The Arab conquests in Central Asia

The Expeditions into the Jaxartes Provinces.

It might perhaps have been expected that Qutayba’s next object after the capture of Samarqand would be to establish Arab authority in Sughd as firmly as had been done in Bukhārā. It would probably have been better in the end had he done so, but for the moment the attractions of the “forward policy” which had already proved so successful were too strong. Instead of concentrating on the reduction of Sughd, it was decided to push the frontiers of the Empire further into Central Asia, and leave the former to be carried out at leisure. Qutayba therefore crossed to Bukhārā, where 20,000 levies from Khwārizm, Bukhārā, Kish, and Nasaf had been summoned to meet him, and marched into Sughd. If there was a Turkish army wintering in the country, it offered no considerable resistance to the advance of the Arabs. In Sughd Qutayba divided his forces into two corps. The Persian levies were sent in the direction of Shāsh, while he himself with the Arabs marched on Khujanda and Farghāna. Our information is brief and lacking in detail. Of the northern expedition we are told only that they captured Shāsh and burnt the greater part of it. Qutayba’s own force had to overcome some resistance at Khujanda, but eventually reached Kāsān, where it was rejoined by the other. The geographers refer also to a battle fought by Qutayba at Mīnak in Ushrūsana, but against whom is not clear[63]. Tabarī (1440. 7) preserves a tradition that Qutayba appointed an Arab resident, ʿIsām b. ʿAbdullah al-Bāhilī, in Farghāna. If this is true, as seems not unlikely, the appointment was probably made during this year. The details of the tradition are quite unacceptable, however. No Arab governor would ever have taken up his residence in a hill-pass in the remotest district of Farghāna, completely cut off from his fellow-countrymen. One of Balādhurī’s authorities carries this or a similar tradition further by crediting Qutayba with the establishment of Arab colonies as far as Shāsh and Farghāna. Here again at most only temporary military outposts can be in question. On the other hand, the extraordinary success achieved by the Arabs on this expedition is apt to be overlooked, and Qutayba might well have imagined, as he returned to Merv, that the latest conquests were as permanently annexed to Khurāsān as Samarqand and Khwārizm.

The helplessness of their Turkish suzerain in face of the victorious Arabs, however, caused a revival in Transoxania of the tradition of Chinese overlordship. Appeals to the Khāqān were of no avail, and in the minds of the Sogdian princes, seeking for some counterpoise to the rapid extension of the Arab conquests, the idea of appealing directly to the Emperor was slowly maturing. Though no definite steps in this direction had as yet been taken, some inkling of it may have reached Qutayba. The Arabs were now familiar with China through the sea-borne trade of the Persian Gulf and at least after, if not before, their conquest of the cities which were already becoming the headquarters of Central Asian commerce, must have become aware of the close commercial relations which these cities maintained with China. Under these circumstances, Qutayba (or possibly Hajjāj) decided to send a mission overland to the Chinese court, possibly to prevent their intervention in the West, but more probably with the intention of promoting trade relations. As the princes of Sogdiana and Tukhāristān were much more alive to the advantages of preserving their commerce and to the dangers which might befall it under the new government than the Arabs could have been, it was probably on their suggestion that the embassy was sent. They would, of course, have no difficulty in persuading governors of the character of Hajjāj and Qutayba that their own interests also lay in safeguarding and encouraging the trade which brought such wealth to Transoxania. If the intervention of the Turks had been caused by their concern for Sogdian trade, it became doubly important for the Arabs to show their practical interest in its welfare. Apart from the immediate gain to the treasury which would accrue, such an action might reasonably be expected to secure the acquiescence of the Sogdians in Arab rule. The date of the mission is fixed as 713 by the Chinese records, which add also that in spite of the refusal of the envoys to perform the customary kow-tow it was favourably received by the Emperor. Both statements are confirmed by Tabarī’s remark that the leader was sent to Walīd on his return, which must therefore be dated between the death of Hajjāj and the end of 714[64]. Unfortunately the Arab records of the mission have been confused with the legendary exploits of Qutayba two years later, becoming so disfigured in the process as to be almost worthless. The wisdom of this step must have been justified by its results, though there are no effects apparent in our histories and the relentless march of Chinese policy was not affected. This embassy is mentioned by the Arabic historians as if it were an isolated incident, but it was, as I have shown elsewhere[65], only the first of many such sent by the governors of Khurāsān to maintain friendly relations with the Chinese court. It cannot be doubted that in the majority of cases at least the object of these missions was commercial, particularly where joint embassies were sent with one or other of the Sogdian principalities.

In the following year 95/714 the raids on the Jaxartes provinces were renewed. It would seem on comparing Balādhurī’s account with Tabarī that Qutayba made Shāsh his headquarters and worked northwards as far as Isbījāb. The prince of Shāsh appealed to China for assistance, but without effect[66]. Qutayba’s plan therefore was to follow up the important trade-route which led from Turfan down the Ili valley, along the northern edge of the Thian-Shan mountains, through Tokmak and Tarāz into Shāsh and Samarqand. Though the economic importance of controlling this trade-route may have had its part in this decision, especially in view of their new patronage of Sogdian trade, it is probable that this was less in the mind of the Arabs than its strategic value as the road by which the Central Asian Turks debouched on Transoxania. Towards the end of the summer, the expeditions were abruptly interrupted by the news of the death of Hajjāj, which had occurred in Shawwāl (June). Deeply affected by the loss of his patron and not a little uncertain of the effect on his own fortunes, Qutayba disbanded the army, sending garrisons to Bukhārā, Kish, and Nasaf, and returned to Merv. Walīd, however, allayed his fears by an encouraging letter, and made his province independent of ʿIrāq. But the death of Hajjāj had affected Khurāsān too deeply for such a simple remedy. The Arabs had gained wealth in their expeditions, they were weary of the constant campaigns and anxious to enjoy the comforts of peace. Factional feeling was merely slumbering, and a new element of unrest had been added by a Kūfan corps under Jahm b. Zahr, which had been transferred to Khurāsān from India by Hajjāj in his last year. All parties among the Arabs were alienated from Qutayba; even Qays had been estranged by his highhanded action in the first place with the house of Al-Ahtam and again by his feud with Wakīʿ b. Abī Sūd, the chief of Tamīm[67]; moreover, they were suspicious of his medizing tendencies. Amongst the Persians he was popular, but Hayyān an-Nabatī, though restored to his position in command of the Persian troops, had not forgiven Qutayba for his disgrace at Khwārizm. It seems extraordinary that the general himself should have been blind to any internal danger and was entirely confident in the loyalty of his army.

