristan, south of the Caspian Sea. Possibly on Great Sultan Sancar’s vizier Fakhr al-Mulk’s instigation they then declared Sultan Tapar’s remaining son Süleyman-Shah sultan in Nishapur (4 September 1153), before attempting to engage the Oguz who now began to pillage Khurasan. Presumably unsuccessful, Süleyman-Shah appears to have fled to Iraq (April–May 1154). The Oguz entered Mashhad and Tus (November 1154), then Nishapur (December 1154), where they may have appointed their one and only governor (Köymen 1984: 419, 428–30, 434–5, 449; Kafesoğlu 1992: 62–3; Sümer 1992: 105; Houtsma IA/8: 480–1). Köymen came to the conclusion that at least a quarter of the urban population of Khurasan died at the hands of the Oguz in the eighteen months after their victory near Balkh (Köymen 1984: 438–45). Although the Khwarazmshah Atsız appears to have taken advantage of the situation and sent his brother Yınal-Tigin to besiege Beyhak and raid its environs with such ferocity the region is said to have suffered from famine for two years after (from December 1153 to May or September 1154), it was Atsız who appears to have tried to form a coalition against the Oguz (1154). He is thought to have approached the Saffarid al-Fazl in Sistan, the Gurid Ala al-Din Jihansuz and 138 Dissolution (1105–94) the Bavendid Shah Gazi in Tabaristan. While he was thus occupied, the Seljuq commanders declared Sancar’s nephew the Karakhanid Mahmud Khan sultan but contrary to custom he does not seem to have announced his investiture to the heads of neighbouring states. Certainly throughout this period in Baghdad the khutba continued to be read in Great Sultan Sancar’s name. Although Mahmud Khan does not seem to have had any success, the Oguz are thought to have withdrawn to Marv and made peace (July 1155). Mahmud Khan seems to have approached Atsız who apparently responded by arriving in Khurasan with his son Il-Arslan (May 1156). Here he learnt that Kamac’s grandson Muayyad Ay-Aba had freed Sancar who was in Tirmiz (April 1156), his wife Terken Khatun having died in captivity. When Great Sultan Sancar notified the heads of the neighbouring states of his release, Atsız proclaimed his allegiance, dying of a stroke soon after (30 July 1156). Before Great Sultan Sancar could re-establish his dominion he also died and the Great Seljuq Empire came to an end (22 or 29 April 1157). Great Sultan Sancar had no sons and Khurasan was divided between Mahmud Khan, who though married to Sancar’s daughter at one point tried to join the Oguz, Kamac’s grandson Muayyad Aba and the Khwarazmshah Atsız’s son Il-Arslan (r. 1156–72) (Köymen 1984: 422–6, 449–51, 454–75; Kafesoğlu 1992: 65–72). Elsewhere Jihansuz, who had been allowed to return to Firuzkuh in Gur on proclaiming allegiance to Great Sultan Sancar, marched on Ghazna. When Sultan Bahram-Shah withdrew, he instated his remaining brother Sayf al-Din. Sultan Bahram-Shah returned and having taken Ghazna by force hanged Sayf al-Din. On Sultan Bahram-Shah’s death his son Khusraw-Shah succeeded to the throne. When he was in Lahore, Jihansuz took Ghazna by force and was declared sultan (1155). Jihansuz died soon after Great Sultan Sancar’s release from captivity (1156) (Köymen 1984: 424–5, 465; Kafesoğlu 1992: 63).

Collapse Sultan Mascud had declared his nephew Muhammad heir apparent and married him to one of his daughters. On his death, the commanders ignored Muhammad and instated the sultan’s son Malik-Shah. Apparently they regretted their choice and three months later Beg Arslan instated Muhammad in his stead. Sultan Muhammad appears to have had Beg Arslan decapitated in order to win over the commanders in Azarbayjan and Arran, but these turned against him and gathered around Süleyman-Shah, who had fled the Oguz (April–May 1154). Süleyman- Shah seems to have been an irredeemable alcoholic (Houtsma IA/8: 480–1). Meanwhile Caliph al-Muqtafi had taken Hilla and Wasit (1154). When Süleyman-Shah came to Lihf and established contact with him, the caliph agreed to have his name included in the khutba on the condition he did not interfere in the affairs of Iraq al-Arab (February–March 1156). However, Muhammad returned from Isfahan and having defeated Süleyman-Shah, imprisoned him (June–July 1156). Sultan Muhammad besieged Baghdad (February 1157). When Il-Deniz occupied Hamadan with Malik-Shah and his charge Arslan-Shah (Tuğrul’s son), the sultan lifted the siege. Possibly because he could not gain the support of Dissolution (1105–94) 139 Sultan Muhammad’s commanders, Il-Deniz withdrew and the sultan reoccupied the capital. In the meantime the Seljuq palace in Baghdad was pillaged and burnt. Presumably since he was caught between the two, namely Il-Deniz in Azarbayjan and the caliph in Iraq al-Arab, Sultan Muhammad does not appear to have attempted to campaign against either again (1159) (Köymen 1984: 481; Turan 1993a: 199–200; Kafesoğlu IA/10: 377–8; Zetterstéen IA/8: 576). After this the Seljuqs of Iraq came under the authority of the atabeg Il-Deniz who kept Sultan Arslan-Shah (r. 1160–77) a virtual prisoner in Hamadan with his sons Jihan-Pahlavan (the sultan’s stepbrother) and Kızıl-Arslan. Malik-Shah was poisoned and Süleyman-Shah put to death (1161), but those commanders ill- disposed toward the atabeg, namely Inanç Beg, Satmaz and Argu, the governors of Rayy, Isfahan and Kazvin respectively, approached the sultan’s brother Muhammad’s atabeg Sungur in Fars. Il-Deniz defeated them outside Hamadan and Muhammad was forced to flee to Khuzistan (1161). Although with the encouragement of the new caliph these commanders and others tried to instate Malik-Shah’s son Mahmud, they were also defeated (1162). Finally all the commanders in Azarbayjan, Arran and Iraq al-Ajam appear to have journeyed to Isfahan to proclaim their allegiance to Sultan Arslan-Shah (1165), who remained in his stepfather’s shadow. On Il-Deniz’s death (1175), Sultan Arslan-Shah is said to have tried to campaign against Jihan-Pahlavan who had declared himself atabeg in Azarbayjan, but having fallen ill he was forced to capitulate and died soon after (1177). Although his eight-year-old son Tuğrul became sultan he also remained in Jihan-Pahlavan’s shadow until the latter’s death (1185). When Sultan Tuğrul tried to assert his sovereignty over Jihan-Pahlavan’s brother Kızıl-Arslan with the help of the Türkmen affiliated to Kıpçak’s grandson Hasan, whose sister he appears to have married, Caliph al-Nasir (r. 1179–1225) became involved against him. Eventually Tuğrul lost his life in a battle near Rayy against the Khwarazmshah Tekiş (r. 1173–1200) and the Seljuqs of Iraq also passed into history (25 March 1194) (Yınanç IA/1: 610ff; Sümer 1992: 113; Kafesoğlu IA/10: 377–8 and 1992: 73–80; Köymen IA/12/2: 19ff). 8 Evaluation

According to social anthropological theory, the use of force defines the difference between primitive and advanced societies. In the former, individuals have the right to seek redress; in the latter, the state is sovereign. Tribesmen regulate political organization, economic exchange and so forth through kinship and descent. Civilized people regulate these and more complex relationships within a society, through specialized institutions (Sahlins 1968: 4–13; Barfield 1993:1–18). According to modern histories, not having developed beyond tribal organization, nomadic pastoral empires either stem from the charisma of one person or are basically a confederation of tribes that mimic a powerful sedentary neighbour in their attempt to regulate relationships with it (Khazanov 1994: 228–62, 295–302; Jagchid and Symons 1989; Waldron 1990: 36; Barfield 1996, 1993: 149–58; Kradin 2002, 2005). They remain barbarians until fully assimilated into sedentary civilized society (Grousset 1970). In the main Turkish historians likewise treat the characteristics of pre-Islamic Turkic polities as tribal, such as Ibn Fadlan’s most recent translator Şeşen (1975: 111–42; see also Divitçioğlu 2005), even if they see the Hsiung-nu, Kök Türk and Uygur empires as the most powerful examples of Turkic statecraft prior to the Great Seljuq and Ottoman empires. This view is most prevalent in Kafesoğlu’s Türk Milli Kültürü (Turkish National Culture). In this work western European historiography is reviewed from the perspective of the Vienna Diffusionist School, with the result that the structure and functions of statecraft proposed as Turkic are from polities not only centuries apart but also thousands of miles from each other (1994). The book’s final section is devoted to the adaptation of Turkic culture to the Irano- Islamic world by the Karakhanids, Ghaznavids and Seljuqids, but adds nothing new to the views Kafesoğlu expressed in his article on the Seljuqs in the IA. Nevertheless, the work is useful for identifying which institutions prevailed in a specific Turkic polity. More recently Togan and Ögel’s former student, the Sinalogue Çandarlıoğlu, a specialist on the Uygurs, has attempted to distinguish a common Turkic socio- political structure (family, clan, tribe, nation, state), political ideology (freedom, country, folk, law), concept of sovereignty and divine favour, statecraft (councils, head of state, First Lady, regent, other offices, the army), religious beliefs Evaluation 141 (ancestor worship, Celestial Supreme Being), economic life, etc. in pre-Islamic times (2003: 91–101), but she does not refer to her sources. In contrast, the economist Divitçioğlu’s in-depth anthropological analysis of the Kök Türk Empire (553–745), now in its third edition, is period-specific (1987, 2000 and 2005). The work’s strength lies in its application of what is known historically; this enables Divitçioğlu to avoid many of the pitfalls of modern social anthropological theory as applied to nomadic pastoral polities (Sneath 2007), even though he accepts most of the structural functionalist school’s segmentary kinship model. As a result, Orta-Asya Türk I˙mparatorluğu: VI.–VIII. Yüzyıllar (Central Asian Türk Empire: VI.–VIII. Centuries) can be said to provide a thorough analysis of Kök Türk politics, society, economics and religion. Divitçioğlu has applied this methodology to aspects of Karakhanid, Seljuq and early Ottoman history as well (1987, 1992, 1994 and 1996), but with mixed results. His opus on the establishment of the Great Seljuq Empire is too brief to be of extensive use in addressing the problems their history raises. There are also İnalcık’s papers on the influences of Iranian and Turko-Mongol ideology on the Ottomans (1993a, 1993b and 1998). These are of interest only because of their interpretations of the eighth-century Orkhun stone inscriptions and the eleventh-century Kutadgu Bilig by Yusuf Has Hajib, both of which are also referred to by Turkish historians on the Great Seljuqs with regard to their concept and execution of statecraft. The Oguz referred to in the Orkhun stone inscriptions were situated near Lake Baykal. Although most accepted Kök Türk supremacy, according to recent excavations, some appear to have migrated from Inner Asia, reaching the region between lakes Balkash and Issyk-Köl during the sixth to seventh centuries. However, none of the names of the lineages known to have made up these Oguz reappear among their namesakes between the Caspian and Aral seas in the tenth century. According to Barthold and Minorsky (see Esin 1980a), the tenth-century Oguz may have been lineages united by the Şul (from Çöl or desert), who were in Jurjan, south-east of the Caspian Sea, from at least as early as the fifth century. This does not mean they were not part of the Kök Türk Empire, quite the opposite. These latter Oguz included lineages known to have been affiliated to the Kimek and the Peçenek from Inner Asia. Consequently, the Orkhun inscriptions can be referred to for Great Seljuq ideology and Türkmen cultural expectations, more so since Kıpçak lineages were also affiliated by the Seljuqs. As for Yusuf Has Hajib; he was from Balasagun, which is thought to have been near Lake Balkash (Bozkurt 1992: 100). His opus, Kutadgu Bilig, which was finished in Kashgar, was dedicated to the Karakhanid Tabgaç Buğra Kara Khan (r. 1056–1103). Kashgar was situated on the westernmost part of the Tarim Basin, where the trade routes from China converged after branching off to the Indus Valley at Yarband. Not surprisingly, the influences in Kutadgu Bilig have been traced to the Iranian, Chinese and Indian literary traditions, as well as specifically to the Islamic philosophers Farabi (870–950) and Ibn Sina (980–1037) (Dilâçar 1988; Arat IA/6: 1038ff; İnalcık 1993a; Divitçioğlu 1992: 130–207). Although Kashgar is known to have come under Türgiş rule (632), this was short-lived. 142 Evaluation Initially on the Left Wing of the western half of the Kök Türk Empire, the Türgiş broke away soon afterwards and established an independent kaganate (658–766). Situated between Lakes Zaysan and Urungu at the time, with a branch east of the Irtysh River, the Karluk began to push the Türgiş out of Kashgar from 670 (Sümer 1992: 18; Arat IA/6: 351–2). When the Uygur, who had ended the Kök Türk Empire, defeated them, the Karluk migrated to Transoxania where they ended the Türgiş kaganate. Consequently, Kutadgu Bilig seems more relevant to Karakhanid than Great Seljuq ideology and expectations. This is of particular importance as the Karluk, the main constituents of the Karakhanids, were at log- gerheads with the Seljuqs throughout their reign. Besides the Orkhun stone inscriptions and Kutadgu Bilig, there is also Mahmud of Kashgar’s the Divanü Lügat-it-Türk to consider. It is thought to have been finished in Baghdad (1072–3) because allegedly presented to the caliph (Atalay 1992/I: xiv–xv), possibly al-Muqtadi. Be that as it may, Mahmud, who appears to have been from a ruling lineage (Atalay 1992/I: xi–xiv) (though contrary to Dilâçar (1988: 23) it is not certain he was related to the Karakhanids (840–1212) (Bozkurt 1992: 92)), seems to have travelled throughout the Tarim Basin and Central Asia, visiting Turkic people and their cities (Atalay 1992/I–IV). Consequently, his opus is not only a reliable lexicon of the Turkic languages at the time of the Seljuqs but in contrast to Kutadgu Bilig provides invaluable insights into the cultural background of the Great Seljuqs’ Turkic affiliates as well. This is verified by Köymen’s studies of everyday life throughout Central Asia with regard to Turkic housing, dress, food, culture and social life based on the Divanü Lügat-it-Türk (1971a, 1971b, 1971c, 1971d and 1975a). Last but not least, there are Ibn Fadlan’s observations of the Oguz (922). These confirm Akçura’s precepts rather than the conclusions Turkish historians have drawn from the Orkhun stone inscriptions and Kutadgu Bilig, or indeed from their reviews of all that is presently known of Turkic history. In his travelogue, Ibn Fadlan remarked that the Oguz commander-in-chief had been chosen because he was considered to be the best rider and archer. In view of the numerous households Ibn Fadlan observed in the commander-in-chief’s camp, it is clear he had to merit his office although of noble descent. It is also clear that nobility was not enough to ensure loyalty among the Oguz. Ibn Fadlan mentions a noble, Küçük Yınal, who had become a Muslim but had to renounce Islam due to his affiliates’ disapproval. In this context, even though he thought the Oguz heathens, relying on their judgement rather than on religious teachings, Ibn Fadlan related that like Muslims they believed there was one Tengri (Celestial Supreme Being). In other words, rather than the Qur’an, the Oguz appear to have conducted their affairs according to the töre or customary law. The töre appears to have benefited from a consultative process because not made up of hard-and-fast rules. In an example given by Ibn Fadlan, a plaintiff is consulted by his patriarch with respect to a ruling. This appears to have applied in most decisions. According to Ibn Fadlan, an objection by their lowliest member could hold up an Oguz council (Şeşen 1975: 30–40). Ibn Fadlan’s observations seem to confirm Akçura’s precepts. Evaluation 143 To understand Seljuq and Türkmen expectations fully however, it is best first to characterize nomadic pastoral organization. There can be little doubt polities sustained by agriculture were more complex than those reliant on pastoralism. Hydraulic agriculture in particular provided gainful employment for urban specialists – administrative, military and religious, not to mention craftsmen and merchants. Nevertheless, as the Seljuqs demonstrated, customs and institutions specific to life on the Eurasian steppes retained their viability in sedentary environments where pastoralism could be practised.

Nomadic pastoral organization According to Khazanov pastoralism is not autarkic. Apart from the need for man- ufactured goods, such as utensils for cooking, carpentry and so forth, almost all pastoralists include vegetables in their diet (1994: 39, 81). Both the Hsiung-nu and the Oguz are known to have cultivated crops. Setting aside the difficulty of breaking open the thick sod or pebbly ground, at least until the advent of the double plough and later still of tractors (Barfield 1993: 136), agriculture on the open steppe remains prone to drought for forty per cent to seventy per cent of any decade (Khazanov 1994: 45). Consequently, it is the ratio of agrarian activity that defines the nomadic component of pastoralism. Where the cultivation outweighs animal husbandry seasonal migrations are shorter, even when separate pastoral households are involved. In more sedentary environments still, animals are pastured in the mountains during summer but brought back in winter and penned, which requires laying-in of fodder (1994: 17–25). Given the different environments of Inner and Central Asia and surrounding zones (Krader 1955; Grousset 1970: xxi–xxx; Ligeti 1946), it is clear that prior to industrialization the determining factor of the type of pastoralism practised was geography.

Herds and horses Cattle do not travel well and go hungry if the grass is not the right consist- ency (Khazanov 1994: 47). Moreover, north of the fifty-first parallel oxen cannot winter in the open. Horses, on the other hand, can survive north of the sixtieth parallel (Jankovich 1971: 16). The hardy steppe horse is known to uncover grass up to 40 cm under snow with a few kicks (Khazanov 1994: 49–50). Besides cattle and horses, nomadic pastoral herds include sheep, goats and Bactrian camels. Although pastoralists use all their animals for subsistence as well as transport, sheep and goats are the mainstays. They not only breed faster than the other three (Barfield 1993: 139), but co-exist throughout the Eurasian steppes and Middle East (Khazanov 1994: 26–7). However, as goats are favoured in marginal pasturelands, sheep tend to be paired with horses. As for cattle, these are favoured in wetter regions while camels predominate in the deserts (Barfield 1993: 137). 144 Evaluation Of these five, the horse is historically the most important. Horses were not only domesticated well after the others, between the fifth and fourth millennia BCE, but also do not appear as mounts until much later. Although argument still rages as to how and where riding developed (Barfield 1993: 132–4; Khazanov 1994: 90–7; Jankovich 1971: 11–32; Mallory 1994: 24–65; Kafesoğlu 1994: 18–38), consensus is that it must have happened sometime before the end of the eighth century BCE, when the Cimmerians and Scythians appear in the Middle East. In Inner Asia the Hu appear later, toward the end of the fourth century BCE (Barfield 1993: 134). Soon after, however, the Hsiung-nu Empire was formed, something that did not occur in the West. The reason for this must be sought in the quality and quantity of pastures, that is, the geography of Inner Asia. It has been calculated that compared to Mongolia, the Hungarian Plain could support at most only 323,000 animals of which a mere 50,000 to 70,000 would have been horses. The figures for Mongolia differ dramatically. According to Chinese annals, in peacetime Hsiung-nu herds averaged between fifteen and twenty-five animals per head; this figure rose to 300 per head among the richest households in good times (127 BCE) and fell to less than two per head among the poorest during plagues, droughts, etc. (CE 46). Given that the Hsiung-nu are thought to have numbered at most one and a half million people, their herds would have consisted of between one and forty-five million heads of livestock. Taking an average of thirty million, it is thought there would have been four million horses (twelve to fifteen per cent of a herd). However, in wartime horses could make up to forty per cent of a herd (CE 46), that is twelve million, or five to six per household. These figures more than verify separate observations that every Hsiung-nu warrior had at least three fresh mounts (Roux 1995: 81–3). The need for horses in order to counter the mounted archers’ self-evident martial supremacy is underlined by China’s near disastrous but ultimately successful expedition to Fergana for their specially bred horses (Jankovich 1971: 29–38; see also Grousset 1970: 35–6). The need for horses may also go a long way towards explaining the deterioration of agriculture in the Middle East under the Seljuqs and later the Mongols. In order to realize his ambition of conquering the world, Chinggis Khan is said to have contemplated turning China into grassland. Not surprisingly, this may also explain why neither the Seljuqs nor the Mongols were able to sustain their conquests in Syria or present- day Iraq, both of which are ill-suited to horses, let alone achieve their ambitions of conquering Egypt (Jankovich 1971: 83).

