Sec. iv] FROM SARÏKOL TO KASHGAR 43
that Hsüan-tsang tells us suffices to fix the limits of Wu-sha with fair accuracy 9. We saw that the river Hsi-to, i. e. the Zarafshan or Yarkand river, formed its southern border. From a subsequent notice we learn that on the north it was adjoined by the kingdom of Chia-sha or Kashgar, which the pilgrim reached from Wu-sha after a march of 500 li northward across stony hills and desert plains. Within the limits thus indicated, the map shows us a narrow belt of cultivable ground, extending in detached oases between the foot of the spurs which descend from Murtagh-Ata and the westernmost portion of the great sandy desert. In the extreme south-east of this area we find the great Yarkand oasis, while the fertile tract of Yangi-Hisar, next in importance and size, marks the north-western end of this belt.
The region of the present Yarkand, situated at the point where the greatest river of Eastern Turkestan debouches from the mountains, must at all times have enjoyed exceptional advantages in the matter of irrigation. It may consequently be assumed to have been in Hsüan-tsang's days, as it is now, the most populous and fertile portion of the territory above defined. Yet it is difficult to believe that Yarkand possessed in ancient days the political importance which it has claimed during recent centuries, especially during periods of Muhammadan rule. The Chinese records, which permit us to trace from the times of the Han dynasty onwards the fortunes of Kashgar, Karghalik, and Khotan as separate and fairly powerful states, do not know Yarkand as the centre of a distinct territory. Mirza Haidar, whose history of his race, the Tärikh-i-Rashid% is by far our best Muhammadan authority on Eastern Turkestan, distinctly tells us that in the days of his ancestors 'Yarkand was a companion city to Yangi-Hisar'. It was his uncle Mirza Aba Bakr who first made Yarkand his capital, and turned it into what it has since remained, the largest and probably richest of the ` Six Cities' of Turkestan to,
In view of these facts we cannot feel certain about the identification of the chief town of Wu-sha with Yarkand, tempting as it would otherwise appear. The distance to Chtia-sha, which Hsüan-tsang puts at 500 li, or five days' march, and which, after his usual practice, may be taken as the distance from capital to capital, would well agree with the five (rather long) marches ordinarily reckoned nowadays between Yarkand and Kashgar 11. Less so the direction, which is actually north-west instead of being north, as recorded by the pilgrim.
It is the direction of march here indicated, as well as the relatively short distance recorded between the Chichiklik plateau and the confines of Wu-sha, which make me believe that Hsüan-tsang followed the direct route towards Kashgar, via Chihil-Gumbaz, Ighizyar, Yangi-
sion that K`ivan-yu-mo (Ch`üan-yü-mo) and K`o-p`an-Co are both designations of the identical territory of Sarikol or Tâsh-kurghan, an opinion already expressed by a Chinese commentator. The close similarity between the names Wou-tch`a (Wu-ch`a) of the Pei shih and Wou-cha (Wusha) of Hsüan-tsang (Beal : U-sha) is incidentally noted by
M. Chavannes. May we not reasonably account for it by the assumption that the name variously spelt Wu-sha and Wu-ch`a properly applied to the territory described by Hsüan-tsang, but was at one period extended also to Sarikol, with which the former was politically connected during centuries ?
g V. de Saint-Martin showed just perception for the bearing of the topographical evidence when, in Mémoire analytique,
p. 427, he proposed to locate Wu-sha at the present YangiHisâr (called by him Inggachar, after the nomenclature of the
modern Chinese sources used by Timkowski and Klaproth ; comp. Ritter, Asien, v. pp. 400, 417). The identification was too narrow, but as close as the cartographical materials available in 1858 would permit.
On the other hand, the identification with Och-Takht Soleyman', i. e. with Osh in Farghana (1), which Julien's translation suggests, and which Beal has thought fit to reproduce (Mémoires, ii. p. 216 ; Si yu-ki, ii. p. 304), shows curious disregard for plain geographical facts, and rests on nothing but a deceptive similarity of sound.
10 See Târikh-i-Rashidii, p. 296. Regarding Yarkand, comp. below, chap. v. sec. i.
11 Compare Yarkand Mission Report, p. 43o, which puts the total distance from Yarkand to the ` Old City ' of Kâshgar at 12o miles.