72 FROM THE OXUS TO KHOTAN [Chap. III
the chapels surrounding the main Stiipa court do not show cella interiors much exceeding six feet on each side.23
It only remains to see whether we cannot trace some earlier mention of this structure, the pre-Muhammadan origin of which has thus become highly probable. On ground so exceptionally bare of permanent structures as the Pamirs, this solidly built little shrine, insignificant as its size is, must have had a far better chance of attracting notice than it would have anywhere in regions crossed by regular routes. Now, when discussing above the several routes by which Kao Hsien-chih moved his three columns to their concentration before Sarhad, I have already shown that one of them, described as the route of ' the hall of the red Buddha' (Mk-JO-tang), must have led down the valley of the Ab-i-Panja from the Wakhjir or the Little Pamir.24 I have also explained why collateral evidence connected with Kao Hsien-chih's return journey from Little P`o-lü to Sarhad led me to look for the spot that gave its name to the route in the vicinity of the point where the route down the Ab-i-Panja Valley is joined from the south by that which descends from the Khora-bohrt and Irshad Passes leading towards Gilgit and Hunza respectively. This point on the Dasht-i-Mirza Murad is not more than two or three miles distant from Karwan-balasi, and the identity of the latter ruin with the Ch`ih fo-tang of the Chinese record becomes still more probable on taking into account that the term t fang, translated above after M. Chavannes' version by ' hall ', is regularly applied also to Buddhist Viharas or chapels, however sma11.25 The reason for the erection at this particular spot of the Kârwân-balasi shrine or the ' Chapel of the red Buddha', as I take it, can no longer be conjectured after the lapse of ages. But that this little chapel should have given its name to the locality, and through this to the route leading past it, is fully in keeping with the use which modern local nomenclature on the Pamirs makes of the few artificial landmarks which that desolate region affords.26
SECTION III.--ON HSÜAN-TSANG'S ROUTE TO KASHGAR
After crossing, on May 27, the Wakhjir Pass, under difficulties which my personal narrative describes,1 I found myself on Chinese soil and at the head of that great Sarikol Valley with which my first journey had already rendered me familiar. As my route down to Tash-kurghan was necessarily the same as the one I followed in July, I9oo, and as the early geography, history, and antiquities of Sarikol have already been fully discussed by me in Ancient Khotan,2 it will suffice here to supplement that account by the survey of two old sites which I was now for the first time able to visit in person. The record of the ancient local traditions relating to both these sites is due to Hsüan-tsang, who, on his return journey about the summer of A. D. 642, passed from Wakhan across the Great Pamir to the Taghdumbash Pamir and thus down to Tash-kurghan, the Sarikol capital.3
The story of the first of the sites is told by the pilgrim in connexion with the origin of the royal family of Chieh-p`an-t`o or Sarikol.4 ' The king gives himself the title of Chih-na-t`i p`o-cla`ii-tan-lo (Cina-devaggotra), meaning the descendant of China and the sun-god. Formerly the country was
23 Cf. Foucher, L'Art du Gandhdra, i. p. 124. about 1845 by Kanjûti raiders. Compare also Abdullah
24 See above, p. 54. Khan, i. e. his tomb, on the Alichur Pamir ; Mazar-lepe and
R5 This use of the term is well illustrated by the name Yol-mazar, both tombs, on the Great Pamir, etc.
Chien fo-t`ang, borne by the agglomeration of Buddhist cave ' See Desert Cathay, pp. 83 sqq.
temples, big and small, ' The halls of the Thousand Buddhas', 2 See Ancient Kholan, i. chap. n (pp. 22-40).
at the famous site south-east of Tun-huang ; see below. 3 Cf. ibid., i. pp. 3o sqq.
26 Thus, some six miles above Karwàn-balasi, Bôzai- ` Cf. Julien, Mlmoires, ii. pp. 210 sqq. ; Watters, Yuan
gumbaz, a widely known local name, is derived from the Chwang, ii. pp. 285 sqq.; Ancient Kholan, i. p. 34•
poorly built tomb of a Kirghiz chief who was killed here