Pasturage and ownership Although the ownership of animals among nomadic pastoralists is private, that of pastures is corporate. On the Eurasian steppes the need to regulate migration routes led to the best pastures being grazed by the strongest lineages at the best times of year (Khazanov 1994: 123–6; also Barfield 1993: 144). This could easily lead to disputes between lineages, such as one accusing another of vacating a pasture too Evaluation 145 slowly. Disputes could also be external as in peacetime it was acceptable to move through another polity’s pastures (Barfield 1993: 144). An oft-cited story concerning Modun (209–174 BCE), the founder of the Hsiung-nu Empire, underlines the concept of corporate ownership. When the Tung-hu requested first his finest horse and then his favourite consort, contrary to advice Modun is said to have complied, saying he wanted to continue good relations. Encouraged, the Tung-hu next requested territory. When counselled to acquiesce because it was wasteland, Modun executed his advisors, saying: land is the foundation of the state, how can you give it away? Nevertheless, the survival of any nomadic pastoral camp depended on their winter pasturage (Barfield 1993: 145). While the minimum number of animals a household needed to survive changed according to the species composition of its herd (Khazanov 1994: 28–33), availability of grazing and water determined the maximum number of animals. This put an added constraint on winter camps because they were more crowded. Camps were set up preferably in windswept areas free of snow, in low-lying mountain or river valleys and depressions. In forested areas houses were built with barns, the surrounding land being cultivated during the summers for grain and for winter fodder (Barfield 1993: 142, 144). Despite their need for secure and plentiful winter pasturage, nomadic pastoral groups found economic and hence political regulation potentially restrictive of their fortunes. As Barth (1986) and also Divitçioğlu (1987) have noted, livestock provided the means of both production and consumption (meat, dairy products, felt, wool, etc.). This made nomadic pastoralists particularly vulnerable since their herds were determined as much by bad winters, crippling spring frosts (jute), summer droughts, or indeed disease, as they were by their reproductive capac- ity. If the ratio of animals to people was upset then survival came under threat regardless of the quality of grazing available by affiliation, because if you ate your capital you could no longer produce and consume. Whatever the minimum number and mix of animals required for survival, therefore, social fragmentation into camps and hence independent economic units must have ensured that not all members of a nomadic pastoral polity were affected by adverse conditions, whether climatic, biological or human. However, it does not follow that ownership of livestock was limited by the number of animals a household could rear and herd. A herd could be distributed to camps where those with insufficient animals of their own, commoners, servants or even slaves, looked after them in return for a share of their produce, most likely under the supervision of a family member or trusted lieutenant (Sneath 2007: 17, 136, 139, 152; also Khazanov 1994: 152–64). The importance of safe and adequate winter pasturage may go a long way towards explaining the early history of the Seljuqs. Apart from Arslan Beg near Nur, it is the most likely reason why Selçuk Beg’s youngest son Yusuf does not receive mention until after Arslan’s imprisonment, and why Musa and his charges may have wintered in Khwarazm throughout this period, not to mention Kutalmış and his brother’s sudden appearance in Khurasan by their cousins’ side. 146 Evaluation Kinship and descent The smallest unit on the Eurasian steppes was the camp, thought to have ideally consisted of a patrilineal extended family. When a son married, he took his share of the herd but did not move away unless at odds with other members (Barfield 1993: 145). This was because large kin groups gave protection, in the same way grouped camps enabled precious winter pasturage to be defended. This did not mean families and hence camps did not segment. There could be many reasons for splitting off besides familial strife, such as overpopulation and more dramatically the demise of relatives and/or animals. With their group depleted, unless able to attract new households, those remaining joined other camps where they had forged a previous relationship, mostly through marriage. If unable to do so, members became sedentarized (Barth 1986), or as noted above more likely reduced to herding a richer relative’s livestock. Patriarchal units and their camps are considered to have formed the building blocks of socio-political organizations forged to deal with the outside world, whether constituted of other pastoral affiliations, empires or sedentary states. Grouped internally along bloodlines, affiliation by descent provided a political genealogy through which rights to pasturage and social and martial obligations were defined (Barfield 1993: 147). In other words, while kinship defined family, descent identified affiliation. Arslan Yabgu’s affiliates, Yağmur, Kızıl, Gök-Taş and Buka were not related to Selçuk Beg or of noble descent. They were known as the Yabgulu because they had accepted Arslan’s authority. Likewise, the Seljuqs were in Çağrı and Tuğrul Beg’s service and later in the service of the rulers of the Great Seljuq Empire. Likewise, the Yınallı were those who served Selçuk’s youngest son Yusuf Yınal and later his descendants, namely Ibrahim Yınal, Er-Taş or Er-Basgan’s. According to social anthropological theory, lineages are ranked according to their genealogical closeness to the founding father, in many cases a mythical figure for political reasons. Also, when the head of a lineage died his eldest son was expected to take over, although a patriarch could choose any one of his sons to succeed (Sahlins 1968: 24–5, 49–50; see also Barfield 1993: 147–9). This is what Sultan Alp-Arslan did when he appointed Malik-Shah heir apparent, although Ayaz was the eldest. Furthermore, it is noteworthy that on Arslan Yabgu’s imprisonment Musa did not take up the Seljuqs’ leadership, instead it was Yusuf Yınal. With the exception of Sümer, Turkish historians have proposed that Çağrı and Tuğrul Beg attacked their uncle because their guardian Musa was the senior Seljuq and Ali Tekin had broken with tradition or been divisive for reasons of his own. In doing so they ignored what they were more than willing to accept with regard to Çağrı and Tuğrul Beg, namely that the eldest member did not necessarily want to lead because it was his birthright and occasionally deferred to a more willing or able relative. Although the Seljuqs called their uncle Musa Yabgu Kalan, it would seem Musa did not want to lead the Seljuqs although he was the eldest. It is, therefore, highly unlikely that he rebelled when Alp-Arslan defeated Kutalmış and succeeded to the throne. Evaluation 147 Turkish historians are wrong to claim that Kutalmış challenged Tuğrul Beg and then Alp-Arslan because he was senior, a consideration they extend to Kutalmış’ son and grandson, Süleyman-Shah and Kılıç Arslan. On the contrary, as the eldest son’s sons, Çağrı and Tuğrul Beg were the senior branch and Çağrı Beg’s descendants from the senior lineage. Moreover, Sultan Malik-Shah confirmed that Seljuq succession was from father to son when he objected to his uncle’s claim to the throne. Although the existence of a conical clan structure is strongly contested (Khazanov 1994: 146–7; Sneath 2007: 47, 60), Sahlins (1968: 49–50) and Barfield (1993: 147–9) consider nomadic pastoral clans also to be conical, in other words the eldest of eldest sons ranks the highest and is listed to the left of the clan’s founding father. At first glance the Oguz seem to be a good example of a conical clan. Oguz Khan’s eldest son Gün’s eldest sons’ affiliates, the Kayı, Bayat, Alka Evli and Kara Evli were respectively senior to the lineages affiliated to Oguz Khan’s remaining sons, Ay, Yıldız, Gök, Dağ and Deniz, in that order. So the Boz-Ok lineages or Left Wing, which claimed descent from Gün, Ay and Yıldız, were regarded senior to the Üç-Ok lineages or Right Wing, which claimed descent from Gök, Dağ and Deniz (see Appendix 2). However, whether won on the battlefield or in council, it appears merit played havoc with genealogies that were little more than political artefacts. As the Seljuqs were of the Kınık, for a time the Kınık rose to the top even though they were the lowest Oguz lineage. If the Seljuqs’ success had lasted longer, perhaps the Kınık’s rise would have been permanent. Nor is there any evidence that the Oguz yabgu always came from the senior Kayı lineage (Togan 1982; also Divitçioğlu 1994). Nevertheless, the division of Turkic and Mongol nomadic pastoralists into Akbudun or noble lineages and Karabudun or common folk (Barfield 1993: 148) appears to be a defining characteristic of conical clans whose structure of commoners and aristocrats seems to allow the emergence of a third level at the apex, namely that of a royal or imperial lineage (Leach 1970; see also Maisels 1993: 236–54), such as the Kök Türk A-shih-na, which separated itself from what Khazanov calls the genealogical cliché (1994: 146). Such lineages could be of remarkable endurance; Modun’s descendants ruled for 600 years, Chinggis’ for 700 years and Osman Beg’s for 600 years (Barfield 1993: 149). The Hsiung-nu ruling lineage, the Luan-ti, is thought to have married the daughters of at least three other lineages, the Hu-yen, the Lan and Hsü-pu (Barfield 1996: 37, 41–2). In the case of the Kök Türk, the Hsiung-nu’s descendants, members of the A-sheh-tê seem to have given their daughters in marriage to the men of the A-shih-na (Divitçioğlu 1987: 169). In social anthropological theory such marriages are classed as matrilateral cross-cousin or ‘mother’s brother’s daughter’ marriages. According to Sahlins, in such a marriage lineage B receives its wives from A and passes its daughters to C, who likewise pass their daughters to D. The system can be closed by allowing D to marry its daughters into A, or it can remain open. The latter allows lineages to be ranked. If wife-receiving rather than wife-giving was the noble prerogative, as 148 Evaluation it seems to have been among the Kök Türk, then the lineages in the above example would be ranked as D, C, B and A. Whereas in a closed system D outranks A, A outranks B and B outranks C, and C outranks D. There are also patrilateral cross- cousin or ‘father’s sister’s daughter’ marriages, in which case the flow of women reverses each generation. However, the prevailing form of cross-cousin marriages is neither matrilateral nor patrilateral, but bilateral. It allows for political or economic choice without making it look expedient since the marriage reinforces a previous union (1968: 58–63).

Nobility and the töre Turkic nobility were known as Akbudun (Ak Kamığ), the common folk as Karabudun (Kara Kamığ) and the slaves as kul. They made up a Turkic affiliation, such as the Oguz, Karluk, etc., noble lineages that had come together with their common folk and slaves. Some affiliations had royal lineages; the Türk had the A-shih-na, the Uygur the Yağlakar and later the Hsieh-tieh, the Kırgız the A-jih. In total the Türk consisted of twelve lineages, which in the context of their empire Divitçioğlu has classified as the inner lineages. The outer lineages are those conquered by the inner lineages. The related lineages are those joined to the inner lineages by marriage, effectively an alliance. Also, there are thought to have been slave lineages, such as the Türk may have been until they rebelled and defeated their masters, the Avar. Last but not least, there also seem to have been groupings of up to 15,000 warrior households from different nations. The Tarduş are a case in point. Originally the Kök Türk Tardu Kagan’s élite corps, they continued to serve the Uygur Empire. Finally, under the Kök Türk, the outer, related and slave lineages were ruled by members of their royal lineage, the A-shih-na (Divitçioğlu 1987: 162–72). When Great Sultan Sancar marched on the Oguz because they had killed his tax collector, they are said to have offered money and Turkic slaves. Nomadic pastoralists are thought to have had no use for large numbers of slaves (Khazanov 1994: 160), particularly as these could easily escape (Maenchen-Helfen 1973: 199), but this does not seem to be true given the number of commoners of slave ancestry among the Türkmen in the nineteenth century (Sneath 2007: 154, 186). Certainly, the existence of slave lineages suggests that slavery was widespread. The relationship between noble lineages and their common folk appears to have been only slightly better than outright slavery. There is no doubt that Turkic nobility expected absolute obedience from their affiliates as well as their slaves. The Orkhun stone inscriptions are quite clear. On the East Face of the inscription erected in his name, the Kök Türk Bilge Kagan states that although he had clothed the naked and made the poor rich (Tekin 1988: 45), by being disobedient the common folk had betrayed him, the state and the töre (Tekin 1988: 43). After Arslan Beg was imprisoned, his affiliates, the so-called Türkmen of Iraq refused to go into service with Çağrı and Tuğrul Beg, claiming persecution by them (1025). After Çağrı and Tuğrul Beg had come to Khurasan, where the Türkmen of Iraq were in full revolt against the Ghaznavids, and Evaluation 149 were successful at Dandankan (1040), they moved away to Iraq, where they gained their name of distinction, and then to south-eastern Anatolia when the Seljuqs expanded into Iraq al-Ajam too (1043–4). It was only toward the end of his life that one of them, Kızıl, was reconciled with Sultan Tuğrul, marrying his sister (1040–1). Separately, Sultan Tuğrul is said to have had a group of Türkmen affiliated to him caught and beaten when they refused to come to northern Syria as ordered (June 1056). Clearly, the relationship between Turkic nobility and those affiliated to them was one of master and servant, whether or not the nobility was the founder of an empire. The töre appears to reflect this. In the Orkhun stone inscriptions the töre is linked to statecraft in the sense of the rule of law (Tekin 1988: 9, 13, 15, 17, 37, 41, 43, 51), being referred to also as Turkic custom (Tekin 1988: 11, 41). When Kapagan Kagan dispensed with his commander-in-chief and counsellor Tonyukuk, he is said to have become a member of the Kök Türk high court (705–16). Kafesoğlu lists the following crimes as punishable by death in such a court: murder, theft of livestock, drawing a sword in peacetime, adultery and rape (1976: 239–40). As is clear from Ibn Fadlan however, a centralized authority was not necessary for the rule of law. On the Eurasian steppes the nobility decided and applied the code of laws by which they and their affiliates and slaves conducted themselves. They did not need a sovereign or even a contiguous realm. As late as the seventeenth century (20 September 1640), various Mongol lineages held a kurultay or general assembly at which they agreed on a law code that resembled one drawn up in the sixteenth century by Altan Khan. The code remained intact among the Oirats until 1892 and was applied by judges in courts (Sneath 2007: 181ff).

The Seljuqs’ nobility and aspirations In keeping with Akçura, Kafesoğlu argued that Selçuk Beg must have been of a ruling lineage as his father’s name of distinction was Temir Yalığ (Iron Bow) (IA/10: 354). Sümer agreed that the bow symbolized sovereignty as the Orkhun stone inscriptions state that the Kök Türk kagan in the East had sent three arrows to his counterpart in the West, but he did not draw an inference from this as concerns the Seljuqs’ lineage. Turan thought that since Selçuk Beg is referred to as sübaşı or commander-in-chief he must have been of noble blood but concluded there was no evidence he came from a ruling lineage (1969: 31–4). Köymen based his view on the Seljuqs’ letter to the caliph. Written after Dandankan, in it they do not mention their ancestry, only that they had been in the service of Islam until their paternal uncle and leader was imprisoned by Sultan Mascud’s father and died seven years later in captivity. Consequently, Köymen thought the Seljuqs free men (1979: 359–61). The Seljuqs clearly were Akbudun or aristocrats. Their affiliates, such as Yağmur, Kızıl, Gök-Taş and Buka, were Karabudun or commoners in their service. This is confirmed by the Ghaznavid vizier; on hearing of the Seljuqs’ arrival in Khurasan, he is said to have remarked that until then they had been dealing 150 Evaluation with mere shepherds but now would have to deal with their masters. Furthermore, Arslan Beg’s followers are said to have numbered 4,000 tents. This can be taken to indicate that like other nomadic pastoral noble lineages since the Hsiung-nu, the Seljuqs also organized those in their service decimally. As for their political aspirations, in keeping with Akçura, both Köymen (1979: 43–78) and Kafesoğlu (IA/10: 353–8) argued that from the outset the Seljuqs were intent on establishing their own polity. Selçuk Beg’s possible usurpation of taxes due to the Oguz yabgu from Jand and Arslan’s eventual appropriation of the title of yabgu seem to support this argument, as possibly does Sultan Mahmud’s failure to entice the Seljuqs into Ghaznavid service. Nevertheless, a desire to remain independent explains their actions better. Upon their arrival in Jand, the Seljuqs are said to have decided that to have any standing in the eyes of the locals they would have to adhere with their customs. Köymen was of the opinion that Selçuk Beg requested a Muslim jurist from the nearest Samanid governor (1979: 21–3). Certainly, having converted to Islam, the Seljuqs would have been in a position to attract other Muslims to their banner and resist the Oguz yabgu’s tax collector. Particularly as in the first quarter of the tenth century Türkmen of the Oguz and Karluk are known to have resided in Süt- Kent, near Lake Karakul north of the Syr-Darya. There may also have been a large number of Türkmen near Tashkent (Sümer 1992: 59). Whether Selçuk Beg achieved his position by converting to Islam or by forcibly resisting the Oguz yabgu’s tax collector or both, it seems to have enabled the Seljuqs to negotiate pasturage around the village of Nur, between Bokhara and Samarkand. However, the necessity for this is not clear; possibly the Türkmen who flocked to the Seljuqs’ banner were put in Arslan Beg’s service, or having fallen out with Mikail, Arslan Beg decided to leave, possibly with the newly joined Türkmen, or Shah-Malik forced the Seljuqs out of Jand and they separated. Whatever the case, clearly there was not enough for all of the Seljuqs around Nur. After Jand, Çağrı and Tuğrul Beg are thought to have approached a junior member of the Karakhanid dynasty but failed to agree terms. However, rather than Çağrı Beg’s reputed journey across Iran into the Caucasus as a result, it is more likely they and their uncle Musa remained in Transoxania with Yusuf Yınal, possibly wintering in Khawarazm. Contrary to Turkish historians’ assumption that Çağrı and Tuğrul Beg attacked their uncle Yusuf Yınal because the Karakhanid Ali Tekin overlooked the older Musa when he appointed him to lead the Seljuqs, it is clear that Musa never had any such ambition. Most likely the said attack took place after Ali Tekin’s death, when his commander-in-chief killed Yusuf Yınal. The Seljuqs needed Ali Tekin’s support not only because he was at odds with the branch of the Karakhanid dynasty Çağrı and Tuğrul Beg had fallen out with, but also in order to remain in Transoxania. After the demise of the Samanids, the Oguz, Türkmen and Kıpçak all seem to have been vying with the Karluk for pasturage. What better way of securing grazing rights and migratory schedules than through martial association with a local ruler, particularly while the Karakhanids and Ghaznavids fought over the Samanids’ spoils? Evaluation 151 There were Oguz or Türkmen groups other than the Seljuqs in Transoxania, such as Hasan Tak who deserted Ismacil Muntasir on the battlefield for the Karakhanid Ilek Khan Nasr. Hasan Tak emerges two years later, in Khurasan, this time at the head of Sultan Mahmud’s forces, which are known to have contained an Oguz contingent, presumably his (Sümer 1992: 70). While Yusuf Yınal most probably remained in the environs of Nur during the fifteen years prior to their appearance in Khurasan, Musa and his charges may have wintered in Khwarazm, which was under Ghaznavid governance. If so this could have served to reinforce the Seljuqs’ knowledge of Irano-Islamic expectations gained while in Jand. This interpretation of events helps explain not only the Seljuqs’ loyalty to Ismacil Muntasir but also their affinity to the Khwarazmshah Harun and Ali Tekin whose attempt to dislodge the Ghaznavids from the region they appear to have supported. In short, the Seljuqs cannot be said to have even tried to establish their own polity after leaving Jand, not until they crossed into Khurasan. After Dandankan their expansion westward into Iraq and Azarbayjan and from there into Anatolia and Syria appears to have been eased by their understanding that Irano-Islamic expectations differed from theirs. They would have gained this knowledge from the Samanids and later the Ghaznavids. Western Iran, Mesopotamia and Syria allowed nomadic pastoralists to co-exist with sedentary farmers, at least to a certain degree, but the size, number and proximity of urban centres required a centralized authority.

The problems of conquest Turkish historians agreed that the Türkmen caused the Seljuqs severe problems and that they were redirected to Anatolia to undermine Byzantine defences.

Migrations and the Türkmen Togan thought because the Kırgız did not take up the Kök Türk’s imperial mantle from the Uygur when they ended the latter’s empire (840), the Mongol Khitay were able to take over in Inner Asia (924), precipitating the westward migration of the Kimek (Kuman/Kıpçak) (1981). What Togan appears to have overlooked is that the Khitay subsequently founded the Liao dynasty in northern China (936–1122). When the establishment of the Sung dynasty in the remainder of China (960–1004) stopped them from expanding south, the Khitay turned their attention westward. It is their failed attempt to conquer Kashgar and the Issyk-Köl region that seems to have started the Kimek migration (1017) (Arat IA/6: 713ff). Reports of a Kıpçak presence in Khwarazm coincide with the governorship of Altun-Taş (1017–32) (Barthold 1990: 316). According to still other reports, during severe winters the Kıpçak are said to have migrated south when not at war with the Oguz (Sümer 1992: 64). Originally situated around the Irtysh River, the Kimek appear in Eastern Europe after they had driven the Peçenek and Uz into the Balkans (1050). This is 152 Evaluation more than a century later than Togan proposed and coincident with the Seljuqs’ ascendancy. By the beginning of Sultan Alp-Arslan’s reign, Jand was ruled by a Kıpçak beg and other Kıpçak were mingling with the Oguz/Türkmen in Üst-Yurt (1065–6) (see Sümer 1992: 92). If there was internal pressure on their resources, the Oguz could not have allowed the Kıpçak to share winter pasturage. It is more than a century after this that the Oguz were forced to migrate into Transoxania and from there into Tocharistan, coincident with the Khitay’s offshoot, the Kara-Khitay’s westward displacement of the Kanglı, which put pressure on the Karluk. Karluk presence in Transoxania brought the Great Seljuqs in contention with the Kara-Khitay and the Oguz (1141 and 1153), resulting in the destruction of the Great Seljuq Empire. Consequently, the reasons for the Türkmen influx into the Middle East need to be sought elsewhere. Sultan Mahmud had cleared Khurasan of Arslan’s affiliates (1028), but his eldest son Mascud invited Yağmur back in order to wrest the throne from his brother (1030). Having succeeded, Yağmur persuaded Sultan Mascud to allow also Kızıl, Gök-Taş and Buka to return. Although the so-called Türkmen of Iraq are said to have served Sultan Mascud well, their conduct seems to have remained the focus of local complaints. As a result Sultan Mascud had Yağmur and fifty lesser leaders killed – ostensibly for refusing permanent military service (1033). This caused a series of retaliatory raids by their sons and Kızıl who were supported by the Türkmen from around Balkhan Mountain, Khwarazm and Transoxania. These raids coincided with the Seljuqs’ falling out with the Karakhanids and the assassination of their mentor the Khwarazmshah Harun by no other than Sultan Mascud. Presumably this is why within months of crossing into Khurasan with a handful of mounted archers the Seljuqs are said to have been able to field thousands of Türkmen, though these did not include Arslan Beg’s affiliates (1035). While the Seljuqs can be argued to have provided leadership, the situation in Khurasan seems to have been dire under the Ghaznavids. Local notables are said to have repeatedly asked for help from the Karakhanids (Barthold 1990: 313). A report sent to Sultan Mascud cited Governor al-Fazl Suri’s cruelty as the reason for support of the Seljuqs. After the Seljuqs’ initial victory, the Ghaznavid commander for Buzgan defected with his men (1038). Other Turkic mamluk commanders such as Böri Tekin Yusuf, Ali Karib, Gazi and Eryaruk apparently followed (1039) (Köymen 1979: 260–78). Caliph al-Qa’im complaints about Türkmen depredations most probably concerned Arslan Beg’s affiliates (1043/4). However, Tuğrul Beg’s reply that Seljuq realms had grown too small can be taken to confirm that the Seljuqs’ success intensified the Türkmen migration into Khurasan. After Dandankan the Seljuqs are thought to have directed a large group to Azarbayjan via Tabaristan (1047). Soon after an Oguz group is likewise said to have been advised by Ibrahim Yınal that Khurasan could not support their number (1048) (Kafesoğlu IA/10: 393–6). Khurasan is almost completely without water and vegetation, even in comparison to the Gobi and Kara-Kum, whereas pasturage in both Azarbayjan and Anatolia is similar to Central Asia (Kafesoğlu IA/10: 385ff; also Cahen 1984: 77–9). Nor must it be forgotten that this westward route appears to have Evaluation 153 been familiar. Turkic gazi bands are thought to have attempted passage into Azarbayjan and eastern Anatolia in the middle of the previous century (Cahen 1984: 26). In summary, it would seem that Mongol expansion into Inner Asia and from there into Central Asia caused a westward displacement of Turkic affiliations from the middle of the tenth century. However, this did not become severe until two centuries later. The increasing number of Türkmen groups in Transoxania and Khwarazm, therefore, may reflect the spread of Islam as much as demographic pressure on the Eurasian steppes. What appears certain is that the Ghaznavids’ rehabilitation of Arslan Beg’s affiliates in Khurasan precipitated a further influx of Türkmen. The Seljuqs merely took up their leadership, albeit successfully. Having conquered Khurasan, the Seljuqs then directed groups attracted by their success westward into Azarbayjan and Anatolia. This served to relieve pressure on the indigenous population of Iraq while severing the overland route between the Byzantines and the Fatimids.