On re-opening the campaign in 96/715, therefore, his only precautions consisted in the removal of his family and personal property from Merv to Samarqand and the posting of a guard on the Oxus, in view of a possible restoration to favour of Yazīd b. Muhallab. It is unlikely that Qutayba could have had in mind the possibility of Walīd’s death; what he feared was more probably a rapprochement between the Caliph and his heir Sulaymān, who was his bitter enemy.

The object of this last campaign was probably the complete subjugation of Farghāna. Having established his authority over the important section of the Middle Jaxartes and its trade route, it remained now to round off his conquests by extending it also over the central trade route between Farghāna and Kashgaria. The account which Tabarī intends to convey, however, is that Qutayba marched first into Farghāna and from there led an expedition against Kashgar, with complete success. In an article of mine published in the Bulletin of the School of Oriental Studies (II. 467 ff.), all our evidence for this expedition has been critically discussed, and shown to be against the authenticity of the tradition. It is unnecessary, therefore, to do more than summarise very briefly the arguments there put forward. (1) None of the historians earlier than or contemporary with Tabarī contain any reference to a raid on Kashgar, and even Tabarī’s own statement is not borne out by the authorities on which it professedly rests. Only one of these relates an expedition to Kashgar, and that under the command of an unknown leader. (2) The interval between the opening of the campaign and the death of Qutayba in Farghāna in August or September does not allow time for such an expedition, especially in view of the mutinous attitude of the army after the death of the Caliph. (3) The Chinese account of Arab interference in Farghāna cannot refer, for chronological reasons, to Qutayba’s expedition, and in any case is silent on any attack on Kashgaria.

That an expedition of this sort should have been attributed to Qutayba is not surprising, in view of the tradition of the embassy to China, and of the great renown which attached to his memory. Later tradition[68] recounted that Hajjāj pledged the governorship of China to the first to reach it of his two governors in the East, Muhammad b. Qāsim and Qutayba. “Sīn” was, of course, not the sharply defined country of our days, but rather a loose term for the Far East, including even the Turkish lands in the North-East. Qutayba had probably done little more than make preparations for his campaign, perhaps to the extent of sending out minor raiding expeditions, when the news of the death of Walīd brought everything to a standstill.

The historians give the most contradictory accounts of the events that followed; according to Balādhurī the new Caliph Sulaymān confirmed Qutayba in his command but gave permission to the army to disband. Tabarī’s narrative, with which Yaʿqūbī’s in general agrees, is fully discussed by Wellhausen (274 ff.), together with a valuable analysis of Qutayba’s position. The story of his highhanded negotiations with Sulaymān is too well known to need repetition. Finding the army disinclined to follow him, he completely lost his head and roused the mutiny in which he was killed. The Persian levies, who were inclined to side with him, were dissuaded by Hayyān an-Nabatī, and at the last only his own family and bodyguard of Sogdian princes remained faithful.

The death of Qutayba marked not merely the end of the Arab conquests in Central Asia for a quarter of a century, but the beginning of a period of retrogression. Under Wakīʿ b. Abī Sūd, his successor[69], the armies melted away. Mukhallad, the son of Yazīd b. Muhallab and his lieutenant in Transoxania, carried out summer raids on the villages of Sughd, but an isolated attempt on the Jaxartes provinces by ʿOmar’s governor, Al-Jarrāh b. ʿAbdullah, met with ignominious failure. It is possibly to this that the tradition, mentioned by Barthold (Turkestan 160), of the disaster met with by a Muslim army refers. On the other hand an embassy was sent in the name of the Caliph to renew relations with the Chinese court, and a third in concert with the kingdoms of Tukhāristān and Samarqand, etc., during the reign of ʿOmar[65]. There is mention also of an expedition into Khuttal which regained some territory. But it was Qutayba, with Hajjāj at his back, who had held his conquests together, and when he disappeared there was neither leader nor organisation to take his place. The history of the next decade clearly shows how loose and unstable was the authority of the Arabs. It was force that had made the conquests, and only a settled policy of force or conciliation could hold them. The first was absent. “Qutayba in chains at the world’s end is more terrible to us than Yazīd as governor in our very midst” is the graphic summary put into the mouths of the conquered, while of Rutbīl, king of Zābulistān, we are told expressly that after the death of Hajjāj “he paid not a cent of tribute to any of the governors of Sijistān on behalf of the Umayyads nor on behalf of Abū Muslim.”[70].

Nor was ʿOmar’s policy a true policy of conciliation, based as it was not on the maintenance of the Arab conquests but on the complete evacuation of Transoxania. His orders to that effect were of course indignantly rejected by the Arab colonists in Bukhārā and Samarqand, but together with his appointment of the feeble and ineffective ʿAbdur-Rahmān b. Nuʿaym al-Qushayrī as governor, such a policy was naturally construed by the Sogdians as mere weakness, and an invitation to regain their independence. In addition to the embassies to China, to be related in the next chapter, and possibly also some negotiations with the Türgesh, Ghūrak sought to win back his capital by playing on ʿOmar’s piety. The Caliph sent envoys to the princes of Sogdiana calling on them to accept Islām, and Ghūrak, outwardly professing his adherence, sent a deputation to ʿOmar urging that as “Qutayba dealt with us treacherously and tyrannically, but God has now caused justice and equity to reign” the city should be restored to the Sughdians. The commonsense of the judge appointed to try the case on ʿOmar’s instructions by the governor of Samarqand, Sulaymān b. Abiʾs-Sarī (himself a mawlā), solved the problem in an eminently practical manner, and we are told that his decision, so far from being “malicious,” was satisfactory to both the Arabs and the Sughdians, if not perhaps to Ghūrak. Beyond the remission of kharāj, it is doubtful whether ʿOmar’s administration benefited the subject peoples in the slightest, and the reaction which followed his brief reign only aggravated the situation. Already before its close the Sughdians had withdrawn their allegiance[71].

