6o FROM YASIN TO KASHGAR [Chap. 1I
by the knoll, which makes the spot exceptionally well adapted for the purpôses of a travellers' shelter, had seemed sufficient on the occasion of my previous visit to justify my recognizing in the ruined enclosure ` the last remains of the ancient structure to which Hsiian-tsang's record refers '. The close survey of the ground that more favourable conditions as to time and assistance now rendered possible fully confirmed this conclusion. It showed me that the south-western portion of the large enclosure was once occupied by a block of quarters, forming a rectangle whose outside measurement was 58 by 55 feet (see plan, Pl. 1). The lines of the main walls, two and a half feet thick for the most part and of the same construction as the outer enclosure, could be traced quite clearly on the ground even without excavation, though broken down almost to the surface level. The ground plan comprised two rows of five rooms each, ranged on either side of an inner courtyard measuring approximately S3 by 262 feet. The rooms showed a uniform dimension of nine and a half feet square and were divided from each other by walls one and a half feet thick. The disposition of these rooms leaves no doubt about their having been intended for travellers' :quarters and the inner court for the storage of caravan goods, &c., while the unoccupied portion of the larger enclosure outside would serve for the transport animals. The worst winds over the Chichiklik-maidân seem to blow from the north-east, and whoever planned this ancient Sarai may have left the northern and eastern sides of the outer enclosure unoccupied for the purpose of allowing the animals to seek shelter under those walls from the cutting blasts.
The fact that the grave mounds are now found uniformly scattered over the whole area, including the portion occupied by the quarters, proves the structure to have been in its present state of complete decay for a very long period. It is possible that the ancient hospice was already in ruins when seen by Hsiian-tsang; for his account, as contained in the available translations of the Hsi yii-chi, gives indeed a full account of the legend about its saintly foundation as he heard it, but leaves us in doubt whether the pious traveller himself enjoyed its protection. It only remains for me to add that plentiful fragments of pottery found outside the enclosure and on the adjoining ground prove that the spot was occupied as a usual halting-place for caravans, & c. The two huts that a Chinese `Amban' had had built some 200 yards away to the north-west, which in 1906 afforded modern proof of the suitability of the spot for a hospice, were now found completely in ruins though only erected since 1903. Their rapid disappearance serves to illustrate the severe climatic conditions and also, by contrast, the solid construction of the ancient hospice, which has left its clear traces even after the lapse of so many centuries.
From the Chichiklik eastwards I had to follow once more the route which leads down to Tar-bâshi and through the extremely confined rock gorges of Tangi-tar (Fig. 68). I have fully discussed on a previous occasion the antiquarian and quasi-personal interest attaching to this trying defile, owing to the adventures experienced there by Hsiian-tsang and nearly a thousand years later by Benedict Goës.e I need only add that on my descent to Tar-bâshi I noticed clear indications of ancient glaciation in large ancient moraines and in a succession of plateau-like steps, which seemed to mark the terminal points reached at different geological periods by the glacier that once filled the valley. There can be little doubt that the Chichiklik-maidân, too, owes its peculiar configuration to the former presence of a big glacier completely covering the head of the Shindi valley.
On September 15th our routes divided at Toile-bulung, a Kirghiz grazing ground with patches of cultivation, where the stream from the Tangi-tar defile joins the one descending the Burams5..l valley from the north (Map 2. D. 4). Lai Singh moved off by rapid marches eastwards •in order to reach, via Yarkand and Khotan, that portion of the main K`un-lun range above Kapa and
9 Cf. Serindia, i. pp. 78 sq.