The ‘Türkmen problem’ In view of the rebellions led by Tuğrul Beg’s stepbrothers and cousins, clearly all was not well between the Seljuqs or between Tuğrul Beg and the Türkmen. On first entering Nishapur, Tuğrul Beg is said to have asked the jurist Sacid not to withhold his advice as the Seljuqs were strangers to Iranian ways (Sümer 1992: 82; Köymen 1979: 277–8; Kafesoğlu IA/10: 386–7; Turan 1993a: 62; Divitçioğlu 1994: 80). On joining his brother, however, Çağrı Beg is said to have proposed the city’s plunder. Having prior to their capitulation reassured the inhabitants of Nishapur that the city would not be sacked, allegedly Tuğrul Beg refused. Çağrı Beg insisted that unless permitted the Türkmen would defect, whereupon Tuğrul Beg is said to have threatened suicide. The brothers appear to have compro- mised; the populace is said to have been taxed and the proceeds distributed to the Türkmen (Sümer 1992: 91–2; also Köymen 1979: 271). With the exception of Köymen, Turkish historians have taken this as proof that the Seljuqs allied with the indigenous aristocracies as Köprülü postulated, settling the Türkmen on the marches from where they could be recalled when required. In this section the problem will be looked at from the point of view of the Türkmen. When taken in conjunction with analyses in later sections of the roles played out by the Seljuqs’ Iranian viziers and their Turkic military, this approach will show that rather than adopting the ways of the Iranians, the Seljuqs adapted their Turkic heritage to their subjects’ sedentary expectations. Divitçioğlu is the only historian to have attempted an analysis of Türkmen expectations (1994: 85–109). Having argued, as other Turkish historians, that Ibrahim Yınal and Kutalmış rebelled because Tuğrul Beg did not allow them a share of the Seljuq domains, Divitçioğlu went on to propose that Türkmen pillaging arose from the nomadic pastoral inability to produce all the goods necessary for survival. This is very much in keeping with modern studies on the nature of war and trade between China and its northern neighbours. These maintain that nomadic 154 Evaluation pastoralism has never been autarkic, but they add that when trade was possible in markets established along the Great Wall raiding was sporadic. When northern China came under the sway of a strong dynasty however, war or better still the threat of war ensured that China traded more rather than less silk for the number of horses offered as a tribute to satisfy protocol (Jagchid and Symons 1989; Barfield 1993; Khazanov 1994; also Divitçioğlu 1987: 252–66). Silk not only passed for money on the Inner Asian steppes (Bekin 1981), but had ceremonial significance as well (Allsen 1997). In Central Asia the situation appears to have been reversed. The Oguz are said to have accepted trade only when the Samanids were strong enough to resist their raids (Barthold 1990: 275). Possibly this is because in comparison to Inner Asia, urban centres in Transoxania were much closer. Moreover, given the proximity of summer pasturage in the T’ien-shan, Pamir and Hindu Kush, migration cycles in Transoxania were also that much shorter. Mounted archers did not have to travel far to access the self-evident riches of Tashkent, Fergana, Bokhara and Samarkand. When Ibrahim Yınal sought the Türkmen’s backing on rebelling a second time, their first request was that they should not be forced to campaign in Iraq al-Arab (Sümer 1992: 96). Quite apart from the fact that the shortage of grazing would have been a severe cause of hardship for their horses, the heat would have also affected their pack animals, the Bactrian camel. By all accounts the Türkmen were not paid soldiery. If they had been Ibrahim Yınal would not have requested land from Sultan Tuğrul to support his affiliates (Köymen 1976: 61). Prevented from pillaging, they would have had to replace any losses from their herds. When in preparation for a campaign against al-Basasiri in northern Syria Sultan Tuğrul ordered his affiliates to bring their families to the environs of Baghdad, the Türkmen are said to have refused because the countryside could not provide for them. Instead they asked to go home so they could rest their horses and replenish supplies (June 1056). Although the sultan had a group caught and beaten, he is said to have rescinded once the Türkmen gave their word they would return in the autumn (Köymen 1976: 104–5). This they must have done, because when Mosul fell to the sultan’s forces (May 1057), which appear to have included mamluk soldiery, the Türkmen wanted to sack the city. Apparently the sultan at first resisted, as he had done in Nishapur nineteen years earlier, but then he had the city emptied and the inhabitants compensated at his expense (Köymen 1979: 271; also Sümer 1992: 87, 91–2). According to Sümer, the Türkmen’s second request, that Ibrahim Yınal was not to reconcile with the sultan, arose from the Seljuqs’ growing preference for mamluk soldiery (1992: 96). His argument was that although initially distrustful of the Ghaznavid mamluks’ loyalty, the Seljuqs had not only rewarded them after Dandankan but also increased their number at the Türkmen’s expense. Toward the end of Sultan Tuğrul’s reign Turkic commanders such as Erdem, Gawhar-A’in, Humar-Tekin and Ay-Tekin Süleymani were all mamluks (also Köymen 1967: 7–8). Gawhar-A’in is known to have been inherited from the Buyids (Köymen 1967: 9–10). Sümer thought the Türkmen’s last request, that the Seljuqs’ viziers Evaluation 155 should not be appointed without their consent must be related to this preference since the viziers would have considered mamluk commanders and their soldiery more obedient and better sedentarized. To take Sümer’s last point first; there were other reasons why the Türkmen may have been angry with the Seljuqs’ Iranian administrators. Although the so-called Türkmen of Iraq appear to have offered only their martial services in return for being allowed into Khurasan, they had been taxed by the Ghaznavids, losing not only their livestock when unable to pay but also their children. This may go some way towards explaining the Türkmen’s request for administrators who recognized them as their conquerors rather than as their subjects. As for their second request; this seems related to Ibrahim Yınal’s first rebellion (1050–1), not to the Seljuqs’ growing preference for mamluk soldiery at the expense of the Türkmen begs and their affiliates, as Sümer argued. The immediate cause may be traced to when Tuğrul Beg appropriated Rayy as his capital (1042). This is when Ibrahim Yınal is thought to have first made clear his disappointment at not being allowed to keep what he had conquered (Köymen 1963: 58–9); a situation that must have become particularly galling in view of Kavurt’s position in Kirman and Alp-Arslan’s in Khurasan. After all, Ibrahim Yınal and for that matter Kutalmış were of Çağrı and Tuğrul Beg’s generation. Given that a ruling Turkic dynasty shared sovereignty (Ibid.; also Kafesoğlu IA/10: 353ff; Turan 1993a: 72ff and Divitçioğlu 1987: 192–5), Köymen’s argument was that when Tuğrul Beg later also asked for Hamadan and lesser fortifications toward Rayy, Ibrahim Yınal refused and having left Tuğrul Beg gathered ‘his forces’ (Ibid.). These ‘forces’, however, seem to have been the Türkmen situated between Hamadan and Rayy. On being defeated, Ibrahim Yınal is said to have asked Tuğrul Beg to forgive him, which he seems to have done. It can be no surprise, therefore, when Ibrahim Yınal asked the Türkmen for support a second time, that they stipulated he was not to reconcile with Sultan Tuğrul. Ibrahim Yınal did not and was strangled with his own bowstring. Sümer seems to have overlooked yet another point, which was brought out by Köymen (1963: 158–67); namely that Ibrahim Yınal’s rebellions in particular appear to have been crushed with the Türkmen who remained affiliated to Çağrı and Tuğrul Beg. Consequently, regardless of any preference the Seljuqs’ leadership may have developed for mamluk soldiery, even as late as 1060 the Türkmen seem to have formed by far the larger part of Seljuqid forces. Although possibly of an earlier date, there is also a story concerning Çağrı Beg. Apparently when Çağrı Beg chided his brother for the disrepair of his realms, Tuğrul Beg responded that he (Çağrı Beg) had devastated a built-up region like Khurasan in order to subdue it and now that it was his he was obliged to rebuild it. Tuğrul Beg is said to have gone on that not only had Iraq al-Ajam been devastated prior to his arrival but also it remained surrounded by enemies, which necessitated Türkmen forces to be situated at important crossroads (most likely between Hamadan and Rayy), making further depredations in the region unavoidable (Sümer 1992: 87). As noted, on the Orkhun stone inscriptions erected in his name, Bilge Kagan states that he clothed and enriched the poor, and that in return he expected their 156 Evaluation unerring loyalty (Tekin 1988: 43, 45). Some Turkish historians have elevated this to an ideological tenet of Turkic statecraft (İnalcık 1993a: 10–15; Turan 1993a: 72). Kafesoğlu on the other hand saw this in reciprocal terms, as a framework of obligations (IA/10: 389). Put another way, given their military contribution and support, some of the Türkmen must have felt betrayed by the Seljuqs’ concern, if not preference, for their non-Turkic vassals’ and subjects’ welfare; regardless of whether or not like them they were Muslims. Although Kafesoğlu can be said to have come close, neither view does justice to the patrimonial character of nomadic pastoral polities. Rather than bureaucrats, the heads of noble lineages had households (Allsen 1997: 53); immediate family members that included in-laws, such as Tonyukuk in the Kök Türk case, chamberlains and guards. When the head of the lineage became kagan, the chamberlains and guards became the central government and élite corps (Ibid.). Members of family were given an appenage, such as that of the Seljuqid Kavurt in Kirman or of Alp-Arslan in Khurasan, which is what Ibrahim Yınal wanted. Nobles were obliged to their households; they fulfilled this by making available a regular supply of clothing, food and drink (Ibid.). As in the case of the Mongol Code of 1640 which in 1677 was amended to provide for paupers (Sneath 2007: 184), clearly the concept could include the common folk affiliated to a noble house. Rather than the anthropological concept of reciprocity between members of a tribe, it is this patrimonial character of nomadic pastoralism that explains the Seljuqs’ expectations. This is not to deny that the Türkmen most likely regarded the domains conquered as theirs to plunder rather than to rule. The Oguz called settled people yatuk or lay-about (Atalay 1992/III: 14). Merchants fared no better; sart (Atalay 1992/I: 342, III: 13), which was derived from sarı-it or yellow-cur, came to be applied to all sedentary people in Central Asia, not only merchants (Barthold IA/10: 236–7).

The conquest of Anatolia There can be no doubt that the Seljuqs diverted those Türkmen unwilling to keep public order in accordance with the customs and laws of Islamdom westward into Anatolia, which was suitably Christian. However, the state of Byzantine defences in Anatolia cannot be ignored in any analysis leading up to Malazgirt (1071). After all, before the arrival of the Türkmen, the Byzantines had successfully reversed attempts by both the Umayyad and Abbasid caliphates.

Incursion The Chinese graded nomadic pastoral incursions as follows: looting border areas, robbing border areas, border encroachment, invasion, a great invasion, deep invasion and large-scale invasion (Jagchid and Symons 1989: 24). Like their kinfolk in Inner Asia, at first the Türkmen appear to have merely looted and robbed border areas from the Mukan steppe (Appendix 1, Map A1.4), Evaluation 157 possibly with the help of the Shaddadids through whose domains they most likely returned (1042–5). Although this may have alerted them to the parlous state of Byzantine arms, Köymen’s argument was that at the time even Azarbayjan was not safe. Accordingly, Tuğrul Beg assigned Musa’s son Hasan and Çağrı Beg’s son Yakuti to secure the region for the Türkmen. When Hasan (possibly with Yakuti) ventured across the border into Anatolia, however, he lost his life in a battle with Katakalon and Aaron (1047), the generals for Ani and Van respectively, which although Armenian were Byzantine protectorates at the time. Tuğrul Beg seems to have assigned Ibrahim Yınal to avenge Hasan’s death, which Ibrahim Yınal did by invading Anatolia. Whether or not the Türkmen continued to raid the Byzantine Empire after Tuğrul Beg made peace, however, is not clear (1050). What is known is that Tuğrul Beg was otherwise occupied in Iraq al-Ajam with the Buyids and Ibrahim Yınal’s rebellion. Four years later Tuğrul Beg took command of the Türkmen at the head of his own forces (Appendix 1, Map A1.4). Although various dynasties in eastern Anatolia proclaimed their allegiance and raids were conducted as far north as the Caucasus, Tuğrul Beg was unable to take the fortress of Malazgirt despite Armenian help. Clearly the Seljuqs already regarded this redoubt as central to their operations. Tuğrul Beg appears to have left behind one of his commanders, Sabuk Beg, with 3,000 men. Yakuti and Sabuk Beg are said to have won every engagement against Briennios (1057). The following year Ani and Kars were besieged and Malatya was sacked. On Sultan Tuğrul’s orders, two years later the Türkmen once again invaded Anatolia. This time Sivas was sacked and Urfa besieged by Salar-i Khurasan, possibly a mamluk commander. After another Byzantine army was defeated (1061), Sultan Tuğrul journeyed to Azarbayjan to review Yakuti’s conduct of operations (1062). The first problem for the Seljuqs seems to have been securing the Türkmen’s northern flank and rear to ensure their safe return, their main pastures being in Azarbayjan. Another consideration may have been control of the trade routes through the Caucasus and Arran. It is not surprising, therefore, that during his first campaign (1063), having cleared the Caucasus as far north as Tbilisi, Sultan Alp-Arslan’s forces not only took numerous fortifications and walled towns situated in north-eastern Anatolia, including the walled city of Ani, but also torched them (Appendix 1, Map A1.5). Soon afterwards the Seljuqs seem to have taken the fortress of Malazgirt and established a base in Ahlat, under the command of Gümüş-Tekin a mamluk commander (1066). From here, until they fell out, Gümüş-Tekin and Afşin Beg are known to have taken the fortifications between the Euphrates and Tigris rivers, which impeded safe return to Ahlat from sorties into central and south-eastern Anatolia. It was only when Afşin Beg fell out with Gümüş-Tekin and killed him that Türkmen raids are reported also from Aleppo, which is where Afşin Beg took refuge after he had sacked Kayseri and raided as far inland as Konya (1067) (Appendix 1, Map A1.5). When Romanos Diogenes became emperor, therefore, Anatolia was under attack not only from the north-east and east but also the south-east. More importantly, however, the looting of border areas from Azarbayjan had turned into a 158 Evaluation large-scale invasion from eastern Anatolia. The emperor’s first two campaigns (1068–9) clearly reflect this fact (Appendix 1, Map A1.6). However, in order to appreciate the remarkable consequences of Sultan Alp-Arslan’s resounding victory at Malazgirt, the Byzantine Empire’s provincial administration also needs to be examined. As noted, this was not the first time the Byzantines had to defend Anatolia from raiding, quite the opposite. The Arabs had been doing so regularly since the establishment of the Umayyad caliphate and had gained a foothold in eastern Anatolia that had survived well into Abbasid times.

Defence Faced with a sharp fall in tax revenue after losing more than half of its domains at the beginning of the seventh century, it seems that during the reign of Emperor Constans II (641–68) the Byzantine Empire was organized into themes or provinces under military governance. It is thought that by granting most of the remaining imperial estates on a hereditary basis Constans II made Byzantine soldiery self- supporting, selling them arms from state-owned depots (c. 659) (Treadgold 1995: 21–5). Initially there were seven such provinces in Anatolia and possibly another three in the Balkans, which had been mostly overrun by the Slavs. Subsequently the number of themes rose, but it is doubted that the rise in the number of soldiery was proportionate. This increase appears to have been caused as much by the subsequent expansion of the Byzantine Empire as by the breaking-up of rebellious themes (Treadgold 1995: 25–8; Oman 1991: 179–84; also Dupuy 1984: 52ff). In short, the frontier themes defended the Byzantine Empire’s borders while the soldiery in the inner themes maintained order, both being supported by the tagmata or heavy cavalry troops formed by Constantine V (741–5), which were independent of the provincial garrisons. On their south-eastern border, the Byzantines attempted to keep the passes in the Toros and Antitoros mountains under regular observation to gain advance warning of Muslim raids launched from Tarsus and Adana, and later more frequently from north Syria. It is thought that on being alerted the nearest battalion tried to track the raiders while the main cavalry corps gathered. At the same time infantry were sent to the passes by which the Muslims were expected to return; the Cilician Gates and those of Adata and Melitene. Once gathered, the Byzantine cavalry either attacked or if the raiding party was too large defended until reinforcements arrived from neighbouring provinces. If the raiders retreated before they could be engaged, the Byzantines tried to attack their camp at night with infantry and/or dismounted cavalry. The best results seem to have come when infantry were deployed at the right passes at the right time. A case in point was Nicephorus Phocas’ brother Leo’s victory over Sayf al-Dawla, the Hamdanid Amir of Aleppo (963). In this instance Sayf al-Dawla tried to return through the same mountain pass by which he had entered Anatolia, losing all his booty and many men in the process (Oman 1991: 209–17; also Treadgold 1995: 30). Evaluation 159 After this, however, Muslim raids seem to have fallen off sharply. Possibly this is why towards the end of Basil II’s reign (976–1025) and even more so at the start of Constantine IX Monomachus’ (1042–55) when civil administrators replaced the governors general whose rebellions had to be suppressed. Monomachus also debased the coinage, possibly to reduce army pay, and relieved the Armeno-Cappadocian provinces of their obligation of universal military service, tax- ing them instead (Treadgold 1995: 40; also Oman 1991: 180). While many soldiers appear to have been reduced to plundering such that warlords arose in castles with subject villages whose inhabitants served as infantry (Oman 1991: 180), it is the latter which above all else appears to have proven catastrophic. Without their so-called Iberian Army, said to have totalled 50,000 men from the themes stretching north-east from Urfa and Kayseri into the Caucasus, the Byzantines were no longer able to withstand the Türkmen incursions (Oman 1991: 214–19). Nor must it be forgotten that during their incorporation into the Byzantine Empire the Armenians had suffered persecution and relocation, which is thought to have made them less than loyal. In short, Malazgirt appears to have been a ‘do or die’ battle for the Byzantine Empire in Anatolia.

Malazgirt Turkish scholars estimated that the force Romanos Diogenes gathered on the banks of the Sakarya River before Malazgirt was almost 100,000-strong, including 30,000 auxiliaries and 3,000 supply wagons. According to the Tactica, a military manual compiled during the reign of Leo VI (886–912) (Dupuy 1984: 53), there should have been two carts and a pack horse for every sixteen foot soldiers (decury) in a battalion or regiment (banda or numerus) consisting of 480 men (Oman 1991: 191). Consequently, 3,000 supply wagons suggest 24,000 infantry. Each battalion could also expect support from a surgeon with six to eight stretcher-bearers (Ibid.), or 3,450 medical staff. In addition there would have been the engineering units attached to each thema or corps (Ibid.), plus a siege machine that required 1,200 men to operate. A corps consisted of two to three turma or divisions, which in turn were made up of five to eight regiments or battalions (Dupuy 1984: 54–5). In short, 30,000 infantry and auxiliaries seem consistent with 3,000 supply wagons. Again according to the Tactica, it is thought that the number of infantry in a Byzantine formation varied between eleven per cent and fifty-four per cent (Oman 1991: 196; also Dupuy 1984: 57–9). To take the latter percentage first, it is unlikely the emperor would have embarked on a campaign against an enemy wholly consisting of cavalry with only 25,000 of his own. A cavalry force of 300,000, let alone one of half a million as some Muslim sources report (Sümer and Sevim 1971) seems equally unlikely, despite the addition of mercenaries and soldiery from metropolitan and Balkan provinces (Oman 1991: 219). There is, however, the report that Romanos Diogenes brought over the five Balkan tagmata, all told 12,000 men. These are thought to have been in addition to the 12,000 in 160 Evaluation Anatolia and the Varangian Guard of 6,000 mercenaries, which makes a total of up to 30,000 heavy cavalry. Separately there are the 5,000 cavalry known to have been garrisoned in Tarsus and as roughly a quarter of each provincial corps elsewhere seems to have been cavalry, another 1,000 in Antakya. In addition there is the cavalry of the disbanded Iberian Army, which Romanos Diogenes tried to drill back into a fighting force. Of these 12,000 are thought to have been of the Christianized Banu Habib, who deserted to the Byzantines in 934 and are assumed to have kept up their numbers. Since the Iberian Army seems to have totalled 50,000 men, this suggests that there were another 9,500 cavalry besides the Banu Habib (Treadgold 1995: 34, 79–80, 83–5, 116–17, 216). This would make a possible 57,500 cavalry without the soldiery from Cappadocia, which is known to have been on the field. Nor does this include the Uz and the Peçenek. Consequently, 55,000–60,000 Byzantine cavalry with 15,000 auxiliaries seems a reasonable assumption (Oman 1991: 191, 219). The force dispatched to help the Bagratid dynasty may have been mostly infantry in order to strengthen their defences against the Seljuqs. After Sultan Alp-Arslan’s campaign in Georgia (1066), the Türkmen appear to have raided along the rivers and streams stretching from Tbilisi into Anatolia south of the Pontic Mountains (1070) (Appendix 1 Map A1.7). Likewise, infantry would have held the fortress of Malazgirt. There is also the infantry under Trachaniotes, who with Roussel was dispatched to take Ahlat (Oman 1991: 194). Although Roussel’s Normans most likely were heavy cavalry, it seems, therefore, that Romanos Diogenes had almost all his cavalry with him on the battlefield. Consequently, even if as Turkish scholars estimated the Byzantines were 100,000-strong when embarking on their last campaign, it is unlikely they had more than half that number by the time they had reached Malazgirt. As for the Seljuqs, it is doubtful they numbered 40,000–55,000 men, let alone 100,000 or more as some European military historians have suggested (Oman 1991: 219). In all probability even the twelve core tribes of the Kök Türk could not field more than 60,000 men. It can only have been with the addition of the Tokuz Oguz and other affiliated tribes that the Kök Türk Empire was able to put 400,000 men in the field (Divitçioğlu 1987: 210). The Seljuqs and the Türkmen made up but a fraction of the lineages once affiliated to the Kök Türk. To be sure, the sultan’s reported route from Aleppo suggests he may have recruited additional soldiery before being joined by Sabuk Beg and the Türkmen in Ahlat (Appendix 1, Map A1.8), thus the assumption by Turkish scholars of Kurdish cavalry at Malazgirt, but if Sultan Alp-Arslan had outnumbered Diogenes it is doubtful the emperor would have risked a purely cavalry battle, let alone pursued the Seljuqs on the field. This seems to be confirmed by Sultan Alp-Arslan’s reported attempt at reconciliation before battle. As for the battle itself, despite reports concerning the Uz and Peçenek defec- tions, Trachaniotes and Roussel’s refusal to rejoin as ordered, or indeed Andronicos Ducas’ dislike for the emperor, it appears Emperor Diogenes disregarded all that the Byzantines had learnt of steppe warfare during previous centuries (Kafesoğlu IA/7: 242ff; see also Oman 1991: 219 and Dupuy 1984: 61). This knowledge Evaluation 161 was readily available in military manuals (Dirimtekin 1993: 230ff). Besides the Tactica, there is the earlier Strategicon, which is thought to have been compiled during the reign of Emperor Maurice (582–602) (Dupuy 1984: 53), and the work by Nicephorus Phocas (c. 980) (Oman 1991: 215). According to the Tactica, which is accepted as having been based on the Strategicon, when combating a Turkic foe, generals were advised to protect their rear (with a natural obstacle if possible), take care of their flanks, never pursue blindly (or become separated), but engage at close quarters as soon as possible (Oman 1991: 206). More importantly, the Tactica also recommended a front line of heavy infantry, supported by foot archers whose bows were bigger than those the cavalry had and hence shot further (Ibid.). The Turkic composite bow, on the other hand, included a layer of split horn on the stave toward the archer. This gave it greater penetrative power and a range over 300 m. Although at that distance the arrows were ineffective against chain mail, at least until the Ottomans developed a superior bow in the fifteenth century (see Dupuy 1984: 4–5), the Byzantine heavy cavalry could not match their firepower. If foot archers were able to shoot their mounts from underneath them, however, Turkic horsemen became vulnerable to both bowshot and lance as they wore little if any armour. Nor must it be forgotten that cavalry cannot close if the infantry stand steadfast, as horses will shy away (also MacDowall 1995: 26–7). The problem was that mounted archers continually harassed with interchange- able groups, until death, fatigue and mounting frustration forced gaps in the enemy line. The gaps were then charged in wedge-shaped formations known since the Scythians (MacDowall 1995: 23; also Maenchen-Helfen 1973: 201ff). However, this was difficult to effect against heavy infantry supported by cavalry, since the latter could parry flanking manoeuvres. Moreover, well drilled foot soldiers were more likely to keep to prescribed distances in the heat of battle, something com- pressed cavalry formations could not do, as horses tend to panic when close to each other (MacDowall 1995: 24). Consequently, the feigned retreat, which was an integral part of steppe warfare, invariably lured wholly cavalry formations into giving chase, particularly as their aristocratic members remained eager to inflict shock combat. Mounted archers gave the appearance of fleeing in disparate groups rather than executing an orderly retreat. As both riders and horses had different temperaments and battle experience, pursuing groups tended to disjoin in such circumstances (MacDowall 1995: 24–5), breaking up the chain of command. Having separated and tired their heavily armoured foe by feigning flight (MacDowall 1995: 21), mounted archers turned, encircled and began to shoot in an arcade. This could demoralize the most battle-hardened troops thus forced to stand their ground. Usually, therefore, when not able to fall back on their heavy infantry, the more inexperienced riders panicked such that those better disciplined were captured or annihilated by ever larger groups that gathered for the kill with sword, pike and mace. There is no doubt that armour impeded mobility (Jankovich 1971: 29), quickly tiring both horse and rider, but mounting a horse did not give the sedentary rider parity with those off the steppes. Hsiung-nu children rode calves and foals, 162 Evaluation the smallest riding sheep and goats while shooting arrows at field mice, small birds and poppies (Kafesoğlu 1994: 209). Nearly two centuries after Malazgirt, according to the oft-cited report by Johannes Plano de Carpini who was on an abortive ambassadorial mission to Mongolia (1245–7), two–three-year-old children could not only sit in the saddle but also gallop a horse. Nor could sedentary horses match the mobility of the hardy steppe mounts (Köymen 1967: 56–7, 59). Although smaller, these were more manoeuvrable because of their high neck carriage, their long, thin limbs, and hard hooves (Kafesoğlu 1994: 206; see also Maenchen-Helfen 1973: 203–21), which gave them speed and resilience. Taken in conjunction with the large hunting drives conducted each autumn (kurt oyunu, or the wolf’s game) that served as war games (Kafesoğlu 1994: 274; also Ögel 1993), it is clear that no matter how well drilled, sedentary cavalry remained hard put to combat superior Turkic horsemanship and archery. Consequently, it is more than likely Romanos Diogenes would have survived to fight another day if he had retained his infantry units rather than relied on the numerical superiority of his cavalry. Arguably he attempted to close because his heavy cavalry was no match for the Seljuqs. Once gaps opened on his flanks and units were separated, they had no steadfast infantry to fall back on. In the confusion he seems to have been unable to effect an orderly retreat, losing touch with his reserve. Forced to face about and stand his ground, possibly he dismounted and turned his heavy cavalry into infantry as prescribed in the Strategicon (MacDowall 1995: 24). Even if so, his heavily armoured men would have been tired by then if not also demoralized and the Seljuqs’ armoured mamluks comparatively fresh, which would have made all the difference in close combat. Whatever the case, encircled, gradually most of the remaining Byzantine heavy cavalry seems to have been annihilated with the exception of the reserve, thought to have consisted mainly of mercenaries and Armenian nobility (Oman 1991: 219–21; see also Dupuy 1984: 61).