Thus within six years from the death of Qutayba, much of his work was undone. He had laid the foundations on which the later rule of Islām was built, and laid them well, though his own superstructure was too flimsy to withstand the tempests of the years ahead. But the fault was not entirely, perhaps not even chiefly, the fault of the builder. He was snatched away before his work was done, even if in his latter years he tended to neglect everything else for military glory. As we shall see, there was no peace in Transoxania until other men arose, great and strong enough to adopt and carry out the best of his plans. The ruthlessness and ferocity of his conquests, however, have been much exaggerated. He was always ready to use diplomacy rather than force if it offered any hope of success, so much so that his lenience was misconstrued on occasion by both friends and foes. Only in cases of treachery and revolt his punishment came swift and terrible. That he did not hesitate to take vengeance on his private enemies is to say no more than that he was an Arab. It was not without reason that in later days the Muslims of Central Asia added Qutayba’s name to the roll of martyrs and that his tomb in Farghāna became a favourite place of pilgrimage[72].

To sum up the position in Central Asia in the years immediately following Qutayba’s conquests:—

Notes

IV. THE TURKISH COUNTERSTROKE.[74]

The princes of Transoxania had so long been accustomed to regard the Arabs as mere marauders that it was some time before they could realise the loss of their independence. Though necessity forced them at first to adopt a conciliatory spirit (as, for example, in their acceptance of Islām under ʿOmar II), they were dismayed to find all the machinery of permanent occupation set in motion, and their authority flouted by tactless and greedy Arab officials. Such a state of affairs was tolerable only in the absence of any countering force. The situation was not stationary for long, however; even before Qutayba’s death other and disturbing factors had begun to enter. Our best clue to the complications in Transoxania during this period is the attitude of Ghūrak, king of Sughd, of whose movements, fortunately, sufficient indications have been preserved. In maintaining a precarious balance between the Türgesh and the Arabs, his true statesman’s instinct seldom misled him in judging how and when to act to advantage throughout his troubled reign. In addition to this we have the evidence, unreliable in detail but confirmatory in the mass, of the embassies sent by the subject principalities to the Chinese court. Doubtless they were despatched in the guise of commercial missions and in many cases were truly so, but that they frequently possessed a political character can hardly be denied. The dates of these embassies as given in the authorities translated by Chavannes fall naturally into four periods. In the following list all embassies have been omitted in which the Arabs are known to have participated or whose object is known to have had no connection with the Arab conquests, as well as those which appear to be duplicated, and those from the minor states:

Number of Embassies from:—

These four periods, as will be seen, closely correspond to the fluctuations of Arab authority in Transoxania.

In the same year, 713, that Qutayba first led his army across the Jaxartes, a new era of westward expansion opened in China with the accession of Hiuen-Tsong. In 714 the Chinese intervened in the affairs of the Ten Tribes and obtained their immediate submission, while in the following year they restored the deposed king of Farghāna. In 716, on the death of Me-chuʾo, Khan of the Northern Turks, the powerful tribes of the Türgesh asserted their independence, and under their chief Su-Lu established, with Chinese assistance, a new kingdom in the Ili basin. The princes of Transoxania eagerly sought to profit by these developments to free themselves from the Arab yoke. In 718 a joint embassy was sent to China by Tughshāda, Ghūrak, Narayāna king of Kumādh, and the king of Chaghāniān. The first three presented petitions for aid against the Arabs, which are given in full in Chavannes’ Documents. Tughshāda asked that the Türgesh might be ordered to attack the Arabs, Ghūrak related the capture of Samarqand and asked for Chinese troops, Narayāna complained of the seizure of all his treasures by the Arabs and asked that representations might be made to induce them to remit their crushing taxation. It is significant that the king of Chaghāniān, acting for his suzerain, the Jabghu of Tukhāristān, did not compromise himself by joining in these requests. But beyond “fair words” the son of Heaven took no action, and no Chinese forces appeared West of the Jaxartes, in spite of the repeated entreaties addressed by the princes to their self-elected suzerain.

The Türgesh, however, were not long in intervening on their own account. Whatever opportunity the Arab government had to pacify the Sughdians was lost by a succession of incompetent governors. Already in the reign of ʿOmar II, as has been seen, they had withdrawn their allegiance from the weak ʿAbdur-Rahmān b. Nuʿaym. For a moment the situation seemed to improve at the beginning of the governorship of Saʿīd “Khudhayna” (102/720) owing to the firm handling of Samarqand by his lieutenant Shuʿba b. Zuhayr. But disturbances broke out and Shuʿba was recalled, perhaps in a vain attempt to appease the insurgents. It would seem that the Sughdians appealed to the new Turkish power in the East and Su-Lu, unable to make headway against the growing influence of China, willingly seized the opportunity of diverting his armies into Transoxania. A small Türgesh force was sent under Köl-chur (called by Tabarī Kūrsūl)[75] to make common cause with the Sughdian rebels in the following spring (end of 102). Saʿīd awoke to find the whole country in arms, a Turkish force marching on Samarqand, and the local princes, with few exceptions, aiding the invaders. The Arab commanders could not rely on their levies and a small garrison at Qasr al-Bāhilī was evacuated only with the utmost difficulty. The tale of their relief by a small force of volunteers is one of the most spirited narratives of adventure in Tabarī. But such episodes did not affect the general success of the Turkish forces. Kūrsūl continued his advance through Sughd without opposition, avoiding Samarqand, until at last Saʿīd was roused by public reproach to march against the Turks. After a small initial success, which he refused to follow up, he was severely defeated and confined to the neighbourhood of Samarqand. The Turks were not strong enough to undertake a siege of the city, as the whole operation seems to have been little more than a reconnaissance in force combined with a raiding expedition. As the Türgesh retired, the Arab cavalry followed them up as far as Waraghsar, the head of the canal system of Sughd. Ghūrak appears to have refrained from committing himself by openly aiding the rebels, and doubtless recognised that the Arabs were not so easily to be dislodged. From the fact that Saʿīd’s camp was pitched at Ishtīkhan, in close proximity to him, it may even be conjectured that he outwardly supported the Arabs.

