always attached, also, to the position of Maral-bàshi, as the most convenient point for the bifurcation of the routes from Ak-su to Kashgar and Yarkand respectively.
It is therefore curious that no definite mention of Maràl-bàshi can be traced in the Chinese historical records so far accessible in translation. The identification of Maràl-bàshi proposed by M. Chavannes with Wo-shê-1ê 6 x fe, which Kao Hsien-chih's biography in the Tang Annals refers to as a locality passed on his march from Kuchà to the Pamirs, rests upon this being placed ten marches from Po-huan or Ak-su and the same distance from Su-lé or Kàshgar 16 Nor is it possible to recognize a distinct mention of Maràl-bàshi in the itinerary which the Tang shu gives of the route from Po-huan to Su-I6.17 I will therefore defer the analysis of such data as may otherwise be gathered from this itinerary until we have completed our rapid survey of the actual high road to Kashgar.
SECTION III.—A TANG ITINERARY FROM AK-SU TO KASHGAR
Modern Of the five long marches under a very hot sun which carried me from Maràl-bàshi to the
Kashgar. neighbourhood of the ' New Town ' of Kàshgar, the first three lay mostly close to the present
main bed of the Kashgar-daryâ.. This route, as the Maps (Nos. 8. A. I ; 5. C, D. I) show, represents indeed the shortest line between the two places ; but there is no archaeological or other evidence to indicate that it was the line followed in ancient times. The vagaries of the river-course meandering in a flat alluvial plain, and the difficulty inevitably caused by inundations at the time of the summer floods, raise a clear presumption against this. I was in fact informed that whereas the present road keeps to the river's left bank between the stations of Charge and Lung-kou, a distance of close on 6o miles, it had lain, as it does elsewhere, to the south of the river until long after the close of Yàqûb Bég's régime. This statement was supported by the distinctly new look of the straggling patches of village land that we passed at rare intervals along this portion of the road. We were told that Ördeklik, the largest of these colonies, comprising about 15o households, was only about thirty years old. Here, as elsewhere along the lower course of rivers passing through the flat of the Tarim basin, cultivation is much hampered by constant changes in the river-beds, and the consequent difficulty of maintaining canal heads. Evidence of this could be seen at more than one point, where fields laid out not many years ago had been abandoned owing to irrigation failing from this cause.
Return We crossed the narrow and deep-cut bed of the river to its right bank at Lung-kou. The ground
past here presented the same forlorn appearance, until at Yangi-àbàd we reached the eastern extremity
to Kâshgar. of the practically continuous cultivated tract of Kàshgar. Fine orchards and groves of big poplars from here onwards attested prolonged occupation, such as might be expected on ground to which facilities for adequate irrigation are assured by its position on the alluvial fan of a considerable river, not too far away from where it leaves the mountains. The comparative nearness of these, and of their vast stores of snow and ice, to which the oases below owe moisture and life, was strikingly brought home to me by the magnificent view which a morning of exceptional clearness revealed on the march from Lung-kou to Faizàbàd. It comprised the whole of the high glacier-clad range dividing the Pamirs from the Tarim basin, and extended from the dome of Muz-tàgh-atà in the south to the head-waters of the Kàshgar river. It seemed a visible proof that my long journey round the huge basin of innermost Asia was drawing to a close. Faizàbàd, where I met with a kindly official welcome and halted for a night, was the last of the district head-quarters which I still had to visit in the Tarim basin. From here, on May 3oth, a ride of some thirty-two miles,
16 Cf. Chavannes, Turcs occid., p. 153 note. 17 See below, p. 839.