Sec. ill] DISCOVERIES IN AN ANCIENT REFUSE-HEAP, L.A. VI. ii 383
dwellings and refuse-heaps of approximately the same period did not yield the smallest fragment of paper. Even at the Lou-lan Site some inference may, perhaps, be drawn as to the comparative rarity of paper from the fact that in several instances the reverse of papers bears writing by a different hand and, in one case at least, of an official character.8
In striking contrast with the abundance of Chinese records is the small number of Kharosthi documents which have found their way into this general deposit of refuse. This disproportion seems to me quite consistent with the prominently Chinese character of the station represented by the L.A. ruins and of the traffic and administrative activity which it served. Besides four fragmentary tablets and three small torn pieces of paper with Kharosthi script (L.A. v1. ii. 0102, 0103, 0236, Plate xxxviii) there was found the large and almost complete paper document L.A. vi. ii. 0234 (Plate xxxIX). In shape and in the arrangement of the writing it closely resembles the Kharosthi documents on leather which another precious refuse-heap, N. xv, had yielded at the Niya Site.° A novelty is presented by a strip of fine undyed silk, L.A. vi. ii. 0235 (Plate XxxIX), inscribed with two lines of Kharosthi. It has furnished the first tangible confirmation of the Chinese antiquarian tradition that silk was one of the ancient writing materials used before the invention of paper, about A. D. 105.10 The material in this case, too, just as in that of the Kharosthi documents on paper, must have come from China. It deserves notice, therefore, that the other records on silk which I subsequently discovered, two of them in Kharosthi and Brahmi respectively, came from ruined watch-stations along that very route which the ancient Chinese silk trade had followed through the desert west of Tun-huang." The fragment L.A. v1. ii. 059 (Doc. No. 918, Plate xXVIIi) is of some interest because the torn piece of paper bears on one side remains of a Chinese epistle written at Tun-huang, and on the other three lines in Kharosthi.
Glad as I was of these plentiful manuscript remains, which confirmed or expanded previously gained archaeological knowledge, I was at the time even more interested in a small torn scrap of paper, L.A. vi. ii. 0104 (Plate CLIII), which raised a fresh problem. Barely 4 inches long with a maximum width of two inches, it retained small portions of four lines in what was a truly ` unknown ' writing. It manifestly. ran from right to left, and some of the characters distantly recalled Aramaic. It was obvious that no decipherment could be hoped for from so tiny a fragment. Yet this did not prevent
my thoughts from instinctively connecting this script of clearly Western look with people from ancient Sogdiana, or even more distant Iranian lands, who might have followed this early high road to the country of the silk-weaving Seres. I could not foresee then that a lucky discovery, made four months later in a ruined watch-station at the eastern end of this ancient desert route, would put into my hands quite a number of complete documents in the same script. Still less did I imagine that their subsequent partial decipherment by the philological penetration of that lamented Iranian scholar, M. Robert Gauthiot, would prove them to be in Early Sogdian writing and language.12
That all these multifarious records had found their way to the large refuse-heap as sweepings
from the neighbouring quarters and offices was made plain by the plentiful remnants of worn-out articles of clothing, furniture, and implements of all sorts which were mixed up with them in the layers of rubbish. A reference to the descriptive list below will show the variety of the miscellaneous relics recovered from L.A. VI. îi.13 More vividly perhaps than anything else, they