sign in sign up
<<
Home , Alp Arslan, David Ayalon, Daylamites, Dirgham, Ghilman, Hamdanids (Yemen)

The Mamlūks of the Seljuks: 's Military Might at the Crossroads Author(s): Ayalon Source: Journal of the Royal Asiatic Society, Third Series, Vol. 6, No. 3 (Nov., 1996), pp. 305-333 Published by: Cambridge University Press on behalf of the Royal Asiatic Society of Great Britain and Ireland Stable URL: http://www.jstor.org/stable/25183239 Accessed: 09-04-2017 00:48 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://about.jstor.org/terms

Cambridge University Press, Royal Asiatic Society of Great Britain and Ireland are collaborating with JSTOR to digitize, preserve and extend access to Journal of the Royal Asiatic Society

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The of the Seljuks: Islam's Military Might

at the Crossroads*

DAVID AYALON

General overview

The study of the Mamluks under the Seljuks is of pivotal significance, because those Mamluks formed the essential connecting link between their predecessors in the 'Abbasid and their successors in the Sultanates of the Zangids, the Ayyubids and the Mamluks of and on the one hand, and in the Sultanate of the Ottomans on the other. They were also part of an exceptionally important stage in the ethnic transformation which those Mamluks underwent with the progress of time. The full dimensions and centrality of the factor under the Seljuks can be seen, in my view, only on the basis of the following premises, guidelines and considerations. (a) The Mamluk socio-military institution, in its various forms, had been the mainstay of Islam's military might throughout the greatest part of its existence, with the ethnic transformation just mentioned having no mean share in that phenomenon. The pagan Turks and their like formed for many centuries the backbone of the elite Mamluk armies because of their remarkably superior military qualities, according to the criteria of those times. By force of circumstances they were very gradually replaced by Mamluks of Christian (mainly Eastern) origin, and this is what gave that institution its relative adaptability to the advance of technology and, consequently, a longer lease of life. (b) During the early centuries of Islam the centre of gravity of its might moved steadily eastwards and north-eastwards, until it eventually drew its major military manpower from Central and the areas adjacent to it. (c) The preponderance of the eastern lands of Islam (including Egypt), vis-a-vis its western ones was decisive. This is true in spite of the fact that without those western lands Islam could never have become that great religion and civilization which we know today. The Seljuks and their coming from merged into that eastern part. (d) The Mamluk and the eunuch institutions became, at least from the end of the ninth century onwards, that is about a hundred years before the Seljuks set foot on Islamic soil, closely connected or even intertwined with each other. One of the reasons for that closeness

* This study is an enlarged version of my lecture which was delivered to the Royal Asiatic Society on 12 October 1995. A considerably larger one is in preparation. [Editorial note: JRAS, 1946, part 1, contained, on pp. 67?73, an article on "The plague and its effects on the Mamluk army", by David Neustadt, as he then was. This was Professor Ayalon's first published learned article in a European language. The Society is very pleased to be able to mark the fiftieth anniversary of its publication in so uniquely appropriate a way.]

JRAS, Series j, 6, 3 (1996), pp. 305~333

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 306 David Ayalon

was that the eunuchs were then already assigned the task of supervising and bringing up the young Mamluks. In this context it would be worth mentioning that the eunuchs were made part of the ruler's court, without fuss, by Caliph Mu'awiya, the founder of the , as early as the third quarter of the seventh century. The Mamluks, on the other hand, were made the major military element of the ' only in the second quarter of the ninth century, by Caliph Mu'tasim (218-27/833-42), an act which was accompanied by much publicity and severe internal struggle. From that crucial event onwards the surest way to reach sound scholarly results on both Mamluk and eunuch research is to study them together. (e) When the Seljuks entered Islamic territory towards the end of the tenth century or thereabouts, both the Mamluks and the eunuchs were already everywhere, strongly entrenched, holding key positions and extremely powerful. In order to avoid any misun derstanding, I would like to emphasize in advance that the dominant designation of "eunuch" at that time, and long before and after it, was the euphemism khddim (pi. khadam), meaning literally "servant".1 If one replaces khddim in the historical sources by khasi (= "the castrated"), quite a different history, and the correct one, will emerge. Now the Seljuks came from Central Asia into the lands of Islam as vassals of stronger Muslim rulers, and gathered strength only gradually, until they reached the peak of their power. During all that process, there is mention of no systematic attempt to suppress the Mamluks, and none whatsoever to uproot the eunuchs. If such an attempt had been made, it would have been recorded in considerable detail, and it would certainly have ended in ultimate failure, as did so many other attacks on the Mamluks that were made in the Muslim East. Furthermore, the Seljuks had no long-range reasons for eliminating the Mamluk military element. On the contrary, they had every reason to preserve and strengthen it. Otherwise, there would have been no empire, at least not the great and enduring one we know, with all the far-reaching consequences which would have resulted from that.2 Basing oneself on such premises, guidelines and considerations, one is bound to find that the dimensions of the Mamluk military establishment were immense under the Seljuks, at least from the reign of onwards, although the picture will unfold only gradually. We are still at the beginning of this process. Within the framework of those premises and within the limits of the theme of the present study, I would like to highlight some specific aspects of my general view of the Mamluk institution. I shall start by singling out two most indicative instances, completely independent of each other, which belong to a rather late period in Seljuk history. The first is that of the famous Nizam al-Mulk (died 1092); and the second is that of Caliph

1 I repeatedly dealt with this term in a number of studies of mine, and not only those confined to the subject of eunuchs. In a book which I finished recently, called Eunuchs, Caliphs and - A Study in Power Relationships, now in the Magnes Press of the Hebrew University of , I discuss that and other terms relating to eunuchs in a very detailed Appendix. Earlier studies where eunuch terminology is discussed at considerable length are: "The eunuchs in the ", Studies in Memory of Gaston Wiet (Jerusalem, 1977), pp. 267?9; "On the eunuchs in Islam", JSAI, I (1979), pp. 74-93; "On the term khddim in the sense of'eunuch' in the early Muslim sources", Arabica, XXXII (1985), pp. 289-308. 2 See also my "Aspects of the Mamluk phenomenon", part A, Der Islam, LIII (1976), pp. 209-18.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 307

al-Muqtafi (530-55/1136?60). I chose the first because it shows that that vizier's theory and practice concerning the Turkish Mamluks were in complete accord. The second was chosen because of its centrality in the history of both the 'Abbasid Caliphate and the Seljuk Sultanate (it reflects with unique clarity the attitude of the Caliphs and the Sultans to pagan Turkish Mamluks vis-a-vis Eastern Christian ones); because of its far-reaching implications beyond that history; and, last but not least, because it has been given an erroneous interpretation which has deprived it of both its purport and its focal significance. Afterwards I will present and very briefly analyze a small selection of source evidence relating to the period preceding the Seljuks, which proves the great superiority of the warlike qualities of the Turks inside and outside the Mamluk military society. This will be followed by an indication of the close relations existing between the Turks and the Islamic region bordering their homeland, and of the immense impact of that region on the Mamluks who entered through it into the lands of Islam. Then will come a case in point: a brief examination of the role of the in the East, as compared with that of the Turk. Then the gap between the pre-Seljuk evidence and that for the two opening instances will be closed, showing once again the superiority of the Turkish Mamluks at that time. This will be followed by two consecutive sections dedicated to the different kinds of ethnic changes which the two main Mamluk branches of the Seljuk tree underwent: the Ottoman on the one hand and the Zangid-Ayyubid-Mamluk on the other; and to the ultimate outcome of those different changes. Then a summary of the role of the eunuchs under the Seljuks ? an inseparable part of the Mamluk phenomenon ? will follow. The article will close with an affirmation of the unquestionably Mamluk origin of the Ottoman military slave system.

Nizam al-Mulk

He was not only the detached admirer and protagonist of the Mamluk system, dominated by the Turkish ethnic group, as it had developed under the Samanids, which he lauded so highly in his Siydsat-ndma. He himself owned thousands of Turkish Mamluks (uluf min al-Atrak), about whom nothing is said in the accounts of the chronicles until they suddenly appear in 476/1084 when the career, and even the life, of their patron is threatened. By their mere appearance they immediately remove that threat.3 Without that private Turkish Mamluk army the story of that undoubtedly great personality, who dominated the Seljuk scene for so long, would have been quite different. At that crucial moment in his career the presence of units other than the Mamluks is not even hinted at.4 What must be considered as quite certain is that the case of Nizam al-Mulk's Mamluks

3 Ibn al-Athir, al-Kamil (Beirut edition), x, p. 131. According to Sibt Ibn al-Jawzi the number of these Turkish Mamluks was more than one thousand (Mir at al-Zamdn, Ankara, 1968, pp. 227, 1. 18-225,1. 17). 4 In the obituary notices about Nizam al-Mulk the fact that he drew his power and authority from the thousands of Turkish Mamluks he possessed (kathrat mamdlikihi; wa-malaka min al-Atrdk ulufan) is clearly stated. See Ibn al-Athir, al-Tdrikh al-BdhirJt al-Dawla (, 1963), p. 10; and Abu Shama, who cites Nizam al-Mulk's contemporary or near-contemporary, the historian b. 'Abd al-Malik al-Hamdani, who died early in the sixth/twelfth century (Kitab al-Rawdatayn, Cairo, 1288H, i, p. 26). For more detail see "Aspects of the Mamluk phenomenon", pp. 213-14. The Siyasat-ndma certainly reflects Nizam al-Mulk's personal attitude to the Turkish Mamluks and the spirit of the time, even if it was edited after his death.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 308 David Ayalon

cannot be an isolated one, but part and parcel of a wider and longer-established phenomenon.

Caliph al-Muqtafi

There are three sources known to me which give an account of a matter in which the Caliph and the Seljuk Sultan were involved: al-Bundari, Ibn al-Furat and al-Husayni. Sibt Ibn al-Jawzi provides us with a much earlier account which is a most illuminating forerunner of the story of al-Muqtafi. We shall present the evidence of all four in that order. Al-Bundari: "The Caliph [al-Muqtafi], when he was made Caliph, was made to swear (or, more precisely, an oath was exacted from him), that he would not buy any Turkish Mamluk. [Therefore] he used to buy throughout his reign either an Armenian or a Greek Mamluk. Of the Turks he had only Terkesh, whom he owned before he became Caliph" (wa-kana al-imam lammd ustukhlifa ustuhlifa 'aid annahu la yashtari mamlukan Turkiyyan wa-kana yaqtani muddata khildfatihi immd Armaniyyan aw Rumiyyan wa-lam yakun lahu min al-Atrak ilia Tarshak malakahu qabla al-imdma)5 Ibn al-Furat:6 When Caliph al-Muqtafi ascended the throne in , the Seljuk Sultan Ghiyath al-Din Masud (527-42/1133-52), the ruler of and Western , made him swear that he would not buy any Turkish Mamluks, and that is why he bought throughout his reign Armenian and Greek Mamluks. Of the Turkish Mamluks he had only one, whom he bought before he became Caliph. It was Tershek. He appointed him as commander-in-chief of the Caliphal army (istahlafahu [Mas'ud] 'aid annahu la yashtari mamlukan Turkiyyan fa-kdna yaqtani muddata khildfatihi al-mamdlik al-Arman wal-Rum wa-lam yakun lahu min al-Atrak ilia Tarshak malakahu qabla al-khildfa fa-walldhu imrat [imriyyat] al-umara). He selected a group of those Armenians and Greeks, gave them priority in his service and raised them in rank. This build-up of the army was part of a wide strategic plan of the Caliph. He fortified the walls of Baghdad and deepened its ditches. He appointed governors in the districts and improved the information network which resulted in much greater safety on the roads. For his Mamluks he bought swift horses [al-khuyul al-sawabiq). He spent much money on that whole beneficial programme. He was able to carry it out fully and without hindrance because whoever remained of the Seljuk sultans were occupied with each other (ishtighdl ma baqiya min al-saldtin al-Saljuqiyya bdduhum bi-bdd). The idea behind the whole scheme was to free the tiny Caliphal state from the oppressive hand of the Seljuks. Their expected reaction did come. But owing to the thorough preparations just prescribed, the defenders of Baghdad succeeded in withstanding the heavy siege laid on the town. "And this was the end of the Seljuks' interference with Baghdad's affairs" (kdna hddhd dkhar tdarrud al-Saljuqiyya li-Baghddd) 7

