PREFACE
The first draft of this work was presented to the University of London in December 1921, under the title of “The Arab Conquest of Transoxania”, as a thesis for the degree of Master of Arts, and was approved by the Senate in January 1922, for publication as such. During the year my attention was taken up in other directions and, except for the publication of two studies on the subject in the Bulletin of the School of Oriental Studies, nothing further was done until by the generosity of the Trustees of the Forlong Bequest Fund an opportunity of publication was offered. In its present form the work has been largely rewritten and revised. It makes no claim to present a complete historical account of the Arabs in Central Asia, but is intended solely as a critical study of the authorities in greater detail than has hitherto been made. Much is therefore omitted because it has already been dealt with in the standard histories. In order to keep down the cost of publication, the extensive references which originally accompanied the text have been cut down to a few notes at the end of each chapter. No references are given when, as in the great majority of cases, the authority for the statements made can easily be found in the appropriate place either in Tabarī or Balādhurī.
I regret that several works which are indispensable for a thorough study of the subject have, for linguistic reasons, been inaccessible to me. Such are van Vloten’s Opkomst der Abbasiden, and almost the whole range of Russian research work. Through the kindness of Sir Denison Ross, however, I have been able to avail myself of a draft MS. translation of the most important and valuable of them all, Professor W. Barthold’s Turkestan, as well as of his as yet unpublished London lectures on “The Nomads of Central Asia.” My sincere thanks are due to Sir Denison Ross also for his continued interest and material assistance ever since he first introduced me to the subject; to Sir Thomas Arnold for much encouragement and helpful counsel; to Professor Barthold, who has read the MS. through and made a number of valuable suggestions; to the Trustees of the Forlong Bequest Fund for their kindness in undertaking the publication; and in no small measure to my wife, who has given much time and labour to preparing the MS. for publication.
London, April, 1923.
I. INTRODUCTION
THE OXUS BASIN
Early History.
The Oxus is a boundary of tradition rather than of history. Lying midway between the old frontier of Aryan civilisation formed by the Jaxartes and the Pamīr and the natural strategic frontier offered by the north-eastern escarpment of the plateau of Īrān, it has never proved a barrier to imperial armies from either side. It was not on the Oxus but on the Jaxartes that Alexander’s strategic insight fixed the position of Alexander Eschate, and when the outposts of Persian dominion were thrust back by the constant pressure of the Central Asian hordes, their retreat was stayed not on the Oxus but on the Murghāb. Thus when the tide of conquest turned and the Arabs won back her ancient heritage for Persia, they, like Alexander, were compelled to carry their arms ever further to the East and all unknowing re-establish the frontiers of the Achaemenid Empire. It was from the legends of Sāsānian times, enshrined in the pages of the historians and the national epic of Firdawsī, that the Oxus came to be regarded as the boundary between Īrān and Tūrān.
Through all the centuries of invasion, however, the peoples of Sogdiana and the Oxus basin remained Iranian at bottom, preserving an Iranian speech and Iranian institutions. But the political conditions of the country at the period of the Arab conquests were so complex that it is necessary to trace briefly the course of their development.
The second century B.C. was a period of upheaval in Central Asia: the powerful Hiung-Nu peoples were dispossessing weaker tribes of their pasture lands and forcing them to migrate westwards. Between 150 and 125 B.C. a succession of nomadic tribes, the last and most powerful of which were a branch of the Yueh-Chi, were driven down into Sogdiana. It is now generally held that these tribes were of Aryan origin, though the question is not perhaps settled with absolute certainty. Before long, however, a second group, the K’ang, possessed themselves of Sogdiana, driving the Yueh Chi on into Bactria and the Afghan mountains[1]. In these districts they found, alongside the Iranian peasantry, a settled population of Tukhari (in Chinese, Ta-Hia), already noted in the Chinese annals for their commercial enterprise[2], and while at first the nomad tribes introduced complete confusion, it would seem that they rapidly absorbed, or were absorbed by, the native elements, and thus assimilated the Hellenistic civilisation of Bactria. From this fusion arose, about 50 A.D., the powerful Kushan Empire which spread into India on the one side and probably maintained some form of suzerainty over the K’ang kingdoms of Sogdiana on the other. Under the new empire, Buddhism was acclimatised in Turkestan, and Sogdiana developed into a great entrepôt for Chinese trade with the West. Towards the close of the third century the Kushan Empire, weakened by attacks from the new national dynasties in India and Persia, reverted to its primitive form of small independent principalities, which, however, retained sufficient cohesion to prevent a Persian reconquest. It is practically certain that Sāsānian authority never extended beyond Balkh and rarely as far. In the fourth and fifth centuries references are made to a fresh horde of nomads in the north-east, the Juan-Juan (Chionitae, Avars)[3], but it does not appear that any new settlements were made in the Oxus countries.
