Sec. ii] FIRST EXCAVATION OF BUDDHIST SHRINES 243
Apart from the ancient structures thus attacked in their very foundations, there was ample Effects of
evidence at the site of the havoc worked by erosion both at the present and during earlier erosion.
periods. Wherever the tops of posts or other parts of the timber framework of houses still rose above the protecting cover of sand, they showed by their curious splintered and abraded appearance the force of the wind and driven sand to which they were exposed during the Burân periods. The same was the case with all walls which the operations of treasure-seekers or the shifting of dunes had within recent years deprived of their protective layer of sand. Finally, we have to recognize manifest results of erosion, probably from an early period, in the numerous places between the ruins where the ground was thickly strewn with fragments of coarse pottery, small corroded pieces of metal and similar débris, while no trace of structural remains survived. In view of our previous observations on ` Tatis ' it seems safe to assume that these remains mark the positions occupied by less pretentious buildings, which, like the houses of Khotan cultivators at the present day, were built wholly of sun-dried bricks or stamped clay. These were likely to crumble away far more quickly than buildings with a solid timber framework covered by hard plaster, and erosion could complete their destruction before the drift-sand arrived in sufficient masses to cover them up and afford protection. This observation helps to explain, at least partly, why at a site like Dandân-Uiliq, which for various reasons may be supposed to have been occupied by a comparatively large settlement, the structural remains traceable at present are limited to small groups of timber-built structures widely scattered over the ruined area.
I propose to describe the excavations carried on among these structures group by group, Ruined
though the limited time at my disposal did not allow me systematically to follow the same shrine D. r.
order in the course of the work. I may, however, make an exception in the case of the small structure with the clearing of which I commenced my explorations on the morning of December 19. On the previous evening I had noticed much-decayed remains of a small square building, D. 1, immediately to the south-east of my camp. Turdi knew it as a ` But-khâna', or ' temple of idols ', and well remembered having searched it in his own fashion on previous visits. The last of these must have been his ` prospecting ' expedition on my behalf, for the several fragments of stucco reliefs brought back by him, and now described in the list of Dandân-Uiliq antiques (see below) as D.T. oo1—o14, agree exactly with those excavated by me at D. 1, and in one instance supply a portion missing in one of the latter pieces. There can be no doubt either that among those who had searched the ruin before was also Dr. Hedin, for the small stucco reliefs obtained by him from a much-decayed building, where ` the sand was quite shallow', comprising seated and standing Buddhas (the latter described as draped women), as well as decorative fragments, must, as the illustrations show, have been cast from the same moulds as many of the pieces described in my list 2. But the sand, though only 2 to 3 feet high, had not been removed, and by laying bare the foundations and floor I might expect to gain useful preliminary knowledge as to the general construction and arrangement of such shrines.
My hope in this respect was not disappointed. A careful examination of the remains of walls Structural
brought to light on the north and west sides showed that there had been an inner oblong or
square cella enclosed by equidistant outer walls, forming a kind of corridor or passage on each side about 5 feet wide. The walls to the south and east had completely disappeared to the very