5 Zubdat al-Nusra wa nukhbat al-'usra, ed. M. T. Houtsma in Recueil de textes relatifs a l'histoire des Seljoucides, ii (Leiden, 1889), p. 235. 6 The evidence of Ibn al-Furat is presented by me as a combination of literal citations, paraphrases and connecting sentences of mine. This does not affect in any way the unequivocal meaning of that evidence. 7 Ibn al-Furat, Ta'rikh, iii, MS Vienna, A.F. 119, fols. 55b, 1. io-56a, 1. 8. See also ibid., fols. 47a; 72b; 98a-i03a; 103b; 131a.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 309

Al-Husayni: "When Caliph al-Muqtafi reached the age of puberty the Seljuk Sultan Mas'ud demanded from him insistently to commit himself by oaths and agreements not to buy Turkish Mamluks, and he [al-Muqtafi] made a pledge that he would comply with this demand. Sultan Mas'ud's representatives in Baghdad perpetrated evil deeds there, most of them being against al-Muqtafi's will. When he forbade them from doing so, they did not heed him. The Sultan's deputy in Baghdad was Mas'ud al-Bilali, a stupid and brainless eunuch; irreligious; far away from being good and very near to being bad. He was doing things most of which were unlawful and contradictory to the accepted political conduct. By all that he intended to antagonize and grieve Caliph al-Muqtafi. Letters of complaint against him were sent continuously from the Caliph's court to the Sultan. Sometimes the Sultan would upbraid him [the eunuch] and sometimes he would abstain from doing so. As a result of all that a feeling of alienation and rancour filled al-Muqtafi's heart. And when Sultan Mas'ud died he set out briskly to repel the foreigners from Baghdad. He had Mamluks, part of them being Greek and part being Armenian. He made them commanders and entrusted each of them with [the rulership: defence:] of part of the land of Iraq. [The eunuch] Mas'ud al-Bilali was defeated and retreated from Baghdad" (kana al-sultdn Mas'ud lamma balagha al-imam al-Muqtafi li-amri akhadha 'alayhi al-'uhud wal-mawdthiq annahu la yataqaddam bi-shira al-ghilman al-Atrak ja-'dhadahu 'aid dhalika wa-kana ashab al-sultdn Mas'ud yatasarrafuna bi-Baghdad tasarrufat fasida aktharuha yaqa 'aid ghayri wafqi al-Muqtafi li-Amr Allah wa-rubbamd kana yanhdhum 'anha fa-Id yantahuna wa-yazjuruhum wa-la yanzajiriina wal-naib 'an al-Sultan bi-Baghdad kana Mas'ud al-Bilall khzdim sakhif al-aql wal-ray qalll al-din bald 'an al-khayr qarib min al-sharr kana yatamid ahwalan aktharuha khdrij 'an al-shar bald min rusum al-siydsa al-ma'quda yaqsid bi-dhalika Ihdsh al-imam al-Muqtafi li-Amr Allah wa-kanat al-murasalat min al-dlwan al-'azlz tatawdla Ud al-sultdn Mas'ud bil-shikaya minhu fa-taratan kana yazjuruhu 'anftlihi wa-taratan yumsik 'anhu fa-hasalafi qalb al-imam al-Muqtafi li-Amr Allah min dhalika wahsha wa-intawa Ud hiqdfa-lamma tuwuffiya al-sultdn Mas'ud tashammara li-daf al-a'ajim 'an Baghdad wa-kana lahu mamallk ba'dhum Rum wa-ba'duhum Arman fa-ja'alahum umara wa-fawwada Ud kulli wdhid minhum jdniban min jawdnib al-'Iraq wa-inhazama Mas'ud al-Bildll min Baghdad)8 Sibj Ibn al-Jawzi: In Muharram 471/July 1078 Caliph al-Muqtadi (467-87/1075-94) received from Sultan Malikshah (465-85/1072?92) a letter of demands and restrictions, one of which was that members of his inner circle, his eunuchs and his followers would not possess Turkish Mamluks within the precincts of the Caliphal palace (la yakunfiha [ft Dar al-Khildfa]ghilman Atrak lil-khass wa-la lil-khadam wal-atba).9 The three accounts dealing with al-Muqtafi are not equal in importance (by far the most important of them is that of Ibn al-Furat). But they complement each other very nicely. What is absolutely certain about the evidence of all three of them is that both the Seljuks and the 'Abbasid Caliphs considered the Turkish Mamluks to be the very best ones, with the Armenians and the Greeks definitely inferior to them (the Seljuk sultan must also have believed that these ones would not endanger his hold on Baghdad). The only Turkish Mamluk that al-Muqtafi possessed was promoted by him to the topmost post in the military hierarchy. Al-Muqtafi does not seem to have been bothered by the possibility that

8 Al-Husayni, Akhbdr al-Dawla al-Saljuqiya (Lahore, 1933), p. 129. 9 Sibt Ibn al-Jawzi (Ankara ed.), p. 196, 11. 1-2.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 310 David Ayalon a single Turkish Mamluk at the head of a Mamluk body composed of non-Turkish ethnic groups would arouse dangerous antagonism among the members of those groups. Such was the awe and respect for the Turks within the army ranks (see also below: the evidence of Ibn Hassiil). At the same time it is equally clear that the Mamluk socio-military system remained the unquestionably preferred one. The Caliph did not hesitate to rely on what were then believed to be second-rate Mamluks as the mainstay of his military might. Reliance on another, namely non-Mamluk, military element was not even considered. It was with this kind of Mamluks that he succeeded in freeing Baghdad and himself from the Seljuks. The Caliph, like any other Muslim ruler basing his power on a Mamluk kernel, had other armies in his service as well. In the final struggle over Baghdad, only the are mentioned in passing in Ibn al-Furat's account, and not in the main part of it.10 From the account of al-Husayni we learn two important things: (a) how central was the role of the eunuch representative of the Seljuk Sultan in the relations between the two courts; (b) how heavy was the pressure which the Sultan had to exercise in order to dissuade the Caliph from buying Turkish Mamluks. And this brings us to the much earlier account of Sibt Ibn al-Jawzi. What we learn from this is that the presence of Turkish Mamluks in the Caliphal court was a cause of great concern to the Seljuk rulers, and that that concern was shared by both the Great Seljuk (Malikshah) and the smaller one (Mas'ud) who had Baghdad under their sway. The combined evidence of the four accounts also shows us that that problem consti tuted a long-standing issue of no small consequence between the two courts for at least fifty-eight years. It stands to reason as well that a controversy about it arose more frequently than the two widely separated occasions on which it surfaced in the sources. What can also be deduced from the accounts under discussion is that at first the Seljuks tried to erode the power of the Turkish Mamluk body in the Caliphal court by attacking its fringes (namely, by eliminating that part of it which was owned by the Caliph's entourage or retinue), and only at the end did they succeed in liquidating the part belonging directly to the Caliph himself.11 Finally, it would be most interesting to know how the military success achieved by "second-rate" Mamluks in such an important struggle affected their future recruitment. It may be that further research will provide an answer to this question.

10 Ibn al-Furat, Tdrikh, iii, fol. 72b. 11 C. E. Bosworth has made very important contributions to the study of the Turks and the Turks' recruitment as Mamluks during the first half millennium of Islam's history. He was also the first to call attention to crucial evidence relating to those subjects. Yet I have a fundamentally different view than his about a number of major issues regarding the Mamluk institution. In the present context I cannot share his opinion about Caliph al-Muqtafi's Mamluk policy. According to him that Caliph "disliked Turks" (my italics - D. A.) and therefore "merely recruited Greeks and Armenians instead", in order to preserve the Mamluk character of his army ("Ghaznevid military organisation", Der Islam, XXXVI (1961), p. 43; The (Edinburgh, 1963), p. 100). All the sources, including the two he cites (al-Bundari and al-Husayni) appear to tell us something different. I have already stressed the unique importance of the al-Muqtafi story in my article "Egypt as a dominant factor in Syria and during the Islamic period", Egypt and Palestine ? A Millennium of Association, Jerusalem, 1984, p. 30, n. 25. But at that time I was not aware of Bosworth's earlier mentioning and interpretation of it. For my disagreement with him about other essential parts of his view of the Mamluks, see "Aspects of the Mamluk phenomenon", part A, p. 206, n. 18; "On the eunuchs in Islam", pp. 94, 100, 105. Since, however, all these parts belong to one whole (which can be epitomised in Bosworth's own words: "That any of the nations customarily supplying slaves were conspicuously superior to the rest is problematical" - see the two references cited in this note), I shall discuss them, together with other parts, in the fuller version of this study. There any points of agreement will also be referred to.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 311

The Pre-Seljuk Evidence

What is significant and unique about the Turks is that the unparalleled number of highly appreciative statements about their warlike ability and the quite consistent policy of the Muslim rulers in enlisting them militarily are in full agreement. No other ethnic group could pride itself with anything resembling even remotely that combination of those two elements.12 The innumerable statements about the fighting qualities of the Turks and the Turkish Mamluks still need to be collected more systematically than has been done up till now. Of the numerous ones which I as well as other Islamicists have already collected, I shall cite only very few, which seem to me to be particularly relevant to our subject. I do not think that one can gather evidence about any other ethnic group in the Muslim sources which will even vaguely resemble the few instances presented here. Here are those statements: I. Speaking of the Turkish Mamluks of the 'Abbasid Caliphs, al-Istakhri says: "And the Turks constituted their [the Caliphs'] armies because of their superiority over the other races in prowess, valour, courage and intrepidity" (wa-kanat al-Atrak juyushahum li-fadlihim 'aid sair al-ajnasft al-bas wal-jara wal-shajaa wal-iqddm)P II. Speaking of the same Turkish Mamluks of the same 'Abbasid Caliphs, a considerably earlier authority, al-Jahiz, says: "They became to Islam a source of reinforcement and an enormous army, and to the Caliphs a protection and a shelter and an invulnerable armour as well as an innermost garment worn under the upper garment" (fa-sdru lil-Islam maddatan wa-jundan kathlfan wa-lil-khulafa wiqdyatan wa-mawilan wa-junnatan haslnatan wa-shi'aran duna al-dithar14) ,15 III. The same al-Jahiz furnishes us with a most instructive instance of which he was an eyewitness. Not less instructive is his comment on it. He says: "I [i.e. al-Jahiz] tell you that I witnessed from them [i.e. from the Turks] something wonderful and extraordinary. I saw in one of al-Ma'mun's campaigns (ghazawdt) two lines of horsemen on both sides of the road near the halting place [at which the Caliph and his army intended to camp]. The line on the right-hand side of the road was composed of one hundred Turkish horsemen. The line of the left-hand side of the road was composed of one hundred horsemen of 'others' (min sair al-nds). All were arrayed in battle-order, awaiting the arrival of al-Ma'mun. [But the Caliph seems to have been delayed on his way and meanwhile] it was midday and the heat became intense. When al-Ma'mun reached the place, he found all the Turks sitting on

12 The only parallel which I can find (and this is a partial one) is the repeated praise, expressed grudgingly, which the and Franks bestow on each other's military prowess during the Crusading period. But that is beyond the present subject. 13 Al-masalik wal-mamdlik (Leiden, 1927), p. 291, 11. 16-17. This statement was already cited in my The Military Reforms of Caliph al-Mu'tasim - Their Background and Consequences, which appeared in the form of a stencilled brochure (Jerusalem, 1963 - reprinted, with changed pagination, in Islam and the Abode of War, Variorum (London, 1994), article I). The reference here (p. 25) and throughout the present study is to the Variorum pagination. Ibn Hawqal expresses the same idea in different words (Surat al-Ard, Leiden, 1939, p. 448, H.9-10). 14 This is a very high distinction. In a tradition relating to the Medinan supporters of the Prophet Muhammad (the Ansdr) they are addressed: antum al-shi'dr wal-nds al-dithdr, which Lane translates: "Ye are the special and close friends [and the people in general are less near in friendship]". It is noteworthy that shi'dr means also a coat of mail [worn under the garment]. See Taj al-Ariis under dir. 15 Mandqib a\-Atrdk, ed. van Vloten (Leiden, 1903), p. 49, 11. 7-8. "Military reforms", p. 26.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 312 David Ayalon