But the new governor of ʿIrāq, ʿOmar b. Hubayra, was not the man to stand idly by in face of the danger that threatened Khurāsān. The weakness shown by Khudhayna and the complaints of oppression from his subjects, were sufficient reason for his recall, and Saʿīd b. ʿAmr al-Harashī, a man of very different stamp, was installed in his place. The transfer may be placed in the late autumn of 103/721. The new governor’s first act was to summon the rebels to submit, but a large number of nobles and merchants, with their retainers, either fearing that they could expect no mercy, or anxious to free themselves altogether from the Arab yoke, prepared to emigrate to Farghāna. Ghūrak did his utmost to persuade them to remain, but without effect; their absence would no doubt affect the revenues, and a certain emphasis is laid on the point in Tabarī’s account. Leaving hostages behind, the malcontents marched towards Farghāna and opened negotiations with the king for the occupation of ʿIsām. The majority settled in the interval at Khujanda, but other parties actually entered Farghāna, and one body at least occupied a fortified position on the Zarafshān. Al-Harashī followed up his demands by marching into Sughd and encamped near Dabūsia, where he was with difficulty persuaded to stay until sufficient contingents arrived. On advancing, he was met by a messenger from the king of Farghāna, who, outwardly professing to assist the Sughdians, had secretly decided to rid himself of them by calling in the Arabs against them. Al-Harashī eagerly seized the opportunity and pressed forward, receiving the allegiance of Ushrūsana as he passed. The emigrants, although urged by their leader Karzanj either to take active measures or to submit, decided to risk a siege in Khujanda, trusting to the protection of the king of Farghāna. But when Saʿīd set about the siege in earnest, and they realised that they had been betrayed, they surrendered on unexpectedly easy terms. Saʿīd divided them, placing the nobles and merchants in a camp apart from the soldiers. By the execution of Thābit, a noble from Ishtīkhan, he provoked a revolt, under pretext of which he massacred the nobles and the troops, sparing the merchants, who numbered four hundred, only in order to squeeze them of their wealth. Tabarī’s account very thinly veils al-Harashī’s responsibility for this wanton act of atrocious cruelty, which could not fail to embitter the feelings of the whole population of Transoxania. It is curious that the Persian Tabarī (Zotenberg IV. 268) has an entirely different story, which is found in none of the Arabic authorities. The refugees who escaped eventually took refuge with the Khāqān of the Türgesh, where they formed a regiment (no doubt continually recruited from new emigrants) which particularly distinguished itself in the war against the Arabs[76].

The expedition to Khujanda may be put in the spring and summer of 722 (end of 103, beginning of 104), though the chronology here, and indeed for all this period, is uncertain. The piecemeal reduction of the fortresses in Sughd occupied the remainder of the year, a series of operations whose difficulty is sufficient witness to the effect of the news from Khujanda in stiffening the resistance to the Arabs. The first fortress to be attacked was that of Abghar, in which a band of the emigrants had settled. The attack was entrusted to Sulaymān b. Abiʾs-Sarī, with an army composed largely of native levies from Bukhārā, Khwārizm, and Shūmān, accompanied by their princes. Sulaymān persuaded the dihqān to surrender, and sent him to al-Harashī, who at first treated him well in order to counteract the effect of the massacre of Khujanda, but put him to death after recapturing Kish and Rabinjān. The most inaccessible fortress and the crowning example of Al-Harashī’s perfidy were left to the last. The dihqān Subuqrī still held out in the fortress of Khuzar, to the south of Nasaf; unable to take it by force, Al-Harashī sent Musarbal b. Al-Khirrīt, a personal friend of Subuqrī, to offer him a pardon. On his surrender, he was sent to Merv and put to death, although the amnesty, it is said, had been confirmed by ʿOmar b. Hubayra.