In the middle of the fifth century, another people, the Ephthalites (Arabic Haytal, Chinese Ye-Tha), perhaps a branch of the Hiung-nu, not only completely overran the former Kushan territories, but by successive defeats of the Persian armies forced the Sāsānid Kings to pay tribute. The Ephthalites appear to have been a nomadic people organised as a military caste of the familiar Turkish type, and the existing institutions and principalities, in large part at least, continued side by side with them[4]. Their rule was too transitory to produce any lasting effects, or to inflict any serious injury on the commerce and prosperity of Sogdiana.
The rise of the Central Asian empire of the Turks proper (Tu-Kueh) dates from their overthrow of the Juan-Juan in Mongolia in 552, under their great Khan, Mokan. His brother Istämi (the Silzibul of the Byzantine historians), the semi-independent jabghu of the ten tribes of Western Turks, after consolidating his power in the Ili and Chu valleys, formed an alliance with Khusrū Anūshīrwān, and in a joint attack between 563 and 568 the two powers completely overthrew the Ephthalite kingdom and divided their territories. For a brief moment the Oxus was the actual boundary between Īrān and Tūrān; under pressure from the silk traders of Sogdiana, however, the alliance was broken and the weaker successors of Anūshīrwān could scarcely do more than maintain their outpost garrisons on the Murghāb. From this time the Ephthalites, like the Kushans, were gradually assimilating to the Iranian population[5], though the change was less rapid in the Cisoxine lands of Lower Tukhāristān, Bādghīs, and Herāt, where Ephthalite principalities were re-constituted, probably with Turkish support, and continued to give Persia much trouble on her north-eastern frontiers[6]. On the other hand the Turks of the five western tribes (Nu-she-pi), who became independent after the break up of the Great Khanate about 582, maintained their suzerainty over Sogdiana and the middle Oxus basin by frequent expeditions, in one case at least as far as Balkh. There is no trace in our records of extensive Turkish immigration into the conquered lands; at most, small groups of Turks wandered south with their herds, especially, it would seem, south of the Iron Gate[7]. In general, Turkish interference in the administration of the subject principalities was at first limited to the appointment of military governors and the collection of tribute. Thus, in the semi-legendary account given by An-Naysābūrī of the Turkish conquest of Bukhārā the Bukhār Khudāh is represented as the chief dihqān under the Turkish governor. It is possible also that the native princes maintained guards of Turkish mercenaries.
At this period, therefore, so far from the Oxus being a barrier, there was considerable intercommunication between the peoples on either side, and at least the elements of a racial and cultural unity, in spite of political divisions. This is a factor of importance in the history of the Arab conquests: the conquest of Transoxania is intimately linked with the fortunes of Lower Tukhāristān, and only became possible when the latter country was completely subdued. On the other hand, the Jaxartes formed a natural racial and political frontier. “Shāsh and Sughd have seldom run together” says Vámbéry, and in spite of nominal annexations on more than one occasion Muslim rule was not effectively imposed on Shāsh and Farghāna until some time after the final conquest of Transoxania. Their chief importance for the history of Transoxania is that they formed the jumping-off place for counter-invasions from the East. It is not without significance that of the two battles which were decisive in establishing Arab rule in Sogdiana one was fought to the west of Balkh and the other on the Talas river, far into the Turkish lands beyond the Jaxartes (see pp. 84 and 96).
Political Divisions.
Researches into Chinese records have now made it possible to obtain a more definite idea of the political conditions of these frontier provinces in the seventh century. All the principalities acknowledged the Khan of the Western Turks as overlord and paid tribute to him under compulsion, though, as will appear, there is good cause for doubting whether a Turkish army ever came in response to their appeals for support until the rise of the Türgesh power in 716.