the backs of their horses, with the exception of three or four, whereas 'all that medley1 (jdmi' tilka al-akhldt) have thrown themselves on the ground, with the exception of three or four. [Remembering that episode] I said to a friend of mine: see what happened to us! I swear that al-Mu'tasim knew them better than we did when he gathered them and fostered them (fa-qultu li-sdhib liunzur ayy shay ittafaqa land ashhadu anna al-Mu'tasim kdna a'rafbihim hina jama'ahum wa-istana' ahum) }6 IV. Al-Ya'qubi, in his account of the building of Samarra, which forms the basis of our study of that capital of Islam, tells us that al-Mu'tasim, after a long search for a suitable place, reached the desirable locality, where he found only a Christian convent. He alighted there and asked its monks about the place. They told him that it is said in their old books that its name is surra man rdd and that in the distant future (ba'da al-duhur) it will be built by a great and victorious king, whose companions' (ashdb)17 faces will be like those of the birds of the desert (wujuhuhum wujuh tayr al-faldt), and that he and his children will inhabit it. Al-Mu'tasim retorted: "By ! It is I who will build it and inhabit it and my children will inhabit it as well. One day [my father Harun] al-Rashid ordered his sons to go out hunting. I went with [my brothers] Muhammad [al-Amin] and al-Ma'mun, together with other adult sons of his. Each one of us caught his quarry. Mine was an owl (buma). Then we went back and displayed our game in front of him [i.e. Harun al-Rashid]. The eunuchs (khadam) who were with us started telling him: 'This is the game of so and so', until they presented my game. Actually, they were afraid to display it, because they thought he might consider it to be an evil omen, or that he would treat me harshly. Harun al-Rashid asked: 'Who caught this?' they answered: 'Abu Ishaq [al-Mu'tasim]'. He was delighted and laughed and then he said: 'This one will become Caliph and his army and his companions and those who will prevail on him will be a people whose faces are like the face of that owl' (innahu yali al-khildfa wa-yakun junduhu wa-ashdbuhu wal-ghdlibuna 'alayhi qawman wujilhahum mith la wajhi hddhihi al-buma). Then his descendants will inhabit it.' Nothing of the game captured on that day made him so happy as he was with my catching of that owl".18 V. Ibn Hawqal: "The most precious slaves are those arriving [in Khurasan]19 from the lands of the Turks. There is no equal to the Turkish slaves among all the slaves of the earth".20 VI. Al-Istakhri, speaking of Transoxania (md ward' al-nahr) says: "As for the slaves, they arrive to them [the people of Transoxania] from among the Turks who surround them [in quantities which are] beyond their need (ma yafdul 'an kifdyatihim). From their country [i.e. Transoxania] they are transported to the remotest parts of the earth (yunqal ild al-dfdq min bilddihim). These are the best slaves [from the region] which surrounds the whole East (wa-huwa khayr raqiq yuhitu bil mashriq kullihi)".21 VII. Al-Istakhri: "Samarqand. Its people are distinguished by their dazzling beauty and sedateness. They are excessive in demonstrating their manliness and in defending them

16 Ibid., p. 38, 11. 21-39, 1. 8. 17 Quite often ashab fulan, relating to a ruler or a commander, means his army or his armed followers. See also below, in the same passage (no. IV). 18 Al-Ya'qubi, Kitab al-Bulddn (Leiden, 1892), p. 257, 11. 4-21. 19 In the Muslim sources Khurasan sometimes includes Transoxania. See e.g. passage no. VII. 20 Ibn Hawqal, p. 452. 21 Istakhri, p. 288, 11. 5-18. See also Ibn Hawqal, p. 465, 11. 6-8.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 313

selves [against the enemy?] more than any other people in Khurasan, which takes a heavy toll of their wealth (hatta yujhif dhalika bi-amwalihim). Samarqand is the slaves' place of assembly in Transoxania. The best slaves in Transoxania are those brought up in Samarqand" (wa-bi-Samarqand majmd raqlq ma wara al-nahr wa-khayr al-raqlq bi-md ward' al-nahr tarbiyat Samarqand).22 VIII. Al-Istakhri on the Samanid army: "In all the lands of Islam there is no king who is stronger and who possesses more military equipment and who is more suitable to rule than they [the Samanids] are. Because in all those lands of Islam the armies are composed of odd people belonging to all kinds of tribes, countries and remote regions (shudhdhadh al-qabail wal-buldan wal-atrdf). If they disperse as a result of defeat or break up as a result of some event, they never gather again. This is not the case with these [Samanid] kings. For their armies, both Turkish Mamluks and freemen (juyushuhum al-Atrak al-Mamlukun wa-min al-ahrar) [are different]. The home and place [of each soldier] is known. If any number of them is killed or dies, their ranks can be filled because of their numerousness. If they disperse as a result of some event, all of them will gather again in one place. Thus they cannot be blamed for what the other rabble armies are blamed for. It is impossible for them to be scattered among other armies and move from one kingdom to another, as does usually happen to the vagabond armies and their like. In the 'Abdallah b. al-Mu'tazz ["the one day Caliph"] incident, Bars, the Mamluk (ghulam) of [the Samanid ruler] Ismail b. fled from his patron [and went to Baghdad]. The Caliphate's [court] was astounded by the number of his soldiers, their equipment and their accompanying animals. In the Caliphal capital there was not such an army (wa-lam yakun bi-hadrat al-khilafa jaysh mithlahu). And after all [this Bars] was only a slave of the Samanids. The people of Khurasan did not notice his absence".23 IX. Al-Muqaddasi on the Samanids and their army: "This is the greatest and mightiest fortress of Islam. ... Its army is the best of all armies. ... its people are very courageous. It is the rampart against the Turk and the shield against the Ghuzz" (huwa hisn al-Islam al-dzam. . . . junduhu khayr al-junud wa-qawmuhu ulu bds shadld. . . . huwa sadd al-Turk wa-turs al-Ghuzz).24 X. Speaking about the taxes under the Samanids, al-Muqaddasi states: "As for the taxes [there], they are light, but they make them heavier on the Oxus bank as far as slaves are concerned. They do not let a Mamluk pass without an official permit from the ruler. They levy together with the permit between seventy and one hundred dirhams. The same goes for the slave-girls: [they do not let them pass] if they are Turks. On every woman they levy twenty to thirty dirhams" (wa-amma al-daraib fa-hayyina wa-yus'ibuna bi-hdfat Jayhun fl al-raqlq wa-la yu abbiriina ghulaman ilia bi-jawaz min al-sultdn25 wa-ydkhudhuna ma a al-jawaz sab'lna Ud mi afa-kadhalika 'aid al-jawarl[la yu abbirunaha] bi-lajawaz idhd Atrdkan wa-ydkhud huna 'aid al-mara 'ishrlna Ud thalahlna dirhaman)26 In another manuscript of the same opus it is

22 Istakhri, p. 318, 11. 14-17. 23 Ibid., pp. 292-3. 24 Al-Muqaddasi, Ahsan al-TaqdsimfiMdrifat al-Aqdlim (Leiden, 1906), p. 260,11. 1-11. See also my "Egypt as a dominant factor . . .", pp. 26-7, and p. 26, n. 19. 25 Sultan in the second half of the fourth/tenth century does not yet mean unequivocally Sultan. Loosely speaking, it is the equivalent of "the Higher Authority". 26 Al-Muqaddasi, p. 340,11. 12-17.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 314 David Ayalon

said: "The taxes are light. The heaviest are on the Oxus. No Turkish Mamluk is allowed to pass without an [official] permit. Seventy dirhams are levied on him" (wa-ammd al-daraib bihi khafifa athqaluhd 'aldjayhun wa-ld ya'bur ghuldm Turki ilia bi-jawdz wa-yukhadh 'alayhi sab'una dirhaman).27 XI. Ibn Hassul, after having lavished praises on the Turks in general, goes on praising the Turkish Mamluks in particular. He says: "One of the indications of the noble character of the Turks and of the intensity of their ambitions [is the following fact]: Islam enjoined [the Muslims] to carry out campaigns against the infidels among them [i.e. among the Turks], as it enjoined them to do the same thing against the infidels of the rest of the races in the rest of the countries. And behold! Any one of them [of the Turks] who is taken captive is not content with less than a status of equality with his patron in [whatever relates to] food, drink, dress and mounts (fa-man subiya minhum lam yarda ilia bi-an yusdwiyahu sayyiduhu fi mat'amihi wa-mashrabihi wa-malbasihi wa-markabihi). He does not have to do menial work which members of other infidel races have to do when they become slaves or are taken captive, like sweeping the house or tending animals. And we have never seen an Indian or Rumi or Armani or any other one of all the kinds of people who were made slaves [in Islam], the limits of whose ambitions are not known, and the size of which cannot be measured. Whereas the Turk, when he is freed from the shackles of (idhd kharaja min wathdqihi)28, is not content with less than the command of an army, or a very high office, or leadership of a group. And this phenomenon is not confined to Khurasan,29 which borders their [the Turks'] lands, and adjoins (muldsiqa) them, but you encounter the same thing, if you like it [ = if you want to have additional examples], in Egypt, the country which is the remotest from their [the Turks'] homelands, and which is ignorant of their affairs and of their language; and the same goes, if you like it, for Iraq, where there are various armies, like the Daylamites, the Kurds and the . The Caliphs and the Daylamite kings [i.e. the Buwayhids] tried, generation after generation, to harm and weaken them, but did not succeed.30 As for their high value, their bravery and their might, suffice it to mention what the Prophet Muhammad said about them: 'Make truce with the Turks as long as they leave you alone' (tdriku al-Turk ma tarakukum)". A few lines later he continues: "The most amazing thing about them [the Turks] is that nobody had ever seen a pure Turk (Turkiyyan khdlisan) who had been afflicted by effemination (takhnith), and this in spite of the fact that this disgrace is general, and this affliction is common among all the peoples we saw, especially those of Gilan. If, however, one does find traces of effemination (ta'nith) in any Turk in his speech, in his hints, in his dress or in his jewellery, he is proved to be one of the Turks of a mixed breed, who thoroughly mingled with the race of their neighbours, the local inhabitants of those lands [= the lands of Islam]" (fa-innahu 'an muwalladi al-Atrak alladhina sardfxhim 'irq al-mujdwirin lahum min ndbitat hddhihi al-diydr). Here our author turns to another subject with these opening words: "After having

27 Ibid., p. 341, note. 28 This statement about the of the Mamluk before he started his great career is of great weight, because it constitutes unconscious evidence and because it is so general. The central problem of manumitted Mamluks and those of them who were lifelong slaves even when they rose in rank and office has still to be studied systematically. 29 Here is yet another instance of Khurasan in the wider sense (including Transoxania). 30 This is a free translation.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 315

referred to some of the merits of the Turks: about which there cannot be any objection or controversy, we shall return to the subject of their kings, who were either from the selfsame kings of Khurasan of the house of Saman [and others] and Sabuktikin and his sons and descendants, like Mahmud, Muhammad and Mas'ud, and how they overcame their adversaries and attained their aims, relying on their own ability".31 XII. Mahmud al-Kashghari: "God said: I have an army whom I called Turk. . . . And this is a great advantage of theirs over the rest of humanity. . . . He [God] settled them in the highest place with the best climate of the whole earth, and He called them His own army (sammahum junda nafsihi). This on top of their beauty and their gracefulness, their good manners and respect for the elder, their keeping their pledge, their avoidance of conceit and their bravery, as well as other innumerable praiseworthy qualities".32 This very partial list, which goes up to about the middle of the eleventh century, could easily be continued up to the end of the fourteenth and beyond. It culminates in Ibn Khaldiin's detailed and most appreciative evaluation of the Turkish Mamluk institution since its inception and up to his own time, an evaluation which is, in my view, of unequalled importance.33 The twelve passages quoted above speak for themselves. They will be very briefly analysed in this section as well as in the next to the following one, and not in the order of their appearance. But before starting on that analysis it should be pointed out that the Mamluk regiment of al-Mu'tasim was composed of Turks, as testified not only by al-Jahiz (passages II, III), but also by all the other sources speaking about it and about the building of Samarra. From al-Ya'qubi, our best source on this subject, we learn that al-Mu'tasim had collected, already in the reign of his brother al-Ma'mun, 3000 Turkish Mamluks as a preliminary to his grand scheme.34 Thenceforward, the Turkish Mamluk element remained the dominant one under the 'Abbasids, in spite of the numerous internal struggles and ups and downs in their history.35 The evidence of al-Istakhri (passages I, V, VI, VIII) covers a good number of central aspects, which will be referred to here and in the next to the following section. He points up, in the most unequivocal language, the military superiority of the Turks over all the other ethnic groups, as the sole reason for their forming the [main] armed forces of the 'Abbasids (passage I). The earlier al-Jahiz holds exactly the same view about the Turks in general (passage II). At the same time he gives most illuminating eyewitness evidence, which he

