of such remains of mural paintings as still survived in a few accessible cave-shrines. Accordingly I decided to begin by clearing an area situated at the foot of the northernmost large group of ruined shrines on the left side of the gorge (Fig. 310). Niâz, an old frequenter of the site, said that in his youth he had seen at this spot remains of small structures, some vaulted, which had since been completely covered up and hidden by the debris thrown down on the fan-like slope in the course of the clearing of the big central temple above. Excavation was started from the foot of this slope, where it ends in precipitous cliffs of clay rising above the irrigation cut that carries water from the stream to the eastern portion of Toyuk cultivation. On this lowest level two rooms were laid bare, with remains of brick walls built against the vertically cut clay face eastwards (see plan in Pl. 25). Small torn pieces of Chinese manuscript rolls, apparently containing Sutra texts, were found in one of them. Separated from these rooms by a narrow passage which has once been vaulted, a hall or court, ii, extended northward, of which only the back wall cut into the natural clay and a small portion of the southern wall survived. Within this hall there were found on the floor and covered by fallen brickwork many Chinese manuscript fragments of the kind just mentioned, but including also large pieces, among them a few with Uigur or Tibetan writing on the reverse. This use of the reverse of Chinese Sutra rolls for Tibetan or Uigur texts is similar to that observed at Ch`ienfo-tung and points to late occupation of the ruined shrines in this group, a conclusion fully borne out by other observations.13 Beyond Toy. 1. ii, excavation on the lowest level was stopped when the natural surface was reached.
As the excavation proceeded eastwards up the slope, work became more difficult owing to the heavy debris, including masses of brickwork, overlying the original slope to a depth of 8 or 9 feet. In two small rooms cleared to the south of the area small fragments of Chinese and Tibetan manuscripts were recovered, and besides them a few fragments of stucco relievo that probably belonged to some image destroyed in a shrine higher up. Two shoes, Toy. 11. 02-3 (Pl. LXVI), of excellent make, one woven in string and the other quilted, were also discovered here, besides the string sandal, 04. The clearing farther to the north had to contend with increasingly deep masses of debris, but after heavy labour reached a cella, about 17 feet square, whose walls near the entrance still showed faint traces of paintings. Two vaulted passages which adjoined the cella on the north and east were found to be filled with hard mud that had been carried down by rain. Within and outside the cella a number of manuscript fragments in Chinese and U igur were recovered. The miscellaneous finds included numerous pieces of pottery vessels, some of superior make with decorated or polished surface (Toy. III. 06, 8, 9, 12, 17-19, &c., Pl. CIII) ; fragments of a woollen pile carpet, 01, and numerous fragments of silk fabrics, among them several damasks and figured silks, 033-4 (Pl. XLIII, LXXXV). Here too was recovered the fairly well preserved knife, 111. i. 02, in its lacquered silver-bound sheath, which also still contained two black wooden chopsticks.
These excavations, which were much hampered by the constant sliding down of debris from the higher slopes, showed that the structures to be found here were already much decayed before they had been buried by the earth, brick debris, and boulders thrown down from above. As the finds offered no adequate compensation for the labour involved, the clearing was stopped on the fourth day. Before this, however, we had come upon masonry inserted in what was evidently a natural fissure in the rock face, in such a way as to suggest that it may have served to support a ramp that once led to the shrines and monastic cave-quarters above.
I next turned my attention to the large group of shrines, situated on a terrace about two hundred feet above the right bank of the stream and about three-quarters of a mile from the Toyuk Mazâr, which is the most conspicuous among the ruins on the western side of the gorge (Figs. 309, 311). 13 Cf. e. g. Grünwedel, /oc. cit., p. 324.