The Seljuqs of Anatolia Sultan Malik-Shah clearly continued from where his great-uncle and father left off by repeated campaigns into the Caucasus (1075–6, 1078–9, 1084 and 1086). Nevertheless, he is considered to have been at best ambivalent if not outright hostile towards the Türkmen. Thus his latter two campaigns are interpreted as attempts to control Türkmen migration into Anatolia (Turan 1993c: 77; also Cahen 1984: 93–4). Although Sultan Alp-Arslan seems to have taken responsibility for the Caucasus and Anatolia – possibly due to the death of Yakuti who is not mentioned in the apportionment after his accession – with the unification of the Great Seljuq Empire under one ruler, the Türkmen could no longer have been led ‘from the saddle’ as it were. Arguably this is why soon afterwards the sultan appointed a mamluk commander, Gümüş-Tekin to supervize Türkmen raids into Anatolia, which came to be launched from Ahlat (1066). The Türkmen, however, appear to have expected a Seljuqid. This possibly explains Gümüş-Tekin’s death at the Evaluation 163 hands of the Türkmen beg Bekçioğlu Afşin, who was quite happy to report to Sultan Alp-Arslan once pardoned. So soon after Malazgirt, Sultan Alp-Arslan’s death arguably created a power vacuum that was filled by Kutalmış’ sons while Sultan Malik-Shah dealt with Kavurt, then the Karakhanids and Ghaznavids. This is why Sultan Malik-Shah’s approach may appear ambivalent. To be sure, there are the various reported attempts at reconciliation between the Seljuqs and the Byzantines, whether through the good office of the caliph or directly. On the other hand, there were also the repeated campaigns conducted by mamluk commanders. These invariably ended with inconclusive sieges of Iznik, which suggests that, like his great-uncle and father, Sultan Malik-Shah considered the Türkmen in Anatolia in his dominion – even if Kutalmış’ sons did not. It is also unlikely that the caliph unilaterally proclaimed Süleyman-Shah sultan, as this would have been feasible only prior to Kavurt’s defeat. Sultan Malik- Shah did not move against the Karakhanids until the caliph confirmed his status (1073). Nor is Süleyman-Shah’s elder brother Mansur thought to have died until much later (1078). As for the reports that Süleyman-Shah was appointed sultan by Malik-Shah, why were neither his brothers nor his sons released upon his death? It was only upon Sultan Malik-Shah’s death that Kılıç Arslan emerged in Anatolia, very much as his father and uncles did when Sultan Alp-Arslan died. Moreover, despite his assurances to the contrary, when the Seljuqs’ Uqailid vassal Muslim asked for the annual tax Philaretos Brachamios had promised for Antakya, if, as Süleyman-Shah claimed, he was also Malik-Shah’s vassal, arguably he would not have been asked to pay or for that matter attacked the Uqailids, let alone appointed a Fatimid jurist (1085–6). Therefore, contrary to Köymen (1963: 102–7) and Kafesoğlu (1953: 75 and IA/10: 391), as Turan argued (IA/11: 210– 12), most likely Süleyman-Shah appeared subservient to Sultan Malik-Shah in order to rid himself of his brothers, particularly the older Mansur, and thus gain dominion in Anatolia. Once confident of his gains, Süleyman-Shah appears to have unilaterally declared himself sultan. As for the argument that Süleyman-Shah behaved like a Byzantine rather than a Seljuqid vassal and hence the Türkmen in Anatolia would have become incorporated into the Byzantine Empire (Cahen 1984: 88–92), this must be considered doubtful to say the least. Emperor Alexius’ peace agreement with Süleyman-Shah strongly suggests that in the wake of their defeat at Malazgirt, the Byzantine Empire had no central authority left in Anatolia (1082). No matter how temporary, Süleyman- Shah’s customs’ bureaux on the Bosporus certainly appear to confirm this. Consequently, on Süleyman-Shah’s death (1086), although he did not journey to Anatolia, Sultan Malik-Shah appears to have made sure that neither al-Kasim nor another beg gained dominion over the Türkmen (Köymen 1963: 109). Under pressure from the Byzantines who seem to have used Porsuk and Bozan’s campaigns to regain territory, when at last al-Kasim did journey to Isfahan to proclaim allegiance, he was strangled with his own bowstring, with or without Sultan Malik-Shah’s consent. The sultan’s reported offer of some seaports in his request for Emperor Alexius’ daughter in marriage to his son, as well as the reputed proclamations of allegiance by the begs of Konya and Aksaray, further 164 Evaluation strengthen the argument that after Malazgirt the Seljuqs became established not only in the countryside but also in the towns of Anatolia – whether or not they had Byzantine help such as from Nicephorus Melissenos. Despite the self-evident enmity between the two main branches of the Seljuqs, namely Arslan’s and Mikail’s descendants, Sultan Malik-Shah’s continued concern with the Caucasus also suggests there was a steady migration through here into Anatolia. These campaigns were in the main against the Georgians, but it must not be forgotten that Sultan Alp-Arslan also returned here during Romanos Diogenes’ first and second Anatolian campaigns (1068/9). This was when the region is said to have first come under pressure from the Kimek migration. This is thought to have pushed the Alan, Komuk, Sabir and various remnants of the Khazars into the Caucasus. Certainly, the mountainous neck of land between the Black Sea and the Caspian appears to have been the favoured route for Turkic migrations into Anatolia. According to Togan the Bulgars from the Volga region were the first Turkic people to arrive in north eastern Anatolia (149–129 BCE). Next the Sabirs moved into Azarbayjan from the south Caucasus (305). According to Byzantine sources, during the fifth century some of the Huns also centred themselves on the Mukan steppe (445). Two decades afterwards the Ağaçeri, a ruling lineage of the Huns, arrived in Azarbayjan (465). In short, the argument is that Turkic people were situated in the Caucasus, Azarbayjan and north-eastern Anatolia from before the start of the Common Era (Togan 1981: 168–71). The Byzantines’ known resettlement of some Avars in Anatolia (577, 620) (Kalafat 1992: 29), as well as the Abbasids’ eastern Anatolian Sugur Army, which is thought to have consisted of Türkmen (760) (Kalafat 1992: 26), lend further support to this line of reasoning. Possibly the following villages in present-day Turkey were originally Avar settlements: Avadan (Tarsus and Eskişehir), Avaduri (Midyat), Avakent (Kulp, Diyarbakır), Avak (Trak, Hisn Kayfa), Avalama (Konya), Avan (Şirvan, Hisn Kayfa), Avana (Borçka), Avanoğlu (Kırşehir), Avanuşağı (Pazarcık, Kahramanmaraş), Avora (Niksar, Tokat), Avarek (Van), Avarık (Egin), Avas (Bakırköy, Istanbul), Avasorik (Erciş, Van), Avason (Manavgat) and Avasor (Muradiye, Van) (Ibid.). The Seljuqs’ conquest of Anatolia, therefore, may have been facilitated not only by the existence there of people of Turkic descent, but also by the familiarity of its Greek, Armenian, Slavic and other local populaces with immigrants per se. Yet another factor that must be considered in this context is the seeming hostility to Byzantine rule in western Anatolia, which appears to have remained mainly Greek. It is difficult to explain how else Çaka Beg and al-Kasim could have undertaken to build navies that were clearly intended to challenge those of the Byzantines in both the Aegean and Marmara seas. As much seems to be indicated also by the inhabitants of Izmit; although only 90 km from Istanbul, before they let in the Byzantine commander Eustathios they reportedly had to be persuaded that al-Kasim and Emperor Alexius Comnenus had agreed to jointly defend the city against Sultan Malik-Shah’s commander Porsuk. In short, immediately after Malazgirt the Seljuqs seem to have benefited further not only from contention among Byzantine generals for Emperor Michael VII Evaluation 165 Ducas’ throne, but also from those local populaces ill-disposed to being ruled from Istanbul, especially since Byzantine governance in Anatolia had clearly fallen into disarray even before most of what remained of its military was destroyed on the battlefield. Familiarity with previous Turkic re-settlements throughout Anatolia may have further simplified matters for the Seljuqs and the new immigrants, Oguz or otherwise.

The basis of empire As has been noted several times, Turkish historians agreed with Köprülü that the Seljuqs had adopted Irano-Islamic institutions. They thought this tendency most prevalent during Nizam al-Mulk’s vizierate (1063–92), particularly during the reign of Sultan Malik-Shah (1072–92). Having looked at Great Seljuq history from the point of view of the Türkmen, in this section the premise of Köprülü’s framework is analysed with regard to the Seljuqs’ expectations; their martial tradition, use of the iqtac and their view of legitimacy and succession. The role Nizam al-Mulk and his sons played in the Great Seljuq Empire is discussed in ‘The power struggle’ below, which covers the internecine warfare that ensued on Sultan Malik-Shah’s death. The Seljuqs were mounted archers; this is discernible not only in the way the Türkmen overran Anatolia under their leadership, but also in the martial and secular character of their governance. The reasons for this are to be found in the mounted archers’ predominance prior to the advent of gunpowder, which served to reinforce Seljuqid expectations. A weapon with superior range tends to dominate battle formations (Fuller 1998: 21–3). Since speed and mobility likewise govern offensive considerations (Ibid.), it is not surprising that pastoral cultures quite naturally combined archery with equestrian pursuits, raiding and wearing down their foe before delivering the finishing blow. By comparison, in sedentary cultures archery cloaked the deployment of other weapons (Ibid.), mainly because agriculturists remained well adapted to infantry action or shock combat. The problem with this was that foot soldiers could not attack cavalry (Fuller 1998: 68; also MacDowall 1995: 3–6). In order to defeat the growing armies of mounted archers, therefore, sedentary people were forced to rely more and more on cavalry (Dupuy 1984: 36ff). As equestrian pursuits were not part of their daily concerns, however, horses had to be bred specially (MacDowall 1995: 10–14). The development of armour, which few could afford, reinforced this tendency as it required special breeds of horses (Dupuy 1984: 38; also Oman 1991: 126–30). Besides the weight of its chain mail, a mount bore a rider with a steel casque, mail shirt, gauntlets, steel shoes, large iron stirrups, shield, long lance and broadsword. Byzantine heavy cavalry were equipped with dagger, axe, bow and quiver as well (Dupuy 1984: 56). Consequently, right from the start cavalry and/or archers in sedentary cultures were almost without exception of noble birth, when not hired. Although the Assyrians were first to establish a professional army, their cavalry were drawn from nobility (c. 700 BCE) (Dupuy 1984: 7). In ancient Greece, archers 166 Evaluation came from nobles who could not afford horses (Fuller 1998: 38). Alexander the Great’s Companions, his heavy cavalry of young Macedonian noblemen, were supported on the flanks by mercenary Thessalian cavalry and other similar units (356–23 BCE) (Dupuy 1984: 36–7). When the Romans were forced to switch emphasis from infantry to cavalry after defeats by the Parthians (363), the Goths (378) and later the Huns, their armies came to include allies (foederati) or mercenaries (bucellarii) (Oman 1991: 8–21, 176–7; also Dupuy 1984: 36–41). This appears to have given well-off nobles as well as Roman generals the opportunity to form personal armies despite legal sanctions (476) (MacDowall 1995: 6–7). The Byzantines were able to settle these, the bulk of their heavy cavalry becoming ‘small free landowners’ (Oman 1991: 189). The turning point for Muslim arms seems to have come during the Abbasid Civil War (809–37), when Caliph al-Amin’s (809–13) army was defeated outside Rayy by a much smaller number of cavalry sent from Khurasan by his brother al-Macmun. Despite his victory and subsequent elevation to caliph, while al-Macmun prevaricated whether to govern from Marv or Baghdad, his younger brother al-Muctasim seems to have started to build a similar, dedicated force (814–15). In the main made up of Turkic captives supplied by the Samanids, the army appears to have consisted of others such as the slave Ithak, a Khazar cook, and Afşin, a Transoxanian prince. Although many in the caliphal armies of this period hardly spoke Arabic, their children appear to have become integrated into Muslim society. In the latter half of the tenth century such warriors were replaced by ghilmans or mamluks as they came to be known. Recruited mostly by Turkic men of arms who remained responsible for their upkeep, these so-called ‘slave soldiers’ seem to have led celibate lives, going so far as to become eunuchs. However, even they were not all captives purchased on the slave markets. For instance, Anuş-Tekin al-Dizbari from Khurasan is thought to have become a mamluk to further his prospects, retaining his mentor the Dailamite Dizbar’s name as a mark of respect (Kennedy 1994: 148–60, 206–7).

Warrior élites The origins of the martial traditions governing the mamluks’ expectations may be traced back to peaceful attempts at resolving tribal disputes through the good offices of elders who did not have coercive powers or lead raids. Unlike elders who lost face when unable to resolve a quarrel within the common code however, warrior bands could seek out certain types of wrongdoers and compel redress or mete out punishment. Consequently, when confronted with new kinds of dispute, such men could also make law (Gluckman 1977). On the Eurasian steppes warriors are thought to have banded together in stock- ades called pi-yung by Chinese chroniclers of the first millennium BCE, although the word is not considered to be of Chinese origin. Such fortifications, which consisted of an outer and an inner wall wherein the warlord lived, came to be called ordu or palace; thus its application to the cities of Turkic rulers. The Chou Evaluation 167 dynasty (c. 1122–1256 BCE) is thought to have emerged from such a warrior band (Esin 1980b: 9–25), as did the Hsiung-nu Empire (209 BCE–CE 155). According to an oft-repeated story, the founder of the Hsiung-nu Empire, Modun, whose name is thought to redact to Batur (Lattimore 1967: 450), Turkic for hero or brave, trained his band of warriors by taking aim at one of his favourite horses first and shooting it. Although he had those who failed to follow suit killed, when next he shot one of his consorts some warriors still appeared to hesitate. They too were killed. As a result, when Modun aimed his whistling arrow at one of his father’s horses, his warriors shot it without second thoughts. At the next opportunity Modun is said to have aimed his arrow at his father. He was duly killed and Modun replaced him, becoming shan-yü. Modun appears to have been typical of his ilk. After defeating the Yüeh-chi he is said to have drunk their ruler’s blood from his lopped skull, which was lined with gold and encased in leather. Initially recorded among the Scythians, this practice is also known among the Oguz. An Oguz warrior who wore a drinking cup on his belt to signify his martial prowess was called an alp (Esin 1980b: 9–25). Also alpagu, alpagut, alpavut and yılpagu; it is thought to go back at least to the fifth century (Köprülü IA/1: 379–84). Today it still means hero, brave or stout-hearted, as it did in Mahmud of Kashgar’s time (Atalay 1992/IV: 21). Some warrior bands went so far as to demand to be buried alive with their leader on his death. The practice, known among descendants of the Kanglı and the Töles (Esin 1980b: 9–25), is thought to have also prevailed among the Kök Türk (Divitçioğlu 1987: 206), the Angles and Saxons (Divitçioğlu 1992: 113–29), and Germanic cultures such as the Visigoths (Oman 1991: 177). Whether or not of such an extreme form, fealty obliged band leaders to feed and clothe their warriors. Besides their patrimonial character, their very nature encouraged warrior bands to raid both their pastoral and their sedentary neighbours. More so in Central Asia perhaps; here captives could be sold to slave-traders who marketed them throughout Islamdom. Those suited to the martial arts became mamluks, as did many captive warriors. However, the latter are thought to have formed the nuclei of warrior élites, such as among Kök Türk, being regarded as a kul or servant rather than a slave (Divitçioğlu 1987: 191). Among the Huns in Europe, for instance, although ‘enslaved’, captive warriors could purchase their freedom with their share of booty (Maenchen-Helfen 1973: 199). The mamluk Gawhar-A’in’s behaviour is particularly noteworthy in this context. Having been transferred by the last Buyid ruler to his son Abu Nasr, Gawhar-A’in is said to have remained by Abu Nasr’s side until his death when imprisoned by Tuğrul Beg. Subsequently appointed commander of Baghdad, Gawhar-A’in was wounded attempting to prevent Sultan Alp-Arslan’s stabbing, which caused the sultan to die. Gawhar-A’in went on to serve Sultan Malik-Shah and Sultan Berk-Yaruk (Köymen 1967: 9–10). According to Nizam al-Mulk, it took twenty years to train a mamluk like Gawhar-A’in. This is why Köymen thought at least initially the Seljuqs inherited their mamluks. He based his argument on the increased mention of mamluk commanders toward the end of Sultan Tuğrul’s reign, names that remained at the fore 168 Evaluation in the first years of Sultan Alp-Arslan’s reign. Nevertheless, Köymen also pointed out that given the importance Nizam al-Mulk attached to the training of mamluks in his Siyasat-Nama it was doubtful their preparation was institutionalized even during Sultan Malik-Shah’s reign. He thought rather that ‘slave’ soldiers were trained and equipped by a Turkic ‘master’ who served in the palace before being assigned to campaigns or regional posts (Köymen 1967: 12–15 and 1966), as was the case under the Samanids. According to Özaydın, the Hajib-i Buzurg, the sultan’s chief of staff and most senior member after the vizier, was responsible for two sets of commanders who were known as hajib or ağacı. One group received correspondence from dignitaries throughout the realm and verbally relayed the sultan’s orders to his vizier, which suggests each was assigned to a region. The other group of commanders were chamberlains. The Jandar was responsible for the security of the palace. The Silahdar looked after the palace arsenal and the sultan’s weapons, which he carried for him. The Alam took care of the sultan’s banners. The Ahur tended to the sultan’s horses. The Tashtdar supplied the palace’s ewers and bowls. The Jamedar was responsible for royal and ceremonial robes. The Sharabdar procured the sultan’s medicines and stocked the palace cellar. The Chashnigir oversaw the sultan’s table and royal banquets. As for the remaining palace posts not filled by the military, in time these seem to have lost their status except for the sultan’s personal secretary (2001: 202–7). References to Amir-i Ahur or commander of horses, such as Inanç Yabgu in Hamadan, suggest that having proved their worth in the palace commanders were promoted as vacancies arose; from dealing with regional notables or chamberlains to policing regions, conducting campaigns and finally governorship, such as Artuk, Aksungur, Yağı-Sıyan, Porsuk and Bozan. The first three were Turkic begs and can be assumed to have resided on their iqtacs, as the Porsuks seem to have done in the environs of Hamadan. While on campaign, however, it is also likely that each would have benefited from the provisions stockpiled on crown lands for this purpose. Sultan Malik-Shah’s royal mamluk corps is said to have numbered 46,000 (Kafesoğlu 1953: 156–8). It is not clear, however, how many were auxiliaries or novices, or if they included foot soldiers capable of shock combat and siege warfare (Köymen 1967: 43–73). Steppe tactics did not preclude shock troops. The stone inscription erected for Tonyukuk, commander-in-chief and counsellor to the Kök Türk kagans, clearly refers to foot soldiers (Tekin 1988: 49). There is a reference to foot soldiers also on the stone inscription erected for Bilge Kagan (Tekin 1994: 3). According to Kafesoğlu, soldiery from the empire’s core regions augmented the royal mamluk corps billeted in Isfahan. He thought the Seljuqs were able to field 300,000 men, excluding soldiery from vassal states or the Türkmen (1953: 156–63). This figure is derived from Nizam al-Mulk’s Siyasat-Nama (Köymen 1999). Yınanç (1944: 174–6), on the other hand, and possibly therefore Turan (IA/5/2: 949ff) argued that the 300,000 soldiery mentioned in the Siyasat-Nama were in the main the Türkmen in Iraq. Turan thought that by adapting existing Evaluation 169 structures of land holding to Turkic expectations, Nizam al-Mulk had given the Türkmen means of support through agriculture and commerce rather than pillage (also Lewis 1993: 161). Some of Sultan Malik-Shah’s most famous commanders, Artuk, Çubuk and Yağı-Sıyan appear to be cases in point (Köymen 1972: 93–5; also Sümer 1992: 97). Aksungur is thought to have been of the Kıpçak Sabyu lineage. Since his father El-Turgan was named Abdallah, presumably the clan had converted to Islam upon entering the Seljuqs’ service. Said to have been trained with the mamluk soldiery on Sultan Malik-Shah’s orders (Sevim, 1990b: 72), Aksungur became governor general of Aleppo. Likewise Sultan Malik-Shah appointed Yağı-Sıyan governor general of Antakya. Rather than choosing to adopt mamluk soldiery because of their Iranian viziers’ preference for ‘slaves’, as the most illustrious alps of their time the Seljuqs appear to have brought to the fore the martial traditions that they and the mamluks off the Eurasian steppes were familiar with.

The iqtac Because Nizam al-Mulk worked for the Ghaznavid administration in Khurasan, Turkish scholars have concluded that the institutions he promoted under Sultan Alp-Arslan and Malik-Shah are traceable to the Samanids. However, the iqtac was first used by the Abbasids. It was originally known as qatica or land vacated by the vanquished, in particular by the Sasanians in Iraq al-Arab. From the reign of Caliph Umar (634–44) such lands appear to have been granted to Arabs on a hereditary basis. They were mostly individuals but sometimes a tribal group. Where the land continued to be managed by its indigenous owner, the grant allowed its Arab owner to pay the alms tax collected from Muslims (ushr) from the land tax collected from non-Muslims (kharaj), the balance being his remuneration as a citizen warrior (jund). Possibly due to non-Arabs in the military, in time the latter type of holding appears to have fused with the former and become private property. Under the Buyids, this latter type of land grant was known as iqtac al-istighlal. The muqtac or holder of the grant, in this case a mamluk commander, was permitted to collect taxes calculated to approximate his remuneration. The iqtac al-tamlik, on the other hand, denoted private ownership or more correctly the hereditary land grants (Turan IA/5/2: 950–2; also Kuçur DVIA/22: 47–8 and Kennedy 1994: 192–3), as in the original qatica. Under Mucizz al-Dawla (932– 67), having depleted their grants, some mamluk commanders returned them. Since those who retained theirs also complained that the taxes they collected did not approximate to their needs, Mucizz al-Dawla was forced to otherwise reward the former while extending the latter’s grants. Despite appointing relatives to oversee them however, Mucizz al-Dawla appears to have failed to prevent extortion by his commanders. In Turan’s opinion (IA/5/2: 951ff; also Kafesoğlu 1953: 165), commanders who did not reside on their grants during peacetime could not be expected to 170 Evaluation have had any interest in developing them. This is why, according to Turan, under the Seljuqs land grants to military personnel acquired a hereditary character, encouraging land management and policing while ensuring the availability of fully trained and equipped soldiers. If on the death of the commander his son was a minor, a regent was appointed until he came of age. In the absence of direct descendants, the grant was allocated to the commander’s eldest brother or failing that to one of his mamluks. However, the grant could not be gifted or sold to another, nor put into a pious foundation (waqf), so that it never became private property as in the case of the qatica. Moreover, according to Turan the local inhabitants remained free not only to appeal directly to the sultan, if they felt their rights were being usurped, but also to move elsewhere for whatever reason. Not surprisingly, many grants were reallocated to others during a holder’s lifetime, either because of mismanagement or due to reappointment. Although it is not clear which periods he was referring to, according to Turan under the Seljuqs large administrative iqtacs prevailed in Iraq al-Arab and smaller military iqtacs in Iraq al-Ajam where they were incorporated in hierarchical fashion into larger administrative grants of which some had to remit a percentage of the taxes collected to the central treasury (Turan IA/5/2: 951ff). This system does not appear to have been inherited from the Buyids under whom, according to Lambton (1980: 368), iqtacs where the holder had administrative functions normally associated with a governor remained the exception. Nor do the Buyids’ military iqtacs appear to have been granted for life, except that if the muqtac died his family received a pension, thus forestalling the grant from defaulting to private ownership (Turan IA/5/2: 951ff). In order to find previous examples of large administrative iqtacs it is necessary to look at land grants under the Abbasids. During the reign of Caliph al-Macmun (813–33), Azarbayjan seems to have been offered to several commanders with the intent of suppressing the prolonged Babak rebellion (816–37), which took place during the Abbasid Civil War (809–33). In return for an agreed remittance, the commanders were expected to govern the region by financing the requisite forces from tax revenue. Under Caliph al-Muctasim (833–42), the Turkic mamluk commander Afşin finally appears to have succeeded. While there he seems to have incorporated Armenia, too, as both provinces are known to have been governed by Turkic men of arms, first by Afşin’s commander Moyun Çor, then al-Bokhara and finally by the al-Saj, of whom Yusuf is said to have agreed once again to remit a nominal amount to Caliph al-Muqtadir (908–32) (Kennedy 1994: 156, 166, 184, 192–3; Togan IA/2: 100). Interestingly, after an attempted coup against Caliph al-Muqtadir (908–32), his vizier is said to have granted iqtacs to Turkic mamluk commanders as before, that is, the holder collected the kharaj but only paid the ushr in return for the provision of fully trained and equipped soldiery, though it is not clear if the commanders resided on their grants or in Samarra near Baghdad. Possibly as yet unaware of Köymen’s research, which was not published until a decade later, Turan took the granting of some of the caliph’s iqtacs to the Türkmen (1073) as his starting point for the use of land grants under the Seljuqs, claiming the system was fully established by the end of Sultan Malik-Shah’s reign (IA/5/2: Evaluation 171 952–3). According to Köymen (1976: 61), however, Ibrahim Yınal is known to have asked Sultan Tuğrul to grant him an iqtac in order to support his forces. This brings the use of such grants forward by nearly two decades. Turan’s exposition left many questions unanswered. For instance, there are the lands along probable campaign routes used to stockpile provisions and fodder for the royal corps, likewise called iqtacs (Kafesoğlu 1953: 156–9; also Lambton 1980: 372). However, these may have been hass or crown lands. Besides the 200-strong royal guard, which were paid in cash (Ibid.), trainees (kara gulam) and mamluk soldiery in the royal corps not yet of appropriate rank are also thought to have been salaried (Köymen 1967: 14–15, 24, 40–1). This suggests the royal corps were paid directly out of the Private Treasury, which included incomes from crown lands, tax from non-Muslim vassals and presumably plunder gained on campaign. Taxes from the provinces and possibly tribute from Muslim vassals went into the Public Treasury, any shortfall in the latter being supplemented ‘temporarily’ from the former (Kafesoğlu 1953: 145–6, 164, 166). Then there are the caliphs’ and the lesser Seljuq khatuns’ land grants; although the latter seems to have required the holder to train and equip a required number of mamluks (Kafesoğlu 1953: 164), as exemplified by Terken Khatun with her own diwan and mamluk soldiery. There were also Kavurt’s numerous daughters whose iqtacs were their dowries (Köymen 1972: 88). Moreover, it seems the system whereby the muqtac collected taxes calculated to approximate to his or her financial requirements, as under the Buyids’ iqtac al-istighlal, was used by the Great Seljuqs regardless of whether the taxes were due to the Private or the Public Treasury. Khwarazm was governed by Arslan-Argun, but one of Sultan Malik-Shah’s chamberlains, the Khwarazmshah Anuş-Tekin, is also said to have been referred to as ‘governor’. This error may have arisen because he was granted a percentage of the taxes levied in Khwarazm as remuneration for procuring the sultan’s ewers and bowls (see Kafesoğlu 1992: 36–7).

