130o FROM MAZAR-TAGH TO MARAL-BASHI [Chap. XXXII
the town of Uch-Turfân. The way led mainly through a belt of fairly old cultivation skirting the foot of a barren hill range to the south known as Kara-teke. The more careful appearance of the fields and the frequency of fine arbours were explained when I learned_ later from the scholarly Chinese magistrate at Uch-Turfân, P`êng Ta jên, an old acquaintance, that the population in this tract was descended from colonists brought here from Kashgar and Khotan in the time of the Emperor Chien-lung after a rebellion which ended with the extermination of all previous settlers.18' That these were of Kirghiz or allied stock was suggested by a number of large domed tombs, or ` Gumbaz ', in ruins which I passed near Sugetlik-mazâr (Map No. 19. A. 4), and which local tradition vaguely ascribed to Kipchak princes.
To the north, beyond the narrow belt of cultivated ground, there stretches a wide grassy plain which receives abundant-moisture from the Tushkan-darya, and which is bound to prove attractive as a winter grazing-ground for the Kirghiz. These form the great majority of the population higher up in the valley, where numbers of them appear to have taken to agriculture also. There were plenty of Kirghiz tobe seen at Uch-Turfân itself, and having established my quarters at the country residence of their Beg, outside the flourishing little town, I was able also to use my two days' stay for anthropometrical work among them.1° Most appeared to retain still the characteristics of true Turk stock in their physical features.20 But in others it was impossible to mistake evidence of intermixture with ` Sarts ', as the Kirghiz here call the settled Muhammadan population. The use of this term, so common in Western Turkestan but heard by me here for the first time in the Tarim Basin, was apt to recall the close vicinity of the Russian frontier and the Sir-clarya headwaters.
The usual dust haze of the spring was hiding the view of the great snowy range of the Tienshan northward. It was thus impossible to obtain even a distant glimpse of the Bedel Pass, by which Hsüan-tsang had once gained the Issik-kul region and thence Sogdiana.21 But even without that imposing background Uch-Turfan presented itself to me as the most picturesque and pleasant of any district headquarters I had visited in Chinese Turkestan. The view of the fertile green valley, set off vividly by the chain of barren grey hills which encircle the town from the south, was particularly striking from the height of the Chinese citadel (Fig. 337). This crowns the top of a precipitous rocky spur, which adjoins the west wall of the town and projects beyond it like a huge natural ravelin, rising with its westernmost cliffs to a height of some 25o feet. The citadel and the flanking defences joining it to the town walls are recent, having been built in the place of fortifications destroyed when Uch-Turfân was besieged and taken during the Muhammadan rebellion. But this natural stronghold is bound to have been utilized since early times.
Its position may well be assumed to account for the name Ï a-shih ch`êng 3 ` the Big
Stone Town', by which the itinerary of the Tang Annals previously referred to mentions Uch-
Turfan.22 It acids : ` It is also called Yü-chu jig, or else the district town of Yuen-su
.' As noted before, it is by the last name or its variant Wen-sat a rd that the ` kingdom' and town are referred to in the Former Han Annals, the Hou Han Mu, and the Wei 110.23 The Former Han Annals ascribe to it a population of 2,200 families, which seems proportionate, and indicate its position quite correctly with reference to the Wu-sun capital which lay 610 li to the
181 ForsythMissionReport,p.42, gives the date as A.D. 1765. ' See the photograph, Fig. 333.
20 Regarding these cf. Joyce, J. Anlhrop. Inst., xlii. pp. 453 sqq., 462 ; below, Appendix C.
21 Cf. Julien, Mémoires, i. p. r i ; Watters, Yuan-Chwang,
i. p. 66. The Bedel Pass and not the Muz-art is clearly indicated by the recorded direction to the north-west of Po-luchia, or Ak-su ; cf. Chavannes, Turcs occid., p. 9. The
Bedel Pass is undoubtedly meant by the Po-ta-ling mountain chain, the crossing of which the Tang itinerary mentions on the route from Wên-su to the Issik-kul ; see Chavannes,
22 See Chavannes, Turcs occid., p. 9.
23 Cf. Wylie, Notes on the Western Regions, J. Antlrrop. Inst., xi. p. 94 ; Chavannes, T`oung pao, 1906, pp. 227, 231, 233, 252, 256; 1905, PP• 553 sq.