| || || |
Ruined quarters within fort.
348 THROUGH THE LOP DESERT [Chap.%
which a layer of sand and fine gravel had accumulated, evidently blown across by the prevailing north-east winds. It was here, not far from the inner north-east angle which had afforded special protection, that a few wooden posts rising a foot or so above the gravel surface had attracted Tokhta Akhttn's attention when he visited the ruin in the preceding spring.
While digging here single-handed he had come upon what looked to him like the brushwood layer of a roof, and amongst it he discovered the piece of paper inscribed with Tibetan which he had brought to me at Charkhlik. It was obviously the best place for my intended trial excavation. So I set my men to work on a line stretching along the east wall and soon had the satisfaction of seeing a row of small apartments emerging from below the cover of sand and gravel (Fig. 114). • They were all built of brick, with rough posts of Toghrak wood to support a roofing made of tamarisk branches with earth above. Except that they were ranged more or less parallel to the east wall of the fort, as seen from the plan (Plate 30), the disposition of these little rooms was very irregular. Among those cleared in the course of that and the following day's trial excavation (M. 1. i—vii), the largest measured only sixteen by eleven feet, and in some of the smaller ones the width was barely five feet.
The small size and roughness of the half-underground hovels were compensated by the rich antiquarian contents of the refuse which seemed to fill them to the roof. No sooner had the digging begun than numbers of pieces of paper and wood inscribed in Tibetan cropped up. The layers of straw, chipped wood, and decomposed dirt which the occupants had left behind, or which had been thrown in when these hovels were turned into dustbins, continued to yield such records, complete or fragmentary, right down to the bottom. From the room M. 1. i, where Tokhta Akhttn had made his find, over forty pieces were recovered, while in the group of small closets marked iv the number of such ` finds ' rose to 136 by the second day.
The great majority of these Tibetan records consisted of narrow slip-like tablets of wood, from six to eight inches in length and up to two inches wide. When complete, they usually showed an oblong seal cavity at the left end, thus clearly proving the secular nature of the contents. Documents of this kind seemed to predominate also among the records on paper, consisting mostly of small and rather flimsy sheets, which recalled my corresponding finds at Dandan-oilik. In some instances these thin sheets were found still neatly folded up, as though meant for dispatch, and bearing seal impressions in red paint. But by the side of such miscellaneous remains of correspondence, &c., oblong leaves, with neatly written text between regularly ruled lines, also turned up. In these it was easy to recognize fragments of Pbthis with religious contents, as script, shape, and material alike closely resembled those of the Kanjur fragments which I recovered in 1901 from the Tang fort at Endere.
Remains of implements of all sorts, articles of clothing, arms, &c., were recovered in equal
abundance from these deep deposits of rubbish. Leaving the description of them and of the Tibetan records from this site for a later chapter, I will here only mention the many curious pieces of scale armour, made of hard leather and lacquered in red and black. These were often still fastened together with the original interlacing thongs of leather, and by their arrangement proved that I was right in recognizing as parts of scale armour the few detached pieces of hard leather, with threading holes, which had come to light during my first excavations at the Niya Site.4 Some chronological interest attached to the discovery of two large and intact jars (Fig. 87), each measuring two feet in diameter and bearing below the neck an engraved "Tibetan character. The manner in which these jars were fixed in small raised recesses between the rooms ii and iii, v and vii showed clearly that they were used for the storage of liquids while the quarters were actually
4 Cf. Ancient Kholan, i. pp. 374, 411 ; also p. xvi.