Seljuqid legitimacy Possibly taking their cue from the caliphs’ treatment of Seljuq sultans as their protectors, Turan, Kafesoğlu and Özaydın thought of the Seljuqs as saviours of Sunni Islamdom and their regard for the caliphs was reverential. Turan included the marriages between the two houses as evidence of the latter, while Kafesoğlu and Köymen treated this aspect in dynastic terms. However, the Seljuqs’ relationship with the Abbasids was concerned with power rather than reverence and service in the name of religion. Nor was their attitude unique. Before the Seljuqs, the Twelver Shicite Buyids in Iraq had the khutba read in Caliph al-Qa’im’s name while the Sunni Marwanid and Mirdasid dynasties in south-eastern Anatolia and northern Syria had it read in the Fatimid caliph’s name. When on Sultan Tuğrul’s death Caliph al-Qa’im tried to re-assert his political authority, inviting the local dynasts to discuss how best to usurp Seljuqid rule, after taking up residence in the Seljuqs’ palace, the Sunni Uqailid Muslim pillaged the caliph’s palace. Muslim left Baghdad only after learning of Alp-Arslan’s victory 172 Evaluation over Kutalmış. Nearly a century later, the hostility the ruler of Hilla, Dubays al-Sadaqa, showed toward Caliph al-Mustarshid seems to have been based on a similar concern: curtailing the caliph’s temporal ambitions. Despite Tuğrul Beg’s request for recognition immediately after Dandankan (1040), Caliph al-Qa’im does not appear to have responded until some years later, when he is said to have appealed for peaceful relations on behalf of the Buyids (1043/4). Tuğrul Beg’s attitude seems unambiguous; daughters were not exchanged in marriage between the Seljuqs and the Buyids until Caliph al-Qa’im had issued an edict legitimizing the Seljuqs’ sovereignty (1046/7). Even after Buyid rule began to unravel in internecine warfare (1048), such that Arslan al-Basasiri, their commander in Baghdad, had the khutba read in the Fatimid caliph’s name (1050), Tuğrul Beg is said to have declined the caliph’s invitation and instead sent an envoy (1052). It appears Tuğrul Beg arrived in Baghdad only after the caliph had persuaded the Buyid ruler al-Rahim to proclaim his allegiance (1055). Having brought his army, Tuğrul Beg imprisoned al-Rahim despite the caliph’s protests on the pretext that he had started the riots sparked by the Seljuqs’ presence. Arguably, therefore, Tuğrul Beg used the caliph’s good office to put an end to the Buyids, rather than championing Sunni Islam against the Shicite heresy as Turan proposed. This is not to deny that Tuğrul Beg dispatched Kutalmış against Arslan al-Basasiri, or that on Kutalmış’s defeat he advanced personally on al-Basasiri with Ibrahim Yınal and Yakuti (1056), forcing al-Basasiri to withdraw to Syria. It was after this that the caliph declared Tuğrul Beg sultan (1058). Even Turan thought this was their agreement: Tuğrul Beg saves Caliph al-Qa’im from al-Basasiri and in return the caliph declares him sultan. Köymen claimed that as sultan Tuğrul Beg was resolute that the caliph should hand over all temporal rights (1963: 178–9). He pointed to the events said to have unfolded after the caliph’s unilateral intervention on learning of Ibrahim Yınal’s intention to rebel. Having gained Ibrahim Yınal’s personal reassurance through his vizier, Caliph al-Qa’im is said to have persuaded Sultan Tuğrul to remain in Baghdad so as to protect him from al-Basasiri. When Ibrahim Yınal rebelled instead of dealing with al-Basasiri, Sultan Tuğrul was forced to pursue him and al-Basasiri took Caliph al-Qa’im prisoner. However, on Ibrahim Yınal’s demise, Sultan Tuğrul does not appear to have attempted to free the caliph, at least not until his offer of peace had been rejected by al-Basasiri who insisted the khutba was read in the Fatimid caliph’s name. As this would have undermined the Seljuqs’ legitimacy, al-Basasiri can be said to have forced Sultan Tuğrul’s hand; the sultan won over the caliph’s captor, Quraysh, and having secured al-Qa’im’s release, advanced on al-Basasiri for the final time. Nor must it be forgotten that having secured the caliph’s wife Arslan Khatun’s release before the caliph’s, Sultan Tuğrul refused to return her until al-Qa’im had agreed to give his daughter to him in marriage. Çağrı Beg’s daughter Arslan Khatun’s marriage to the caliph and Sultan Tuğrul’s to the caliph’s daughter were childless. Sultan Alp-Arslan’s daughter Fülane’s marriage to the then heir apparent al-Muqtadi also appears to have been Evaluation 173 childless. It is noteworthy that this marriage ceremony is said to have taken place in Nishapur, not Baghdad. As for Sultan Malik-Shah, he is said to have sent the chief jurist for Khwarazm to arrange the marriage of his daughter Mah Melek Khatun to the caliph. The jurist, however, advised the Caliph al-Muqtadi not to take a ‘coarse’ Türkmen into his household. When confronted by Nizam al-Mulk, he is said to have retorted that he was not one to sell his religion for worldly concerns. After the exchange of daughters during Sultan Tuğrul’s reign, however, the caliph could not have ignored Sultan Malik-Shah, particularly as he had married Sultan Alp-Arslan’s daughter when he was still heir apparent. At first Terken Khatun is said to have demanded a sum commensurate with that offered by the Karakhanids and the Ghaznavids. After Arslan Khatun intervened however, Terken Khatun seems to have accepted less on the condition that the caliph did not maintain or entertain other women, and kept Mah Melek Khatun by his side at all times (1081). For this reason it is thought that the marriage took place after Caliph al-Muqtadi’s first wife died (1087). Unlike previous marriages between the two houses, Mah Melek Khatun’s marriage to Caliph al-Muqtadi was not childless. Even though Jacfar was the youngest of the caliph’s sons, soon after his birth Sultan Malik-Shah demanded that he be appointed heir apparent. In response, the caliph expelled Mah Melek Khatun and Jacfar. Once back in Isfahan, the Seljuqs are said to have referred to Jacfar as Amir al-Müminin – commander of all Muslims – a title reserved for the caliph (Kitapçı 1994: 109–56; Işıltan IA/8: 573ff; Kafesoğlu 1953: 207; Turan 1993a: 169–73). In view of the above it seems more than likely that Sultan Malik-Shah was poisoned by the caliph as Kitapçı and Işıltan have suggested, rather than by his wife Terken Khatun as Kafesoğlu, Turan and Özaydın did. Kafesoğlu thought Terken Khatun conspired with the caliph because it was Nizam al-Mulk who had persuaded Sultan Malik-Shah to make his eldest Berk-Yaruk heir apparent. She was willing to return Jacfar to the caliph so that her infant son Mahmud was proclaimed sultan instead of Berk-Yaruk (1953: 208–10). Although it does not explain her willingness to return Jacfar, Turan was of the opinion that Terken Khatun wanted to gain control not only of the empire but also the caliphate (1993a: 170). Köymen appears to have preferred silence, mentioning only that a Seljuqid princess had married a caliph during Sultan Malik-Shah’s reign (1963: 193). After Sultan Malik-Shah, Seljuqid relations with the Abbasids deteriorated steadily. During the interregnum Seljuqid claimants sought to legitimize their ascendancy over each other by gaining the caliph’s seal of approval, which was by all accounts freely given. After the interregnum the Abbasids began to play off one Seljuqid against the other, encouraging the sultans in Iraq al-Arab against Great Sultan Sancar in Marv. This worked out badly for them as two caliphs lost their lives in the process. When the Great Seljuq Empire collapsed, the Abbasids were finally able to re-establish a temporal if minor domain. The Seljuqs’ intention to incorporate the caliphate may be traced to the insistence of Turkic people that as there is one Supreme Celestial Being in the Heavens there can be only one ruler on the Earth (Roux 2001: 61). Arguably this is also the source of the Turkic concept of world dominion noted by Köprülü and propounded by both 174 Evaluation Kafesoğlu (IA/10: 392) and Turan (1979). The Kök Türk, for instance, waged war for universal peace and unity – good relations between the Heavens and the Earth. But warfare was not the only way. The earliest known Kök Türk stone inscription at Bugut, which was erected for Taspar Kagan (572–81), a devout Buddhist, is considered to indicate that at least during his reign the Kök Türk thought Buddhism would better enable world dominion (Roux 1994: 17–18). The later Orkhun stone inscriptions refer to Kök Tengri or Supreme Celestial Being (hence the Turkish Tanrı for Allah) and a pantheon of deities and spirits (Tekin 1988 and 1994), but not to shamans. Arguably when all was in order under a kagan who merited divine favour, there was no need for shamans or indeed monks. Once the Kök Türk Empire dispersed, along with Buddhism Kök Tengri paled into insignificance and local spiritual beliefs once again came to the fore (İnan 1995: 26; also Roux 1994: 53, 92). The Kök Türk kagans do not appear to have been sacred, but the Khazar kagans who likewise stemmed from the A-shih-na were. The Uygur and possibly the Kırgız rulers were also taboo (Divitçioğlu 1987: 129). Among the Oguz, sovereignty appears to have been based on felicitous merit, but there is no evidence that religious and political functions were separate. On the contrary, these seem to have been shared alongside martial functions. Nevertheless, as affiliates of the Oguz, the Seljuqs must have been aware of the basis of dominion under both the Kök Türk and the Khazar. The former were their ancestors and the latter their contemporaries.

Seljuqid succession The Irano-Islamic expectation that the eldest accedes to the throne cannot be used to evaluate Turkic expectations with regard to succession. On the Eurasian steppes older sons were expected to take their inheritances and set up on their own, even if they continued to share water and pasturage. When the father died, the youngest son inherited his share in addition to the family hearth. At Turkic campsites, the eldest son’s tent was pitched next to the father’s, where the youngest son stayed, then came the father’s brothers’ tents starting from the eldest (İnalcık 1993b: 51, 54; see also Barfield 1993: 145–6). The eldest son took over on his father’s death; the youngest took possession of that under his father’s direct rule, the so-called ‘hearthlands’ (İnalcık 1993b: 49–52). It was possible for a father to choose a son other than his eldest to take over, as it was for the eldest to voluntarily yield to a younger brother. Çağrı and Tuğrul Beg’s father Mikail is considered to have been the eldest of Selçuk Beg’s sons. He is thought to have died before Selçuk Beg; as a result Arslan appears to have appropriated the title of yabgu. Although on his imprisonment Musa was senior, Ali Tekin seems to have appointed the youngest Yusuf Yınal leader. When Yusuf was killed after Ali Tekin’s death, Musa came to be referred to as Yabgu Kalan, literally Yabgu Remaining. Despite this and though he was younger than Çağrı Beg, the Seljuqs chose Tuğrul Beg to lead them, which was also why he became sultan. Evaluation 175 In the absence of sons of his own, Sultan Tuğrul is said to have chosen his brother’s youngest son Süleyman to become heir apparent. Rather than the Great Seljuq Empire however, it seems more likely that Sultan Tuğrul was bequeathing his domains to Süleyman, whose mother he had married upon his brother’s death (leviratus), the remainder of the empire having been apportioned among Çağrı Beg’s sons. The Seljuqs appear to have almost completely ignored Süleyman, except for Kavurt who had the khutba in Shiraz read in Süleyman’s name after Alp-Arslan’s victory over Kutalmış. Since Kavurt did not challenge Alp-Arslan, this can be taken to indicate that as the sovereign of Kirman he wanted Süleyman to be the sovereign of Shiraz. When Sultan Alp-Arslan prevented him from expanding his domains at the expense of the Shebankarids however, Kavurt challenged him but was finally unsuccessful although some of the sultan’s commanders favoured him. On learning this Sultan Alp-Arslan withdrew, leaving Kavurt to rule Kirman. The heir apparent Malik-Shah was the second of Sultan Alp-Arslan’s sons. Although the sultan is said to have expected his eldest son Ayaz to object to Malik- Shah’s accession, it was Kavurt who objected immediately on grounds that as the eldest Seljuq he was senior. In his response Sultan Malik-Shah replied that the Great Seljuq Empire was his to rule because inheritance was from father to son. Sultan Malik-Shah designated his eldest, Terken Khatun’s son Ahmad (1077–87), heir apparent (Özaydın 2001: 6), but when he died Berk-Yaruk was appointed. Terken Khatun’s insistence that her son Mahmud acceded on the sultan’s death is related to his seniority by blood (Özaydın 2001: 16). She was not only a Karakhanid, but also Sultan Malik-Shah’s first wife. The sultan’s marriage to his first cousin, Berk-Yaruk’s mother, Zubaydah Khatun may have been a deriva- tive of the principle of leviratus in that Yakuti and his wife may have died before their children had come of age. Nobility of birth may also explain the ‘bloodless’ defeats Muhammad Tapar suffered at the hands of Berk-Yaruk after Muayyad al-Mulk’s death. Muhammad Tapar’s and his uterine brother Sancar’s mother appears to have been a concubine. Of the two brothers who founded the Kök Türk Empire, when Bumin Kagan (542–52) died the younger Istemi Kagan (552–76) did not come east and claim seniority. Instead Bumin Kagan’s sons, Kara (552–3), Mukan (553–72) and Taspar (572–81) followed as Great Kagan, although according to one report Bumin Kagan’s brother A-i became Great Kagan before Kara (Taşağıl 1995: 18). Problems did not arise until Taspar Kagan’s death. As Great Kagan Kara had willed Mukan should follow him and in his turn Mukan had willed that Taspar should follow him (Taşağıl 1995: 19, 27). According to Taspar Kagan, however, since Mukan Kagan had passed over his son Ta-lo-pien, the favour should be returned. The problem with this seems to have been that Mukan Kagan’s Türk wife was childless. Ta-lo-pien was not of a Türk mother, being the offspring of a marriage of dynastic convenience. At the council gathered to decide the matter, Kara Kagan’s son She-t’u is said to have supported Taspar Kagan’s son An-lo for this reason. Although An-lo was confirmed, he was not able to establish order and the council reconvened, She-t’u 176 Evaluation becoming Işbara Kagan (582–7) with An-lo’s consent (Taşağıl 1995: 34–5). Ta-lo- pien, who became A-po Kagan, and more importantly Istemi Kagan’s son Tardu Kagan (576–603), with the death of his uncle’s sons now the most senior of the ruling A-shih-na lineage, both objected, pitching the Kök Türk Empire into a protracted period of inner strife that may have precipitated its first collapse. Although Işbara Kagan was succeeded by his brother Baga Kagan (587–8), when his son Tou-lan Kagan (588–600) took over, An-lo’s son Jan-Kan, who became Ch’i-min Kagan (599–609), also refused to accept Tou-lan or Tardu as Great Kagan (Taşağıl 1995: 35–63; also Barfield 1996: 133–8). According to Divitçioğlu (1987: 137–50; also Barfield 1996: 136), Kök Türk succession was inherently problematic because they regarded their brothers’ sons to be brothers while their brothers’ grandsons reverted to being cousins. In social anthropological theory this is known as the equivalence of brothers (FaBrSo = Br) (Sahlins 1968: 11). Since according to Barfield nomadic pastoralists have a conical clan structure (1993: 147–9), at least in his case this judgement seems contradictory. Nor does it necessarily explain the war of succession during the rejuvenation of the Kök Türk Empire. It is not known whether Bilge Kagan (716– 34) and his father Kutlug Ilteriş Kagan’s (682–92) younger brother Kapagan Kagan’s (692–716) sons were ‘brothers’ or ‘cousins’. What can be said, therefore, is that Kül Tekin, Bilge Kagan’s younger brother, killed all of Kapagan Kagan’s clan, thereby possibly ensuring accession by seniority from within the ruling A-shih-na lineage until the demise of the empire. Despite reports that the Oguz yabgu was chosen from all their nobles (Divitçioğlu 1994: 17–19), the rulers listed in both Rashid al-Din and Bahadır Khan’s versions of the Oguz-Nama appear to have been mostly from the senior Kayı lineage which, according to legend, derived its genealogy from Oguz Khan’s eldest son Kün Khan (also İnalcık 1993b: 54–5). According to Togan this was why the Karakhanids gave their daughters in marriage to the Kayı. In both versions of the Oguz-Nama the appellation Inal or brother-in-law appears frequently as a ruler’s name alongside that of Kayı and Yavquy (yabgu) (1982: 133–4). Upon the demise of the Uygur Empire, Bilge Kül Kadir Kagan took the title of Kara Khan, Oguz Khan’s father’s name, and related himself to the A-shih-na (Kafesoğlu IA/12/2: 187). As Sunni Muslims the Seljuqs’ sovereignty was legitimized by the Abbasid caliphs, but as a Turkic clan they appear to have been well aware of the role fictitious genealogies played. This is why they are thought to have appropriated the rumour that they were Tonga Alper’s descendants (Mansuroğlu IA/4: 192–3). Tonga Alper was none other than Afrasiyab in Firdausi’s (c. 940–c. 1020) Shah- Nama (Togan 1982: 123), a compendium of Pahlavi myth, legend and history presented to Sultan Mahmud of Ghazna (Ritter IA/4: 643–9). Like the legendary Afrasiyab, in Rashid al-Din’s version of the Oguz-Nama Oguz Khan arrives over the Volga (Atil, Itil, Etil or Edil), through the Caspian Lowland, having crossed the Caucasus from Derbend into the Mukan steppe, and returns through Demavend and Nishapur in Khurasan (Togan 1982). According to Togan, Afrasiyab was considered to be Turkic (1981: 18–19, 108–10, 134–5, 167–8, 174). Mahmud Evaluation 177 of Kashgar unmistakably refers to him as Alp Er Tonga and more correctly as Tonga Alper (Atalay 1992/I: 41 and III: 368). Another of the Seljuqs’ contemporaries, Yusuf Has Hajib, idealized Tonga Alper as a model leader in Kutadgu Bilig (Dilâçar 1988). The Fergana and Vienna manuscripts of the Oguz-Nama confirm that the Karakhanids also claimed to have descended from Tonga Alper. The Seljuqs’ and the Karakhanids’ genealogical claims further illustrate the connection between descent and martial traditions on the Eurasian steppes. Kafesoğlu argued that dynastic members thought victory in combat signified divine favour. İnalcık went further still; he suggested that in view of the magico- religious source of Turkic sovereignty, any human attempt to regulate succession was considered to have been in vain (1993b: 41–3). Certainly, the Hsiung-nu rulers’ title Cheng-li Kutu Shan-yü is thought to translate as Tengri Kutu Shan-yü (Roux 1994: 134), or Shan-yü, the Celestial Supreme Being’s Felicity. This is particularly noteworthy given that Tengri is the oldest word identified as Altaic (Roux 1994: 91). However, before the role that meritorious felicity played in Turkic succession can be discussed, it is necessary to gain a perspective on the Turkic concept of divinity. The origins of the Hsiung-nu ruling lineage are not known, but the two legends of lupine origin associated with the Kök Türk (Taşağıl 1995: 11–12; also Divitçioğlu 1987: 72–3) provide a clear indication. One legend reflects a shamanistic character (İnan 1995: 4, 160–5), the other refers to the Türk’s prowess as smiths. Before Bumin Kagan founded their empire, the Kök Türk are said to have been the Juan- Juan’s ironsmiths (Taşağıl 1995: 17). The smith’s craft is accepted as second only to the shaman’s vocation of leaving his or her body during a self-induced trance so as to ascend to the Heavens or to descend into the Earth (Divitçioğlu 1987: 83ff; see also Roux 1994: 49–51, 64–5 and Eliade 1989: 5, 470ff). Current consensus is that shamanism spread among Turkic people long after their belief in a Celestial Supreme Being emerged (İnan 1995: 1; also Ocak 1983: 33–6).This is of some importance because shamans are not known to have represented the Celestial Supreme Being (İnan 1995: 87), most likely because they did not believe in one (İnan 1995: 27–8). As under the Hsiung-nu, therefore, so also among the Kök Türk; sovereignty was further legitimized through kut rather than trance. This is confirmed by the Orkhun stone inscriptions, according to which Kök Tengri pulls Kül Tekin’s parents by their hair to the Heavens where the Türk sacred lands and water are in order. Arguably the ritual strangulation of Kök Türk kagans during inauguration emphasizes the point, because it was not self- induced as in the case of a shamanistic trance. Not surprisingly, a divine mandate is thought to have precluded a shaman from rising to the Heavens and eclipsing his or her ruler (Roux 1994: 53, 118). To return to the topic of meritorious felicity; as İnalcık pointed out (1993b: 41; see Atalay 1992/IV: 388), Mahmud of Kashgar defined kut as uğur (good luck or omen, also auspiciousness), devlet (state or government, also prosperity, success or good luck), baht (destiny or fortune), talih (good fortune) and saadet (happiness, also prosperity). Although deemed essential for life (İnan 1995: 37), not only could kut be transmitted, visited or gifted, but also those in receipt of it 178 Evaluation could be censured for their actions (Roux 1994: 31, 131–4; also Divitçioğlu 1987: 59; see Atalay 1992/I: 92 and Tekin 1988: 17, 51). Arguably this is very much in keeping with the alternative meanings of Ogur (the clans), namely to split a bone (as in Ak and Kara Kamag Budun) and felicity/statecraft. In short, kut was not a divine right. The expectation that divine favour must be renewed is indicated in Turkic accession rituals from at least as early as the Tabgaç (T’o-pa) or Wei dynasty (386–534). Their emperor was seated on the ground while being told of the virtues expected of him and of the harmony anticipated between the Heavens and the Earth. Subsequently, the high officials raised both the emperor and the empress several times (Roux 1994: 118). Consequently, kut seems best defined as a life force, which having been divinely gifted could be renewed as well as ruined (Roux 1994: 132, 133, 207; also İnan 1995: 37). While it may have originally denoted victorious strength or the light of victory as in the Indo-Iranian hvarnah (Roux 1994: 134), this suggests a celebration of victory rather than a justification of the pursuit of power. So what determined Turkic succession? Was it seniority by age, blood or ability? The answer is they all played a role in helping royal claimants gain the support of the majority of those who enabled their dominion. The Luan-ti was the ruling Hsiung-nu lineage. Unlike the Kök Türk A-shih-na however, members of which seem to have married the daughters of the A-sheh-tê, the Luan-ti married the daughters of at least three other lineages, namely the Hsü-pu, Hu-yen and Lan. Until 59 BCE Hsiung-nu succession was from father to son, or in ‘unusual situations’ such as that of an infant shan-yü from elder to younger brother. After the Hsiung-nu interregnum, succession was from younger to elder brother (Barfield 1996: 41–2 and Divitçioğlu 1987: 168, 192–8). This is reflected in the Kök Türk legends of origin. In the one that emphasizes their prowess as smiths the eldest son becomes ruler, while in the one that emphasizes their shamanistic powers the youngest becomes the ruler through merit (Taşağıl 1995: 11–12; also Divitçioğlu 1987: 72–3). Whatever the underlying principle, arguably Hsiung-nu succession appears to have been more stable than that of the Kök Türk because more noble lineages had a vested interest in its success. Like the Kök Türk council (kengeç), doubtless the Oguz council (kengeş) had a say in who became yabgu and in his subsequent decisions. According to Ibn Fadlan (Şeşen 1975: 30), the Oguz took decisions on a consultative basis. It is known that the Seljuqs also held councils. Although after Dandankan the Seljuqs apportioned their newly won realm and those they intended to conquer at a kengeş, Sultan Alp-Arslan and Malik-Shah assigned family members to various parts of the empire. Not surprisingly perhaps, rather than a kengeş the Seljuqs are better known for toys, the feast held at a kengeş or marriage. In Nizam al-Mulk’s Siyasat-Nama, Sultan Tuğrul is noted to have had a daily feast laid out for the poor. But this custom also appears to have been relaxed under Sultan Malik-Shah. According to Nizam al-Mulk, the Çiğil complained behind the sultan’s back that he had not held a toy for them when in Transoxania (Köymen 1999: 89). Evaluation 179 Selçuk Beg was of a noble Kınık lineage that broke away. Those who joined the Seljuqs in Khurasan were not of noble origin but breakaway kin groups of Turkic descent that had superficially accepted Islam. They could not have a say in Seljuqid succession except when asked for military support by a contender. Eventually Türkmen lineages emerged, either from the breakaway kin groups or from breakaway nobility like the Seljuqs. In time these were joined by local Turkic lineages that emerged from the Great Seljuqs’ mamluks, some of which at least appear to have been of noble descent. Until then the Seljuqs’ Iranian administrators thought they had a say in who should succeed to the throne. For instance, there is the case of Sultan Tuğrul’s Khwarazmian wife Altuncan Khatun’s son Anushirvan. While Sultan Tuğrul was besieged by Ibrahim Yınal in Hamadan, possibly because of its duration Kunduri is said to have tried to replace him with Anushirvan (1058–9). Kunduri is also attributed with trying to declare Süleyman sultan when Sultan Tuğrul died. Although Kunduri is referred to as his vizier, Sultan Tuğrul is said to have consulted his wife not Kunduri in matters of state. The vizier is thought to have merely offered counsel though he must have been in charge of the sultan’s administration (Köymen 1976: 72), which did not include Çağrı Beg and his sons’ domains. Kunduri’s first letter to Alp-Arslan confirms this. Çağrı Beg and his sons ruled independently, as did Sultan Tuğrul. They came to each other’s aid when requested. While Tuğrul Beg helped Çağrı Beg in Khwarazm, Kavurt and Alp-Arslan joined Sultan Tuğrul and Yakuti to defeat Ibrahim Yınal. Kavurt was assigned Kirman and Kuhistan (1048) (Merçil 1980: 21). Subsequently Alp-Arslan was assigned Balkh, Tocharistan and Tirmiz with his father’s vizier Shadan (1058) (Köymen 1972: 5). Nevertheless, Alp-Arslan is reputed to have been victorious against the Ghaznavids well before, when only fourteen or fifteen years old (1043–4) (Köymen 1972: 4). Said to have taken command due to his father’s illness, most likely the story is based on Alp-Arslan’s later victory over Sultan Mawdud’s son Mascud’s forces (1057). Clearly, however, Alp-Arslan’s martial energy was regarded favourably by the military. One of Süleyman’s mamluk commanders, Erdem, and others originally under Sultan Tuğrul are all said to have defected to Alp-Arslan whose resolve to include Sultan Tuğrul’s realms was demonstrated when he journeyed from Balkh to Nishapur on hearing rumours that the sultan was on his deathbed (1063) (Köymen 1972: 10). Not surprisingly, it was Alp-Arslan and not Kavurt who received Kunduri’s allegiance when the vizier failed to deal with Kutalmış. At the time Kavurt is known to have advanced as far as Isfahan. Kavurt did not challenge Alp-Arslan, but had the khutba in Shiraz read in Süleyman’s name. When Sultan Alp- Arslan returned Fars to the Shebankarids however, Kavurt clearly opted for full independence (Köymen 1963: 65–8). Despite the support of some of Sultan Alp- Arslan’s commanders this proved unsuccessful, although the sultan was forced to withdraw. When Malik-Shah succeeded, the military once again supported Kavurt but, unlike Kunduri, Nizam al-Mulk was able to retain the loyalty of enough soldiery to defeat Kavurt and have him killed. The commanders that supported Kavurt did not switch sides; they laid down their weapons. 180 Evaluation As the sultan’s vizier, the unification of the Seljuq realms under Sultan Alp- Arslan enabled Nizam al-Mulk to place his numerous sons, sons-in-law and no doubt others worthy of his trust in administrative positions and more importantly to build up military support, including his own forces. As a result, even if at the beginning the Seljuqs’ administrative servants were unable to determine who acceded, by the end of Sultan Alp-Arslan’s reign they can be said to have become at least as important as the Seljuq military. Under Sultan Malik-Shah this created contention between the men of the sword and those of the pen. Worse still for Nizam al-Mulk and his dependants, the vizier made an enemy of Sultan Malik-Shah’s wife Terken Khatun, insisting on seniority by age rather than by blood on the death of her first son, the heir apparent Ahmad. Although the norm in Islamdom, this went against Turkic ideology, also because Turkic khatuns were expected to rule when their husbands went on campaign or died, leaving behind an underage heir (Eberhard 1942: 89; Kafesoğlu 1993: 257–8; Barfield 1993: 146–7). What Nizam al-Mulk and his sons appear to have overlooked in their claim to power is that while the Seljuqs only needed the caliph’s word to legitimize their rule, they needed the Seljuqs’ military more, despite public support. At the time of the vizier’s assassination and the sultan’s death, Nizam al-Mulk’s sons were at odds not only with the Seljuqs but also with their military. This deprived Great Seljuq succession of consensus at a critical conjuncture, when the Abbasid caliphate’s legitimacy had been challenged by the demand that a Seljuqid become the next caliph. In short, rather than indeterminacy inherent to Turkic succession, the internecine warfare that followed Sultan Malik-Shah’s death reflected a power struggle between two ideologies.

