showed a varying thickness from three to four feet. With the exception of the refuse deposit within II. ii already described, the most curious feature of these small apartments was their narrowness, particularly disproportionate to the thickness of the dividing walls. The one furthest to the northeast, II. ii, measured only four feet across, and the other two, iii and iv, nine feet six inches and four feet respectively. In the absence of distinctive finds—none were made in any of the three apartments—it seems difficult to guess their original purpose. But it is unlikely that they could have been constructed to serve as quarters. In view of the remarkable. thickness of the walls it has occurred to me that they might have been intended for store-rooms or possibly prison cells. It is obvious that an ancient Chinese ` Ya-mên ' at an important station would have needed both, just as they are usually provided in similar modern structures.
Dr. Hedin had found both rooms iii and iv filled with sand to a height of over three feet, and Documents
on clearing this had come upon ` only two or three pieces of torn paper '.' I myself had the floor, found ba tir.
over which a fresh layer of drift-sand had accumulated, carefully searched again without any result. It was left for Mr. Tachibana, on his rapid visit paid to this site in 1910, to discover here the interesting Chinese document dating from A. D. 324 and representing the draft of a letter from the Chang-chili Li Po, to which I shall have to recur below. From the verbal explanations that the young Japanese explorer was able to give me in the autumn of that year, he appears to have found it in an interstice of the brickwork of a wall within room iv, and at some height above its floor. The examination I was able to make of the little apartment in February, 1914, showed in two places shallow niches or holes which appeared to have been roughly broken out from the wall. In the absence of further and more exact information, I assume that Mr. Tachibana had made his interesting discovery in a small fissure or hole between two bricks and had subsequently enlarged it in the hope of finding more. It is impossible, of course, to guess how the crumpled-up sheet of paper had found its way into that fissure. But in any case its discovery in such a place suggests that the interior walls had at the time already lost the plastered surface which they must, no doubt, once have had, the structure being more or less in a state of ruin. Hence the discovery does not affect the doubt expressed above as to these rooms having been constructed for ordinary occupation as quarters.
Within the angle of the main wall and the north-eastern cross-wall of L.A. II lay the room v, Finds in
measuring about 3o by 12 feet inside. From the fact that along a portion of its north wall Îoom L.A.
there were found remains of a narrow platform, about two feet high, and in the wall behind it a cupboard-like recess about eight inches deep, it may, perhaps, be concluded that it had been used as a living apartment or office. In this room were found four Chinese documents on wooden slips, . among them two complete (Doc. Nos. 75o and 820) and one dated A. D. 263 (L.A. it. v. 3, Doc. No. 738). Embedded in the floor just below the platform lay the oblong tablet L.A. II. v. 5, showing remains of Kharosthi writing. It had probably reached this position some time before the site was abandoned ; for the wood was rotten and eaten by insects and the surface badly perished by damp. Here were also recovered a well-preserved bronze bolt, L.A. IL v. 002 (Plate XXXVI), with a square head at one end, and the fragment of a kiln-fired clay bowl, L.A. II. v. ooi, with a deep leaf-green glaze recalling Han pottery.
Immediately to the west lay a small detached structure built of timber and plaster, of Constructive
which two rooms, L.A. II. vi and vii still showed the lines of their walls marked b ri ht osts features in
> Y u P g P rooms L.A.
and remains of horizontal reed wattle (see Fig. io1, on extreme right). The room vi had at its II. vi, vii. north-east end a sitting platform built in clay, two feet eight inches high and five feet wide, approached by three steps. The wall dividing it from room vii showed remains of a panelled
' Cf. Hedin, Central Asia, ii. p. 633. 6 Cf. below, p. 4o9.
1174 3 C