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

continued in occupation considerably longer than Chong-tim, and well into the Muhammadan
period. Besides two T'ang pieces they comprised two Sung dynasty coins of A.D. 1034–8
and 1068–78 respectively, as well as a small silver coin of the Golden Horde struck at Khwārizm
in A. H. 743.
Within a mile or so of these little 'Tatis' a belt of close-set tamarisk-cones was entered, such
as usually marks approach to an old river-bed, and a little beyond we struck the well-defined
dry bed of the Göre-akin, lined by big Toghraks, live as well as dead. It was known to my guides
as an old terminal branch of the Kāshgar River which passes through the defile between the two
rocky hills west of Tumshuk (Map No. 15. D. 1) and was said to be traceable far away eastwards.
Beyond this riverine belt, in all details recalling ground I had seen so often where the rivers
of Keriya, Niya, etc., die away in the Taklamakān, I came upon groups of ruined dwellings rising
on low mounds amidst tamarisk-cones of small size. But they were obviously not of any antiquity;
they probably mark an attempt to reclaim this area from the jungle made during a period relatively
recent when the Göre-akin may for a time have again carried water. We had reached the northern
edge of the terminal delta of the Kāshgar River, and the constant changes to which cultivation in
such areas is subject have been noted too often before to need more than a passing reference.
Arrived at the village of Tumshuk by the high road I found myself on ground close to ruins
which had repeatedly before received attention from European travellers and archaeologists.
Professor Pelliot in 1906 had devoted several months' work to the extensive ruins of Buddhist
shrines found at the foot of the picturesque rocky hill which rises to the north-west of Tumshuk.
On the rapid visit I paid to them I had the satisfaction to see with what systematic care and
thoroughness his clearing had been carried out, even though it was obvious from the first that the
temples and monastic dwellings of the site had all suffered badly through burning. On the opposite
side of the hill chain and not far from the ancient circumvallation of Tāpa-shahr and the left bank of
the Göre-akin I inspected the scattered ruins of dwellings which are known as Tokuz-sar āi. Coins
of Muhammad Arslān had been brought to me at Tumshuk as finds from the latter locality, and
suggested that the abandonment of this site could not have taken place earlier than the fourteenth
or fifteenth century. This dating has its special interest with reference to the previously noted
'Tati' remains lower down the Göre-akin, and a rapid examination of the ruins confirmed the
chronological evidence of the coins. A day's halt at Tumshuk allowed me to visit also the
ruins to be found at both ends of the long-stretched ridge which continues the line of the pre-
viously mentioned hill to the south-east beyond the gap passed by the high road. Those at
the southern end manifestly belong to a pre-Muhammadan stronghold (Fig. 339) and might have
tempted exploration had it been possible for me to spare time for the protracted labours there
needed.
As it was, the necessity of keeping adequate time in hand for the manifold tasks of the summer
months now urged a return to Khotan as early as possible. A variety of considerations imposed
the route via Yārkand, and in order to cover the 350 odd miles which on it separated me from
Khotan without too great a sacrifice of time, a succession of forced marches became necessary. The
first of them, on May 21, carried me to Marāl-bāshi over ground which offered varied geographical
interest. Right through from Tumshuk the road leads transversely past the foot of a series
of rugged hill chains, all striking parallel to each other from north-west to south-east (Map No. 15.
C, D. 1). Each of them is now broken up by erosion into what looks like a line of isolated rock
islands cropping out from the flat plain. Yet uniform geographical formation shows clearly that they
are but the remnants of one ancient range which started at a sharp angle from the outermost T'ien-
shan north of Marāl-bāshi and once had its continuation to the south-east far away into the

Terminal
delta of
Kāshgar
River.

Ruins near
Tumshuk.

Remains of
Tokuz-sarai.

Hills near
Tumshuk.