Sec. ii] EXCAVATIONS AROUND AND WITHIN TANG FORT OF ENDERE 279
I now had the bank of ancient refuse carefully cleared and ascertained that it extended right down to what seemed the natural soil of hard loess, six feet lower. On either side it was embedded between the stamped clay layers of the later rampart. There could be no possible doubt that the rampart, with characteristic indifference to solidity of construction, had been built up here encasing a small rubbish mound belonging to the settlement abandoned centuries earlier. Its contents consisted mainly of sheep-dung and twigs, with a plentiful admixture of rags of miscellaneous fabrics, including silk, felt, and coarse woollen materials (E. Fort. 0012-0018).
It is a matter of distinct archaeological interest to ascertain whether among all these fabrics there are any of cotton ; for the use of the latter, as far as observations based on the results of my excavations go, has up to the present been proved only for sites dating from the Tang period or later. In view of this criterion it is important to record that the expert report received from Dr. Hanausek on the representative specimens which were submitted to him for analysis, conclusively proves the absence of cotton among the rags excavated from below the rampart of the Endere fort. This negative evidence helps further to strengthen the conclusion already arrived at as to the early origin of this rubbish mound.1' Here was found also a bone knife-handle (E. Fort. 001. c ; Plate XXVIII), closely resembling the one found in ruin N. xxvl of the Niya Site. It seemed probable that the refuse deposit had attained considerable consistency long before the builders of the Chinese fort in the seventh century thought fit to embed it in their rampart, and I wondered how much more of the débris of the earlier site might still rest safely hidden under other portions of the circumvallation.
Owing to the heavy accumulations of drift sand which this circumvallation had helped to catch and retain, it had been impossible on my previous visit to clear all the rooms of the large structure forming the principal quarters of the fort. The greater number of labourers now available enabled me to complete this heavy task. From room iv in the south-east portion of this structure, E. III, we recovered the fine wooden pillar, six feet four inches high, seen on the right in Fig. 70, which once must have borne a double-bracket supporting the ceiling. The rich mouldings it shows can only have been produced by the turning-lathe—a remarkable feat seeing that the maximum diameter of the pillar was over thirteen inches. The equally well-worked though less massive pillar, seen on the left of Fig. 70,12 was the only find which rewarded the excavation of the large hall in the north-west corner. This measured 46 by 27 feet and had a sitting platform 4 feet broad and 21 inches high along its north wall. The extant portions of the walls had here preserved their plaster facing, over six inches thick. Such plaster may be assumed to have once covered the massive brickwork in the other rooms also, where now it has fallen off by exposure. In the north-west corner of the adjoining court, vi, the whitewashed wall plaster survived to a height of about five feet and showed traces of small sketches in colour, including that of a kneeling figure robed in blue. There were remains of some Tibetan characters, too, but none sufficiently clear to be read with certainty.
In the area covered with stable refuse, to the north of the main quarters, we now discovered two small underground apartments completely filled with sand (see E. vii, viii in plan, Plate 20). They had no door or other opening and were evidently approached from above where the foot of timber and plaster walls belonging to an upper story could just be made out. Below this, the walls up to a height of six and a half feet were built of brick with a plaster facing. In both rooms there were elaborately modelled fire-places, as seen in the photograph of E. viii (Fig. 72),13 showing that these were meant for shelter in winter. Just as in the dwellings of Niya and•Khâdalik the fire-place was
" For the details of Dr. Hanausek's painstaking analysis cf. below, E. Fort. 0012-00x4, oor8, in Descriptive List.
n For elevations of these ttvo pillars, see plan, Plate 2z.