The power struggle Terken Khatun’s dealings with the caliph could have secured her regency until Mahmud was of age, despite Nizam al-Mulk’s sons. In return for having the khutba read in Mahmud’s name, the caliph stipulated that Terken Khatun’s commander was appointed commander-in-chief. This may have come too soon after Sultan Malik-Shah’s appointment of her former vizier in place of Nizam al-Mulk. The vast sum Terken Khatun is said to have spent in accession money did not secure the loyalty of the royal mamluk corps billeted in Isfahan. At Burujird some of the commanders defected to Berk-Yaruk who was supported by the Nizamiyya, the deceased vizier’s mamluk soldiery. This set the pattern for the interregnum. The caliph legitimized whichever Seljuqid was on top while their soldiery switched allegiances as and when it suited them, actively encouraged by Nizam al-Mulk’s sons. Besides Sultan Malik-Shah’s royal mamluk corps in Isfahan and the regional commanders and governor generals with their soldiery, Terken Khatun, Tutuş, Ismacil, Arslan-Argun and even Nizam al-Mulk all had their own soldiery. Even if Berk-Yaruk’s first two encounters in the field had not expended the royal Evaluation 181 mamluk corps, it must have diminished them substantially, strengthening the regional commanders’ standing. The situation appears to have become acute after Berk-Yaruk’s battle with Ismacil. Although Üner and the commanders in Isfahan refused Tutuş’ invitation, despite Terken Khatun’s intention for them to join him, Berk-Yaruk needed the regional commanders’ support to defeat his uncle Tutuş. Nizam al-Mulk’s son Muayyad al-Mulk is credited with rallying them to Sultan Berk-Yaruk’s banner. It is not surprising that the commanders then openly resisted the young sultan’s ‘strongly centralist’ vizier, al-Balasani, who like Nizam al-Mulk is said to have wanted a civil administration supported by a central force that was financed from direct taxation. Without his own royal mamluk corps, Berk-Yaruk could not prevent the commanders from murdering al-Balasani despite having refused to hand him over. Having escaped with his life, Berk-Yaruk seems to have been forced to recruit soldiery from elsewhere. In his first battle with Tapar, he is thought to have been supported by numerous Türkmen. In their second encounter however, Berk-Yaruk is said to have had a large contingent of infantry (April 1101). Soon after he appears to have led a massacre of Ismacilis (June 1101), killing their commander Düshmenziyar. Apparently his commanders complained that even on the field Tapar’s men were taunting them for being Ismacilis. This suggests that at the very least they had infiltrated Berk-Yaruk’s forces (Hillenbrand 2001: 208). Earlier in Khurasan Ismacilis also seem to have fought on Berk-Yaruk’s side, as Habashi’s soldiery, when Sancar defeated them. The warfare between the Seljuqs lasted for eight years before it began to abate (1092–1100), a period coincident with the rise in Nizari activity. During this time the Ismacilis terrorized notables throughout the Great Seljuq Empire by highly public assassinations. The year before the killings peaked they took the citadels of Girdkuh and Shahdiz near Damghan and Isfahan respectively (1096) (Hillenbrand 2001: 206). The latter citadel was of considerable importance as besides its proximity to Isfahan it controlled the road to Khurasan. To these must be added the various fortifications in Dailam, where for instance Ghutinar appears to have been visible from Alamut, Lamasar and Maymündiz (Hillenbrand 2001: 215), as well as those in Kuhistan, Khurasan and Syria, all of which seem to have exerted control over their immediate environs. As for the Türkmen, they seem to have become spread too thin to be a force. They are mentioned as the affiliates of various dynastic members. There is Çağrı Beg’s son Arslan-Argun in Khwarazm and Khurasan, Sultan Alp-Arslan’s son Böri-Pars in present-day Afghanistan, Çağrı Beg’s grandson Ismacil in Azarbayjan, Sultan Malik-Shah’s brother Tutuş in Syria and south-eastern Anatolia, and last but not least Muhammad Tapar in Ganja. Arslan-Argun seems to have been unable to take Nishapur with his Türkmen. It is only after he gained the support of the mamluk commanders in Marv that he was able to take not only Nishapur but also Balkh and Tirmiz. Mamluk soldiery also supported Böri-Pars and his vizier ‘Imad al-Mulk. It is noteworthy that Arslan-Argun appears to have been content for the Great Seljuq Empire to revert 182 Evaluation to bipartite rule as in his father’s time. Although arguably Böri-Pars supported a unified empire, he cannot have been too concerned with the state of affairs in Iraq either given that he retired to Herat as soon as he had defeated his uncle. As for Azarbayjan, given that Terken Khatun sent her commander Kür-Boğa to support Ismacil with mamluk soldiery, there could not have been that many Türkmen there either. Although Tutuş is known to have advanced on Iraq al-Ajam through eastern Anatolia and Azarbayjan, on the first attempt he was forced to withdraw when Aksungur and Bozan defected. On his second and last attempt he waited for Terken Khatun’s mamluk soldiery. Even though Berk-Yaruk had the support of the Türkmen in Iraq al-Arab, Tapar defeated him. Moreover, during their subsequent confrontations Tapar seems to have repeatedly tried to rally those Türkmen in Azarbayjan and Arran but failed. The contenders’ dependence on the various forces mentioned would have been much reduced if they had the funds to recruit their own soldiery. Possibly with both the Public and Private Treasury in Terken Khatun’s possession, Berk-Yaruk for one seems to have been hard pressed for money from the start. Despite his victory at Burujird, he is said to have accepted money from Terken Khatun immediately after, ceding Isfahan to her son Mahmud. Most likely due to the regional commanders’ resistance to central taxation, Berk-Yaruk appears to have run out of funds altogether by the end of 1099, if not earlier. Reports of his pillaging the caliph’s estates near Wasit and demands of back taxes from the caliph’s vizier began in 1100. There is the plundering of Baghdad soon after, despite an infusion of funds from the caliph, and the demand for back taxes from the ruler of Hilla, Sayf al-Dawla Sadaqa. That same year the commanders imposed their first agreement on Tapar and Berk-Yaruk (1101), wherein they stipulated how much tax Tapar would pay into the Public Treasury. Tapar seems to have been equally short of funds, likewise confiscating disposable wealth from the caliph’s vizier and the inhabitants of Rayy. The second and final agreement between Berk-Yaruk and Tapar clearly underlined the regional commanders’ independence, confirming as it did that they were free to change allegiances whenever they wanted. Given the rift between the men of the sword and those of the pen it seems clear that in the absence of an ongoing campaign no one Seljuqid had a sufficiently large force with which to ensure their sovereignty on Sultan Malik-Shah’s death. What forces were available outside those stationed in the regions seem to have been dissipated in five of the first six battles between Berk-Yaruk and his chal- lengers; these are said to have been quite bloody. Without funds, Berk-Yaruk was completely at the mercy of the commanders. As a warrior élite however, the importance of the commanders’ fealty must not be underestimated. This is what Great Sultan Sancar appears to have achieved. There is no mention of him ordering commanders on campaign as in the case of Sultan Tapar. Great Sultan Sancar always seems to have led his forces personally, which most likely was the reason underpinning their loyalty. Evaluation 183 Common ownership of the realm As noted at the outset, when older sons of a patriarch left they did so with their inheritance. This enabled them to set up on their own with their families. As long as they remained with their father they continued to share water and pasturage according to his position within the polity. If the camp was overpopulated or became so during their father’s lifetime, then they might strike out on their own. In this case they and their affiliates shared grazing rights according to their position, which would be below their father’s. While the youngest son was expected to remain with his father and inherited the family hearth, his father’s position was inherited by the eldest, if another had not been designated. After Dandankan Çağrı Beg retained Khurasan while Musa Yabgu and presumably his sons were assigned to the conquest of Herat and Sistan. Tuğrul Beg took upon himself the conquest of Iraq al-Ajam from Nishapur, with Ibrahim Yınal and Kutalmış. It is not clear what happened to Musa Yabgu’s sons Hasan, Yusuf and Böri, one of whom may have rebelled in Herat on Sultan Alp-Arslan’s accession. As Herat was subsequently assigned to Sultan Alp-Arslan’s son Togan- Shah however, they may have died earlier without heirs. Çağrı Beg seems to have retained Alp-Arslan with his other sons, Ilyas and Osman, assigning the eldest Kavurt to the conquest of Kirman and Yakuti (Alp-Sungur) to that of Azarbayjan. Kavurt and Yakuti’s descendants ruled Kirman and Azarbayjan to the end. Tuğrul Beg was clearly reluctant to allow either Ibrahim Yınal or Kutalmış to gain a foothold in Jibal and Jurjan respectively, which they conquered. As a result Ibrahim Yınal rebelled twice, first when Tuğrul Beg took Hamadan and its environs from him and second after he had been assigned Mosul, which he seems to have considered an inadequate reward for his invaluable services. His brother Er-Taş was not allowed to keep Sistan either, although he helped Musa Yabgu to conquer it. Er-Taş is thought to have returned to Khurasan where he died soon after. His sons Muhammad and Ahmad joined their uncle Ibrahim Yınal during his second rebellion and were killed. Ibrahim Yınal’s brother Er-Basgan migrated to Anatolia after having taken part in Kavurt’s first rebellion, which was sparked by Sultan Alp-Arslan returning the Shebankarids’ lands. Having been forced to seek refuge with the Byzantines, Er-Basgan appears to have joined Kutalmış’ son Süleyman-Shah. Kutalmış and his brother Resul-Tekin also rebelled against Sultan Tuğrul. They survived to challenge Alp-Arslan for the sultan’s realms, losing their lives. Kutalmış’ sons appear to have been released on Sultan Alp-Arslan’s death, migrating to Anatolia where Süleyman-Shah established an independent polity. It is doubtful he accepted Sultan Malik-Shah’s suzerainty as some Turkish historians have claimed. When the Great Seljuqs’ vassal, the Uqailid Muslim lost his life in a battle with Süleyman-Shah, his descendants asked Sultan Malik-Shah whether to hand Aleppo over to him or Tutuş. Not surprisingly Sultan Malik- Shah chose his brother and Süleyman-Shah lost his life in the ensuing battle with Tutuş. By comparison, the city dignitaries in Mosul invited Süleyman-Shah’s son Sultan Kılıç Arslan, as well as Sultan Tapar’s vassal Sayf al-Dawla Sadaqa 184 Evaluation and commander Porsukoğlu Aksungur. However they did so without informing Sultan Tapar, possibly because he had assigned Mosul to Çavlı Sakavu who had defeated and killed their governor, Çökürmüş. In any event, Çavlı defeated Kılıç Arslan who lost his life while in flight. Kavurt lost his life challenging Sultan Malik-Shah’s right to the throne, but his descendants continued to rule Kirman. The nature and scope of Sultan Malik-Shah’s brother Tekiş’ rebellions are not clear. Possibly like Arslan-Argun during Sultan Berk-Yaruk’s inconclusive reign, he merely wanted to gain a sustainable or larger dominion of his own in Khurasan, as his nephews had in Kirman and Anatolia. For their part, Tutuş’ sons Ridwan and Dukak, the so-called Seljuqs of Syria, became isolated after their father lost his life in the attempt to gain the throne from his nephew Berk-Yaruk. Certainly they appear to have accepted Sultan Berk- Yaruk’s sovereignty even if their relationships with Sultan Tapar seem ambivalent. However, clearly Dukak wanted a realm of his own. When Ridwan was forced to accept this, it seems Dukak was quite happy to have his elder brother’s name in the khutba before his. Great Sultan Sancar did not unify the Great Seljuq Empire after he challenged and defeated his nephew Mahmud. Instead he established the Seljuqs of Iraq alongside those of Kirman and Anatolia. The sultans of Iraq, Mahmud and Mascud, seem to have resented Great Sultan Sancar’s suzerainty. Mascud took the field, challenging Great Sultan Sancar’s appointment of his brother Tuğrul as sultan. Nevertheless, the almost continual warfare between Sultan Tapar’s descendants seems to have been due to the ambitions of their commanders and the caliphs as much as their objections to Great Sultan Sancar’s suzerainty, which is thought to have left them in financial difficulties because he appropriated a good portion of Iraq al-Ajam. To their credit, however, as they grew older they appear to have tried but invariably failed to wrest control back from the commanders, one of whom always managed to designate themselves their atabeg. In sum, contention between the Seljuqs arose in the quest for a share of the Seljuqs’ sovereignty. Direct challenges to the throne only arose when this was denied. Consequently contention cannot be said to have stemmed from the concept of common ownership.

The role of women With the exception of Köymen Turkish historians have treated Seljuq women as meddlesome, focusing in particular on Sultan Malik-Shah’s wife Terken Khatun. Although Köymen noted the role Sultan Tuğrul’s wife Altuncan Khatun played in state affairs, he did not bring out the importance of other Seljuq women. It would seem that Turkish historians have been influenced by Muslim men of letters who appear to have been adversely affected by the Seljuqs’ and their affiliates’ rude vigour. A couple of generations before the Seljuqs, Ibn Fadlan had been embarrassed by an Oguz woman who scratched her private parts in his presence. Seeing his discomfort, her husband admonished him that better so than she cover herself in public and commit adultery in private (Şeşen 1975: 31). Evaluation 185 Sultan Tuğrul is said to have consulted with his wife on all matters of state. During Ibrahim Yınal’s rebellion it was Altuncan Khatun who came to Sultan Tuğrul’s aid while his vizier, Kunduri, attempted to put her son Anushirvan on the throne. As a result, rather than Kunduri, it was Altuncan Khatun who fought off Ibrahim Yınal’s advance units and enabled the sultan to withdraw to Rayy, where she joined him. Little seems to be known about Sultan Alp-Arslan’s wife, but it may be indicative that the Great Seljuqs’ vassal in Khuzistan is said to have saved his life by paying part of the taxes outstanding from Sultan Tuğrul’s reign to Sultan Alp- Arslan’s wife. Sultan Malik-Shah’s wife Terken Khatun had her own diwan and soldiery, and it was her vizier who replaced Nizam al-Mulk upon his assassination. Given her other activities, it is clear she played a similar role to Altuncan Khatun and Sultan Alp-Arslan’s wife. Certainly, in his Siyasat-Nama, Nizam al-Mulk counselled at length against allowing women a voice in affairs of state. Not surprisingly, his son Muayyad al-Mulk had Sultan Berk-Yaruk’s mother Zubaydah Khatun strangled. Rather than being meddlesome, the role Seljuq women played appears to have arisen from the fact they had as much right to the empire as the men, although the latter ruled. Sultan Tapar is thought to have had his wife Gevher Khatun killed on his death in order to prevent his brother Sancar from inheriting his realms by marrying her, as was customary. Gevher Khatun was Yakuti’s great granddaughter and after her father Mawdud’s death directly responsible for Azarbayjan. On her death, Sultan Mascud’s wife, who as Sultan Berk-Yaruk’s daughter was Yakuti’s daughter Zubaydah Khatun’s namesake and granddaughter, gained a say in the running of Azarbayjan. Great Sultan Sancar’s former commander Karasungur benefited from her patronage even though Sultan Mascud had defeated and ousted him when Karasungur was Sultan Tuğrul’s regent in Azarbayjan. Sultan Mahmud’s wife likewise had her own vizier and, on her father Great Sultan Sancar’s orders, at one time stood surety for the Seljuqs’ Shicite vassal Dubays al-Sadaqa, whom the Abbasid caliph feared. 9 Conclusion

Köymen remains the main historian on the Great Seljuqs. In Turkey his works are supplemented by Kafesoğlu’s and more recently Merçil’s on the Seljuqs of Kirman, Sevim’s on the Seljuqs of Syria and Palestine, and lastly Özaydın’s on Sultan Berk-Yaruk and Sultan Tapar. Turan concentrated on the Seljuqs of Anatolia and Sümer on the Oguz. After his work on Sultan Malik-Shah, Kafesoğlu sufficed with his book-length article on the Seljuqs. In Turkey the Council for Higher Education’s insistence that all theses presented cover new ground is one reason why no one has followed Köymen or Kafesoğlu. In the West there is no such bureaucratic excuse. Their failure stems from their view that everything nomadic is essentially primitive and Iranian culture or Islam is the civilizing element in all things Turkish. Despite taking care to point out the Turkic character of the Great Seljuq Empire, ironically Turkish scholarship on the Seljuqs shares this view, largely because of Köprülü. In Bizans Müesseselerinin Osmanlı Müesseselerine Tesiri (The Influence of Byzantine Institutions on Ottoman Institutions) Köprülü argued that Ottoman institutions had been derived from the Seljuqs and hence the Abbasids, but then added that the Abbasids may have been influenced by the Byzantines and the Iranians. Köymen, Turan and Sümer were his students. Even Kafesoğlu, who argued that the Seljuqs succeeded in ‘welding’ Turkic characteristics to Islamic ones, could only come up with ‘old Turkish traditions, customs, and usages’ besides noting that the Seljuqs and their military communicated in Turkish. According to Kafesoğlu, the other characteristics were the sultan’s tuğra or monogram on official documents; the bow and arrow, used as symbols of sovereignty; the institution of guardianship (atabeg); the role of women in state affairs; Turan or steppe tactics in battle; yoğ or leviratus; strangling of dynastic members with their own bowstring, not to shed their blood; toy or public banquets; tuğ or horsetail banner; large hunting drives that served as manoeuvres; the ball and stick game; folk dancing; and customary laws, though he did not specify how and where these were applied instead of Shari law. Akçura’s precepts are only partially applicable to the Seljuqs, but Köprülü’s framework fails altogether. The Seljuqs did not ally with indigenous aristocracies and turn their backs on the Türkmen as he proposed. Their Türkmen affiliates could not have considered the Seljuqs first among equals. The Seljuqs were Conclusion 187 Akbudun, nobility. The Türkmen were Karabudun, common folk, little better than slaves. There was no need for the Seljuqs to ally with indigenous aristocracies to become free of the Türkmen. Quite the opposite, Sultan Malik-Shah clearly set out to eradicate local dynasties, replacing them with not only so-called mamluk commanders of Turkic origin but also Türkmen begs, breakaway Turkic noblemen and their affiliates who arrived after the Seljuqs became established. Ultimately it was the Turkic character of the Seljuqs’ statecraft that prevented the sons of indigenous notables such as Nizam al-Mulk from transforming the Great Seljuq Empire into an Irano-Islamic state. Turkic nobility organized their affiliates in thousands, hundreds and tens. The larger groups were commanded by family members and the lesser ones by affiliated nobles. At each level there were households of chamberlains and élite corps. Noblemen led by example and earned their élite corps’ fealty. Those of lesser martial ability held other jobs. They became household administrators and in the case of an empire government officials. Consequently, even a breakaway noble house was ready to rule a realm. The Great Seljuqs most certainly had the will and the aptitude, adapting to an agrarian environment with urban centres. There can be no doubt they would have become Shicites rather than Sunnis if the former had prevailed in Transoxania. They showed the same practicality by not imposing their households on their Iranian and Arabic subjects. Instead they appointed government officials from local noble houses. Regardless of their contributions however, these did not share sovereignty in the Great Seljuq Empire. The Seljuqs only appointed male or female family members, or Turkic commanders from their élite corps. Appendix 1 Maps