The whole of Sughd was thus once more in the hands of the Arabs. The nearer districts, Khwārizm and Bukhārā, had remained loyal and the Oxus basin seems to have been unaffected. But to make a solitude and call it peace did not suit the aims of the Arab government and Al-Harashī found that his “policy of thorough” only provided Ibn Hubayra with an excuse for superseding him. During the winter, therefore, he was replaced by Muslim b. Saʿīd al-Kilābī, who, as the grandson of Aslam b. Zurʿa, came of a house long familiar with Khurāsān. The danger of the movement of revolt spreading to the Iranians of Khurāsān seems to have preoccupied the Arab government during all this period. Saʿīd Khudhayna had poisoned the too-influential Hayyān an-Nabatī on suspicion of rousing the Persians against the government and that it was felt even in Basra may be seen from Ibn Hubayra’s advice to his new governor, “Let your chamberlain be one who can make peace with your mawālī.” Muslim, in fact, favoured the Persians and did all in his power to appoint officials acceptable to them, the Mazdean Bahrām Sīs, for example, being appointed Marzubān of Merv[77]. But all such measures were merely palliatives and could not materially affect the growing discontent in Sughd and Tukhāristān. During his first year of office it is recorded (if the narrative is not, as Wellhausen thinks, a duplicate of the raid on Farghāna in the following year) that Muslim marched across the river but was met and pushed back into Khurāsān by a Turkish army, narrowly escaping disaster. It is not improbable that the local forces were again assisted by Türgesh on this occasion. In the following year, however, before the close of 105, a second expedition gained some success at Afshīna, near Samarqand. Meanwhile Hishām had succeeded Yazīd II as Caliph, and ʿOmar b. Hubayra, whose Qaysite leanings were too pronounced, was recalled in favour of Khālid b. ʿAbdullah al-Qasrī of Bajīla. The transfer took place most probably in March (724), though another account places it some months later. Muslim was now preparing an expedition into Farghāna, but the Yemenite troops at Balkh held back partly through dislike of the campaign and doubtless expecting the governor’s recall. Nasr b. Sayyār was sent with a Mudarite force to use compulsion; the mutinous Yemenites were defeated at Barūqān and unwillingly joined the army. It is noteworthy that troops from Chaghāniān fought alongside Nasr in this engagement. Before leaving Bukhārā Muslim learned that he was to be superseded, at the same time receiving orders to continue his expedition. Four thousand Azdites, however, took the opportunity of withdrawing. The remainder, accompanied by Sughdian levies, marched into Farghāna, crossed the Jaxartes, and besieged the capital, cutting down the fruit trees and devastating the land. Here news was brought that Khāqān was advancing against them, and Muslim hurriedly ordered a retreat. The Arabic accounts graphically describe the headlong flight of the Arabs. On the first day they retired three stages, the next day they crossed the Wādī Sabūh, closely pursued by the Türgesh; a detachment, largely composed of mawālī, which encamped separately, was attacked and suffered heavy losses, the brother of Ghūrak being amongst the killed. After a further eight days’ march, continually harassed by the light Turkish horse, they were reduced to burning all the baggage, to the value of a million dirhems. On reaching the Jaxartes the following day, they found the way barred by the forces of Shāsh and Farghāna, together with the Sughdians who had escaped from Saʿīd al-Harashī, but the desperate and thirsty troops, hemmed in by the Türgesh from behind, cut their way through. The rearguard made a stand, but lost its commander. At length the remnants of the army reached Khujanda, where ʿAbdur-Rahmān b. Nuʿaym took command on behalf of Asad b. ʿAbdullah, and made good his retreat to Samarqand.

This disaster, which is known as the “Day of Thirst,” marks a period in the history of the Arab conquests. It was practically the last aggressive expedition of the Arabs into Transoxania for fifteen years, but of much greater importance was the blow which it struck at Arab prestige. The rôles were reversed; from now onwards the Arabs found themselves on the defensive and were gradually ousted from almost every district across the Oxus. No wonder, therefore, that the memory of the “Day of Thirst” rankled even long after it had been avenged[78]. According to the Arab tradition, the Türgesh armies were led on this occasion not by Su-Lu himself, but by one of his sons. Unfortunately the accounts of Su-Lu in such Chinese works as have been translated are silent on his Western expeditions, and the Arab historians are our only authorities. The immediate result of the Arab defeat, not only in Sughd but in Tukhāristān and the southern basin as well, was to stiffen the attitude of passive resistance to the Arabs to the point at which it only needed active support to break into a general conflagration. From this time, if not before, the subject princes regarded the Türgesh as the agents of their deliverance, commissioned by China in response to the urgent entreaties they had addressed to the Emperor for aid in their struggle. We find this actually expressed in a letter sent three years later by the Jabghu of Tukhāristān, which is, in Chavannes’ words “but one long cry of distress”[79]. “I am loaded with heavy taxation by the Arabs; in truth, their oppression and our misery are extreme. If I do not obtain the help of the (Chinese) Kagan ... my kingdom will certainly be destroyed and dismembered.... I have been told that the Celestial Kagan has given this order to the Kagan of the Türgesh: To you I delegate the affairs of the Far West; you must at once send soldiers to drive out the Arabs.” The point of view here expressed is of course that of the ruling princes, whose resentment at the curtailment of their authority is understandable. Besides making allowance for some natural exaggeration, it would be dangerous to assume that this was as yet fully shared by the people. In all probability, if we may judge from historical analogies, there was also a pro-Arab party in Sogdiana, who felt that the best interests of the country lay, not in an opposition whose final issue could scarcely be in doubt, but in co-operation with their new masters as far as was possible. The tragedy of the Arab administration was that by alternately giving and refusing co-operation on its side, it drove its supporters in the end to make common cause with its opponents.

But though the situation was steadily deteriorating the decisive moment had not yet come. The new governor, Asad b. ʿAbdullah, seems to have seen something of the danger though factional feeling was running so high that the administration was almost helpless in face of it. He tried to continue Muslim’s policy of conciliation by appointing agents of known probity. Tawba b. Abī Usayd, a mawlā who had been intendant for Muslim, and who “treated the people fairly, made himself easily accessible, dealt uprightly with the army and maintained their supplies,” he persuaded to remain in office under him. Hāniʾ b. Hāniʾ, the financial intendant at Samarqand, was unpopular; he was recalled and Al-Hasan b. Abiʾl-ʿAmarrata of Kinda, who was in sympathy with the mawālī, appointed in his place. With him was associated Thābit Qutna, who had been a leader of some repute under Saʿīd Khudhayna, “gallant warrior, distinguished poet, confidant of Yazīd b. Muhallab, and universally popular”[80]. Still more significant is the fact that one of Asad’s earliest actions was to renew the practice, neglected since the days of ʿOmar II, of sending an embassy to the Chinese court. As before, however, the Arabs resented the favour shown to the Persians, and the military weakness of Ibn Abiʾl-ʿAmarrata roused them to open anger. Strong Turkish forces, probably guerilla bands swollen by refugees and malcontents from the wasted districts, spread over the country and appeared even before Samarqand. The governor made some show of opposition, but avoided coming to grips with them, thus intensifying his unpopularity.