31 Ibn Hassul, pp. 41, 1. 8?43, 1. 15. On this author and his work cited here see below. 32 Diwdn Lughat al-Turk (Ankara, 1939-46), i, pp. 293, 1. 13-294, 1. 15. 33 See e.g. the list quoted and discussed in my "The Great Ydsa of Chingiz Khan . . .", SI, XXXVI (1972), pp. 117?30. The translation of the whole passage from Ibn Khaldun's history, Kitab al-Ibar, and its analysis were published in my "Ibn Khaldun's view of the Mamluk phenomenon", JSAI, II (1980), pp. 340?9. For my acquaintance with and treatment of that piece of evidence see ibid., p. 341. The moral qualities of the Turks, mentioned by Mahmud Kashghari (passage XII), are often praised by other sources as well. See e.g. notes 4 and 5. 34 Al-Ya'qubl, pp. 255, 1. 19-256, 1. 1; 262, 1. 19; Tabari (series III), ii, p. 1180, 1. 17; al-Mas'udi, Muruj (Paris, 1861-77), vii, pp. 118, 1. 5; 122, 1. 3. 35 Al-Mu'tasim had three great drawbacks in forming his Turkish Mamluk regiment: (a) his reign was short (218-27/833-42); (b) he started building Samarra several years after he became Caliph; (c) that regiment came into being on the verge of the great disintegration of the 'Abbasid empire. Thus the Mamluks could never be used as part of a general strategy conducted from Islam's central seat of government. Furthermore, the Caliphs who succeeded al-Mu'tasim were not of his calibre, and did not possess the vision for properly following up his great work.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 316 David Ayalon

accompanies with a highly convincing explanation of why al-Mu'tasim preferred the Turks when he formed his Mamluk regiment (passage III; see also al-Kashghari, passage XII). Al-Istakhri, however, puts into bold relief the disadvantage of the 'Abbasids, because of their lack of direct access to the sources of supply of the Turkish Mamluks (passage VIII). Already at a relatively early period, when they were still powerful, the 'Abbasids needed the cooperation of the , Saffarids and Samanids for ensuring that supply. With the weakening of the 'Abbasids, rulers who had better access to the Mamluk homelands, like the Samanids and the Ghaznavids, became much stronger than the central government. Both al-Istakhri and the later al-Muqaddasi (passage X) consider the Samanids, with their Turkish Mamluks, to be the strongest in Islam, militarily. As for al-Muqaddasi, it is quite indicative that while speaking so highly of the Samanid army, he says nothing comparable about the army of the Fatimid empire, which was then at the peak of its greatness, and towards which he showed a growing sympathy.36 From the detailed account of al-Istakhri some meaningful matters emerge, after trimming his obvious exaggerations: the other armies were not as worthless as he describes them, and the story about the Samanid commander Bars might well have been an invention. But the great difference between the might of the Turkish Mamluk army of the Samanids and the Caliphal one, as reflected in that story, seems to me to be basically correct over a certain period. What is also true in al-Istakhri's account is that a strong Mamluk army radiates authority beyond its limits. Besides such an army a military body of free men will become more disciplined and soldierly. From al-Istakhri (passage V) and al-Muqaddasi (passage X) we learn that the Transox anians tended to sell to other countries only the surplus of the numerous Turkish Mamluks who arrived in their country, and even those could pass westwards only with a special permit coupled with a tax payment, clearly implying that non-Turkish slaves were not subject to that procedure. This gave them absolute control of the Turkish slave traffic, and enabled them to choose for themselves the very best, leaving for the neighbouring Muslim countries Mamluks of lower quality. I do not know of any other slaves coming from any part of the Abode of War, and destined for a military career, who underwent such a sifting process as the Turks. The existence of Muslim buffer states between the source of supply and the source of demand was one of the greatest drawbacks of the whole Mamluk system. It was eliminated thoroughly and over a long period only by the Ottomans, who built their slave army on other ethnic groups much more accessible to them and more adaptable to the changes which the Art of War underwent. I brought in al-Mu'tasim's story about his catching an owl while hunting (passage IV) because it seems to me that it reflects correctly the great, and justified, trust which the 'Abbasid Caliphs put in their Turkish Mamluk army, against quite stubborn opposition. Passage no. XI needs some introduction, because of its being, in my view, particularly indicative, and because its author (Abu al-Ala' Muhammad b. 'All b. Hassul, died 450/1058) is much less known than the other authors cited here. He was an Iranian born in Hamadan, who first served the Buwayhid Majd al-Dawla of Rayy. When Mahmud of Ghazna captured Rayy he employed him. By the time of the Seljuk invasion he headed the

36 See "Egypt as a dominant factor . . .", p. 27 and n. 22.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 317

chancellery office (diwdn al-rasail) in that town and was then employed by Tughril, the Great Seljuk.37 Al-Tha'alibi, in his Tatimmat al-Yatima speaks highly of him, and states that he learnt much of him. He wrote a booklet, dedicated to Sultan Tughril, which he called The Preference [or Advantage or Superiority] of the Turks over the Rest of the Armies (Tafdil al-Atrak 'aid Sd'ir al-Ajndd). The purpose of the booklet was to refute an earlier work, which has not come down to us, by Abu Ishaq b. Ibrahim b. Hilal al-Sabi (died 384/994), dedicated to the famous 'Adud al-Dawla, the Shi'ite-Daylamite-Buwayhid ruler, and called Fi Akhbdr al-Dawla al-Daylamiyya. A single manuscript of Ibn Hassul's work was found in a private library in Baghdad, and was published in 1940 with a Turkish translation.38 This is, of course, a partisan text by its very character.39 However, since it was dedicated to Tughril, what is said there must have been in keeping with Tughril's own ideas and policy relating to the Turkish Mamluks, which is verified by what we know from other evidence, part of which is presented here (see below). Secondly, his service to three regimes, two of them Turkish and one Buwayhid, put him in a very good position for forming a view about their armies. Thirdly, some important parts of what he says are in full conformity with what we know from other sources. Furthermore, some aspects on which he sheds light, and which do not emerge as clearly and as forcefully in other sources, seem to me to be most logical and convincing. It is mainly these aspects which will occupy us in the following lines. Ibn Hassul's remark on the behaviour of the Turkish Mamluks in the rulers' courts as compared with that of other Mamluk ethnic groups is both revealing and makes sense. In fact, it could not be otherwise. At that time the Turkish Mamluks had already for two centuries formed the cream of the Mamluk armies. A Turkish Mamluk newcomer to the ruler's court knew exactly what awaited him there. This must have been a matter of common knowledge in his homeland, at least through the constant contact created by the slave dealers going back and forth between there and the lands of Islam. But that was by no means all. That newcomer, provided with that knowledge, did not have to fend for himself in establishing his superior status. In the court he would find numerous Mamluks of his own stock, firmly established in influential positions and having the strongest possible interest in preserving Turkish dominance. There was no chance whatever that this formidable structure would have been weakened, to say nothing of being replaced, by any other military body in the service of the ruler. The three ethnicities of Mamluks which Ibn Hassul mentions by name as inferior to the Turkish Mamluks (the Indians, the Greeks and the Armenians) are very relevant to the thesis presented here. The Indians remained, even at the height of their presence in the armies of Islam, mainly an Eastern regional factor, destined to disappear, or at least to

37 Bosworth, The Ghaznavids, p. 59. 38 Belleten (Ankara), IV. The part in is at the end of the volume, pp. 1-51. It includes an introduction by 'Abbas al-'Azzawi (pp. 3-24), and the text itself is on pp. 24-51. 39 When the 'Abbasid Caliphs were under Buwayhid suzerainty their Mamluk policy was not free from the pressure of that Daylamite dynasty. The success of the Daylamite soldiers in so many eastern armies is really astounding, but in the slightly longer run they stood no chance against the Turkish Mamluks, and they disappeared. Even a Buwayhid ruler preferred Turks over them (see e.g. Bosworth, "Ghaznevid military organisation", p. 42; Ayalon, "On the eunuchs . . .", p. 109).

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 318 David Ayalon

decline sharply in importance. They certainly could not serve as a real constant counterba lance to the Turkish Mamluks.40 Within the framework of our thesis the case of the Armenians and the Greeks is far more consequential. The immense gap separating them from the Turkish Mamluks made it extremely difficult - nay, practically impossible - for their patrons to put them, of their own free will, on an equal footing with them, let alone replacing them. Only the pressure of external factors could bring about such a change: a firm demand of a stronger authority on whom the patron is dependent (as in the case of Caliph al-Muqtafi), or much wider and uncontrollable factors mentioned in other parts of this study. And it should always be remembered that those Greeks and Armenians formed the essential connecting link in the Mamluk ethnic transformation which prolonged the life of that institution as an efficient military tool in the service of Islam. Ibn Hassul's affirmation about the Turkish Mamluk ascendancy (accompanied by a conviction of superiority) in the wide area stretching from the borders of their homelands in the north-east to Egypt in the south-west, could not be more correct. The Khurasan part of that affirmation will be referred to in the next to following section. Here only the remotest and most problematic country at that time, Fatimid Egypt, will be briefly discussed. It would appear that even there, where the Mamluk system was so seriously curbed,41 the military superiority of the Mamluks of that ethnicity was recognised. Caliph al-'Aziz (365?86/975-96), the only ruler of that dynasty who made a concentrated, and temporarily successful, attempt to recruit them, is stated to have preferred them (especially their Hamdanid component) over the Maghribis, because of their fortitude and reliability (wa-kana yamil ild al-Atrak akthar min al-Maghdriba la siyyamd al-Hamddniyya42 li-shiddat basihim wa-fadl al-najdafihim).43

40 "Aspects of the Mamluk phenomenon", part A, pp. 202-3; 206, n. 18. 41 I have already pointed out, in "Egypt as a dominant factor . . .", pp. 30?1, some of the reasons which prevented the Mamluks from taking real root under the Fatimids. There I mentioned especially the fact that, on moving from the Maghrib to Egypt, much of their ruling and military systems had already been crystallized, and they could not replace, or at least replace properly, the already existing and most powerful military bodies. Another factor which I referred to there was the great distance separating Egypt from the Turkish Mamluks' homeland, which was, in addition to that, cut off from Egypt by mainly Sunnite states hostile to the Fatimids. (The antagonism of the Seljuks to the acquisition of Turkish Mamluks by the Sunnite al-Muqtafi must have been at least as strong in the case of the Fatimids. This, coupled with the much greater distance of Egypt from the sources of supply, illustrates well the obstacles which the Fatimids faced in forming a proper Mamluk army.) Yet another major factor was the very inadequate military schools of the Fatimids, when they had them (I discuss this aspect in my book on the eunuchs). The impossibility of the Mamluk system (with the Turkish Mamluks at its heart) developing under the Fatimids as it should, was clearly demonstrated in the second half of the fourth/tenth century, as soon as its great protagonist al-'Aziz billah died (386/996), that is to say, quite soon after their occupation of Egypt. That was one of the reasons why their hold on Syria was even then not strong enough. With the later appearance of the Seljuks and the Crusaders that hold, without a further military force, could not but become more precarious. Note also the military feats of the Turkish Mamluk contingent which sent in 568/1172-3 to the Maghrib ("Aspects of the Mamluk phenomenon", part B, pp. 12-13). 42 The Turkish Mamluks of the , who were incorporated in the Fatimid army, and who seem to have been considered as seasoned soldiers, because of their combat experience. 43 Al-Maqrizi, al-Muqaffd, ed. Zakkar (Beirut, 1973), p. 307 (cited from Yaacov Lev's important article, "The Fatimid army, A.H. 358-427/968-1036", Asian and African Studies, XIV (1980), pp. 171-2. This citation is from the quite late historian al-Maqrizi, but there is good reason for accepting it. Al-Maqrizi copied, on the whole correctly, from earlier authors, showing quite often little inclination to mention their names (see e.g. my "The Great Yasa of Chingiz Khan - a reexamination", part C2, "Al-Maqrizi's passage on the Ydsa under the Mamluks", SI, XXXVIII (1973), pp. 107?56). His writings about the Fatimids are invaluable. Where would we have been in the study of that dynasty without his chronicle Ittiaz al-Hunafa and his topographical work al-Khitat?

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 319

Also worthy of particular note is Ibn Hassul's emphasis on the unequalled manliness of the Turks. As far as this implies that only men born in the steppe or a similar area, and who spent their childhood and early boyhood there, could serve as worthy Mamluks, whereas their sons, born and brought up in an urban environment, could not, it is in keeping with the inevitable and correct policy of barring the sons from joining their fathers in the Mamluk elite (a policy justified also by Ibn Khaldun). But Ibn Hassul's remark goes further than that. It suggests the idea that the Mamluks, especially the Turks, should be segregated from all the rest in order to preserve their ethnic purity. This was, in fact, the policy of al-Mu'tasim of segregating his Mamluks in Samarra, so that they would not have any contact with the same muwalladun mentioned by Ibn Hassul.44 Elements of that approach lasted long in Mamluk society, but certainly not with the same degree of absoluteness.