The maps in this section are in keeping with Turkish historiography. The first map attempts to show Turkic migrations according to Sümer and Kafesoğlu. The second concerns the probable borders of the Great Seljuq Empire at its height during the reign of Sultan Malik-Shah (1072–92). The third map aims to facilitate reference to place names used in the text. The remaining maps try to chart the raids and campaigns that led to the conquest of Anatolia. 0 1000 Miles

OGUR 0 1000 KM BULGAR CE 2 CE 10 Western Siberian Lowland

KIMEK KHAZAR (KUMAN/KIPÇAK) Sayan Mountains KAGANATE l ALTI CE 11 CE 8–10 OGUZ UZ Ura CE CE 11 6 Altay TOKUZ OGUZ PEÇENEK Kirgiz Steppe KIMEK PEÇENEK (KUMAN/KIPÇAK) (inc. PEÇENEK CE 9 and UYGUR) CE 11 UZ CE 10 KIPÇAK OGUR CE 8 CE CE 11–12 10 Aral 3 BCE Sea OGUZ Black CE 10 Sea Caspian Sea Pamiro Gobi Alaya Desert Tar im KARLUK Basin OGUZ CE 7–12 CE 12 TÜRKMEN Mediterranean CE 11–12 Sea

Himalayas

Map A1.1 Turkic migrations and the Seljuqs 0 1000 Miles Caspian Lowland 0 1000 KM Aral Byzantine Sea Black Caucasus Empire Khwarezm Sea Caspian Sea Karakhanids Gobi Desert Anatolia Azarbayjan Tar im Basin Syria Taberistan Jazira Khurasan Mediterranean GREAT SELJUQ EMPIRE Sea Sawad Sistan Khuzistan Fatimids Kirman Himalayas Egypt Fars Ghaznavids

Libyan Arabian Hijaz Desert Sea

Red Sea

Indian Ocean

Map A1.2 The Great Seljuq Empire Aral 0 1000 Miles Caspian Lowland Sea 0 1000 KM Caucasus Mountains Black KYZYLKUM Sea Caspian Kars Tbilisi Sea Gurganj Nur Ani Ganja Bokhara Sivas Erzurum Samarkand Kayseri Karakum Desert Malatya Ahlat Urfa Tabriz Tirmiz Siirt Marv Antakya Harran Nesa Baverd Aleppo Jurjan Tus Alamut Raqqa Mosul Zinjan Nishapur Serakhs Balkh Hama Arbil Mashhad Hims Hamadan Rayy Tekrit Ghazna Kirmanshah Herat Damascus Burujird Baghdad Kavir Desert Isfahan Wasit DASHTELUT Kandahar Basra Kirman Shiraz AN NAFUD Thar Desert Red Persian Sea Gulf AD DAHNA Gulf of Oman

Map A1.3 Urban centres in Seljuq history 0 250 Miles

0 250 KM Caspian Lowland Black Caspian Sea Sea Caucasus Mountains

Tbilisi

Pontic Mountains Kars Izmit Ani Ganja Mughan Erzurum Türkmen Steppe Erzincan (1045–8) Ankara Sivas Eskisehir (1058) (1060) Malazgirt Seljuks (1055) (1058) (1046–8) Kayseri Malatya Ahlat Tabriz Diyarbekir Konya Türkmen of Iraq (1060) (1041 and 1044) Urfa

Mosul

Antakya

Mediterranean Aleppo Sea

Map A1.4 Raids under Sultan Tuğrul 0 250 Miles

0 250 KM Caspian Lowland Black Caspian Sea Sea Caucasus Mountains

Sultan Tbilisi Alp-Arslan (1068)

Kars Izmit Emperor Diogenes Ani Ganja (1068) Erzurum Mughan Steppe Erzincan Ankara Sivas Afsin Eskisehir (1067) Malazgirt

Malatya Kayseri Gümüs-Tekin Ahlat Tabriz (1066) Diyarbekir Sultan Alp-Arslan Konya (1064) Afsin Urfa (1067) Mosul

Antakya Aleppo Mediterranean Afsin Sea (1068) Baghdad

Map A1.5 Raids under Sultan Alp-Arslan and Afşin 0 250 Miles

0 250 KM Caspian Lowland Black Caspian Sea Sea Causasus Mountains

Tbilisi Pontic Mountains Kars Izmit Ani Ganja Erzurum Mughan Steppe Ankara Erzincan Sivas Eskisehir Emperor Diogenes Türkmen (1069) (1069) Malazgirt

Kayseri Malatya Ahlat Tabriz Diyarbekir Konya Afsin (1069) Urfa

Mosul

Antakya

Mediterranean Aleppo Sea

Map A1.6 Emperor Diogenes’ second campaign 0 250 Miles

0 250 KM Caspian Lowland Black Caspian Sea Sea Caucasus Mountains

Tbilisi Türkmen Pontic Mountains (1070) Kars Izmit Ani Ganja Erzurum Mughan Steppe Erzincan Ankara Sivas Eskisehir Manuel Comnenus Malazgirt (1070) Kayseri Malatya Ahlat Tabriz Diyarbekir Konya Er-Basgan (1070) Urfa

Mosul

Antakya

Mediterranean Aleppo Sea

Map A1.7 Manuel and Er-Basgan 0 250 Miles

0 250 KM Caspian Lowland Black Caspian Sea Sea Caucasus Mountains

Tbilisi

Pontic Mountains Kars

Izmit Ani Ganja Erzurum Mughan Steppe Ankara Erzincan Sivas Eskisehir Emperor Diogenes Malazgirt (1071) Kayseri Malatya Ahlat Tabriz Diyarbekir Konya Urfa Sultan Mosul Alp-Arslan (1071) Antakya

Mediterranean Aleppo Sea

Map A1.8 Emperor Diogenes’ final campaign Appendix 2 The Oguz According to Rashid al-Din

Bayat Gün

Döger Ay Dodurga

B o z Kizik

O k Karkin Oguz

Ü ç Beçene Gök O Çavuldur k

Salur Eymur Ala-Yuntlu

Yigdir Bügduz Yiva Appendix 3 Seljuq rulers

Great Seljuqs Çağrı (1040–60) Tuğrul (1040–63) Alp-Arslan (1060–72) Malik-Shah (1072–92) Mahmud (1092–3) Tutuş (1094) Berk-Yaruk (1093, 1094–9 and 1099–1104) Malik-Shah II (1104–5) Tapar (1099 and 1104–17) Mahmud (1117–18) Sancar (1118–57)

Seljuqs of Iraq Mahmud (1117–31) Davud (1131–2) Tuğrul I (1132–5) Mascud (1135–52) Malik-Shah (1152–3) Muhammad (1153–9) Süleyman-Shah (1159–61) Arslan-Shah (1161–77) Tuğrul II (1177–94)

Seljuqs of Kirman Kavurt (1041–73) Kirman-Shah (1073–4) Sultan-Shah (1074–85) Turan-Shah I (1085–97) Iran-Shah (1097–1101) Appendix 3 199 Arslan-Shah II (1101–42) Muhammad (1142–56) Tuğrul-Shah (1156–70) Arslan-Shah II (1170–7) Turan-Shah II (1170–83) Bahram-Shah (1170–5) Muhammad (1175–86)

Seljuqs of Syria Tutuş (1078–94)

Aleppo branch Ridwan (1095–1113) Alp-Arslan (1113–14) Sultan-Shah (1114–17)

Damascus branch Dukak (1095–1104) Tutuş (1104)

Seljuqs of Anatolia Süleyman-Shah (1077–86) Kılıç-Arslan (1092–1107) Malik-Shah (1107–16) Mascud (1116–56) Kılıç-Arslan II (1156–92) Malik-Shah II (1192) Kayhusraw (1192–6 and 1205–10) Süleyman-Shah II (1196–1203) Kılıç-Arslan III (1203–4) Kaykavus (1210–19) Kaykubat (1219–36) Kayhusraw II (1236–46) Kaykavus II (1246–59) Kaykubat II (1249–57) Kılıç-Arslan IV (1248–64) Kayhusraw III (1264–5) Mascud II (1282–3, 1284–93, 1294–1300 and 1302–4) Kaykubat III (1284, 1293–4, 1300–2 and 1304–7) Mascud III (1307) Bibliography

The following abbreviations are used in the bibliography: AÜDTCF Ankara Üniversitesi Dil Tarih Coğrafya Fakültesi AÜY Ankara Üniversitesi Yayınları I˙Ü I˙stanbul Üniversitesi I˙ÜEF I˙stanbul Üniversitesi Edebiyat Fakültesi MEB Milli Eğitim Bakanlığı TDAV Türk Dünyası Araştırmaları Vakfı TDK Türk Dil Kurumu TTK Türk Tarih Kurumu TVYY Tarih Vakfı Yurt Yayınları

Primary sources in Turkish Abü’l-Farac, G. (1945) Abü’l-Farac Tarihi, Cilt I, E. W. Budge publication, trans. Ö. R. Doğrul, Ankara: TTK. —— (1950) Abü’l-Farac Tarihi, Cilt II, E. W. Budge publication, trans. Ö. R. Doğrul, Ankara: TTK. Ahbâr üd-Devlet is-Selçukkiye (1943) M. Ikbal publication, trans. N. Lügal, Ankara: TTK. Aksaraylı Kerimeddin Mahmud (1943) Selçuki Devletleri Tarihi – Müsamerat-al-ahyar, trans. M. N. Gençosman, Ankara. Dede Korkut Hikâyeleri (1993), trans. C. Kudret (ed.), Ankara: Varlık. Ibn Fadlan (1975) I˙bn Fazlan Seyahatnâmesi, trans. R. Şeşen (ed.), Istanbul: Bedir. Ibn Haldun (1988) Mukaddime, Cilt 1 and 2, trans. S. Uludağ (ed.), Istanbul: Dergah. Kaşgarlı Mahmut (1992) Divanü Lügat-it-Türk, trans. B. Atalay (ed.), Ankara: TTK. Nizâmü’l-Mülk (1999) Siyâset-Nâme, trans. M. A. Köymen, Ankara: TTK. Nizamülmülk Siyasetname (1954), trans. M. Ş. Çavdaroğlu, Istanbul: Sermet. Orhon Yazıtları (1988), trans. T. Tekin (ed.), Ankara: TDK. Sâmi, Ş. (1901) Dictionnaire Français-Turc Illustré, 3rd edn, Istanbul: Mihran. —— (nd) Dictionnaire Français-Turc, 2nd edn, Istanbul: Mihran. —— (nd) Kamus-ı Fransevi, 3rd edn, Istanbul. Tunyukuk Yazıtı (1994), trans. T. Tekin (ed.), Ankara: Simurg. Yusuf Hâs Hâcib (1947) Kutadgu Bilig, R. R. Arat (ed.), Ankara: MEB. (First published in 1069–70.) Bibliography 201 Secondary sources Akalın, M. (ed.) (1979) Târihi Türk Şiveleri, Ankara: AÜY. Akçura, Y. (1976) Üç Tarz-ı Siyaset, E. Z. Karal (ed.), Ankara: TTK. (First published in 1904.) Akpınar, T. (1993) ‘Selçuklu Tarihçilerimizde Siyasi ve İdeolojik Düşünceler’, in Tarih ve Toplum, pp. 42–50, Istanbul: Ocak. Allsen, T.T. (1997) Commodity and Exchange in the Mongol Empire, Cambridge: CUP. Arsal, S. M. (1947) Türk Tarihi ve Hukuk – I˙slâmiyetten Evvelki Devir, Istanbul: İÜ. Arsel, İ. (1990) Şeriat ve Kadın, Istanbul: TÜMDA. Ashtor, E. (1976) A Social and Economic History of the Near East in the Middle Ages, CA: University of California. Avcıoğlu, D. (1979) Türklerin Tarihi, Cilt 1, 2, 3, 4 and 5, Ankara: Tekin. Barfield, T. J. (1993) The Nomadic Alternative, NJ: Prentice Hall. —— (1996) The Perilous Frontier, London: Blackwell. Barth, F. (1986) Nomads of South Persia, IL: Waveland. Barthold, V. V. (1990) Moğol I˙stilâsına Kadar Türkistan, trans. H. D. Yıldız, Ankara: TTK. (First published in 1900.) Beckwith, C. I. (1987) The Tibetan Empire in Central Asia, NJ: Princeton University Press. Bekin, A. R. (1981) I˙pek Yolu, Ankara: AÜDTCF. Berkes, N. (1998) The Development of Secularism in Turkey, NY: Hurst. Berktay, H. (1983) Cumhuriyet I˙deolojisi ve Fuat Köprülü, Istanbul: Kaynak. —— (1992) ‘The Search for the Peasant in Western and Turkish History/Historiography’, in H. Berktay and S. Faroqhi (eds) New Approaches to State and Peasant in Ottoman History, pp. 109–84, London: Frank Cass. Bircan, O. (1993) Belge ve Fotoğraflarla Atatürk’ün Hayatı, Ankara: MEB. Birinci Türk Tarih Kongresi: Konferanslar Münakaşalar. (The First Turkish Historial Congress.) (1932), Ankara Bozkurt, F. (1992) Türklerin Dili, Istanbul: Cem. Cahen, C. (1984) Osmanlılardan Önce Anadolu’da Türkler, trans. Y. Moran, Istanbul: E Yayınları. Çandarlıoğlu, G. (1995) Orta-Asya’da Timuriler, Çin’de Ming Münasebetleri, Istanbul: MSÜ. —— (2003) I˙slâm Öncesi Türk Tarihi ve Kültürü, Istanbul: TDAV. —— (2004a) Uygur Devletleri Tarihi ve Kültürü, Istanbul: TDAV. —— (2004b) Sarı Uygurlar ve Kansu Bölgesi Kabileleri (9–11 Asırlar), Istanbul: TDAV. Çavdar, T. (1995) Türkiye’nin Demokrasi Tarihi (1839–1950), Ankara: İmge. Claessen, H. J. M. and Skalník, P. (1981) The Study of the State, The Hague: Mouton. Dilâçar, A. (1988) Kutadgu Bilig I˙ncelemesi, Ankara: TTK. Dirimtekin, F. (1993) ‘Selçukluların Anadolu’da Yerleşmelerini ve Gelişmelerini Sağlayan İki Zafer’, in Malazgirt Armağanı, pp. 231–58, Ankara: TTK. Divitçioğlu, S. (1992) Nasıl Bir Tarih? Istanbul: Bağlam. —— (1994) Oguz’dan Selçuklu’ya (Boy, Konat ve Devlet), Istanbul: Eren. —— (1996) Osmanlı Beyliğinin Kuruluşu, Istanbul: Eren. —— (2005) Orta-Asya Türk I˙mparatorluğu: VI.-VIII Yüzyıllar, 3rd edn, Ankara: İmge. —— (1987) Kök Türkler –Kut, Küc, Ülüg, Instanbul: Eren Diyanet Vakfı I˙slâm Ansiklopedisi 1–22, Istanbul. Dupuy, T. N. (1984) The Evolution of Weapons and Warfare, VA: De Capo. Eberhard, D. W. (1942) Çin’in Şimal Komşuları, trans. N. Uluğtuğ, Ankara: TTK. —— (1987) Çin Tarihi, Ankara: TTK. 202 Bibliography Eliade, M. (1989) Shamanism – Archaic Techniques of Ecstasy, trans. W. R. Trask, London: Penguin. Elvin, M. (1973) The Pattern of the Chinese Past, CA: Stanford University Press. Ergin, M. Büyük Türk Sözlüğü, Istanbul: Hayat. Eröz, M. (1992) Eski Türk Dini (Gök Tanrı I˙nancı) ve Alevilik Bektaşilik, Istanbul: TDAV. —— (1983) Hıristiyanlaşan Türkler, Ankara. Esin, E. (1980a) A History of Pre-Islamic and Early Islamic Turkish Culture, Ankara: Ünal. —— (1980b) ‘Türklerin İç Asya’dan Getirdiği Üniversalist Devlet Mefhumu ve Bunun “Ordu” Mimârisindeki Tezahürleri’, in O. Okyar and H. İnalcık (eds) Social and Economic History of Turkey (1071–1920), pp. 9–25, Meteksan. —— (1994) ‘The Oghuz Epics and Saljuq Iconography’, in R. Hillenbrand (ed.) The Art of the Saljuqs in Iran and Anatolia, Edinburgh. Evans-Pritchard, E. E. (1978) The Nuer, Oxford: OUP. Forard, P. C. ‘The Development of Military Slavery under the Abbasid Caliphs of the Ninth Century (AD) (Third Century AH) with special reference to the reigns of Mutasim and Mutadid’, unpublished PhD Thesis, Princeton University, December 1961. Fox, R. (1967) Kinship and Marriage, London: Pelican. Fried, M. H. (1967) The Evolution of Political Society, NY: McGraw-Hill. Frye, R. N. (1975) The Golden Age of Persia: The Arabs in the East, London: Weidenfeld. —— (1996) The Heritage of Central Asia, NJ: Markus Wiener. Frye, R. N. and Sayılı, A. (1980) ‘The Turks in Khurasan and Transoxania at the Time of the Arab Conquests’, in R. N. Frye Islamic Iran and Central Asia (7th–12th Centuries), Chapter XIII, London: Varorium Reprints. Fuller, J. F. C. (1998) Armament and History, VA: Da Capo. Georgeon, F. (1996) Türk Milliyetçiliğinin Kökenleri: Yusuf Akçura (1876–1935), 2nd edn, trans. A. Er, Ankara: TVYY. Gluckman, M. (1977) Politics Law and Ritual in Tribal Society, London: Blackwell. Golden, P. B. (1992) An Introduction to the History of Turkic Peoples, Mouton: Harrassowitz. Gordlevski, V. (1988) Anadolu Selçuklu Devleti, trans. A. Yaran, Ankara: Onur. Grousset, R. (1979) The Empire of the Steppes – A History of Central Asia, trans. N. Walford, NJ: Rutgers. Günaltay, Ş. (1937) I˙slâm Dünyasının I˙nhitatı Sebebi Selçuk I˙stilâsımıdır? Ankara: AÜ. Harris, M. (1991) Cannibals and Kings: The Origins of Cultures, NY: Vintage. —— (2001) Cultural Materialism – The Struggle for a Science of Culture, updated edition, CA: Altamira. Herodotus (1992) The Histories, trans. G. Rawlinson, London: Everyman. Hillenbrand, C. (1999) The Crusades – Islamic Perspectives, Edinburgh: EUP. —— (2001) ‘The Power Struggle between the Saljuqs and the Ismacilis of Alamüt, 487– {518/1094}–1124: The Saljuq Perspective’, in F. Daftary (ed.) Medieval Ismacili History and Thought, pp. 205–20, Cambridge: CUP. Hodgson, M. G. S. (1974) The Venture of Islam: Conscience and History in a World Civilization, Vol. 1, 2 and 3, IL: University of Chicago. Humphreys, R. S. (1991) Problems in Islamic History, NJ: Princeton University. IA see I˙slâm Ansiklopedisi and note on p. xi. İnalcık, H. (1993a) ‘Turkish and Iranian Political Theories and Traditions in Kutadgu Bilig’, in The Middle East and the Balkans under the Ottoman Empire – Essays on Economy and Society, Indiana University Turkish Studies, Vol. 9, pp. 1–18, IN: Bloomington. Bibliography 203 —— (1993b) ‘The Ottoman Succession and Its Relation to the Turkish Concept of Sovereignty’, trans. D. Howard, in The Middle East and the Balkans under the Ottoman Empire – Essays on Economy and Society, Indiana University Turkish Studies, Vol. 9, pp. 37–69, IN: Bloomington. —— (1993c) Osmanlı I˙mparatorluğu – Toplum ve Ekonomi, Istanbul: Eren. —— (1998) Essays in Ottoman History, Istanbul: Eren. İnan, A. (1995) Tarihte ve Bugün Şamanizm – Materyeller ve Araştırmalar, Ankara: TTK. İnan, Afetinan (1981) A History of the Turkish Revolution and Turkish Republic, trans. A. E. Uysal, Ankara: Pars. Irmak, S. and Çağlar, B. K. (eds) (1994) Cevdet Paşa Tarihinden Seçmeler, Ankara: MEB. I˙slâm Ansiklopedisi, Cilt 1–13 (1997), Ankara: MEB. Jagchid, S. and Symons, V. J. (1989) Peace, War, and Trade along the Great Wall, IN: Indiana University Press. Jankovich, M. (1971) They Rode into Europe, trans. A. Dent, London: George G. Harrap. Kafesoğlu, İ. (1953) Sultan Melikşah Devrinde Büyük Selçuklu I˙mparatorluğu, Istanbul: İÜ. —— (1955) Selçuklu Ailesinin Menşei Hakkında, Istanbul: Osman Yalçın. —— (1976) Türk Milli Kültürü, Ankara: Ayyıldız. —— (1992) Harezmşahlar Devleti Tarihi, Ankara: TTK. (First published in 1956.) —— (1994) Türk Milli Kültürü, Istanbul: Boğaziçi. Kalafat, Y. (1992) Şark Meselesi Işığında Şeyh Sait Olayı, Karakteri, Dönemindeki I˙ç ve Dış Olaylar, Istanbul: Boğaziçi. Karolidis, P. (1993) Romanos Diogenis – Istanbul’a yollar açılırken, trans. K. Dinçmen, Istanbul: İletişim. Kaymaz, N. (1993) ‘Malazgirt Savaşı ile Anadolu’nun Fethi ve Türkleşmesine Dair’, in Malazgirt Armağanı, pp. 259–68, Ankara: TTK. Keegan, J. (1993) A History of Warfare, London: Hutchinson. Kennedy, H. (1994) The Prophet and the Age of the Caliphates: The Islamic Near East from the Sixth to the Eleventh Century, London: Longman. Khazanov, A. M. (1994) Nomads and the Outside World, 2nd edn, trans. J. Crookenden, WI: University of Wisconsin. Kitapçı, Z. (1988) Orta Asyada I˙slâmiyetin Yayılışı ve Türkler, Konya. —— (1994) Abbasi Hilâfetinde Selçuklu Hâtunları ve Türk Sultanları, Konya: Selçuk Üniversitesi. Koestler, A. (1952) The Thirteenth Tribe – The Khazar Empire and its Heritage, London: Random House. Köprülü, M. F. (1938) ‘Ortazaman Türk Hukukî Müesseseleri: İslâm Âmme Hukukundan Ayrı Bir Türk Âmme Hukuku Yokmudur?’, Belleten, Cilt II, Sayı 5–6, pp. 39–72, Ankara. —— (1940) ‘Giriş, İyzahlar ve Düzeltmeler’, endnotes in V. V. Barthold I˙slâm Medeniyeti Tarihi, trans. M. F. Köprülü, Istanbul: Kanaat. —— (1941) ‘Ortazaman Türk-İslâm Feodalizmi’, Belleten, Cilt V, Sayı 19, pp. 319–34, Ankara. —— (1966) Edebiyat Araştırmaları, Ankara: TTK. —— (1981) Bizans Müesseselerinin Osmanlı Müesseselerine Tesiri, Istanbul: Ötüken. (First published in 1931.) —— (1993) Türk Edebiyatında I˙lk Mutasavvıflar, 3rd edn, Ankara: TTK. (First published in 1918.) —— (1994) Osmanlı Devkleti’nin Kuruluşu, 5th edn, Ankara: TTK. Köymen, M. A. (1954) Büyük Selçuklu I˙mparatorluğu – I˙kinci I˙mparatorluk Devri, Ankara: TTK. 204 Bibliography —— (1963) Selçuklu Devri Türk Tarihi, Ankara: Ayyıldız. —— (1966) ‘Alp-Arslan Zamanı Selçuklu Saray Teşkilâtı ve Hayatı’, AÜDTCF Tarih Araştırmaları Dergisi, Cilt V, Sayı 6–7, pp. 1–100, Ankara. —— (1967) ‘Alp-Arslan Zamanı Selçuklu Askeri Teşkilâtı’, AÜDTCF Tarih Araştırmaları Dergisi, Cilt VI, Sayı 8–9, pp. 1–74, Ankara. —— (1971a) ‘Alp-Arslan Zamanı Türk Evi’, Selçuklu Araştırmaları Dergisi, III, pp. 1–14, Ankara. —— (1971b) ‘Alp-Arslan Zamanı Türk Beslenme Sistemi’, Selçuklu Araştırmaları Dergisi, III, pp. 15–49, Ankara. —— (1971c) ‘Alp-Arslan Zamanı Türk Giyim-Kuşamı’, Selçuklu Araştırmaları Dergisi, III, pp. 51–89, Ankara. —— (1971d) ‘Selçuklu Şiirlerine Göre Türkler’in Kültür Seviyesi’, Selçuklu Araştırmaları Dergisi, III, pp. 119–43, Ankara. —— (1972) Alp-Arslan ve Zamanı, Ankara: MEB. —— (1975a) ‘Alp-Arslan Zamanı Türk Toplum Hayatı’, Selçuklu Araştırmaları Dergisi, IV, pp. 1–73, Ankara. —— (1975b) ‘Alp-Arslan Zamanı Selçuklu Kültür Müesseseleri: I. Üniversiteler’, Selçuklu Araştırmaları Dergisi, IV, pp. 75–125, Ankara. —— (1975c) ‘Alp-Arslan Zamanı Büyük Selçuklu İmparatorluğu Dini Siyaseti’, Selçuklu Araştırmaları Dergisi, IV, pp. 127–55, Ankara. —— (1976) Tuğrul Bey ve Zamanı, Ankara: MEB. —— (1979) Büyük Selçuklu I˙mparatorluğu – Kuruluş Devri, Ankara: TTK. —— (1986) ‘Selçuklular ve Anadolu’nun Türkleşmesi Meselesi’, Selçuklu Araştırmaları Merkezi, Aralık, pp. 21–35, Ankara. Krader, L. (1955) ‘Ecology of Central Asian Pastoralism’, Southwestern Journal of Anthropology, Vol. II, No. 4, Winter. Kradin, N. N. (2002) ‘Nomadism, Evolution and World-Systems: Pastoral Societies in Theories of Historical Development’, Journal of World-Systems Research, VIII, III, Fall, pp. 368–88. —— (2005) ‘From Tribal Confederation to Empire: The Evolution of the Rouran Society’, Acta Orientalia Academiae Scientarium Hung, Vol. 58, No. 2, pp. 149–69. Kuran, A. B. (1945) I˙nkılâp Tarihimiz ve Jön Türkler, Istanbul: Tan. Lamb, H. H. (1995) Climate History and the Modern World, London and New York: Routledge. Lambton, A. K. S. (1980) ‘Reflections on the Iqta’, in Theory and Practice in Medieval Persian Government, Chapter X, London: Varorium Reprints. —— (1987) Continuity and Change in Medieval Persia: Aspects of Administrative, Economic and Social History, 11th–14th Century, NY: New York. Lattimore, O. (1967) Inner Asian Frontiers of China, MA: Beacon. Leach, E. R. (1970) Political Systems of Highland Burma: A Study in Kachin Social Structure, London: Athlone. —— (1977) Rethinking Anthropology, London: Athlone. Leiser, G. (1988) A History of the Seljuqs: I˙brahim Kafesoğlu’s Interpretation and the Resulting Controversy, IL: Southern Illinois University Press. Lewis, B. (1993) The Arabs in History, 5th edn, Oxford: OUP. Ligeti, L. (1946) Bilinmeyen Asya, trans. S. Karatay, Ankara: TTK. MacDowell, S. (1995) Late Roman Cavalryman 236–565 AD, Oxford: Osprey. Maenchen-Helfen, O. J. (1973) The World of the Huns – Studies in their History and Culture, CA: University of California Press. Bibliography 205 Maisels, C. K. (1993) The Emergence of Civilization: From Hunting and Gathering to Agriculture, Cities, and the State in the Near East, London: Routledge. Mallory, J. P. (1994) In Search of the Indo-Europeans, London: Thames & Hudson. Mardin, Ş. (1992a) Jön Türklerin Siyasi Fikirleri 1895–1908, Istanbul: İletişim. —— (1992b) Türkiye’de Toplum ve Siyaset, Istanbul: İletişim. McNeill, W. H. (1989) Plagues and Peoples, NY: Anchor. Memiş, E. (1987) I˙skit’lerin Tarihi, Konya: Selçuk Üniversitesi. Merçil, E. (1980) Kirman Selçukluları, Ankara: Kültür Bakanlığı Yayınları. Morgan, D. (1994a) Medieval Persia 1040–1797, London: Longman. —— (1994b) The Mongols, London: Blackwell. Necef, Ekber N. and Berdiyev, A. A. (2003) Hazar Ötesi Türkmenleri, Istanbul: Kanküs. Németh, Gy. (1940) Attila és Hunjai, trans. Ş. Baştav, Ankara: AÜDTCF. Ocak, A. Y. (1983) Bektaşi Menâkıbnâmelerinde I˙slam Öncesi I˙nanç Motifleri, Istanbul: Enderun. —— (2002) Selçukluların Dini Siyaseti (1040–1092), Istanbul: Tarih ve Tabiat Vakfı. Öden, Z. G. (2000) ‘Anadolu Beğlikleri’nde Orduya Genel Bir Bakış’, I˙ÜEF Tarih Dergisi, No. 36, pp. 305–20, Istanbul. Ögel, B. (1962) I˙slâmiyetten Önce Türk Kültür Tarihi, Ankara: TTK. —— (1993) Türk Kültürünün Gelişme Çağları I and II, Ankara: MEB. Oman, Sir C. (1991) A History of the Art of War in the Middle Ages, Volume One: 378–1278 AD, London: Greenhill. Öngül, A. (2000) ‘Tekiş b. Alp-Arslan’, I˙ÜEF Tarih Dergisi, No. 36, pp. 321–34, Istanbul. Özaydın, A. (1990) Sultan Muhammed Tapar Devri Selçuklu Tarihi (498–511/1105–1118), Ankara: TTK. —— (2000) ‘Büyük Selçuklu Emiri Kürboğa’, I˙ÜEF Tarih Dergisi, No. 36, pp. 405–22, Istanbul. —— (2001) Sultan Berkyaruk Devri Selçuklu Tarihi (485–498/1092–1104), Istanbul. Parker, G. (ed.) (1994) The Times Atlas of World History, 4th edn, London: BCA. Ràsonyi, L. (1983) Türk Devletinin Batıdaki Vârisleri ve I˙lk Müslüman Türkler, 2nd edn, Ş. K. Seferoğlu and A. Müderrisoğlu (eds), Ankara: TKAE. Roux, J.-P. (1994) Türklerin ve Moğolların Eski Dini, trans. A. Kazancıgil, Istanbul: İşaret. —— (1995) Türklerin Tarihi – Büyük Okyanus’tan Akdeniz’e I˙ki Bin Yıl, trans. G. Üstün, Istanbul: Milliyet. —— (2001) Orta Asya – Tarih ve Uygarlık, trans. L. Arslan, Istanbul: Kabalcı. Sahlins, M. D. (1961) ‘The Segmentary Lineage: An Organization of Predatory Expansion’, American Anthropologist, Vol. 63, No. 2, April, pp. 322–45. —— (1968) Tribesmen, NJ: Prentice-Hall. —— (1978) Stone Age Economics, London: Tavistock. —— (1987) Islands of History, IL: University of Chicago Press. Sahlins, M. D. and Service, E. R. (1973) Evolution and Culture, NY: University of Michigan. Şapolyo, E. B. (1945) Mustafa Reşit Paşa ve Tanzimat Devri Tarihi, Güven. Sevim, A. (1965) Suriye Selçukluları I – Fetihten Tutuş’un Ölümüne Kadar, Ankara: AÜDTCF. —— (1983) Selçuklu-Ermeni I˙lişkileri, Ankara: TTK. —— (1989) Suriye-Filistin Selçuklu Devleti Tarihi, Ankara: TTK. —— (1990a) Anadolu Fatihi Kutalmışoğlu Süleymanşah, Ankara: TTK. —— (1990b) Ünlü Selçuklu Komutanları – Afşin, Atsız, Artuk ve Aksungur, Ankara: TTK. —— (1993) ‘Malazgirt Meydan Savaşı ve Sonuçları’, in Malazgirt Armağanı, pp. 219–30, Ankara: TTK. 206 Bibliography Sinor, D. (ed.) (1994) The Cambridge History of Early Inner Asia, Cambridge: CUP. Smail, R. C. (1995) Crusading Warfare, 1097–1193, 2nd edn, Cambridge: CUP. Sneath, D. (2007) The Headless State – Aristocratic Orders, Kinship Society, and Misrepresentations of Nomadic Inner Asia, NY: Columbia University Press. Steward, J. H. (1972) Theory of Culture Change – The Methodology of Multilinear Evolution, IL: University of Illinois Press. Strohmeier, M. (1984) Seldschukische Geschichte und türkische Geschichtswissenshaft – Die Seldschuken im Urteil Moderner Türkischer Historiker, Berlin: Schwarz. Süer, H. H. and Çakın, N. (1981) Selçuklular Döneminde Anadolu’ya Yapılan Akınlar, Ankara: Kültür Bakanlığı Yayınları. Sümer, F. (1985) Yabanlu Pazarı – Selçuklular Devrinde Milletlerarası Büyük Bir Fuar, Istanbul: TDAV. —— (1990) Selçuklular Devrinde Doğu Anadolu’da Türk Beylikleri, Ankara: TTK. —— (1992) Oğuzlar (Türkmenler): Tarihleri, Boy Teşkilatı, Destanları, Istanbul: TDAV. (First published in 1965.) Sümer, F. and Sevim, A. (1971) I˙slâm Kaynaklarına Göre Malazgirt Savaşı (Metinler ve Çeviriler), Ankara: TTK. Taşağıl, A. (1995) Gök-Türkler, Ankara: TTK. —— (1999) Gök-Türkler II, Ankara: TTK. —— (2004a) Gök-Türkler III, Ankara: TTK. —— (2004b) Çin Kayanklarına Göre Eski Türk Boyları, Ankara: TTK. Tekin, T. (1993) Hunların Dili, Ankara: Doruk. Temo, İ. (1939) I˙ttihad ve Terakki Cemiyetinin Teşekkülü ve Hidemati Vataniye I˙nkilâbı Milliye Dair Hatıratım, Istanbul: Mecidiye. Thompson, E. A. (1948) Attila and the Huns, Oxford: OUP. Timur, T. (1994) Osmanlı Kimliği, Ankara: Hil. Togan, A. Z. V. (1969) Hâtıralar, Istanbul: Hikmet. —— (1981) Umumi Türk Tarihi’ne Giriş, Istanbul: Enderun. —— (1982) Oğuz Destanı – Reşideddin Oğuznâmesi, Tercüme ve Tahlili, Istanbul: Enderun. Treadgold, W. (1995) Byzantium and Its Army 284–1081, CA: Stanford University Press. Tunaya, T. Z. (1952) Türkiyede Siyasi Partiler, Istanbul: İstanbul Üniversitesi. Turan, O. (1979) Türk Cihân Hâkimiyeti Mefküresi Tarihi, 5th edn, Istanbul: Nakış. —— (1993a) Selçuklular Târihi ve Türk-I˙slâm Medeniyeti, Istanbul: Turan Neşriyat Yurdu. —— (1993b) Selçuklular ve I˙slâmiyet, Istanbul: Boğaziçi. —— (1993c) Selçuklular Zamanında Türkiye, Istanbul: Boğaziçi. (First published in 1971.) —— (1993d) Doğu Anadolu Türk Devletleri Târihi, Istanbul: Boğaziçi. Türkiye Diyanet Vakfı I˙slâm Ansiklopedisi, Cilt 1–22, Istanbul: İSAM. Türköne, M. (1995) Eski Türk Toplumunun Cinsiyet Kültürü, Ankara: Ark. Uzunçarşılı, İ. H. (1988) Anadolu Beylikleri ve Akkoyunlu, Karakoyunlu Devletleri, Ankara: TTK. Waldron, A. (1990) The Great Wall of China: From History to Myth, Cambridge: CUP. Wilkinson, R. G. (1973) Poverty and Progress, London: Methuen. Wittfogel, K. A. (1959) Oriental Despotism, MA: University of Yale. Woods, J. E. (1993) Akkoyunlular, ed. N. Sakaoğlu and trans. S. Özbudun, Istanbul: Milliyet. Yınanç, M. H. (1944) Selçuklular Devri Türkiye Tarihi, Istanbul: İÜ. Index

