cleared, but the process yielded only the fragments of another wooden cupboard, N. xxvi. 04. a-d, like the one referred to above, and the ` striking plate ' of a wooden lock, N. xxvi. oil (Pl. XVI), described in the List below.
One of the purposes for which I had returned to the central portion of the site was the execution
of a task that had remained over ever since my first visit in 1901. In the large ruined residence N. III (Fig. IoI), situated about two miles to the south of the Stûpa, I had then been obliged to leave a few rooms to the west, as well as a large outer hall to the east of the central hall, unexamined. Sand filled them to a great height, and with the comparatively small number of labourers then available their excavation would have entailed a disproportionate sacrifice of time.21 The number of Kharosthi documents found in the cleared rooms had been small in comparison with the size and apparent importance of the residence, which at the time had led us to call it the ` Ya-mên '. In the light of my further experience at this site of the amount of ` waste papers ' to be reasonably looked for in the houses tenanted by persons of consequence, I had subsequently come to suspect that one of the small rooms left unexcavated immediately to the west of the central hall might
possibly have held the ` Daftar ' so far missed.
The ample supply of labour at my disposal on this occasion allowed me to finish the clearing of these rooms on December 17th, together with that of the greater part of the outer hall or loggia to the east, and the result confirmed this suspicion. In one of the two rooms to the west, which the revised plan (PI. 7) of the residence shows marked with x and the north wall of which appears on the extreme left of Fig. 97, as seen before excavation, there were found no less than two dozen Kharosthi documents on wood, several of the rectangular and wedge-shaped tablets being of good size and almost all in excellent preservation (see Pl. XVII, XVIII). This was a substantial addition to the quantity of ancient local correspondence and records previously secured from the site. Among the miscellaneous objects found here and in the adjoining room xi, mostly of wood and of household use (see List), may be mentioned several mouse-traps, N. III. x. oI, 8, 9 (Pl. XVI, XXVII), and four balusters, N. III. x. 014-17 (Pl. XV), with elaborate ball and ring mouldings closely resembling those found at the Lou-lan Site.22
In the large outer hall, N. III. xii, measuring fully 43 feet in width, no finds rewarded the laborious excavation. Of the beams that once carried the roofing only one remained, worked in a single piece of Terek wood but now broken and splintered. N. III. xii. of is a specimen of the capitals borne by the four posts that had supported it. The carved wooden baluster or short pillar, N. Ili. 07 (Pl. XV), is one of those found loose in the sand near the walls of the central hall.23
When this task had been satisfactorily completed after nightfall by the light of big bonfires, there remained for the next day only the complete excavation of the ruined dwelling N. xxxix (see plan, Pl. 5), which had been partially searched in 1906. It brought to light some additional rooms, but no finds of interest. Accordingly, by December i8th I was free to resume my journey eastwards with a good conscience. My renewed visit, rapid as it had to be, had thrown fresh light on the southern extension of the area occupied during the third century A. D. by the settlement
of ancient Niya or Ching-chiieh kt and had enabled me to trace farther south what I believe to
have shown as the contemporary terminal course of the Niya river. But the conclusions previously arrived at and fully set forth in Serindia,24 regarding the history of the site, the great change in physical conditions that has taken place since its abandonment, and the uncertainty as to the direct cause of this abandonment, remain unaltered.
21 For a full description of N. III and the interesting 22 Cf. Serindia, i. p. 449 ; iv. Pl. XXXIII, L.B. v. oo8.
finds it had yielded, see Ancient Khotan, i. pp. 330 sqq., 23 See Ancient Khotan, i. p. 333 ; ii. PI. VIII.
with Figs. 41-3, and plan, ibid. ii. Pl. XXX. 24 Cf. Serindia, i. pp. 242 sqq.