612 AT RUINED SITES OF TURFAN [Chap. XVIII
with smoke and scratched all over that it was impossible to make out whether they were ever painted. On a level about 20 feet higher, a row of six small caves is cut into the crumbling rock
face. The two westernmost are of rectangular shape and retain traces of mural decoration no
longer distinguishable in character. Next follows the small shrine of which Pl. 26 shows a sketch-plan. In it a rectangular pillar spared from the rock, which once, no doubt, as in similar caves at
Chien-fo-tung, served as a backing for images, is surrounded by a circumambulatory passage.
Its walls were formerly decorated with frescoes ; but all that could be made out were remains of a dado showing traces of ornamentation with pendent triangles separated by tassels, which
reminded me of the wall decoration in the hall of the Niya ruin N. 111.5 There were traces of wall-paintings also in two small caves adjoining on the east and communicating with each other. Finally, at the eastern end of the row and projecting farther to the south, a now roofless cave appeared to have served as quarters while the rest were still used for worship.
There remain to be mentioned the ruins of two large watch-towers, manifestly old, which form conspicuous landmarks near the line followed by the high road between Su-bâshi and Pichan.
Judging from their uniform proportions and the use of bricks of the same size, 13" x 8" x 4", they may be ascribed to an approximately identical period. The better preserved of the two stands on
rising ground at a distance of about two and a half miles to the west of the point where the high
road crosses the Lamjin stream, and still rises to a height of over 3o feet. As the sketch-plan (Pl. 26) prepared by Afrâz-gul shows, it appears to have originally consisted of a tower 19 feet
square, which contained a chamber a little over 8 feet square, and was probably once divided into
several stories by a timber flooring. The tower was subsequently enlarged by the addition on all four sides of a mass of masonry 20 feet thick. Room was left between it and the original tower
for two flights of stairs, 4 feet wide, which wound round this core and led up to the top of the enlarged
structure. Access to the stairs is gained by two vaulted passages leading through the walls added on the east and the west. Within the masonry of the enlargement a number of small vaulted
recesses occur, some approached from the stairs, some apparently having had their entrances from the top., They, no doubt, were intended, like the kemers of modern Turfân houses, to give cool shelter during the fierce heat of the summer.
The other tower, appropriately known as . Yoghan-turc, ` the big tower ' (Fig. 306), stands on the top of one of the steep gravel-covered hillocks which fringe the cultivated area of Khandô
on the east (Map No. 31. A. 3). It has suffered a good deal by the burrowings of ` treasure-seekers ',
particularly at its south-eastern corner. Though the masonry had been tunnelled into at various points on all sides, no vaulted passages or stairs were traceable. Here too, however, the present
structure was the enlargement of an earlier core, as was evident from the fact that plastered wall surfaces showed at several points in the centre above the encasing masonry and revealed the existence of an original tower about 16 feet square embedded within it. The tower commands a very distant view both across the bare Sai stretching to the north and east and over the cultivated area westwards.
The route on my return to Toyuk lay down the picturesque valley, gradually narrowing into a gorge, in which the stream of Lamjin breaks through the outer hill range between cliffs of red
clay and sandstone. The cart road descending the gorge below the confluence with the Khandô
stream leads in places through artificial rock cuttings, but it is impossible to say which, if any, of them are old. About a mile below that point the road passes the large detached boulder of sand-
stone known as Tamguluk-tâsh, bearing on two of its faces relievo representations in niches of Buddhist scenes. They have been described in full detail by Professor Grünwedel,6 and no further
5 See Anc. Khotan, i. p. 333 ; ii. Pl. VII. 6 Cf. Grünwedel, Altbuddh. Kultstätten, pp. 315 sqq.