Ahlat 28–9, 70, 75–9, 108, 117, 133, 157, Balasani 37, 104–6, 181 160, 162 Baverd 25, 59–60, 62 Ahmet Vefik Pasha 2 Baudouin 110, 116–17, 119 Aksungur 35–8, 40, 42, 90–4, 98–101, Baudouin II 117, 119 125, 168, 182 Bekçioğlu Afşin 29, 75–7, 89, 157, 168 al-Basasiri 15–16, 27, 29, 66–8, 154, 172 Berk-Yaruk 12, 20, 34–40, 70, 90, 96, 113, al-Muqtadi 32, 96, 142, 172–3 129, 173, 175, 180–2, 184–5; reign of al-Muqtafi 43, 130, 138 97–100, 102–9 al-Mustarshid 16, 41–3, 112, 123–5, Bertrand of St Giles 117–18 127–9, 131, 172 Bohemond 102–4, 109–10, 116 al-Mustahzir 100 Bozan 35–6, 71, 92–4, 98–9, 168, 182 al-Qa’im 15–16, 18, 26, 29, 64, 66–8, Böri-Pars 83, 104–5, 113, 182 71–3, 90, 94, 152, 171–2 Böri-Tekin 113 al-Rashid 43, 112, 130–1, 133–4 Buyids 15, 26, 32–3, 66–7, 74, 154, 157, Alexius Comnenus 86–7, 92–3, 116 169–72 Ali Tekin 24–5, 56–8, 62, 146, 150–1,174 Alp-Arslan 15–17, 20–1, 28–33, 35, 63–4, Cahun, Leon 6 68, 85, 104, 146, 152, 155–7, 160, 162, Cevdet Pasha 2 164, 167–9, 171–2, 175, 178–80, 183, Council for Higher Education 13, 186 185; reign of 70–82 Count Raymond 113 Altuncan Khatun 19, 68, 179, 184–5 Çağrı Beg 21, 23, 32, 53, 56, 58, 60–4, 81, Anatolia, conquest of 156–65; Battle of 155, 174, 179, 183 Malazgirt 159–62; Byzantine defences Çağrı and Tugrul Beg 27, 56, 146, 148, 158–9; Seljuqs of Anatolia 162–5; 150 Türkmen incursion 156–8 Arslan (also Israil, Arslan Beg or Arslan Dandankan 15, 25–6, 53, 61–3, 149, Yabgu) 18, 23–4, 52, 54–8, 145–6, 148, 151–2, 154, 172, 178, 183 150, 152–3, 164, 174 De Linné 5 Arslan-Argun 100, 104–5, 171, 180–1, 184 Dukak 38, 101–4, 110–1, 184 Artuk Beg 32, 84, 89–91, 94 Artukoğlu Sökmen 101–4, 114, 117 Er-Basgan 77–8, 85–6, 146, 183 Âsım 2, 4, 6 Azarbayjan 14, 26, 28, 30, 35, 37–8, 41, Fakhr al-Mulk 36–7, 97, 100, 104, 106, 43, 51, 60–1, 64–6, 70, 73, 77, 83, 95, 117 98–100, 108, 122, 128–30, 134–5, Fatimid 26–7, 32–3, 67, 78, 85, 88–9, 138–9, 151–3, 157, 164, 170, 181–3, 94–5, 163, 171–2 185 Ferave 25, 59–60, 62 208 Index Gevher Khatun 41, 122, 185 Mah-Melek 32, 34, 70, 96, 98, 173 Ghaznavids 9, 18, 24–5, 30, 32–3, 53, Mahmud 112; reign of 121–7 56–7, 59–62, 64, 83, 121, 131, 136, Mahmud of Ghazna 55–60 140, 148, 150–3, 155, 163, 173, 179 Mahmud of Kashgar 18, 142, 167, 176–7 Godfrey of Buillon 38, 103, 109 Malazgirt 26, 28–9, 31, 66, 70, 84, 86, 88, 93, 163–4; battle of 78–81, 156–62 Hamadan 27, 30, 36, 41–2, 60, 65, 67–8, Malik-Shah 6, 11–13, 16, 19–21, 30–8, 40, 71, 79, 81–2, 100, 106–8, 120, 122–30, 42, 70–1, 73–4, 97–8, 100, 104–5, 109, 134–6, 138–9, 155, 168, 179, 183 112–23, 119, 138, 146–7, 162–3, 165, Hasan al-Sabbah 39, 70, 94 167–9, 171, 173, 175, 178–88; reign of 81–96 Ibn Fadlan 10, 18, 140, 142, 149, 178, 184 Marwanids 32, 35, 76, 90, 99 Ibrahim Yınal 21, 26–7, 29, 53, 59, 61, Marv 25, 44, 60–3, 73–4, 81, 96, 104–5, 63–5, 67, 82, 146, 157, 171–2, 179, 127, 131–3, 136–8, 166, 173, 181 183, 185; rebellion 67–8, 153–6 Mascud 112, 119, 122–3, 125–9; reign of Il-Gazi 41, 102, 106–7, 110, 116, 119–20, 129–36 123–4 Mascud of Ghazna 24–5, 44, 58–63 Isfahan 28, 31, 34–7, 39–40, 62, 74, 79, Mengü-Pars 39, 113, 122, 130, 133 83, 85, 90–1, 93–100, 105–9, 113–16, Morton 6 128, 130, 135, 138–9, 163, 173, 179–82 Muayyad al-Mulk 37-8, 97, 100, 104–7, Ismail Gasprinski 4 113, 181 Ismacil Muntasir 23–4, 55–6, 151 Mukan (Mughan) 28, 51, 65, 70, 156, 164, 175–6 Jand 22–3, 53–6, 74, 131, 150–2 Muhammad Tapar 12, 20–1, 37–41, 97, Joscelin of Courtney 110, 117–19 137, 175, 181–6; interregnum 104–11; Joscelin II 133, 135 reign of 112–21 Musa 24–6, 39, 52–3, 56–61, 63, 74, Kafesoğlu’s view 16–17 145–6, 150–1, 157, 174, 183 Kara Khitay 43, 112, 132, 152 Mustafa Celâlettin Pasha 3 Karakhanids 16, 18, 22–5, 30, 32, 42–3, 51, 53–8, 64, 94, 127, 132, 140, 142, Nâmık Kemâl 3 150, 152, 163, 173, 176–7 Nesa 25, 59–60, 62 Katvan 15, 43–4, 55, 112–13, 131–2, 136 Nishapur 16, 25, 34, 44, 60–5, 81, 90, 104, Kavurt 28, 30–1, 63, 68, 70, 72, 76–7, 128, 132, 137, 153–4, 173, 176, 179, 85, 155–6, 163, 171, 175, 179, 183–4; 181, 183 rebellion 74–5, 81–2 Nizam al-Mulk 6, 9, 16, 19, 21, 30, 32–7, Kerükiçi Hoca 51 39, 70–4, 81–2, 90, 95–8, 100, 104, Khwarazmshah Atsız 127, 131–3, 137–8 124, 167–9, 173, 179–80, 185, 187 King Fulk of Jerusalem 133 nomadic pastoralism 143–9; herds and Kılıç Arslan I 39, 91, 110–11, 113; horses 143–4; kinship and descent crusaders 100–1, 116 146–8; nobility 148–9; pasturage and Köprülü 3, 6–7, 11; view of 7–9, 15, land ownership 144–5; the töre 148–9 17–18, 153, 165, 173, 186 Nur 23, 54–5, 73, 117, 134, 145, 150–1 Köymen’s view 15–16 Kür-Boğa 38, 40, 98–9, 106, 109, 182; Orkhun stone inscriptions 2, 4, 17–18, crusaders 101–4 48–50, 141–2, 148–9, 155, 174, 177 Kunduri 9, 19, 27–8, 68–9, 72, 179, 185 Kutalmış 21, 26–9, 31, 53, 63, 65, 70, 74, Pan-Turkism 2–4, 6 79, 84–9, 145–7, 153, 155, 163, 172, Pittard 7 175, 179, 183; rebellion 67–8, 71–2 Porsuk 31, 37, 40, 87, 92–3, 100, 105–7, 163 Lavater 5, 6 Porsukid 114, 116 Index 209 Porsuks 113, 164, 168 Tekiş 32, 81, 83, 90, 99, 113, 117, 139, Porsukoğlu Aksungur 114–16, 119, 122–3, 184 125, 184 Terken Khatun 19, 34–6, 70, 96–7, 105, Porsukoğlu Porsuk 116, 120–1, 127, 129 132, 138, 171, 173, 180–2, 184–5; interregnum 98–100 Raymond 103, 109–10 Tonyukuk 46, 48, 149, 156, 168 Rayy 27, 30, 34–8, 41–2, 59–62, 64–5, Tuğ-Tekin 73, 102–3, 111, 117–20, 126 68–9, 72, 79, 81, 83, 90, 94–5, 98–101, Tuğrul 124–5; reign of 129 106–8, 112–13, 122, 125, 127–8, 132, Tuğrul Beg 15–16, 18, 21, 24–7, 34, 39, 134–6, 139, 155, 166, 182, 185 53, 56, 60–1, 146–8, 150, 153, 155, Reşit Pasha 1 157, 172, 179, 183; reign of 63–9 Richard of Salerne 110, 116 Turan’s view 16 Ridwan 36, 38, 40, 99, 101–4, 109–10, Turkic migrations 151–3 184; crusaders 116–20 Turkish Historical Thesis 3, 6, 10 Robert of Flanders 102 Tutuş 31–2, 35–9, 70–1, 81, 83, 87–9, 94, Roger 119-20 97, 111, 180–4; interregnum 98–102, Romanos Diogenes 29, 70, 75–81, 87, 157, 104–5 159–60, 162, 164 Türkmen ‘problem’ 153–6

Samanids 23, 32–3, 54–6, 150–1, 154, Uqailids 29, 32, 35, 38, 64, 66, 90, 102, 166, 168–9 135, 163 Sancar 16, 20–1, 37–9, 41–4, 97, 105, 112–13, 148, 173, 181–2, 184–5; Yağı-Sıyan 72, 91–2, 98, 100, 114–15, interregnum 107–8; reign of 121–34, 169; crusaders 102–3 136–8 Yakuti 26, 35, 41, 63, 65–8, 73–4, 83, 96, Selçuk Beg 10, 18, 22–3, 51–2, 54, 56, 63, 98, 107, 122, 157, 162, 172, 175, 179, 145–6, 150, 174, 179 183, 185 Selçuk-Shah 42, 122, 127, 129 Young Ottomans 3 Seljuqid organization 166–80; iqtac Young Turks 4–5 169–71; legitimacy 171–4; military Yusuf Akçura 4–5, 7–8, 15, 18, 142, 166–9; succession 174–80 149–50, 186 Serakhs 25, 44, 59–62, 83, 90, 132 Yusuf Has Hajib 2, 18, 141, 177 Süleyman-Shah 31–2, 70, 85–7, 91–2, 98, Yusuf Yınal 24–5, 52, 57–9, 145–6, 100, 131, 137–9, 147, 163, 183 150–1, 174 Sümer’s view 17 Şemseddin Sâmi 3–4 Zengi 42–4, 105, 107, 113, 116, 125–30, 133–5 Taj al-Mulk 34–5, 71, 96, 98 Ziya Gökalp 2 Tancred 109–10, 116–18 Zubaydah Khatun 104–6, 129, 175, 185