Samarqand indeed was gradually becoming more and more isolated, but no assistance could be given from Khurāsān. During his three years of office Asad’s attention was wholly engaged with the situation in Tukhāristān and the South. Even here his constant expeditions, to Gharjistān, Khuttal, and elsewhere, met with no success. Worse still, in 108/726 he found his forces in Khuttal opposed by the Khāqān with his Türgesh. The princes of Tukhāristān had taken to heart the lessons of the “Day of Thirst”, and the powerful chief who had already all but driven the Arabs out of Sogdiana was now called in to expel them from the Oxus basin as well. Asad visited his failure on the Mudarites, whom he may have suspected of treachery, but the indignation called out by his treatment of such men as Nasr b. Sayyār, ʿAbdur-Rahmān b. Nuʿaym, Sawra b. Al-Hurr, and Al-Bakhtarī, made his recall inevitable. Nor had his measures removed the distrust and hatred of the subject peoples. The land was wasted and desolate[81], the crushing taxation was not lightened, and all Persian governors were not of the stamp of Tawba; many of them were but too ready to rival their Arab rulers in greed and cruelty. Asad may have gained the friendship of many dihqāns[82], but that was an easier matter than to placate the population. In such an atmosphere it was only to be expected that Shīʿite and ʿAbbāsid propaganda, though actively combated by the administration, found a fertile field among the Muslim converts in Khurāsān and Lower Tukhāristān, and was already beginning to undermine the whole fabric of Arab government.

For a moment the hopes of a radical change of policy entertained by the mawālī and the clearer-sighted Arabs were raised to the highest pitch by the appointment (in 109) of Ashras b. ʿAbdullah as-Sulami, accompanied by the separation of Khurāsān from Khālid al-Qasrī’s province of ʿIrāq. It is unnecessary to recapitulate here the far-reaching concessions by which he hoped to secure, and actually did for a time secure the allegiance of the Sughdians, or the methods by which the local princes, especially Ghūrak, succeeded in checking the movement[83]. It is generally assumed that the hostility of Ghūrak was due to the serious fall in revenue which would result. Though this was doubtless the plea put forward and accepted by Ashras it can scarcely have been the true issue. Ghūrak’s aim was not to maintain himself on good terms with the Arab governors but to recover his independence. If once the people became “Arabs” all hope of success must have been lost. It was a game with high stakes and Ghūrak won. It must not be overlooked, however, that the account as we have it is traditional and may often be mistaken on the sequence of cause and effect. The astonishing reversal of the measures adopted by Ashras is more probably to be explained by pressure from above, not from below, and our tradition may really present only the popular view of the Caliph Hishām’s reorganization of the financial administration[84]. The Arabs resorted to brutal methods to wring the taxes from the new converts, and with incredible blindness selected the dihqāns for special indignities. It is not unlikely that Narshakhī’s story of the martyrdom of native Muslims in Bukhārā is connected with this event, though there are many other possible explanations, such as, for example, an attempted Hārithite movement (see below, p. 76 f.) The reaction swung the whole population of Transoxania, dihqāns and peasantry alike, into open rebellion. The first small party of emigrants who quitted Samarqand, although supported by a few Arabs, were induced to surrender and return[85], but within a few months the dreaded Khāqān with his Türgesh had joined forces with the rebels and swept the Arabs across the Oxus. Even Bukhārā was lost[86] and only Samarqand with two minor posts on the Zarafshān, Kamarja and Dabūsia, held out. Ghūrak, however, still supported the Arabs, as Samarqand, although besieged, seems to have been in no danger, while his son Mukhtār, doubtless to keep a footing in the opposite camp, joined with the Türgesh.

The pressing danger sobered the Arabs and temporarily united all parties and factions. The army was concentrated at Āmul but for three months was unable to cross the river in the face of the combined native and Türgesh armies. A small body under Qatan b. Qutayba which had already crossed and fortified itself before the arrival of the Turks was beleaguered. The Turkish cavalry even made raids on Khurāsān with an excess of boldness which was punished by a mounted force under Thābit Qutna. At length Ashras got his forces across and, joining with Qatan b. Qutayba, advanced on Paykand. The enemy cut off the water supply, and had it not been for the gallantry and self-sacrifice of Hārith b. Surayj, Thābit Qutna, and their companions, an even greater and more irretrievable “Day of Thirst” had resulted. In spite of their weakness, Qatan and the cavalry of Qays and Tamīm charged the enemy and forced them back, so that Ashras was able to continue his advance towards Bukhārā. In the heavy fighting the Muslim forces were divided, Ashras and Qatan gave each other up for lost, and Ghūrak judged that the time had come to throw in his lot with the Turks. Two days later, however, the armies were reunited and on the retiral of the Turks encamped at Bawādara outside the walls of Bukhārā, whence they prepared to besiege the city. Ghūrak also retrieved his error and rejoined Ashras. The Khāqān withdrew towards Samarqand, but sat down before Kamarja, expecting to take it by storm in a few days at the most. The Arabic narratives of these events are confused in several places, which has given rise to many incorrect statements, such as that Ghūrak was beleaguered with the Arabs in Kamarja and that the garrison consisted of Qatan and his forces. Kamarja was not in the neighbourhood of Paykand, as Wellhausen states, but a few farsakhs west of Samarqand[87]. When the garrison would not yield to assault Khāqān tried other methods. Accompanying his expedition was Khusrū the son of Pērōz and grandson of Yazdigird, heir of the Sāsānid kings. This prince was sent to parley with the garrison, but when he claimed the restoration of his kingdom and promised them an amnesty, it is not surprising that the Arabs indignantly refused to hear him. Nor would the appearance of a Sāsānid prince evoke much enthusiasm amongst the Iranians of Transoxania. As the Sāsānid house had taken refuge in China, however, the presence of Khusrū might be taken as an indication that the rebels were receiving encouragement from China also, though the Chinese records are silent on this expedition. Khāqān’s second proposal, that he should hire the Arabs as mercenaries, was rejected as derisively as the first. The siege was then pressed with renewed vigour, both sides putting their prisoners and hostages to death, but after fifty-eight days Khāqān, on the advice of the son of Ghūrak and the other Sughdian princes, allowed the garrison to transfer either to Samarqand or Dabūsia. On their choosing the latter, the terms were faithfully carried out after an exchange of hostages.

