14o REMAINS OF THE KHOTAN OASIS [Chap. IV
ground near it had undergone the same succession of physical changes. The layer of earth and
mud-brick débris filling the cella was permeated by the tubular passages or cores which roots and vegetable fibres had left behind. These proved that the ruin must have been buried for a time
under a gradual accumulation of cultivated loess layer. By that time the walls, as far as they
reached above the ground had, no doubt, been pulled down and all materials which might be of use, such as the encased timber framework, abstracted. After cultivation had retreated from the site,
wind-erosion must have been at work for a prolonged period. Small banks of loess rising up to six feet in height, and generally bearing on their top heavy pottery débris which had evidently protected them, rise like geological witnesses ' above the general level of the ' Tati ' quite close to the shrine. They mark the extent to which the surface of the ground has been lowered by the winds and the corroding effect of driven sand.
It is obvious that owing to this erosion the pottery débris found on the surface, of which I picked up specimens both at Siyelik and on my way back to the Ak-terek ruin (Si. oo1, ooz ;
A.T.S. ooI-ooIo), may belong to widely different periods. It is probable that much of it dates from early Muhammadan times, since among the coins which my guides had collected for me from the whole Ak-terek site, including Siyelik, specimens of local Muhammadan coinage as well as ' cash ' pieces of the Sung dynasty were largely represented (see Appendix B).13 This may be assumed in particular of the fine glazed ware in various tints of green and blue which plentifully covered certain areas.
From the mass of human bones which mingled with pottery débris around the ruined cella, it was certain that the soil, now again carried off by the winds, had once served for interment. The
inference seems justified that a burial-ground had been established here after the Muhammadan con-
quest, because the site once occupied by a Buddhist sanctuary continued to receive local worship as a Mazâr.14 Also close to the larger temple ruin of Ak-terek I noticed between the dunes large
stretches of eroded ground thickly strewn with human bones, to which the same explanation would apply. Finally, to judge from the rate of erosion gauged at sites like those of Dandân-oilik, Khadalik, Niya, the abandonment of which can be fixed with approximate chronological certainty, I may note that the maximum height of the witnesses ' observed at Ak-terek and Siyelik indicates that cultivation did not cease here until after some centuries of the Muhammadan period.
Apart from the ruined shrine just described the only structural remains my guides could show me at Siyelik consisted of two small ' Tims '. Both were much-decayed mounds marking the
position of Stûpas. One situated about a quarter of a mile to the south of the temple ruin preserved
its lowest base, about sixteen and a half feet square and three feet high, with a portion of a square upper story. A flight of stairs leading up on the south face could just be traced, with a width of
eight feet at the bottom. A second Stûpa mound was visited about a quarter of a mile to the east of the last-mentioned ruin. This was smaller and also badly decayed, showing a base only eight feet eight inches square on the ground and completely broken above its lowest story which measured three feet in height.
The interest of the survey effected during my three days' visit to the Ak-terek Site does not lie so much in the archaeological relics recovered as in the instructive glimpses it yielded of the remains
which may yet lie buried below these vast silent ' Tatis '. Extending from east to west in a line of more than twelve miles this great area of shifting dunes and bare loess undergoing wind erosion