measuring about S3 by 23 feet and completely filled with debris ; this, judging from the refuse found at its NE. corner, may well have served as a monastic dwelling-place.
The complete clearing of the ruins, coupled with the necessary arrangements for labour and adequate transport of water, would have involved a greater sacrifice of time than I was able then to make without endangering my main programme. But even the rapid survey effected sufficed to throw light on some essential points of interest. It showed clearly that the remains of the site belong approximately to the same period as the ruined Buddhist sanctuaries on the hills of Tumshuk, some eleven miles away to the south-east,8 and that their destruction by fire dates, like that of the latter, probably from the very commencement of the Muhammadan period. It is obvious that as long as the Lal-tâgh site was occupied there must have been water in its immediate vicinity, and this can have been brought there only by a branch of the Kashgar river, which still extends its deltaic channels and marshes far beyond Tumshuk. A variety of local observations fully support this conclusion. The ancient dry river-bed passed by us on the way to the L5.1-tagh site and coming from the south-west has already been noticed. Though I could not actually locate it near the ruins, yet the belt of dead Toghraks through which we passed on our way from the ruins to the gap between the Achal-tagh and Bel-tagh (see below) sufficiently proves the continuation southward of this former bed. Similarly the ancient canal, to be noticed presently, that we crossed in the same part of our route, can have received its water only from the south, i. e. from the area west of Tumshuk, which is still reached by flood-beds and irrigation canals from the Kâshgar river.9
Now, on examining the topographical facts recorded by our surveys of 1908 and 1913, we see at once the special antiquarian interest attaching to the Lal-tagh site in its bearing on the question of an earlier route towards Kashgar. My journey from Kelpin to Tumshuk in May, 1908, had enabled me to trace an extensive area of ancient cultivation round the ruined fort of Chong-tim (Map No. 7. B. 4), situated some sixteen miles to the north of Tumshuk and now completely abandoned to drift-sand desert. When discussing the observations there made, I was able to show that this area had been occupied down to Tang times, and that there was evidence that an ancient route led to it straight from the vicinity of Chilan (Map 7. c. 3), where the present Ak-su–Kâshgar high road leaves the foot of the hills."Ô A look at the map proves that the Lâltâgh site lies in the direct continuation of this route, and that the gap between the hills of Lal-tâgh and Chöl-tagh—as the rocky hill chain north of the Tumshuk ruin is called—forms the natural and most convenient passage for such a road. Thus the position chosen for the Buddhist sanctuary at the barren foot of the L5.1-tâgh is also fully accounted for. It flanked a much-frequented route at a conspicuous point, in the same manner as the popular Muhammadan shrine named Okurinazâr marks the point where the present high road between Tumshuk and Marâl-bâshi passes below the picturesque southern spur of the hill chain known as Achal-tâgh or Okur-mazâr-tâgh (Map 8. B. 1)11
Having thus traced the assumed ancient road from Ak-su as far as the L5.1-tagh site, there still remains for us to consider the line of its probable continuation to the south-west. Such a line had necessarily to cross the chain of hills represented by the Bel-tagh and its southern extension, the Achal-tâgh or Okur-mazâr-tagh ; and the map shows that only two passages were available
8 Regarding these shrines of Tumshuk, situated at the southern end of the Chöl-tagh and at the northern extremity of the Tumshuk-tagh facing it (Map 8. B. 1), see Serindia, iii. 1309. A full account may be expected from Prof. Pelliot, who in 1906 effected systematic excavations at the former
site, and from Prof. von Le Coq, who did similar work at the latter in 1913-14.
9 For a brief description of this ground, see Serindia, iii. p. 1310.
lo See ibid., iii. pp. 1307 sq.
11 See ibid., iii. Fig. 344.