The ?aqaliba vis-a-vis the Turk

The Saqdliba are relevant to the thesis presented here, because they serve as yet another decisive proof of the military superiority of the Turks and of the fact that no other ethnic group could compete with them for a long time, and certainly not replace them. This subject will be elaborated at considerable length in the full version of this study. Here only a few salient points will be mentioned. As I have shown in a previous study,45 Dozy had already in i860 brought to the knowledge of students of Islam a passage of the highest importance on the Saqdliba from the geographer Ibn Hawqal, but he severely misrepresented its contents. However, that misrepresentation was copied by later Islamicists who did not check Ibn Hawqal's evidence. My main arguments against Dozy, as expressed in the above-cited study, are these: (a) That he referred only to one part of the passage, and ignored its other part, which is essential for the understanding of its whole purport. (b) That thaj: part to which he did relate says precisely the contrary of what he attributed to it. (c) He confused later and earlier occurrences, thus creating an impossible chronological problem. I shall start with argument (b). Dozy states that according to Ibn Hawqal the Saqdliba in the service of the Umayyad Caliphs of Spain included Galicians, Franks [French and German], Lombards and Cala brians.46 In fact, what that geographer says, in the clearest possible terms, is that all these

44 See "The military reforms of Caliph al-Mu'tasim", pp. 28?9; "Preliminary remarks on the Mamluk military institution in Islam", in V. J. Parry and M. E. Yapp (eds), War, Technology and Society in the (London, 1975), pp. 54-5 (for the relevant passage in Arabic see al-Ya'qubi, pp. 285, 1. 15-259, 1. 10). A strong feeling of affinity, of belonging to "the same race" (jinsiyya, minjins wahid) existed between the Turkish Mamluks and the non-Mamluk Turks (or Turcomans) and the Mongols, all of whom entered Muslim territory through Transox ania and its vicinity from the same wide region, and adopted Islam immediately or ultimately. Yet the barrier separating Mamluks and non-Mamluks of the same privileged stock was never surmounted (see e.g. my "The European-Asiatic steppe: a major reservoir of power for the Islamic world", Proceedings of the 25th Congress of Orientalists (Moscow, 1963), pp. 47-52; "The Wafidiyya in the Mamluk kingdom", Islamic Culture (1951), pp. 89-104). 45 "On the eunuchs in Islam", part B; "Remarks on the Saqaliba within Dar al-Islam", JSAI, I (1979), pp. 92?124, and especially pp. 92?101. 46 Ibid., p. 93.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 320 David Ayalon

Western Europeans were one thing, and the Saqdliba were quite another. Furthermore, Ibn Hawqal attaches to his book a map of the world of his own drawing, on which all those Western Europeans are shown as being clearly distinct from the Saqdliba.47 At the same time I repeatedly stressed that I have no suggestion of my own concerning the ethnic composition of the Saqdliba, and that my main demand is that Ibn Hawqal's evidence should be excluded as a proof (certainly as the backbone of the proofs) of the identity of all those Western peoples with the Saqdliba, as Dozy and his followers regarded it. What I proposed was that other evidence in support of that identity should be gathered, and weighed against that of Ibn Hawqal and that of other Muslim sources who speak in the same or a similar vein.48 Dozy precedes his misrepresentation of Ibn Hawqal's account with yet another grave confusion, this time a chronological one (argument c). He says that at the beginning (dans Yorigine), prisoners taken by the Germanic peoples were [pure] (citing indirectly Ibrahim b. al-Qasim al-Qayrawam) and only later was the same designation widened to include all the Western peoples as well (citing Ibn Hawqal directly49). Now, al-Qayrawam, whose evidence is that of an eyewitness, died after 417/1026, and Ibn Hawqal visited Spain in 336/948, which means, if we accept Dozy's reasoning, that those Saqdliba were Western Europeans dans Yorigine, and only later became Slavs!50 The part of Ibn Hawqal's account which is particularly relevant to the subject of the present study is that which Dozy ignored (argument a), and which tells us about the importation of Saqdliba to the eastern lands of Islam. Our author does not make any distinction, as far as ethnic composition is concerned, between them and the Saqdliba going to the West. He only states that they inhabit the eastern part of their homeland, which is nearer the eastern lands of Islam. On the other hand, he points out that they, in contrast to those of them who are sent to the West, are not castrated (turiku fuhulatan 'aid ahwalihim maqrurln 'aid sihhat abddnihim) .51 Now the existence of Saqdliba slave-traffic to the East on a considerable scale is attested to not only by Ibn Hawqal in the passage under discussion but by a good number of other major Muslim sources.52 In spite of that there is hardly a trace of them in the Mamluk armies of the Muslim East, and above all in those of the 'Abbasid Caliphs and the Seljuks.53 Their availability in abundance did not stimulate those rulers to enlist them in their armies. What could be a better proof of the superiority of the Turks over them? This is in addition to the fact that statements of praise and appreciation of the military ability and prowess of the Turks in the Muslim sources is incomparably more frequent and more varied than similar statements about the Saqdliba. The inescapable conclusion, as I have already stated else

47 Ibid., pp. 97-8. S. Maqbul Ahmad says: "A glance at the maps of Ibn Hawqal shows that they are superior to those of al-Istakhri" ("Kharita", EI2, iv, p. 1079a). 48 "On the eunuchs in Islam", pp. 98, 106, 121, 124. 49 Ibid., p. 93 50 Ibid., pp. 93, 100-1, 107. 51 Ibid., pp. 95-6. 52 Ibid., pp. 102-4. 53 The story of the emasculated Saqdliba is, of course, a different one. It would be interesting to find out, as far as the sources permit, what happened to the numerous unemasculated Saqdliba in the East. In the West, a systematic attempt should be made to distinguish between the Saqdliba eunuchs and non-eunuchs. I am not sure whether such an attempt has been made.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 321

where,54 is that the Saqdliba were able to become dominant only where it had been impossible, or extremely difficult, to obtain Turks, as in Spain and parts of the Maghrib.55 The Fatimids of Cairo did for some time have a Saqdliba component. But this was a military element which they brought over from the Maghrib alongside their Maghribi armies. In concluding this section I would like to say that if the results of further study show, after leaving out the evidence of Ibn Hawqal, that the Saqdliba who went to the West were wholly, or quite substantially, Western Europeans, this will not affect the thesis presented here. What I wanted to prove is that in the East, the heart of Islam's military power, the Saqdliba (and I mean the unemasculated ones among them) were no match for the Turks.56

The Borderlands of the Turkish Home Countries

In order to grasp the full dimensions of the contribution of Turkish Mamluk might to Islam, their study should not be confined to them alone, but must include the contiguous Muslim territory on which the Mamluks set foot as soon as they crossed the border between the Abode of War and the Abode of Islam. The kind of people and the degree of religious awareness which they encountered there formed a decisive factor in shaping their future destiny. The Muslim territory in question included, first and foremost, Transoxania and its neighbour Khwarazmia (for the significance of that fact see also the last part of Ibn Hassul's evidence in passage XI of the preceding section). Transoxania and the adjacent areas had a strong influence on the caliphal and other Mamluks in the following way. These areas are described by the contemporary sources as extremely wealthy, densely populated, thriving in commerce and agriculture, and rich in mineral deposits (mainly Transoxania). Besides, once its population had been finally subjugated, Transoxania accepted Islam and Islamic culture with great and growing zeal. This population was stated to have been deeply religious, immersed in Muslim culture, generous towards foreign Muslims and especially towards those volunteers who flocked to the frontier to fight the infidels; very warlike and imbued with the spirit of jihad against the unbelievers. The number of ribdts in that region is said to have been immense. While

54 Ibid., p. 118, and "Aspects of the Mamluk phenomenon", part A, p. 206, n. 18. 55 This is a very brief summary of my argumentation concerning Turk vs. Saqdliba in "On the eunuchs in Islam". For a more comprehensive picture see ibid., pp. 92-124, with special stress on pp. 114?24 (and within them pp. 115?21). An erroneous comment on my criticism of Dozy's handling of Ibn Hawqal's evidence was published in the article "Sakaliba" in EI2, viii (1995), pp. 879b-88oa. This comment will be exhaustively discussed in the fuller version of this study. 56 Certain aspects relating to the Saqdliba which are relevant to the subject of our study are the following. Saqdliba serve as soldiers already under the Umayyads in the East, especially in the frontier areas, but they do not seem to have formed a central part of the Umayyad army. At that time, and well into the 'Abbasid period, they were still pagans. The process of Christianization, in the full sense of the word, of those of them who did adopt that religion must have been slow, as was the case with the Islamization of pagans of a more or less similar level of civilization. Furthermore, there still remained many pagan Saqdliba in the fourth/tenth century, as stated by al-Mas'udi ("On the eunuchs", p. 113). What was said about people of long-established Christianity as a possible source of Mamluk recruitment does not, therefore, apply to them (for a fuller clarification see ibid., pp. 109-14). The eunuchs from amongst the Saqdliba, who are only briefly referred to here but discussed at length in the above cited article, belong to a different category. What is particularly indicative within the framework of the thesis presented here is that under the Umayyads of the East there was some room for the employment of the Saqdliba in their armed forces. Under the 'Abbasids there seems to have been no such room, and since the Mamluks became the main army of that dynasty, no chance was left for the Saqdliba to join them.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 322 David Ayalon military preparedness and alertness were, on that front, at their height, on the Byzantine front they were on the wane. The geographer al-Muqaddasi, who knew both fronts and both regions, gives us a vivid description to that effect. Another important factor in the relations of the Transoxanians, the Khwarazmians, etc., with the Turks, which should be taken into consideration, is that there was no abrupt division between the lands of the former and those of the latter. The transformation was gradual, and there must have been a great merging between the populations of the two areas, especially in the border zone. The Khwarazmians were described as very similar to the Turks in their physical appearance as well as in their military qualities. Because of that similarity they were sold, for a certain period, as Mamluks, under the pretence of their being Turks. Physical beauty was considered to be an outstanding trait both of the peoples of Farghana and Taraz and of the Turks. Such were the peoples with whom the Turkish nomads had contact during long periods of peace and war, and whom the Turkish Mamluks encountered first, when they crossed the borders of Islam. Many of the choicest Mamluks did not go further west beyond the Muslim frontier states, because they were incorporated in their armies or bought by their well-to-do people. As for the Mamluks who did continue their journey westward, it is only logical to think that quite a number of them did not merely cross the frontier states, but stayed there for varying periods before pursuing their journey. Furthermore, the contact of the Turkish Mamluks with the Transoxanians did not cease on their crossing the Oxus westwards, for when they arrived in Baghdad (and later in Samarra) they found there strong and influential Transoxanian units, which must have received steady reinforcements from their own people. Otherwise these units could not have existed for a considerable number of decades, as they in fact did. Thus, Transoxanian influence on the Mamluks, including even those who reached Baghdad and Samarra, must have been very great and persistent in the early 'Abbasid period. Speaking of Samarra, the capital which symbolizes the beginning of the great history of the Mamluk institution in Islam, it should be pointed out that its founder, al-Mu'tasim, as already mentioned, brought over his Mamluks even before he became Caliph, from Samarqand. This town was the hub of the Turkish Mamluk slave trade, and it is also stated to have produced the best of those Mamluks, because of the upbringing (tarbiya) they received there (passage VII in the preceding section). In this connection I venture to suggest that, if it were possible properly to reconstruct the history of the Iranian, and particularly the Khurasanian, armies of Islam since the beginning of the Muslim expansion and up to the creation of al-Mu'tasim's Mamluk corps, the Iranian-Transoxanian impact on the Mamluk military institution would be shown to have been far greater than can be proved today.57

57 See e.g. my "The military reforms of Caliph al-Mu'tasim", pp. 26-7; "Preliminary remarks on the Mamluk military institution . . .", pp. 52-3; "Egypt as a dominant factor . . .", pp. 9-27. Some citations in our context will be of use. Al-Muqaddasi, p. 261,11. 5-11: "Islam is fresh in it [in Transoxania]" (al-Isldm bihdtariyy). Al-Istakhri, p. 291: "They [the people of Transoxania], in spite of the remoteness of their country, are the first to perform the duty. No one enters the [Arabian] desert before them, and no one leaves it after them". Speaking about the people of Syria (al-Shdm), al-Muqaddasi (p. 152,11. 4?5) says: "They are not as [good as] the Iranians in learning, religiousness and [the observance of] the [religious] prohibitions" (laysu kal-A'djimftal-ilm wal-din wal-nahyi). The same author (p. 32,11. 4-12) says about the people of the East (Iqlim al-Mashriq): "Of all the [Muslim] people their Arabic is the most accurate, because they stuck to learning it and clung to it" (asahh al-nds 'Arabiyyatan li-annahum

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 323

In summing up the last three sections I would say: the military superiority of the Turks and their like was unparalleled. Unparalleled as well were the relations and connections between them and their immediate Muslim neighbours, with their high level of civili zation, economic strength and deep Islamic awareness. How much more so the combin ation of the two. That is why such a great harmony existed between the praise of the Turks and their actual employment, and why that employment lasted so long, for most of the time with full justification. But later on that justification was steadily eroded. The Turks and their like were so strongly entrenched that, for that and other reasons,58 they outlived their purpose for quite some time.

