国立情報学研究所 - ディジタル・シルクロード・プロジェクト
『東洋文庫所蔵』貴重書デジタルアーカイブ

mixed contents. No moisture whatever could rise from the river below to the height of this rocky
ridge, absolutely clear of any vestige of vegetation. The elevated position and the outlying lower
ridge already referred to kept off dunes and prevented erosion by drift-sand.

Clearing of
rubbish
deposits. The clearing of the rubbish deposits was started the first day on the slope just below the
approach to the outer gate of the fort. On the two following days it was extended westwards
to beyond the north corner of v, the sections cleared each day being marked a–c. In each section
the work was carried from the bottom of the slope upwards, and the written records as they emerged
were gathered into bags numbered consecutively with Roman figures, as it was impossible to keep
count of individual pieces, the number of which in the end rose to close on a thousand. The uniform
distribution of the refuse along the whole length of the outer court proved plainly that it must have
been thrown indiscriminately over different portions of its north-east wall, and this makes it impossible
to determine any chronological sequence among the deposits. Before referring to their rich yield of
records I may briefly mention the more interesting among the miscellaneous objects recovered.

Miscel-
laneous
relics of
garrison. Modest finds in themselves, they all help to illustrate the conditions of life prevailing at this
forlorn little guard-station in the desert. The arms provided for its garrison are adequately
indicated by a number of arrow-shafts, M. Tagh. a. 0013–16 ; b. 007–10 (Plate LI), some unfinished
and one with a leaf-shaped sheet of bronze in the place of feathers ; pieces of broken bows made of
tamarisk wood, M. Tagh. a. 0017–18 (Plate LI) ; wooden sheaths for short sword and dagger,
M. Tagh. a. 004–5. There are numerous cast-off shoes worked in various kinds of woollen and
felt materials and sometimes quilted in different patterns, M. Tagh. a. 0039, 0041–5, all suggesting by
their thickness and large size that rigorous cold had to be guarded against. String sandals, too, of
the type familiar from Lou-lan and the Tun-huang Limes were also represented, M. Tagh. a. 0040
being a specimen. By the side of numerous fabric remnants, evidently from clothing, made of wool
and mostly of purple or bright red colour, the absence of finer textiles, especially of silk, was
significant. Of the single silk brocade fragment, M. Tagh. a. iv. 00177, it should be noted that its
weaving closely resembles that of a piece found at the Tibetan fort of Mirān. Wooden eating-sticks
are represented by two pairs, M. Tagh. a. 0019 (Plate LI) ; c. 002. Pieces of netting, M. Tagh. b.
005, 0017, show that the chances of fishing were used when the river brought down its flood-water.
Wooden dies, like M. Tagh. a. 0031 ; iv. 00172, with holes or inked circles arranged in the same way
as at the Mirān fort, point to what obviously was a prevalent pastime among these Tibetan soldiers.
The numerous wooden seal-cases, of a type closely corresponding to those found at the stations of
the Tun-huang Limes, M. Tagh. a. 0025–6 ; iv. 00173, etc., would suffice, even without the
abundance of 'miscellaneous papers', to prove the maintenance of regular communications with and
through this outlying post. The use of wooden keys and locks, similar to those found at the sites
round Domoko, is attested by a number of specimens, M. Tagh. b. 0012–15. etc.

Tibetan
documents. It is, however, to the documents preserved in these deposits of rubbish that we must look chiefly
for data bearing on the character of the ruined fort and the time of its occupation. As regards the
latter the mere fact that the vast majority of the records recovered is Tibetan is a sufficiently clear
indication.⁵ We know from our historical sources that the predominance of Tibetan power in
Eastern Turkestān commenced soon after the middle of the eighth century, that by A. D. 791 the
Chinese 'Protectorates' of An-hsi, i. e. Kuchā and Pei-t'ing (near Guchen), finally succumbed to it,
and that Tibetan supremacy in these regions was maintained until the Uigurs established a great
kingdom about A. D. 860 in the northern territories and westernmost Kan-su.⁶ The Khotan region