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

of some dead trunks of cultivated poplars and fruit-trees, it proved that a small agricultural colony
must have existed near the fortified station.

Excavation The largest of the ruins, Ka. I (Figs 303, 304), was that of a good-sized dwelling with wattle
of ruined and timber walls of a type practically identical with that found in the houses of the Niya Site. Its
house. plan and internal arrangements (Plate 55) showed also very close resemblance to the latter. But
here, as in the dwellings of Lou-lan, the timber used was exclusively of the wild poplar. The
ruined house, in the main portion cleared,¹³ was filled with 6 to 8 feet of sand. Owing to the
immediate vicinity of a dune rising to 12 feet or more, some rooms to the east could not be excavated
with the dozen men or so available. For the same reason no search could be made for any refuse-
heaps outside which might have preserved archaeologically useful relics. The objects brought to
light are described in the List below. Among them may be mentioned the wooden lock and key,
Ka. I. 001–2; a dagger-like iron tool, Ka. I. 003; several amphorae, Ka. I. 0019 (Plate IV;
Fig. 307); strong woollen fabrics of different kinds, Ka. I. 0014–16. All these closely resemble
finds of the same kind made at the Niya and Lou-lan Sites. So does also the piece of an open-
work wooden screen seen in Fig. 307. A rectangular wooden tablet, Ka. I. i. 001, is also of a type
frequent among the wooden documents of a Niya Site, but has lost all traces of writing.

Other sand- At Ka. II were found the remains of a dwelling mainly of timber and wattle, badly destroyed
buried by wind-erosion and subsequently overgrown by tamarisks, which had helped to form a sand-cone
dwellings. some 7 feet high above it, but were dead now. No objects were found on clearing this. The
remains of a third ruined dwelling, Ka. III (Fig. 305), were also buried in a tamarisk-cone, which
was, however, still living; the walls were built here of timber with plastered vertical bundles of
rushes and reeds. Apart from a large pottery jar, a plain wooden plank-bed, 7½ by 4½ feet, was
the only find here. Four more small dwellings, of which the position is marked in the site-plan,
were found either completely eroded down to the foundation beams or else too deeply buried by the
side of high dunes to permit of excavation with the limited number of men available.

Evidence of Scanty as are the newly explored remains of the site and the objects found on it, they yet
agricultural furnish definite evidence that a small agricultural settlement must have existed here far away in the
settlement. desert, and not merely a small frontier guard-post, as I had been led previously to assume. With
regard to the probable date, too, of the site, not merely as regards its character, the new observa-
tions permit us, I think, to form a clearer view. The resemblance in the construction of the houses
and in the type of the objects of daily use found there is sufficiently close to justify the attribution
of the ruins approximately to the period when the Niya and Lou-lan Sites were abandoned, i.e. the
third–fourth century A.D. With this dating the two coins found close to Ka. I fully agree; they

Date indi- are both Wu-chu pieces, apparently of the second–third century A.D. The coins found on the
cated by occasion of my first visit, fourteen in all, were also either Wu-chu pieces or else uninscribed.¹⁴ The
Han coins. few tiny bits of paper found in 1901 among the débris of the ruined quadrangle¹⁵ raise no longer any
chronological difficulty, since the discoveries at Lou-lan have proved that the use of paper by the
side of wood as writing material had reached the Tārim Basin by the middle of the third century,
if not somewhat earlier.

Changes in Even now we cannot determine the immediate cause which may have led to the abandonment
Keriya of the small settlement. But my previous remarks as to the possibility of this having been caused
River bed. by a change in the course of the river which deprived the site of its water-supply¹⁶ have since
received striking illustration by what I was able to observe myself on my renewed visit. The
river, which in 1901 was fully eleven miles away from Kara-dong at its nearest point, flowed seven