The Early Seljuk Evidence

The dominance of the Turkish Mamluks already in the reigns of the first Great Seljuks is, in my view, undeniable. Although much further systematic study should be carried out, what we already know is quite sufficient. Our information about the eunuchs of the early Seljuks (the subject of a later section) strongly upholds that statement. In the following lines two major events in the will be discussed: the final defeat and execution of al-Basasiri (450/1060), and the (463/1071). In the first battle, which marked, inter alia, the victory of the Sunna over the Shi'a in the East, and which took place in the reign of the first Great Seljuk, Tughril (429-55/1038-63), the Seljuk force (or at least its main component) that won the day was composed of 2000 Turkish Mamluks, headed by a eunuch, with other eunuch commanders under him.59 The eunuch aspect of that force will be discussed in a following section. The battle of Manzikert (Malazgird) between the Muslims and the Byzantines took place, as is well known, in August 1071 near . The army of Islam was commanded by Sultan , and that of Byzantium by the Emperor Romanus Diogenes. It ended with a resounding Muslim victory with far-reaching consequences for both religions. It paved the way for the Muslim conquest of and the subsequent rise of the . The accounts of that battle, which started with the disintegration of the Muslim army

takallafuhd takallufan wa-talaqqafuha talaqqufan). This was the kind of people who welcomed the Mamluk novice when he made his first steps in his new world. A combination of fresh and deeply rooted Islam, strengthened by unusual command of the language of the Qur'an, and linked to hardiness and fighting spirit against the Infidel. Nothing could suit the novice more, for, as Ibn Khaldun puts it so succinctly, "They [the Mamluks] embrace Islam with the determination of True Believers, while retaining their nomadic virtues" (yadkhuluna ft al-Din bi-azaim imdniyya wa-akhldq badawiyya) ("Ibn Khaldun's view . . .", p. 345). 58 The difficulty of adaptation to changing circumstances, especially technological innovation, was not confined to the Islamic armies. It lasted even in the Western European armies well into the twentieth century, although not with the same degree of persistence. 59 They are called ghulam or ghilmdn by Sibt Ibn al-Jawzi (Ankara ed., pp. 64, 11. 11, 16, 17, 18; 67, 1. 12) and said to be Turk (p. 67, 11. 8, 14). According to Ibn al-Athir (ix, cp. p. 648, 1. 7, with p. 649, 1. 3), those 2000 cavalrymen were Atrak, see also M. Canard, "Al-Basasiri", ET. The identity of ghulam and mamluk is well established for that period. Very indicative in this context is the account of Sultan Alp Arslan's capture of an immense number of Mamluks in a big foray into the land of the in 456/1064 (Sibt, Ankara, p. 117, 11. 11?13). This event, which took place between the liquidation of al-Basasiri and the victory of Manzikert won by the same sultan, is very meaningful within the framework of building a major Mamluk force by the very first Seljuk rulers. It needs, however, some corroborative evidence.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 324 David Ayalon because it had been taken by surprise, are quite confused. The account which seems to me the most credible is this. Alp Arslan's army deserted him. Those who stayed were 4000 Mamluks (wa-baqiya al-sultdnftarba'at dldfghuldm).60 An army of 10,000 Kurds also joined him, but he relied, after his reliance of God, Be He Exalted, on those 4000 who stayed with him (wa-kana qad ijtima'a ilayhi 'asharat dldfmin al-Akrdd wa-innamd i 'timdduhu ba'da Allah ta'dld 'aid al-arba'at dldfalladhina kdnu ma'ahu).61 I prefer that account for the following reasons. First and foremost, because this was precisely the place where the ruler's Mamluks should have been. As I have already stated, the Seljuks did not weaken the old Mamluk pattern, and with the progress of time that pattern could only be solidified and strengthened. Such a victory could not be won by tribal armies, be they Turcomans, Kurds or Arabs, however great the valour of all of them might be. This is not intended to deny the military contribution of those armies. The Mamluks formed a minority, more often than not a small one, in the general military body of any ruler who employed them, and without the auxiliary forces the Mamluks could not function. Yet with the absence of the Mamluk nucleus, the contribution of those forces could have been, at best, very erratic. As for the Turcomans of the Seljuks, it should be remembered that, on top of their undeniably great military contribution, they served as a major vehicle in spreading the deep into the Muslim East. Secondly, the account appears in Sibt Ibn al-Jawzi's Mir at al-Zamdn, but had been copied by him from Ghars al-Ni'ma Abu al-Hasan Muhammad, the son of Hilal al-Sabi, who was a contempo rary of the battle (he died in 480/1087).62 And that is the only contemporary Muslim evidence we possess about the battle. Thirdly, by contrast with the inflated numbers of soldiers mentioned by the sources as participating in this battle on both sides, the number of Alp Arslan's Mamluks given as taking part in it is reasonable. It is also in accord with the number of Mamluks in some of the big expeditions of the Mamluk Sultanate, which can be proved to be quite correct.63 Thus, according to Alp Arslan's conviction, the only one on earth, after God in heaven, who could and did save Islam, himself and his army from a certain catastrophic defeat, and turn it into a glorious victory, were a relatively small body of faithful Mamluks who stuck to their patron in an hour of desperate need. What certainly increased the glory of that victory was the fact that it was won on a Friday. The opening of the gates of Anatolia, which was the ultimate result of the battle of Manzikert and its antecedents,

60 Sibt Ibn al-Jawzi (Ankara), p. 147, 1. 15. 61 Ibid., p. 148,11. 15-17, for the identity here of ghulam and mamluk, cp. ibid., pp. 147,1. 15 and 148,11. 15?17 with p. 152, 1. 8. See also Ibn al-Jawzi, al-Muntazam (Hyderabad, Deccan, 1358/1939), vii, p. 262, 1. 12; Ibn al-Athir, vi, p. 319, 11. 3-5; Abu Shama, Rawdatayn (Cairo, 1288H), i, p. 173, 1. 1. For the account of the whole battle see Sibt (Ankara), pp. 147-52. I have already referred to this evidence in "Egypt as a dominant factor . . .", p. 29 and n. 24. As far as I know, the decisive role of the Mamluks in the victory of Manzikert is not pointed out in the rich Islamicist literature dedicated to that battle. 62 This has already been proved by Claude Cahen ("Sibt Ibn al-Jawzi", ET, French ed., iii, p. 775b), and after him by Suhayl Zakkar (Mukhtdrdt min Kildbdt al-Mu arrikhin al- Arab, n.p., n.d., pp. 105?6). 63 These numbers are discussed in the as yet unpublished part of my work on the Mamluks, called "The Mamluk army on the march".

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 325 was one of the highest peaks of Islam's success in its struggle against its Christian adversary.64

The Thorough Ethnic Transformation of the Ottoman Branch65

The main reason for preferring the Turks and their like was their superb horsemanship (and marksmanship) and the fact that they were both pagan and on a quite primitive cultural level in comparison with that of the civilization which they were destined to embrace and serve after their arrival from the Abode of War. They could thus easily be moulded as good Muslims and good soldiers.66 With the gradual expansion of Islam into the homelands of the Turks, however, the inhabitants of those lands adopted that religion, and were thus not eligible for slavery.67 Although enslavement and importation of those inhabitants does not seem to have stopped immediately on Islamization, it certainly stopped ultimately.68 And the peoples living beyond those newly Islamized lands might not have been very suitable for recruitment as Mamluks either qualitatively or quantitatively, or for both reasons. The

64 Factors and arguments which tend to reduce the importance of the battle of Manzikert cannot deprive it of the immense effects which it had on future events. A battle on such a large scale, fought out between two great armies headed by their two sovereigns, ending with the rout of one of them and the capture of its king, cannot be minimized. The battle of'Aynjalut (1260), where the Mongol army was, in fact, a garrison of a province, did have its huge impact, and with full justification. How much more so the battle of Manzikert. It is highly improbable that without that battle the fate of Anatolia would have been the same. 65 This part of the present study is confined to the most general outlines of the thesis presented here, and will receive most detailed treatment in the fuller version. 66 The degree of the Mamluks' knowledge of Islam or their strict observance of its precepts was secondary in importance to their Islamic awareness, namely, that Islam and what it stands for is superior to anything else. 67 These sources of supply might dwindle away even before Islam had taken deep root in those lands as a result of internal wars (see e.g. my "The Circassians in the Mamluk kingdom", JAOS (1949), p. 136a, especially Ibn 'Arabshah's evidence); migration of the nomads from one homeland to the other; natural factors (including natural disasters); and the fact that exporting, over a long period, the pick of boys and girls from a certain tribe or group of tribes must have affected negatively that tribe or that group in various ways. 68 Generally speaking, the easiest to convert to Islam (and this is also true, to no small extent, of converts to Christianity) were pagans at a primitive level of civilization, bordering on the Abode of Islam; the hardest - people at a high level of civilization, who adopted monotheism long before the advent of Islam, or at least long before their encounter with it. The best proof that a high level of civilization alone was not enough to deter Islamization of a conquered people, or even to postpone it for a long time, is the relatively quick conversion of the Iranians, or most of them. The adoption of Islam by the Indians, another non-monotheistic people, or peoples, at a high level of civilization, is not a case similar to that of the Iranians. Their immense numbers, the slow Muslim conquest even of parts of that huge sub-continent, and the history of the relations between its Muslim and non-Muslim entities, form a different story. Their being a temporary and relatively local factor in the Muslim Mamluk armies, and the reasons for that, have already been pointed out (see again "Aspects of the Mamluk phenomenon", part A, pp. 202-3). The negative effects of the conversion to monotheism of the peoples of the steppe on the slave trade have already been stressed by al-Itsakhri (p. 223,11. 11?15) in connection with the Khazar. (For the evaluation of that evidence see my "The Mamluk novice: on his youthfulness and on his original religion", REI, LIV (1986), pp. 1-8. For the relative ease with which a pagan tribesman of the steppe parted from his adolescent child see ibid.). Conversion to monotheism (and even conversion to Islam) did not necessarily mean an immediate stoppage of or a drastic decline in the readiness of these peoples of the steppe to sell their next of kin (see e.g. the attitude of the Islamized Mongols of the Golden Horde: al-Maqrizi, Suluk, ii, p. 525). However, in the long run conversion to Islam meant the end of slave traffic from the Islamized region. The under the Ottomans faced a different situation from that of the pagans of the Abode of War. They were forced to raise a quota of their children for their Muslim rulers. There is abundant evidence of their antagonistic reaction to that levy. There were, however, cases where the parents did not lose sight of the great future awaiting their offspring. This brings us to the attitude of relatively new Christians, inhabiting the margins of the Christian world, and whose level of civilization was not high. A case in point is that of the Circassians who were Christians during the latter half of the Mamluk Sultanate. However, Christianity does not seem to have been deeply rooted in them, for ultimately they all became Muslims. The very interesting case of the Saqdliba is discussed in brief in an earlier section of this study, and at greater length in "On the eunuchs . . .".

