Sec. ii] THE SITES OF KOYUMAL AND BASH-KOYUMAL 165
Koy. I. 049 is that of an arm or leg with painted drapery suggesting brocade. From the condition of all these debris remains it was only too obvious that the shrine had at some time been purposely destroyed and perhaps quarried for timber.
A low mound about 3o yards to the south-west of the ruined shrine was found to contain the remains of what evidently was a small monastic structure, Koy. II, with walls built in sun-dried bricks and decayed to within a foot or two of the ground. The detail plan in P1. 8 shows the disposition of the rooms, several of them provided with sitting or sleeping platforms of clay. The southern row of small rooms, resembling cells, had been apparently searched by Mr. Tachibana ; those tb the west and east were cleared without yielding any finds, other than the fragment of a circular bronze disc, probably part of a mirror (Koy. ti. ii. oi). But in the large central room i, measuring 27 feet by 24 and probably used as a place of assembly by the Samgha, there were found a number of fragments of palm-leaf manuscripts in Sanskrit and early Brahmi script of the Gupta type, including the left portion of a very neatly written Pathi leaf (Koy. ii. i. 09, Pl. CXXI) .4 In addition small fragments of birch-bark inscribed in Sanskrit and Gupta characters (PI. CXXI) were recovered in the same room, near the edge of the sitting platform by the south wall.
Apart from the two ruins just described and an almost effaced oblong building to the west of II, no structural remains were traceable within the walled enclosure. Towards its north-western corner a well-marked depression, surrounded by gravel mounds and measuring 70-80 feet across, evidently represented a tank once fed from the river. But far more curious were lines of rough stones, which, as the plan (P1. 8) shows, divide the enclosed area in the fashion of an irregular chess-board. Some narrow lanes, starting from the structures in the centre and running either parallel to, or at right angles across, these lines, seemed to mark off sections of the ground. The lines of stones themselves might at first sight have suggested wall foundations ; but the stones were too loosely placed and the arrangement of the lines too schematic to support this idea. I t occurred to me at the time that the intention might have been to mark thus the lay-out of an encampment, after the manner in which the main thoroughfares of standing camps in the plains of India are often picked out with stones or bricks painted white. There is, however, nothing in the disposition of the lines, as the plan actually shows them, to bear out this conjecture.
The temptation is greater to recognize in these puzzling lines the remains of small stone heaps intended to give support to trellis-work carrying vines, such as is frequently seen in the orchards of modern Turkestan oases and the use of which in earlier times is strikingly attested by the remains I was able to trace in the ancient vineyard of the Niya Site previously described.5 And in this connexion reference may be made here to a curious notice concerning a locality near the ` Stone
Town ', Shih-ch`êng, i. e. Charkhlik, which a Chinese geographical text of A. D. 885 has
preserved. It is found in the MS. Ch. 917 which I brought away from the Caves of the Thousand Buddhas of Tun-huang and from which M. Pelliot has translated short extracts bearing on the
region of Shan-shan or Lops There we read : ` The Grape Town (P`u-t`ao-ch`êng . . 4 ) ;
[from this town] to the south, it is 4 li to the strong place of the Stone Town. It was constructed by K`ang Yen-tien. He planted vines in the middle of the town ; that is why it is called the Grape Town.'
Were it permissible to assume a mistake in the record of the bearing, the distance of 4 li being in reality reckoned to the south of the ` Stone Town ' or Charkhlik, one would be tempted to look for the ` Grape Town ' founded by the Sogdian chief at the site of Koyumal. But in the absence