Geographically the cultivated lands to the west and south-west of the middle Jaxartes are divided by the Hissar mountains into two well-defined areas. The northern area includes the rich valley of the Zarafshān and the lesser streams which descend the northern slope of the watershed, the southern comprises the broad basin formed by the Oxus and its tributaries between the mountains of the Pamīr and the steppelands. The former, which as a whole is called Sogdiana in distinction from the smaller principality of Sughd, was at this period divided between a number of small states, each independent of the others but forming together a loose confederacy in a manner strikingly reminiscent of the Hellenic city-states. The strongest bond of union was formed by their mutual interest in the Chinese silk trade, the chief stations of which were at Samarqand, Paykand, and Kish. The premier city was Samarqand, the pre-eminence of which and high culture of whose population is vouched for by Yuan Chwang. Special emphasis is laid on their enterprise and success in trade, and a number of early embassies, doubtless commercial missions, are recorded from Samarqand and Bukhārā to the Chinese court. The merchant families of Paykand, according to Tomaschek’s rendering of Narshakhī[8], were Kushans, but Iranian elements, reinforced by emigrants from the Sāsānid dominions, formed the majority in the cities. The agricultural population was almost if not entirely Iranian.
A second link between the majority of the cities was formed by the ruling house of the Shao-wu, if, as the Chinese records assert, these all belonged to one royal family. The head of the clan governed Samarqand and was allied by marriage to the Turkish Khan; cadet branches ruled in Ushrūsana, Kish, Bukhārā, and the lesser principalities in the basin of the Zarafshān. In the later lists the rulers of Shāsh and Farghāna as well as the Khwārizm Shāh are shown as belonging to the clan also, though with less probability[9]. Whether the family were of K’ang origin, or, as the Chinese records state, belonged to the Yueh-Chi, they appear in the Arabic histories with Persian territorial titles (Khudāh, Shāh, and the general term dihqān). Some of the princes also possessed Turkish titles, probably conferred on them as vassals of the Khan. The ruler of Samarqand, as king of Sughd, is called the Ikhshīdh or Ikhshēdh, which is easily recognised as the Persian khshayathiya. This title was borne also, as is well known, by the king of Farghāna. It is certain at least from both Chinese and Arabic accounts that these rulers were not Turks. The Turkish names by which they are sometimes called were given out of deference or compliment to their Turkish suzerains, just as Arabic names begin to appear amongst them immediately after the Arab conquests. Particularly misleading is the name Tarkhūn which appears more than once in the list of princes of Samarqand and has been erroneously taken as the title Tarkhān, though it is in reality only the Arabic transcription of a personal name spelt in the Chinese records Tu-hoen. During the six or seven hundred years of their rule all these princes had become fully identified with their Iranian subjects[10]. The “kingship” moreover was not a real monarchy but rather the primacy in an oligarchical system. Their authority was far from absolute, and the landed aristocracy (dihqāns) and rich merchants possessed, as will be seen later, not only a large measure of independence but also on occasion the power to depose the ruling prince and elect his successor. As the succession appears to have been largely hereditary it is probable that, according to Iranian custom, eligibility was confined to a single royal house. In some cases, it would seem, the succession was regulated during the lifetime of the reigning prince by some such method as association in the principate, probably combined with the appointment of the remaining princes to other fiefs[11].
The “confederacy” of these states, however, was in no sense an alliance and probably amounted to little more than a modus vivendi. Besides the more important princes there existed an enormous number of petty autocrats, some possibly Turkish, others probably descended from former conquerors, whose authority may sometimes have scarcely extended beyond the limits of their own villages. In lands subject to the Turks and patrolled by nomadic tribes an effective centralised government was hardly possible. Mutual antagonisms and wars cannot have been uncommon though we have now no record of them, except that during the early Arab period there was hostility between Bukhārā and Wardāna, but the latter cannot be reckoned among the Shao-wu principalities since, according to Narshakhī, it was founded by a Sāsānid prince about 300 A.D. Until the profitable Chinese trade was threatened by the Arabs we find no trustworthy record of combined resistance offered by the country to its piecemeal reduction, and only long after the conquests of Qutayba is there any hint of a concerted rising. At the same time, the strength of the cities and warlike nature of their inhabitants may be gauged from the way in which they not only preserved themselves from destruction at the hands of their successive nomad invaders, but even gained their respect, while this, in some respects perhaps the most highly civilised of all the lands subdued by the Arabs[12], proved also the most difficult to conquer, and most intractable to hold.