The fame of the defence of Kamarja spread far and wide, but it brought little relief to the pressure on the Arabs in Transoxania. Even Khwārizm was affected by the movement of revolt, but at the first symptoms of open rebellion it was crushed by the local Muslims, probably Arabs settled in the district, with the aid of a small force despatched by Ashras. The reference made in Tabarī to assistance given to the rebels by the Turks is probably to be discounted, as is done by Ibn al-Athīr. It is of course quite possible that the movement was instigated by the Türgesh, though no such explanation is necessary, but if any Turks were engaged they were probably local nomadic tribes. Ashras seems to have remained before Bukhārā during the winter, possibly in Paykand; the Türgesh probably withdrew towards Shāsh and Farghāna.

In the following year, 730/111-112[88], the attacks on the army of Ashras were renewed. The course of events can only be gathered from the accounts given of the difficulties experienced by the new governor, Junayd b. ʿAbdur-Rahmān al-Murrī, in joining the army before Bukhārā. His guide advised him to levy a force from Zamm and the neighbouring districts before crossing the Oxus but Junayd refused, only to find himself after crossing put to the necessity of calling on Ashras for a bodyguard of cavalry. This force narrowly escaped disaster on its way to meet Junayd and fought a second severe engagement on the return journey before reaching Paykand. The enemy are variously described as “men of Bukhārā and Sughd” and “Turks and Sughdians”; it may therefore be assumed that they were the same forces against whom Ashras had fought the previous year. Wellhausen is probably correct in supposing that Ashras was practically beleaguered, though not in Bukhārā. The recapture of this city and the retiral of Khāqān took place shortly after Junayd’s arrival, in circumstances which are not described[89]. The attitude of Tugshāda during this episode is not recorded. It is practically certain, however, that he remained in Bukhārā, and after the reconquest was able to make his peace with the Arabs, probably on the excuse of force majeure. At all events he retained his position, possibly because Junayd thought it impolitic in the face of the situation to victimise the nobles in the reconquered territories and thus provoke a more stubborn resistance in the rest of the country. The Arabs seem to have followed up the Turks towards Samarqand, probably to relieve the garrison; the two armies met again at Zarmān, seven farsakhs from Samarqand, where the Arabs claimed a success, one of their prisoners being a nephew of Khāqān. From Sughd the army marched to Tirmidh where Junayd halted for two months in the friendly atmosphere of Chaghāniān before returning to Merv. His intention was no doubt to make arrangements for the pacification or reconquest of Tukhāristān and Khuttal; in the following year his troops were actually engaged in this direction when the Türgesh invasion of Sughd forced him to change his plans. Balādhurī quotes Abū ʿUbayda for the statement that Junayd reconquered certain districts in Tukhāristān which had revolted.

How lightly even yet factional feeling was slumbering was shown after the return of the army, when the Bāhilites of Balkh had a chance to retaliate on Nasr b. Sayyār for their discomfiture at Barūqān. Though Junayd was prompt to punish the offending governor, the incident throws a strong light on one cause of the weakness of the Arabs in these campaigns.

Early in 731/112-113, the Türgesh and Sughdians gathered their forces for the investment of Samarqand. Ghūrak now openly joined the Khāqān. Sawra b. Al-Hurr, the governor of Samarqand, unable to face the enemy in the field, sent an urgent message to Junayd for assistance. The governor hastily recalled his troops, but crossed the river without waiting for them against the advice of his generals. “No governor of Khurāsān,” said al-Mujashshar b. Muzāhim, one of the ablest of the Arab commanders, “should cross the river with less than fifty thousand men.” Accompanied only by a small force, Junayd reached Kish, where he raised some local levies and prepared to march on Samarqand. The enemy in the meantime, after blocking up the water supplies on his road, interposed their forces between Samarqand and the army of relief. Junayd thereupon decided to follow the direct route across the Shāwdār mountains in the hope of avoiding an engagement, but when only four farsakhs from Samarqand was surprised in the defiles by Khāqān. The advance-guard was driven in and the main body engaged in a furious struggle in which both sides fought to a standstill. The Arabs, hemmed in on all sides, were forced to entrench; stragglers, refugees, and baggage, collected near Kish, were attacked by a detachment of Turks and severely handled. Khāqān renewed his attacks on the camp the next day, all but overwhelming Junayd, and settled down thereafter to beleaguer him. In this predicament there was only one course open to Junayd. Had his force perished, Samarqand would certainly have fallen in the end and two disasters taken the place of one. He therefore adopted the more prudent, if unheroic, course of ordering Sawra to leave a skeleton garrison in Samarqand and march out to join him by way of the river: Sawra, however, took the short cut across the mountains, and was actually within four miles of Junayd, when the Turkish forces bore down on him. The battle lasted into the heat of the day, when the Turks, on Ghūrak’s advice it is said, having first set the grass on fire, drew up so as to shut Sawra off from the water. Maddened by heat and thirst, the Arabs charged the enemy and broke their ranks, only to perish miserably in the fire, Turks and Muslims together. The scattered remnants were pursued by the Turkish cavalry and of twelve thousand men scarcely a thousand escaped. While the enemy were engaged with Sawra, Junayd freed himself from his perilous position in the defiles, though not without severe fighting, and completed his march to Samarqand. Tabarī gives also a variant account of the “Battle of the Pass,” the main difference in which is the inclusion of the Jabghu on the side of the Turks. In view of the Arab expeditions into Tukhāristān, it is improbable that the Jabghu, even if he was present personally, which is doubtful, was accompanied by any of his troops. The Persian Tabarī also contains an entirely different version of the Battle of the Pass and the fate of Sawra. The original version is amply attested by contemporary poets, who show no mercy to Junayd. Whatever credit the Arabs gained in this battle is reflected on Nasr b. Sayyār and the mawālī. Junayd remained at Samarqand for some time, recuperating his forces, while couriers were sent to Hishām with the news of the disaster. The Caliph immediately ordered twenty thousand reinforcements from Basra and Kūfa to be sent to Khurāsān, together with a large number of weapons and a draft on the treasury, at the same time giving Junayd a free hand in enlistment.

