lay half a mile to the east and showed only traces of a few rush-and-plaster walls marking the position of stables and cattle-sheds. The ground immediately adjoining, being protected by layers of horse- and sheep-dung, had retained its original level, while the rest of the ` Tati ', measuring altogether about a hundred yards in diameter, had been eroded by the winds to a depth of six to ten feet, as shown by a characteristic group of small ` witnesses ' seen in Fig. 43.
Of more interest was the small site about one and a half miles due north of Khadalik which Mullah Khwaja called Balawaste, and which on examination proved identical with one of the small ` Tatis ' I had passed on March 24, 1901, and then heard collectively designated by my shepherd guides as Ak-tdz.1 The ground passed en route was covered with closely packed tamarisk-cones, but showed here and there small eroded patches with witnesses six to ten feet in height. Balawaste itself was such a patch of open ground, measuring about two hundred and sixty yards in diameter and showing old pottery débris at several points. Near its middle I found what, owing to the erosion effected around, looked like a small plateau, bearing the scanty remains of a roughly built dwelling. Its walls, constructed of vertically-placed rush bundles with a thick outer coating of mud plaster, formed an oblong of about seventy by sixty feet ; within, a series of small rooms could be traced. The layer of sand covering the floor was only a foot or two deep, and apart from some rags (Bal. 007. a-c), including a small piece of silk with a printed colour pattern, no finds resulted from the clearing.
About eight yards to the west of this dwelling and partly covered by the foot of a sand-cone was a smaller ruin of similar construction showing two rooms flanked by sheds which manifestly had served at one time as sheep-pens. Mullah Khwaja declared that it was in one of these rooms he had found the three well-preserved Chinese documents written on wooden tablets, about one foot long and one inch broad, which I had acquired through Badruddin Khan at Khotan and which are
now published and translated by M. Chavannes in his volume of the present Report.2 It will
i be seen there that both in contents and outward appearance these documents agree closely with the
! wooden records I subsequently excavated at the site of Mazar-toghrak to the south-east of Domoko.3
In any case the records clearly emanated from some petty administrative office. A line of
i cursive Brahmi writing appears below the Chinese characters on one of the Balawaste tablets and
thus shows that the issuing office must have been of a local character and was not confined merely to a Chinese garrison. Though the room in question was now filled by sand to a height of some five feet the complete clearing effected yielded no find except the tiny fragment of a wooden tablet
M showing remains of Chinese writing on one side and of cursive Brahmi on the other. This, however,
served to confirm the accuracy of Mullah Khwâja's statement as to the origin of his tablets. Not far from this structure there was found on eroded ground a copper coin badly effaced in which Mr. Allan believes he can recognize a late Wu-shu piece of the Liang dynasty.
A close examination of this ground showed that the pottery covered ` witnesses ' rose here
0! to heights of ten to twelve feet. Yet their flat tops generally remained two or three feet below the
original ground level as indicated by the ruin just mentioned, this difference plainly marking the
0 progress made by erosion since the disappearance of the structural remains to which these ` wit-
nesses ' first owed their survival. Both here and on my subsequent visits to other small ruined sites south there were abundant opportunities for instructive observations on the physical changes which had come over this desert area to the east of the Domoko stream, once evidently occupied by numerous villages. Again and again I noted how the patches of open ground, probably marking the positions where small agricultural settlements had clustered, were being broken up and scooped out
1 See Ancient Kholan, i. pp. 453 sq. 2 See Chavannes, Documents, p. 219 ; PI. xxXVli, Nos. 981-3.
3 See p. 205.