The same lack of unity is apparent in the districts south of the Iron Gate, though nominally subject to a single authority. It is important to bear in mind that the Zarafshān and Oxus valleys were completely independent of one another—that the difference between them was not merely one of government, but also of language, and even, to some extent, of blood, owing to the greater mixture of races in the southern basin. When, occasionally, as in the “Mūsā legend”, reference is made in the Arabic histories to common action by Sughd and Tukhāristān, it is due to a complete misunderstanding of the state of the country prior to the conquest, and it is worthy of notice that no such reference is to be found in any narrative otherwise reliable. On his outward journey in 630, Yuan Chwang found the country divided into twenty-seven petty states under separate rulers, with the chief military authority vested in the Turkish Shād, the eldest son of the Jabghu of the Western Turks, who had his seat near the modern Qunduz. During the period of anarchy which befell the Western Turks in the following years, the whole district was formed into an independent kingdom under a son of the former Shād, who founded the dynasty of Jabghus of Tukhāristān. Minor Turkish chiefs and intendants probably seized similar authority in their own districts, and though the Jabghu was recognised as suzerain of all the lands from the Iron Gate to Zābulistān and Kapisa and from Herāt to Khuttal[13], his authority was little more than nominal except within his immediate district of Upper Tukhāristān. The lesser princes, in Shūmān, Khuttal, &c., many of whom were Turkish, appear to have acted quite independently and did not hesitate to defy their Suzerain on occasion. The name Tukhāristān is used very loosely in the Arabic records, with misleading effect[14]. How relatively unimportant to the Arabs Tukhāristān proper was is shown by the fact that its annexation (see below [p. 38]) is passed over in silence. The brunt of the resistance offered to the early Arab conquests was borne by the princes of Lower Tukhāristān, i.e., the riverain districts south of the Iron Gate, including Chaghāniān and Balkh, together with the Ephthalite principalities in Jūzjān, Bādghīs, and Herāt, and possibly the mountainous fringe of Gharjistān. This explains why the Arabs always regarded Balkh, the old religious capital of the Kushan Empire and site of the famous Buddhist shrine of Nawbahār, as the capital of the “Turks”; it was in fact the centre of what we might almost term the “amphictyony” of Lower Tukhāristān, combining strategic and commercial importance with religious veneration. Long after the Nawbahār had been destroyed by Ibn ʿĀmir this sentiment continued to exist in the country[15].
A chance narrative in Tabarī (II. 1224 f.), which, though of Bāhilite origin, can scarcely have been invented, indicates the situation in Lower Tukhāristān in 710. In the presence of Qutayba, the Shād and as-Sabal (King of Khuttal) do homage to the Jabghu, the former excusing himself on the ground that though he has joined Qutayba against the Jabghu, yet he is the Jabghu’s vassal. The Ephthalite prince of Bādghīs then does homage to the Shād, who must consequently be regarded as the chief prince in Lower Tukhāristān. His identification with the Jabghu himself in another passage (Tab. II. 1206. 9) is obviously impossible. Though certainty on the point is hardly to be expected, the description best suits the king of Chaghāniān (Chāghān Khudāh), who consistently adopted an attitude of co-operation with the Arabs. It would seem too that the king of Chaghāniān commanded the armies of Lower Tukhāristān in 652 and again in 737. Moreover, an embassy to China on behalf of Tukhāristān in 719 was actually despatched by the king of Chaghāniān, which implies that he held a status in the kingdom consonant with the high title of Shād. The conclusion drawn by Marquart and Chavannes that the king of Chaghāniān and the Jabghu were identical is disproved by the Chinese records[16].
