meant to support an upper floor. Here were found a double-wedge tablet, N. xxiv. iv. 2, in perfect condition, unopened and still retaining its clay seal ; a well-preserved double-bracket, N. xxiv. iv. 002 (see Plate xvIII), showing carefully executed wood-carving in Gandhâra style ; also some pieces of turned woodwork seen in the foreground of the photograph, Fig. 62. Of household furniture there are seen also here a large oblong eating-tray, broken, and a wooden cupboard of the type already described. The use of a net, made of the strong fibre rope which is still known at the present day under the name of kôshu, remains doubtful. The walls of this room showed a core of horizontal reeds within their solid plaster. The central pillar once supporting the ceiling had a thickness of one and a half feet ; from a height of about three feet above the floor its wood was completely bleached and fissured by exposure.
Arrange- The size of this room and of another to the south, vii, measuring twenty-five by nineteen feet,
together with other indications, proved that the house must have been that of a well-to-do person.
That he had been an official of some consequence was suggested by the fact that two long and narrow apartments, v and vi, to the east of these halls and manifestly intended only as passages or ante-rooms, yielded finds of well-preserved Kharosthi tablets of respectable size, including an oblong board, N. xxiv. vi. i, nearly three feet in length (see Plate xXV) on both sides of which was writing by different hands suggestive of drafts or office memoranda. As an interesting constructive feature, subsequently met with also in N. xxvi (see Figs. 62, 63), it may be mentioned that the passage, vi, communicated with the inner hall vii, not only by a door, but also by two wide panelled windows, of which the inset in Plate 14 shows the elevation. It is probable that these windows, like their exact pendants still to be found in the larger modern houses of the Khotan region, were originally closed with lattice-work.
Kharo§thi The previous finds of inscribed tablets in this house, scattered as they were, had raised the
tablets from hope of finding more in what might prove the owner's office room. This hope was soon justified office room,
N. xxrv. viii, when the clearing reached room viii, adjoining the main hall on the south and measuring about twenty-six by twelve and a half feet. The photographs, Figs. 61, 62, show parts of it in the foreground. As soon as the excavation proceeding from the east side of the room had reached a point about three feet from the south-east corner of vii, tablets emerged in rapid succession from the sand lying against the north wall of the room. They lay close together, loose, and without any order, from the plastered floor to about one foot above it, just as if files or bundles of these wooden documents had been thrown down haphazard. In the photograph, Fig. 58, a batch of these tablets is seen cropping out from the sand. The total number of pieces recovered here within a few square feet rose soon to fifty-four.
The majority of these are wedge tablets, two of them quite complete, and as detached covering-and under-tablets are represented among them in approximately equal numbers, it is probable that closer examination will lead to the fitting together of more double-wedges. There was found here only a single rectangular document, the under-tablet N. xxiv. viii. 44 b. The remaining twenty-three tablets belonged to that miscellaneous class of records for which in Ancient Khotan I adopted the general designation of ` oblong tablets '.3 The frequency of columnar arrangement in the writing showed that the contents were probably accounts, lists, and miscellaneous office papers ', to use an anachronism. An almost general and very gratifying feature of the records was their excellent state of preservation, without injury from damp or erosion. Evidently we had hit upon files front some official's daftar, thrown down and soon covered by loose drift sand, which ever since had offered them full protection. After this heap had been safely extracted, a careful search in other