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 326 David Ayalon

alternative that remained under these expedient circumstances was the lands inhabited by Christians.69 The Christians, be they Greek, Armenians or others, had a completely different background. On the average, they were much more civilized and cultured; had, of course, been monotheist for a very long time; and, as far as military qualities are concerned, many of them did not live in a purely cavalry milieu and tradition as did the peoples of the steppe. Being part of a long-standing monotheist community was a disadvantage in the Mamluk context, because their monotheism was not only non-Muslim, but to no small extent anti-Muslim. Boys in their early teens cannot simply wipe out that part of their history, and just exchange one brand of monotheism for an antagonistic brand. By and large, their Islamic awareness and devotedness could not be as full and unhesitating as that of a boy who was born into and grew up in a pagan environment.70 In the domain of horsemanship many of them could not reach the high standards of the people of the steppe. The transition from the recruitment of pagans (mainly tribesmen) from primitive areas to that of Christians inhabiting regions of much higher civilization was inevitable for the purpose of the preservation of the Mamluk system. For one thing, there was the gradual dwindling away of the Turco-nomad source already mentioned. For another, there was the slow but growing awareness that the concept of the exclusiveness of cavalry as the foundation of the elite units of the army should be mitigated. For yet another, there was the realization that those more civilized and cultured Christian recruits could be utilised in much wider domains than the original Mamluks could. Under such circumstances it is no wonder that the period of overlapping in the recruitment of the two kinds of religio-ethnic groups was very long, with the Turkish element and its like slowly diminishing.71 One of the great advantages of reliance on those Christian sources of supply was that they were much nearer geographically and, therefore, were very convenient. Caliph al-Muqtafi and other Muslim rulers certainly benefited greatly from that proximity. That brings us to the debate about why the Ottomans recruited by means of enslavement Christians from within the boundaries of their realm (both in Anatolia and in Ottoman ), which is contrary to the Shari'a (the Shari'a permits the enslavement of monotheistic infidels brought over from the Abode of War but not those inhabiting the Abode of Islam). In my view, the major reason was expediency. But that practice seems to me not to have been a purely Ottoman invention. Its origins are to be sought in an earlier period, particularly since the already mentioned battle of Manzikert (1071), which opened the gates of Anatolia to Islam. The shifting fronts between Islam and Christianity and the vague and changing boundaries between the various Muslim principalities in that peninsula made it extremely difficult to draw a clear line between Muslim and Christian territories.72 This must have created in that

69 On the four fronts which Islam faced see my "Aspects of the Mamluk Phenomenon", part A, pp. 198-204 (on the Christian front, pp. 198-9). 70 See the passage referred to in note 23. 71 For the employment of Arman and Rum, especially in Egypt under the Fatimids and earlier rulers, see e.g. B.J. Beshir, "Fatimid military organisation", Der Islam, LV (1978), pp. 43?4. 72 Part of these Greeks and Armenians certainly came from the Abode of War, captured during Seljuk raids on Trebizond and Nicaea to the north and west and on the Armenian kingdom of to the south (S. Vryonis, "Seljuk ghulams and Ottoman ", Der Islam, XLI (1965), p. 227). Still, it is most unlikely that the Seljuks

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 327 peninsula and its vicinity a situation in which it was practically impossible to observe the Sharl'a in that particular domain. I very much doubt whether any Muslim ruler or slave merchant tried hard to do so. It would have been quite a surprise if it had been discovered that all Caliph al-Muqtafi's Armenians and Greeks had been brought over from the Abode of War. It stands to reason that the Ottomans, whose state was born almost two and a half centuries after Manzikert, had already found a well-established practice of recruiting Christians not only from the Abode of War, a practice which must have been greatly accelerated by the Seljuk Sultans' eagerness to keep the Turkish Mamluks for themselves.73 There was a great advantage in that practice, in that it gave the Ottomans full control of the source of recruitment of its own slaves, as contrasted with the absolute, or almost absolute, lack of control by the Muslim rulers of the countries of origin of the earlier pagan Mamluks as well as of a substantial part of the road (or roads) leading from those countries to the Muslim rulers' realms. The Ottomans could and did make much better use of that human material than they could have made of the original Mamluks. In their struggle against their European adversaries they had to adapt themselves to more modern developments. For example, they built a superb infantry based on their slave recruits, and thus could move into the age of firearms more smoothly and effectively than any other Muslim state. They also used those recruits in domains far beyond the purely military machine. In order to make that argument clearer I would like to add that the Ottomans did benefit to no small extent from Western European influences in various ways: by alliance; by importing experts and military equipment; by the movement of Western Europeans into the boundaries of the Ottoman Empire; and certainly in a number of other ways. But the manpower which constituted the backbone of its military forces, and much beyond military might, was overwhelmingly composed of people of Eastern Christian origin, and without that manpower all those influences would have been of very limited use.

The Limited Ethnic Changes in the Other Branch

Whereas the Anatolian-Ottoman Mamluk branch of the Seljuk tree developed in the way just described, the other branch of the same tree developed very differently. It consisted of three regimes, each rising from inside the one which preceded it: the Zangids, the Ayyubids and the Mamluks of Egypt and Syria. All three had Turkish Mamluks as the backbone of their military power up to nearly the end of the fourteenth century. The Turkish-Mamluk component of the two earlier regimes was the subject of a detailed article of mine.74 Here I shall add just one very revealing piece of evidence relating to both the Zangids and the Ayyubids. A contemporary anonymous source, included in Ibn al-Furat's chronicle and published by Claude Cahen, describes the entrance into Egypt of the Zangid contingent headed by Shir kuh and his nephew Saladin. In that relatively short description the members

and whoever succeeded them would be so discriminating as to abstain from enslaving members of those two ethnicities outside these three centres in an area of constantly shifting boundaries (see also Vryonis, op. cit., p. 252). 73 I do not know whether the Seljuks had Rum and Arman Mamluks on the margin. 74 See "Aspects of the Mamluk phenomenon", part B, "Ayyubids, Kurds and Turks", Der Islam, LIV (1977), pp. 1-32; and also "From Ayyubids to Mamluks", REI, XLIV (1981), pp. 43?57.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 328 David Ayalon

of the contingent are thirteen times called Turk or Atrdk, and the commander himself, the Kurd, is called Shirkuh al-Turki. The name of the Kurds is not mentioned even once!75 Of all the Mamluk societies in medieval Islam the ethnic composition of that of the Mamluk Sultanate can be established with the greatest degree of certainty and accuracy. This is also true of the relations between the various ethnic groups. The list which I have reconstructed is this: Turk (or Atrdk), Qifjdq, Tatar, Mughul (or Mughul), Khitaiyya, Riis, Rum, Arman, As, Abaza, Ldz,Jarkas.76 The Qifjdq are rarely referred to and are obviously eponymous with Turk. The As and the Ldz are probably the same. The case of the Mughul and the Tatar is more complicated. The Mughul, who are mentioned only in the Qipchaqi period, seem to have been distinct from the Turk, although a certain degree of overlapping between the two should not be excluded. The Tatar, on the other hand, were often, especially under the Circassians, eponymous with the Turks.77 Thus the real list is a considerably shorter one. The Mamluk Sultanate was actually dominated by the struggle between hardly more than two ethnic groups: the Turk and the Jarkds. The Rum were certainly third in importance. The Arman were much less conspicuous, as also during the Ayyubid regime. The reason for this was possibly their strong support for their Fatimid patrons against the Ayyubid adversary. But there may have been other reasons too. As for the Franks (Faranj, Ifranj), there is a fair number of individual Mamluks who are said to have been of Frankish origin. However, they are never mentioned as a separate ethnic body. This does not support the claim of some medieval European writers that there was a high proportion of Franks in the Mamluk army. One should, however, take into consideration the possibility that there might have been some overlapping between Faranj and Rum. What typifies the ethnic composition of Mamluk society in that Sultanate is the relative weakness of its Christian element, as far as Mamluks coming from civilized countries of long-established Christianity are concerned.78 The Mamluk Sultanate, after the expulsion of the Crusaders and the overcoming of the Mongol danger, did not for a very long time have any enemy who could threaten it on land. Its extreme naval weakness, especially vis-a-vis the growing naval power of the Western Europeans inside and outside the Mediterranean, put it from time to time in considerable straits, but a concerted attack on it from the sea never materialized. The inevitable ultimate clash on land between the Mamluks and the Ottomans was also a clash between two Mamluk systems, both emanating from their Seljuk and 'Abbasid predecessors. There could

75 "Un recit inedit du Vizirat de ", Annales Islamologiques, VIII (1969), pp. 27-46; the Arabic text, pp. 40-6 (see there pp. 42,1. 19 - twice; 43, 11. 1, 12, 19, 13; 44,11. 2, 12, 14; 45, 11. 4, 16, 23; 46; 1. 1. For Shirkuh al-Turki see p. 42,1. 19. 76 Of the chapter dedicated to the Mamluk races in my work on the army of the Mamluk Sultanate, only the part on the Seljuks has been published (JAOS, LXIX, 1949, pp. 135-47). In my article "Mamluk military slavery in Egypt and Syria" in ET, fasc. 103-4 (1987), a summary of the unpublished parts of that chapter is given. That article was, however, abridged by the editors. The unabridged summary appeared under the same title in the Variorum Reprint volume Islam and the Abode of War (London, 1994), no. II, pp. 1-21. The Islamization of the Turks in the Central Asian area did not affect the Mamluk Sultanate, for it recruited its Turkish Mamluks mainly from southern Russia. 77 For the reasons of that synonymy see the unabridged summary, op. cit., pp. 8?9. 78 The relatively new and transient Christianity of the Circassians, living on the margin of the Christian world of that time, has already been pointed out. See note 68.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 329 be no question about the results: an antiquated military system, which did not adapt itself to the development of technology, including its reliance on human material which was incapable, physically and psychologically, of making use of it; and a more flexible one, which did adapt itself to a considerable degree, relying on human material which was far more suitable for that task.79

The Eunuchs of the Seljuks

Eunuchs in key positions under the Seljuks are known from the reign of the first Great Seljuk, Tughril, onwards, and their presence becomes more and more conspicuous later on. I can give here only the briefest possible review about them. A much fuller account is given in my book on the eunuchs, already mentioned, which is now in the press.80 It should also be borne in mind that a eunuch at the top usually represents a pyramid of eunuchs below him. We shall start with the eunuch Khumartakin al-khddim al-Tughral, a young boy with whom Sultan Tughril fell in love to such a degree that he castrated him, so that he could accompany him even on his visits to his chief wife. The pace of the adaptation of that ruler from the steppe to the refinements of a higher civilization requires, perhaps, a systematic study. In that particular domain it seems to have been very quick. The same Khumartakin became so powerful that both the chamberlains and the commanders stood in audience in deference to him (fa-istafhala amruhu wa-sdra al-hujjdb wal-umara yaqifuna 'aid ra'sihi). It was this selfsame Khumartakin who not only headed the already mentioned expedi tion against al-Basasiri, but who also organised it. What makes that double appointment even more impressive is that the whole idea of preparing that expedition and the route it should take was given to Tughril by the chief of the extremely powerful Khafaja confederation, on condition that he (i.e. the Khafaja chief) should command it. but Tughril preferred his eunuch. At least one other eunuch, Sawtakin al-khddim, about whom we shall speak soon, was in the high command of that expedition, composed, as already stated, of 2000 Turkish Mamluks. In the battle of Manzikert the presence of the Sultan himself in the field overshadowed the other commanders. Yet the omnipotent eunuch Kuhraln al-khddim (to whom we shall return) was there, and it was said that a Mamluk of his was the one who captured Romanus Diogenes.81 To come back to Khumartakin, he played a prominent role in the struggle of Tughril with his two brothers; Ibrahim Yanal and, at a later date, Qutulmush. The killing of Ibrahim Yanal was, as is well known, a major event in early Seljuk history. A son of that Ibrahim claimed that Khumartakin was the actual killer of his father. Khumartakin lost favour with Tughril quite suddenly, and, following the advice of'Imad al-Din al-Kunduri, a eunuch belonging to a different category, he was handed over by Tughril's order to

79 See also my Gunpowder and Firearms in the Mamluk Kingdom - A Challenge to a Mediaeval Society (London, 1956, repr. 1978), especially pp. 46-133. 80 On this book see note 1. Only part of the source references collected in the chapter on the eunuchs of the Seljuks is cited here. 81 Ibn al-Jawzi, al-Muntazam, viii, p. 262, 1. 12.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 330 David Ayalon

Ibrahim's sons, who finished him off. That eunuch was only about twenty years old when he was put to death in 454/1062.82 He was replaced by Sawtakin "the special eunuch" (al-khadim al-khass), another commander of the expedition against al-Basasiri, who held the high office of chamberlain (hdjib), and who was also called sarhang (general). He had a much longer career than Khumartakin. As early as Muharram 450/March 1058, that is to say two years before his participation in the battle against al-Basasiri, he was sent by Sultan Tughril to his brother in Mosul, the already mentioned Ibrahin Yanal, with sumptuous gifts which were added to a momentous order, namely that he must come immediately back to him (to Tughril). It was eunuchs who were entrusted so often with the carrying of secret and other messages of the highest order, at least from the reign of Harun al-Rashid onwards. In the struggle for the succession which burst out, after the assassination of Alp Arslan (465/1072), between his two sons, Malikshah and Qawurt, the latter, as is well known, was defeated, captured and executed. It was Sawtakin who arrested him, put him in a tent on the field of battle, and shackled him there. The person who supervised his execution was a much more famous, high-ranking official, Sa'd al-Dawla Kuhra'in, another eunuch (on him see below). Sawtakin fulfilled other important functions until his death in Jumada II 477/September 1084, leaving behind an immense fortune.83 The career of the eunuch Erdem the chamberlain (al-hajlb), who is said to have cut off the head of al-Basasiri (yet another eunuch on that expedition, presumably a commander) is of great interest. He certainly had an important share in putting Alp Arslan on the throne of the Great Seljuks as the successor of Tughril, a fact of the highest importance in the history of that dynasty. Here is the account of Sibt Ibn al-Jawzi. In Muharram 45 5/January 1063 the childless and very ailing Sultan Tughril arrived in Baghdad to announce the appointment of his successor (he died eight months later). There he declared, in the presence of a most impressive audience, that his choice fell on Sulayman, the son of his deceased brother Chaghri Daud, a minor (saghlr), who was then in , and whose widowed mother Tughril married. All present humbly agreed, and the only objector was Erdem, who declared that he would not serve anybody after Tughril, and that he would go to Alp Arslan, another son of Daud. He departed immediately, and this was what turned the scale in favour of Alp Arslan. This is a rather exaggerated version. A more balanced presentation is the concise and excellent analysis of Professor C. E. Bosworth.84 But even according to that analysis Erdem was a major supporter of the successful candidate. The possibility that the result of the battle of Manzikert would have been quite different had the Seljuk army been headed by Sulayman should not be ruled out.85 The eunuch Sa'd al-Dawla Kuhra'in al-khadim was already very prominent under the Buwayhids of Fars (in the reign of'Irnad al-Din Marzuban: 415-40/1024-48). With the