Such conditions of political disunion were naturally all in favour of the Arabs. It might have seemed also that the general insecurity, together with the burden of maintaining armies and courts and the ever-recurring ravages of invasion, would move the mass of the population to welcome the prospect of a strong and united government, more especially as so large a proportion of the Muslim armies were composed of their Persian kin. For the Arabic records in general are misleading on two important points. By their use of the word “Turk” for all the non-Persian peoples of the East, they give the impression (due perhaps to the circumstances of the time in which the chief histories were composed) that the opponents of the Arabs in Transoxania were the historical Turks. The truth is that until 720 the Arab invaders were resisted only by the local princes with armies composed almost entirely of Iranians, except perhaps on one or two special occasions when Turkish forces may have intervened. The other error is in interpreting the conquests as primarily wars for the Faith. Rebellion, for instance, is expressed in terms of apostasy. It is now well established that this conception is exaggerated; religious questions did not, in fact, enter until much later and even then chiefly as expressions of political relationships. To the Iranian peasantry, themselves steadfastly attached to the national cults, the advent of another faith in this meeting-place of all the cultures and religions of Asia at first carried little significance. Two factors in particular combined to provoke a resistance so stubborn that it took the Arabs a century merely to reduce the country to sullen submission. The first of these was the proud national spirit of the Iranians which was eventually to break down the supremacy of the Arabs and give birth to the first Persian dynasties in Islām. The few wise governors of Khurāsān found in this their strongest support, but, outraged again and again by an arrogant and rapacious administration, the subject peoples became embittered and sought all means of escape from its tyranny. The second was the interest of the commercial relations on which the wealth and prosperity of the country depended. This again might have disposed the cities to accept a rule which promised not only stability, but a wide extension of opportunity. The Arab governors, as we shall see, were not indeed blind to this, but the exactions of the treasury, and still more the greed of local officials, combined with the unsettlement of constant invasion to create an attitude of distrust, which deepened later into despair. It must not be forgotten that the commercial ties of the Sogdians were much stronger with the East than with the West, and that this too prompted them to cultivate relations with the Turks and Chinese rather than with the Arabs when the necessity of making a choice was forced upon them.
The Arabic Sources.
The early Arabic sources are remarkably rich in material for the reconstruction of the conquests in Khurāsān and Transoxania. For the earlier period the narratives of Yaʿqūbī and Balādhurī are nearly as full as those of Tabarī, but the special value of the latter lies in his method of compilation which renders the traditions amenable to critical study and thus provides a control for all the others. Moreover, while the other historians, regarding the conquests of Qutayba as definitely completing the reduction of Transoxania, provide only meagre notices for the later period, Tabarī more than compensates for their silence by the enormous wealth of detail embodied in the accounts he quotes from Al-Madāʾinī and others of the last thirty years of Umayyad rule. As a general rule, these three historians rely on different authorities, though all use the earlier histories of Al-Madāʾinī and Abū ʿUbayda to some extent. The monograph of Narshakhī (d. 959 A.D.) based on both Arabic and local sources, with some resemblance to Balādhurī, is unfortunately preserved only in a Persian version of two centuries later which has obviously been edited, to what extent is unknown, but which probably represents the original as unsatisfactorily as Balʿamī’s Persian version of Tabarī. Even so it preserves to us some account of the peoples against whom the Arab invaders were matched, and thus does a little to remedy the defects of the other historians in this respect. It may well be doubted, however, whether some of its narratives merit the reliance placed upon them by van Vloten[17]. The much later historian Ibn al-Athīr introduces very little new material, but confines himself for the most part to abridging and re-editing the narratives in Tabarī, with a tendency to follow the more exaggerated accounts. The geographer Ibn Khūrdādhbih gives a list of titles and names, which is, however, too confused to supply any reliable evidence.
Reference has already been made to certain aspects of the conquests in which the Arab historians are misleading. Their information on the Turks and the principalities of Sogdiana can now, fortunately, be supplemented and parts of their narratives controlled from Chinese sources, chiefly through Chavannes’ valuable “Documents sur les Tou-Kiue (Turcs) Occidentaux.” But there are two other facts which also demand attention: one, that the Arabic authorities, as we possess them, and even with all allowance made for their limitations, are by no means exhaustive; i.e., reliance on omissions in the narratives is an unsafe principle of criticism: the other, that by critical study it is possible to distinguish at certain points several lines of tendentious tradition or legend, directed to the interests of national feeling or of some particular tribe or faction, and centred in some cases round specific persons. These may most conveniently be summarised as follows:
1. A Qaysite tradition, centred on the family of Ibn Khāzim:
2. An Azd-Rabīʿa tradition, centred on Muhallab and hostile to Hajjāj. This became the most popular tradition among the Arabs, and is followed by Balādhurī, but opposed by Yaʿqūbī:
3. A Bāhilite tradition, centred on the tribal hero, Qutayba b. Muslim. In general it found little favour but is occasionally quoted somewhat sarcastically by Tabarī.