The Turks, disappointed in their attack on Samarqand, withdrew to Bukhārā, where they laid siege to Qatan b. Qutayba. Here they were also on the natural lines of communication between Samarqand and Khurāsān. Junayd held a council, and of three alternatives, either to remain in Samarqand and await reinforcements, or to retire on Khurāsān via Kish and Zamm, or to attack the enemy, chose the last. But the morale of the Arabs was sadly shaken; a garrison of eight hundred men for Samarqand was scraped together only by granting a considerable increase in their pay, while the troops openly regarded the decision to face Khāqān and the Turkish hordes as equivalent to courting destruction. Junayd now marched with the utmost circumspection, however, and easily defeated a small body of the enemy in a skirmish near Karmīnīa. The following day Khāqān attacked his rearguard near Tawāwīs (on the edge of the oasis of Bukhārā), but the attack had been foreseen and was beaten off. As it was now well into November, the Türgesh were compelled to withdraw from Sogdiana, while Junayd entered Bukhārā in triumph on the festival of Mihrjān. In Chaghāniān he was joined by the reinforcements, whom he sent on to Samarqand, the remainder of the troops returning to their winter quarters.

Junayd seems to have been content with saving Samarqand and Bukhārā. As no further expeditions are recorded of his two remaining years of office it must be assumed that the situation in Sughd remained unchanged and that the Türgesh irruptions also were suspended. Though the Arabs still held Samarqand and the territories of Bukhārā and Kish, they were in all probability confined to these, while in the southern basin their authority hardly extended beyond Balkh and Chaghāniān. Both sides may have awaited the first move by the other, but were surprised by the appearance of a new factor, which threatened the existence of Arab sovereignty in the Far East more seriously than any external danger. It is noteworthy that in his last year of office (115/733) Junayd resumed relations with the Chinese court. The Turkish title of the leader of the embassy, Mo-se-lan Tarkan, suggests that none of the ambassadors were actually Arabs, but that the governor had commissioned some dignitaries from the subject states to represent the Arab government. The only embassy recorded in this year from a native state, however, came from Khuttal. In the same year Khurāsān was visited by a severe drought and famine, and to provide for the needs of Merv, Junayd commandeered supplies from all the surrounding districts. This, added to the military disasters of the last few years and the insinuations of Shīʿite propaganda, provoked open discontent in the district which had hitherto been outwardly faithful to Merv, namely the principalities of Lower Tukhāristān. The leader of the malcontents was Al-Hārith b. Surayj, who was flogged in consequence by the governor of Balkh. The discontent flared into open revolt on the death of Junayd in Muharram 716 (Feb. 734). Hārith, assisted by the princes and people of Jūzjān, Fāryāb, and Tālaqān, marched on Balkh and captured it from Nasr b. Sayyār. The versions leave it uncertain whether Hārith defeated Nasr and then captured the city or whether he entered the city first and beat off an attempt at recapture by Nasr. (Wellhausen’s reference to the Oxus is due to his so misunderstanding the “river of Balkh” in Tab. 1560. 2. That it refers here, as frequently, to the Dehas river is clear from the distance to the city (2 farsakhs, whereas the Oxus lay twelve farsakhs from Balkh) as well as from the mention of the bridge of ʿAtā.) From Balkh he moved against the new governor ʿĀsim b. ʿAbdullah al-Hilālī, at Merv, capturing Merv-Rūdh on the way. ʿĀsim found a large section of the inhabitants in league with Hārith, but on his threatening to evacuate Merv and to call for Syrian troops, the local forces rallied round him. At the first reverse, the princes of Lower Tukhāristān deserted Hārith, whose army fell from sixty thousand to three thousand. He was thus reduced to making terms with ʿĀsim, but early in the following year renewed his revolt. ʿĀsim, hearing that Asad b. ʿAbdullah was on the way as his successor, began to intrigue with Hārith against him. The plan miscarried, however; Hārith seized the governor and held him to ransom, so that Asad on his arrival found the rebels in possession of all Eastern Khurāsān, and Merv threatened both from the East and from the South. Sending a force under ʿAbdur Rahmān b. Nuʿaym towards Merv Rūdh to keep Hārith’s main body in check, he marched himself against the rebel forces at Āmul and Zamm. These took refuge in the citadel of Zamm, and Asad, having thus checked the insurgents in this quarter, continued his march on Balkh. Meanwhile Hārith seems to have retreated before ʿAbdur-Rahmān towards Balkh and thence across the Oxus, where he laid siege to Tirmidh. Lower Tukhāristān returned to its allegiance; on the other hand Hārith was now supported not only by the kings of Khuttal and Nasaf, but also, as appears from later events, by the Jabghu of Tukhāristān. The government troops were unable to cross the Oxus in the face of Hārith’s army; finding, however, that the garrison was well able to defend itself, they returned to Balkh, while Hārith, after falling out with the king of Khuttal, seems to have retired into Tukhāristān. Here, following the example of Mūsā b. Khāzim at Tirmidh, he made a safe retreat for himself in Badakhshān.

The motives of Hārith’s rebellion have been most variously estimated. In spite of the unctuous sentiments which he is represented as uttering on all occasions, it is hard to find in him the “pious Muslim, ascetic and reformer” whom van Vloten too sharply contrasts with the government officials[90]. In spite too of the prominent position given to him in the Arabic chronicles, it may even be questioned whether he and his small personal following were not rather the tools than the leaders of the elements making for the overthrow of the Umayyad administration in Khurāsān. At all events the weakness of his hold over his temporary followers is much more striking than his transient success. Further evidence of this is given in a most important narrative prefaced by Tabarī to his account of Asad’s expedition into Sughd. Except for the scantiest notices, the Arabic historians have nothing to say regarding the effects of the war in Khurāsān on the situation in Transoxania. Wellhausen’s conclusion (based apparently on Tabarī 1890. 6) that “Hārith first unfurled the black flag in Transoxania in the last year of Junayd” is scarcely tenable. There is further no evidence at all for his assumption that Samarqand had fallen into the hands of the Hārithites, especially as Bukhārā remained loyal to the administration. That Asad’s expedition was not, in fact, directed against Hārith follows in the clearest possible manner from the narrative referred to (Tab. 1585. 6-16).