sculptures or frescoes, such as the well-preserved walls of Rawak or Dandân-oilik had yielded, could not here be hoped for, there was some compensation in the abundance of decorative details and the ease with which their remarkable hardness allowed them to be recovered. At Rawak the same relievo decoration consisted of unbaked clay so friable that many of the pieces broke in the very attempt to remove them.
There was another gratifying experience to reward that day's experimental digging at a site Gilt which had promised so little at first sight. Again and again the practised eyes of my diggers noticed appliqué pieces of appliqué relievo still retaining tiny flakes of gold or else detached flakes—unmistakable relievos. evidence that the greater part of the wall decoration must have once been gilded. For the first time I had here before me definite confirmation of the hypothesis which I had formed in explanation of the leaf-gold washed from the culture-strata of Yötkan. In that gold I had recognized the remains of the profuse gilding which an early Chinese record attests for the sacred buildings of Buddhist Khotan.4 But Yötkan had not furnished so far a single gilt object from a building ; evidently there the wood and friable stucco, which alone appear to have served as material for the decorative features, had completely decayed in a soil kept ever moist by irrigation.
My camp had been pitched at a farm close to the edge of cultivation near the village of Ak-kul, Resumed and thither I made my way in the evening. The dunes to be crossed on our way eastwards were so cultivation
hi h and sterile that I felt surprised when I reached the first fields after less than two miles. a-
g P doned area.
Returning next morning to the site, I could observe how the area of moving sands is steadily being
pushed back here by resumed cultivation. The lands of Ak-kul had been brought under irrigation
only some fifteen years before, and what I saw here was the best illustration of the process by which
the whole sandy waste, separating the Hanguya and Yurung-kash cantons, is now being gradually
recovered from the desert. In the fertile loess soil, to which new cuts carried ample water, the
poplars, willows, and J igda-trees usually planted along the edges of fields were shooting up rapidly.
Hence it was easy to note the new conquests made each year by advancing irrigation. The belt of
luxuriant reed beds and scrub which was moving in front of it, fed by the overflow of the Ak-kul
canal, was fast approaching from the east the ground where earlier occupation has left behind its
pottery débris. Was the time yet to come when, under the pressure of increasing population and
the growing need of land, the oasis would victoriously recover most of the ground lost to the
desolation of the Hanguya Tati'—or would slowly progressing desiccation stop the movement?
This advance of human activity made it easy for me to collect overnight an adequate posse of Structural
labourers for the excavation of the ancient structure. Working with close on sixty men, I soon laid remains of
bare, on September 19, what proved to be the north wall of a temple cella, having an outside shrine.
measurement of fifty-three feet on this side, together with an adjoining passage, five and a half feet
wide. The plan (Plate 4) shows that this passage, as in the case of the Dandân-oilik shrines, must
have extended all round the four sides of the cella. Its outer wall, also of stamped clay, showed
a thickness of three feet, while the wall separating it from the interior of the cella was two feet thick.
On the west side the walls of cella and passage stopped short at a distance of about forty-two feet
from the north-west corner, being there met by the mud flooring of some other structure. This
was on a level about two feet higher than the passage floor, and probably was of later date. On
the east side I had the passage cleared on September 20 for a distance of about twenty-two feet.
At this point the digging was stopped, as the sand slipped in from the dune, which extended over
most of the cella area. The height of this dune would have made complete excavation a difficult
and protracted task. I felt all the more justified in forgoing the heavy sacrifice of time and money