4. A local Bukhārā tradition, followed by Yaʿqūbī, Balādhurī and Narshakhī. It presents the early conquests under the form of an historical romance, centred on the Queen Khātūn in the part of a national Boadicea. Other local traditions, which are frequently utilised by Tabarī, seem to be much more free from serious exaggeration:
5. The few notices in Dīnawarī follow an entirely divergent and extremely garbled tradition from unknown sources, which may for the most part be neglected:
6. The quotations made by Balādhurī (e.g. 422. 10) from Abū ʿUbayda show the influence of a rewriting of episodes with an anti-Arab bias, directed to the interests of the Shuʿūbīya movement, in which Abū ʿUbayda was a prominent figure[18].
7. In the later period, there appears also the fragments of a tradition of which Nasr b. Sayyār is the hero.
Some, if not all, of these traditions developed in some detail, and where they are not balanced by other versions they present a distorted narrative of events, verging in some cases on the fictitious. The most noteworthy examples of this are the Khātūn legend (see below [p. 18]) and the typical story of the exploits of Mūsā b. Khāzim in Transoxania in a style not unworthy of Bedouin romance[19]. It is therefore most important to disentangle these variant traditions and assign its proper value to each. The Bāhilite accounts of Qutayba’s conquests, for instance, contain wild exaggerations of fact, which, nevertheless, have sometimes been utilised in all seriousness by modern historians, amongst other purposes to establish synchronisms with the Turkish inscriptions[20].
With these precautions, it is possible to follow up and reconstruct, with comparative certainty and completeness, that progress of the Arab arms in Central Asia whose vicissitudes are outlined in the following pages.
Notes
(Full Titles in Bibliography)
[1] Franke, Beiträge 41 ff., 67. Cordier, Chine I, 225.
[2] If Marquart’s identification (Ērānshahr, 201 f.) is correct.
[3] Cordier I. 229: Ērānshahr 50 ff.
[4] Yuan Chwang I. 103. Prof. Barthold suggests that the connection between the Ephthalites and the Huns may have been political only, not racial.
[5] Chavannes, Documents 155: Ērānshahr 89.
[6] Tab. I. 2885. 13 and 2886. 3: Yaʿqūbī, History, II, 193: Yāqūt (ed. Wüstenfeld) I. 492: Balādhurī 403: Ērānshahr 65 f., 77 f., and 150. Bādghīs was still a nomad pasture-ground in the XIVth century: Ibn Battūta, III, 67 f.
[7] Yuan Chwang I. 105; II. 266; Chav. Doc. 161: Ērānshahr 250 ff.
[8] Tomaschek, Soghdiana, 170.
[9] See Marquart, Chronologie, 71: Shiratori in Keleti Szemle III (1902) footnote to pp. 122-3.
[10] Cf. Narshakhī 29. 4. On the Iranisation of nomadic elements, Blochet, Introduction à l’Histoire des Mongols, (Leyden, 1910) p. 211 note; Peisker, The Asiatic Background, pp. 353-6.
[11] Chavannes, Notes 91, and cf. below [p. 80].
[12] Cf. Barthold, in Zeitschrift für Assyriologie XXVI (1911) p. 262.
[13] Yuan Chwang I, 75 n. 2, 102 ff: II 270: Chav. Doc. 200 f.
[14] E.g. Tab. II, 1448, 7-10: cf. Ērānshahr 228.
[15] Cf. Yaʿqūbī, Geog. 287: Tab. II 1205. 12: Ērānshahr 66, 87 ff.
[16] Chavannes, Doc. 201, Note 37.
[17] Narshakhī’s unreliability is even more marked in his account of the origins of the Sāmānid dynasty: cf. Barthold, Turkestan 215 n. 3.
[18] See Goldziher, Muhammadanische Studien, I, 195 ff.
[19] Prof. Barthold has drawn my attention to the fact that the story of Mūsā also includes (twice over) an episode from the popular legend of Zopyrus. See his article in Zapiski XVII (1906) 0141, and Wellhausen, Arabische Reich, 257, 265.
[20] E.g. Marquart, Chronologie, p. 8.