82 On Khumartakin see e.g. Sibt (Ankara), pp. 64, 11. 9-20; 74, 1. 14-75, 1. 13; 81, 11. 2-3; 84, 11. 19-22; Ibn al-Athir, ix, pp. 648?9. It should be noted here that while there were numerous eunuch commanders, hardly any eunuch fighting units are mentioned in the sources. I try to explain this phenomenon in my book on the eunuchs. 83 On Sawtakin see e.g. Sibt (Ankara), pp. 27,11. 1-4; 67,1. 3; 161-4; 228-9; Ibn al-Athir, x, p. 92,11. 7-9; C. E. Bosworth, in J. A. Boyle (ed.), Cambridge , v (Cambridge, 1968), pp. 35, 50, 75, 88, 91. 84 Ibid., p. 54. 85 On Erdem see e.g. Sibt (Ankara), pp. 67, 1. 3; 97, 11. 2-17; 102, 11. 6-10, in, 11. 8-10.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 331

extinction of that dynasty he joined the retinue of Alp Arslan, and it was he who tried, unsuccessfully, to protect him with his own body (waqdhu bi-nafsihi), when the Sultan was mortally wounded by an assassin. After that Sultan's death, his son and successor Malikshah sent him to the Caliph's court in Baghdad, from where he brought back to his new master a Caliphal and his certificate of appointment as Sultan (al-khil'a wal-taqlid). This task was often reserved to first-rank eunuchs.

Ibn al-Athir says about Kuhrain: "No eunuch (khddim) before him enjoyed such a great influence; such a full power and such a degree of obedience on the part of the great amirs who readily served him". This was in addition to his other noble qualities, according to the statement of the same author.

In the struggle for the succession to the throne of the Great Seljuks between Muhammad and Barkiyaruk he sided with the loser, Muhammad, and was killed in the decisive battle (493/1100).86 The very famous 'Imad al-Mulk al-Kunduri was castrated by Sultan Tughril at an adult age as a punishment, but he continued to serve his master. He occupied the highest office, that of vizier, in the last eight years of Tughril's reign (422-55/1055-63). Believing that he would become the real ruler of the empire after that Sultan's death, he supported the candidacy of Sulayman the minor against that of Alp Arslan. For that he paid with his life early in the reign of Alp Arslan, and the way was paved for his arch-enemy Nizam al-Mulk to become the central figure in the Sultanate.87 A person who deserves a much more detailed discussion than the brief reference here (and this remark also applies to my book on the eunuchs) is al-Muqarrab , a eunuch (min khadam) of the Great Seljuk Sultan Sanjar (511-52/1117-57), who was murdered by the Shi'ite Isma'ilite Bdtiniyya in 534/1140. He is said to have been the ruler of the whole Sultanate and "the rest of the army of Sanjar [in addition to his own Mamluks] served him and stood by his door" (wa-kana qad hakamafi dawlatihijami'an. . . . wa-kana sd'ir 'askar al-Sultdn Sanjar yakhdumunahu wa-yaqifuna bi-bdbihi). His chief Mamluk 'Abbas, the ruler of the province of Rayy, launched constant heavy attacks on the Bdtiniyya until his death, and inflicted terrible losses on them.88 Another leading eunuch in the Seljuks' service was Sharaf al-Din Kurdbadhu al-khddim. Sultan Sulayman (554?6/1159?61), the Seljuk ruler of Iraq and Western Iran, who was the grandson of Malikshah, was a frivolous person who neglected his duties as a ruler, including his duties to the army. He gave the reins of government to Kurdbadhu, one of the older Seljuk eunuchs, who is said to have been religious and wise and a person who knew how to manage things well. The commanders complained to that eunuch about the Sultan's behaviour, but he used to calm them (wa-kana qad radda jami' al-umur ild Sharaf al-Din Kurdbadhu al-khddim wa-huwa min mashdyikh al-khadam al-Saljuqiyya yarji' ild din wa-'aql wa-husn tabir fa-kdna al-umara yashkuna Hay hi wa-huwa yusakkinuhum). However, the

86 On Kuhra'in see e.g. Ibn al-Athir, x, pp. 66, 70, 73, 79, 90, 100, 116, 144, 162, 164, 165, 176-7, 184, 187, 204, 219-22, 289, 294f., 295,1. 18-296,1. 9, 435, 480. 87 On al-Kunduri see e.g. Sibt (Ankara), pp. 97, 11. 13-14; 124, 11. 12-14; l65> 1- *3; It>n al-Athir, x, pp. 32, 1- 17?33,1- 5; 33, II- 15-17; Ibn Khallikan, Wafayd (Beirut, 1968^72), v, p. 141, 11. 5-22. 88 Ibn al-Athir, xi, pp. 76, 1. 14?77, 1. 5. He was a black eunuch: Jawhar al-khadim al-Habashi al-maruf bil-muqarrab (Ibn al-Jawzi, al-Muntazam, x, p. 87. Ibn al-Jawzi describes him there as the real ruler in Sanjar's reign ? kana mustawliyan 'aid mamlakatihi mutahakkiman fihd (ibid.).

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms 332 David Ayalon

irresponsible deeds of the sultan exceeded all bounds; and when Kurdbadhu called upon him and reproached him, he ordered one of his jesters to uncover his pudenda (kashf sawatxhx) [in order to expose the absence of his genital organs]. After consultation Kurd badhu arrested the Sultan, sent him to jail, and then ordered him killed. The Sultan's vizier and the vizier's private retinue were killed on the spot. Kurdbadhu also took part in the setting on the throne of Arslan Shah (556-73/1161-77), the nephew of Sulayman Shah.89 This is, incidentally, an excellent example of how a eunuch, at the peak of his power, could be exposed, within seconds, to the utmost disgrace because of his mutilation. Returning to the relations between the Seljuks and the Caliphs, within the framework of the subject of the eunuchs, two matters should be pointed out. (a) As already stated, the Seljuks used to have a representative in Baghdad, holding the office of shihna, whose main duty was to watch the Caliph's acts and to make him toe the Seljuks' line. Among these shihnas the eunuchs figure quite prominently. In the reign of al-Mustarshid (512-29/1118-35) it was the eunuch Bihruz, the sponsor of the Ayyubids, who gave the founder of that dynasty the great push forward which culminated in its supplanting the Fatimids in Egypt and the Zangids in Syria;90 and the already mentioned Mas'ud Bilal or Mas'ud al-Bilali, in the reign of this successor, al-Muqtafi.91 (b) In the struggle between the Caliphs, the Seljuks and the Seljuks' allies (or vassals), the presence of eunuch commanders in all camps is very noticeable. Here I shall mention only one of them. In 532/1137, during the reign of our old acquaintance al-Muqtafi, Nazar al-khadim was one of the two commanders who headed an army which repulsed yet another Seljuk attack on Baghdad.92 Under the Syrian branch of the Seljuks and its successors the power of the eunuchs reaches yet another peak. It is recounted in my book on the eunuchs. Here I shall only say that in the competing courts of and they are often the real rulers, as well as the commanders of their respective citadels. Very interesting is the remark of Sibt Ibn al-Jawzi: "Aleppo was afflicted with the following trouble: whenever the ruler dies, one of his Mamluks or eunuchs take charge of his small son [who succeeds him on the throne]. This goes on from the time of [the Hamdanid] Sayf al-Dawla (933-56/945?67) onwards" (buliyat Halab bi-mithli hddhd an yamut sdhibuha wa-yatawalla umiira waladihi al-saghlr bad mamallkihi aw khadamihi min zaman Sayf al-Dawla Ud halummajarran).93 This implies that the Seljuks fit themselves into the already existing pattern in this domain, as they did in other parts of their realm. It also indicates another matter. It is true that alternating of Mamluks and eunuchs did not necessarily repeat itself exactly in the same way everywhere. But the intertwining of eunuchs and Mamluks is forcefully reflected here.

89 Ibid., p. 266,1. 3-268,1. 2. 90 The eunuch Bihriiz's connection with the forefather of the Ayyubids is dealt with in the chapter in the eunuchs under the Ayyubids in my above-mentioned books on the eunuchs. 91 He was also a black eunuch. He is called khasi by Ibn al-Furat, iii, fol. 2b, 11. 6-7; and ahad al-khadam al-khisydn al-Habashiyyin al-kibdr by Ibn Khallikan, Wafaydt, vi, p. 231, 1. 17. On him see also Ibn al-Athir, xi, pp. 118, 131, 161, 189, 195. 92 Ibid., p. 61, 11. 13-16. There is an immense amount of evidence on eunuch commanders in the armies of medieval Islam. Numerous instances are brought forward in that book, which form only a small part of the evidence (this statement excludes the eunuchs of the Mamluk Sultanate). 93 Sibt, Mir at al-Zamdn (Hyderabad, Deccan, 1951), viii, part 1, p. 47,11. 5-6.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms The Mamluks of the Seljuks: Islam's Military Might at the Crossroads 333

The Ottoman Slave System - A Direct Offshoot of that of the Mamluks

My study of the socio-military slave institution under the Seljuks, which certainly needs to be much more comprehensive than it has been up to now, fully upholds Professor H. Inalcik's conclusion in his illuminating article "Ghulam" in the Encyclopaedia of Islam, where he says: "The Ottoman Administration was based on the Ghulam system. The principle of training young slaves for the Palace service and the service of the state had certainly been inherited by the Ottomans from the Seljuk ".941 would add, and its Seljuk predecessors, who inherited the system from the Muslim rulers who preceded them. The conclusions of my study are also in full accord with those of Professor Speros Vryonis, in his brilliant article, "Seljuk ghulams and Ottoman devshirme",95 where he traces the Ottoman slave system to that of the Seljuks and their predecessors. His conclusions are mainly based on a most detailed and insightful reading of Ibn Bibi's Seljuk-ndme, and supplementary material from Muslim insciptions and waqf documents.96 It is true that under the Ottomans the major source for the recruitment of military and administrative manpower became the Balkan peninsula,97 but the recruitment of Greeks and Armenians in Anatolia and its immediate vicinity, which started much earlier, formed a vital link in that process. As for the eunuchs, I tend to think that the Byzantine influence in their adoption by the Muslims in the early decades of Islam's existence was considerably greater than the Iranian. On the other hand, the Byzantine impact in that domain resulting from the Ottoman conquest of in 1453 seems to me to have been quite marginal. Orkhan, the son of 'Uthman, already early in the fourteenth century had prominent eunuchs in his service, as we can learn even from the few documents surviving from his reign.98 As I have already stated, a highly-placed eunuch often indicates a pyramid of eunuchs below him. To sum up: the Ottoman slave system, in spite of all its distinctive characteristics, is a direct descendant of the earlier Muslim one. Without it there would have been no Ottoman Empire, at least not the glorious and formidable one of which we know.

94 ET, ii, 1085b ("Ghulam"). 95 Der Islam, XLI (1965), pp. 224?52. Note also the numerous studies referred to in that article. 96 See especially ibid., pp. 227ff. and the summary, p. 252. There can be little doubt that had we possessed sources similar to that of Ibn Bibi on the Seljuks of Rum for the other Muslim principalities of Anatolia, the same picture relating to their slave systems would have emerged. 97 Vryonis, op. cit., p. 252. 98 ET, "Ghulam", p. 1086a.

This content downloaded from 129.128.216.34 on Sun, 09 Apr 2017 00:48:10 UTC All use subject to http://about.jstor.org/terms