in the walls indicated their position, the wood itself having perished through exposure. The remains of massive roof-beams were found lying in the sand above the floor of the north-east corner room. Both there and in the hall (E. iii.) intermediary free-standing posts had helped to carry them 2.
Owing to the large size of the sand-filled rooms in the northern part of the building their clearing proved a very heavy task, and could, within the limited time available, be accomplished only by keeping every available hand at work until late into the night on the 24th and 25th of February. The offer of additional wages and free rations made the men willing to face the fatigue and the bitter cold of this night labour, while big bonfires kindled with the plentiful dead tamarisk roots gave the light needed. The rooms proved, as I had expected, to have been completely cleared of all movable objects. But in the corner room a sitting platform, 3 ft. broad, as well as a large fireplace, were brought to light, while in the hall (E. iii.), open towards the south, the rough plaster surface of the partly well-preserved east wall proved to be covered with sgraffiti both Tibetan and Chinese. The photograph reproduced in Plate XI shows those to the south of the central pilaster of wood, the scribblings north of it being of a similar type. M. Chavannes' reading and translation, as given in Appendix A, Part iii. shows that the Chinese characters nearest to the pilaster contain a mention of ` the imperial envoy Hsinz Li-ch` an' , whose identity, unfortunately, has not been established. The other Chinese scrawls are too rough and effaced to permit of decipherment.
The Tibetan sgraffiti on this wall are on the whole better preserved than the Chinese ones, across which they are partly written, and bear the appearance of being later. They have been deciphered from my photographs and hand-copies by the Rev. Mr. Francke, whose transcripts and annotated translations will be found in Part iv. of Appendix B. The most interesting among these scribblings is, undoubtedly, the one which Mr. Francke marks with C, and which in the photograph reproduced in Plate XI is seen on the top below some coarse Tibetan and Chinese scrawls. According to Mr. Francke's rendering the main portion of this sgraffito records : ` At Pyagpag (in the) province of Upper o7om lour this army was outwitted, and a tiger's meal was obtained (i.e. many were killed).' To this is added, in coarse big letters, evidently by a different hand : ` (Now) eat until you are fat.' Have we here a record of some victorious Tibetan engagement, perhaps with a Chinese force ? The words, especially those of the postscript, seem to have a strain sufficiently truculent to have been scratched in by some Tibetan brave who took his share in the fighting. But this does not help us to identify either the localities named or the event alluded to,. The remaining three sgraffiti translated by Mr. Francke (one of them A, scratched in close to a small but well-drawn sketch of a charging tiger) seem to be scribblings of a still more casual character, referring to some property lost, a picture possibly once displayed on the wall, &c.
The ruined building, E. nI, was undoubtedly the main structure within the area enclosed by the ramparts, and, judging from the size of its halls and rooms, and the general arrangement of its plan, I think it safe to conclude that it had served as the residence of the officer and establishment, for the protection of whom the fort must be supposed to have been primarily intended. Somehow the spaciousness of the whole structure suggested resemblance to a Chinese Ya-mên, but I am unable to judge how far the disposition of the rooms, &c., might be compatible with Chinese architectural convention.