842 THE ANCIENT SITE BEYOND THE NIYA RIVER [Chap. XI
along the greater portion of the south wall, together with the circular plaster trough, 5 ft. in diameter and 3 ft. high, which was brought to light in clearing the section marked C, necessarily reduced the space available for the accumulation of rubbish.
Those of the men who had been to Keriya or as far as Khotan, and had seen there houses of well-to-do people, thought that this curious trough, with its top hollowed out to a depth of Jo in., was intended, like a similar contrivance still used by the women there, for keeping flowers fresh in water or under moist straw. I am unable to say whether this explanation is correct ; but the fact that the hollow of the trough was found filled with sand only, clearly points to the contrivance, whatever its purpose, having continued in use even while the accumulation of rubbish within the room was proceeding. Another peculiar feature of the room deserves to be noticed here. The door, 22 ft. broad, opening in the south wall and still clearly distinguishable by remains of its jambs, led on to the mud platform here, instead of, as usual in these ruined dwellings, to the level floor of the room.
It appears to me probable that the position of this entrance, as shown in the plan, Plate X X X I I, accounts for the relatively greater frequency with which inscribed wooden tablets turned up near the north and east walls. When once the room had come to be used as a dustbin, it was natural that any one who emptied refuse into it should throw it from the entrance and as far as possible. The inscribed pieces of wood, which plainly represent the sweepings of an ancient office accommodated within some other portion of the house, were by their compactness and relative weight likely to be thrown further than the rest of the refuse, and often to strike against the walls opposite to the entrance. Keeping this in view we need not, perhaps, consider it altogether accidental that of the 24 documents on leather which were picked out from the litter, only 9 were found in the sections AT and E, while the rest lay scattered within C and W, i. e. much nearer to the entrance. Thin pieces of leather, being light, would, when thrown (say, out of a basket), not reach so far as handy pieces of wood similarly treated.
The layers of straw and the fragments of tamarisk matting, which were found embedded deep down in the rubbish, had in all probability belonged to-the roof once covering N. xv. Their position below i to 3 feet of miscellaneous refuse, including waste tablets and leather documents, plainly showed that while the room was used as a dustbin it must have partly at least been exposed to the sky. It is easy to understand that the roof of a room that had once been appropriated for such use would not be kept in repair, or else that an apartment which had become uninhabitable from defective roofing would be permanently utilized for rubbish. The condition of one of the rooms in the large house of Tokhta Akhûn Bég at Khotan, where I put up on two occasions 3, has since recurred to my mind as a very apt illustration. There a large apartment had become disused, owing to a conspicuous hole in the mud roofing which the wealthy owner never thought it worth his while to repair. Being situated conveniently near the kitchen all refuse from the latter had habitually been thrown into it, with results that obliged me to locate my cook elsewhere when I took up temporary quarters there. That the straw and fragments of matting in N. xv. actually came from the original roofing, was proved by a large piece of matting with a mass of well-preserved wheat-straw, which was found
tablets, while they lay exposed until packing became possible after the day's work, had sufficient abrading force to efface pencil marks which, for one reason or other (softness of surface, want of space, &c.), could not be made sufficiently bold and lasting. Such pieces had to be given the. supplemen-