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

respect of numbers was absolutely necessary. At the same time, as several distinct types could
easily be recognized among this multitude of shrines, care had to be taken to make the selection of
those to be searched representative. Fortunately it was easy to realize that the disposition of the
majority of the ruins in long rows of small adjoining cellas, or in groups of detached shrines equally
close together and accessible, would facilitate the employment of a large number of labourers under
adequate supervision. To obtain them rapidly and to keep the work going steadily at high pressure
was made possible by specially favourable circumstances.

At Kara-shahr I had entered again the great division of districts over which my old friend and Effective
patron P'an Ta-jên exercised control as Tao-t'ai of Ak-su. Thanks to his ever-effective recommen- administra-
dation, Chiang T'ai-chin, the energetic prefect of Kara-shahr, had provided all needful administrative tive help.
support for my labours. Fortunately, too, the populous village tract of Korla was within a day's
march to supply large and willing contingents of Turki Muhammadans who knew how to use their
'Ketmans'. Pickaxes needed for the hard débris were secured with equal promptness from the Chinese
military post at the same oasis. Some of the ruins still retained their vaulted roofs and provided
night shelter, such as the men badly needed under the trying climatic conditions prevailing. With
the help of efficient village head-men from Korla it was easy to keep these large bands at work
from the bitterly cold hours of dawn until nightfall, and to relieve them by fresh relays of men as
soon as the effect of long days of strenuous work and of the exposure implied began to tell upon
them.

Valiantly aided by Naik Rām Singh, Chiang Ssŭ-yeh, and after his arrival from the Kuruk- Clearing of
tāgh also by R. B. Lal Singh, I thus managed during my twelve days' stay at the site to get the shrines.
great majority of the shrines and other ruins cleared, and, as the special mark of the broad arrow
used in the plans (Plates 52, 53) shows, in most cases completely. In the few larger temples which
are marked as partially cleared, work was stopped only when it became evident, from the condition
of the heavy masses of hard calcined débris brought to light, that the inward fall of the thick temple
walls after a big conflagration had left but very scanty chance of any remains of interest having
survived the combined effects of fire and such a crushing.

The total number of individual shrines at the main site amounts to over a hundred. In their Con-
dimensions they vary greatly, from miniature cellas of only 4 to 6 feet square to massive rectangular structive
piles measuring up to 80 feet on one side. But the types of construction represented, as reference types and
to the plans and photographs shows, are few, and much uniformity evidently prevailed also in the materials.
arrangement and decoration of the interior. Sun-dried bricks are used throughout these structures.
Their prevailing size is about 12″ × 6″ × 3-4″. In the larger shrines a good deal of timber appears
to have been set in the masonry to give it greater cohesion. In some of those rising on walled-up
terraces, like xxvi, xxvi, I noticed also the insertion of thin layers of reeds. All this points to
climatic conditions not unlike the present, in which masonry of sun-dried bricks alone does not
assure enough strength.

It will be convenient first to describe only the most frequent types briefly in their general Vaulted
features and to leave the mention of details till we deal with individual ruins. The commonest type temple
among the small shrines is the simple cella, either square or rectangular, usually ranged by the side cellas.
of others on a terrace and sometimes approached through a porch. There is reason to believe that
these cellas were always covered with true vaults, which survived in a few near the northern end of
groups I and II. Elsewhere the lower courses of their brickwork could be traced. Another and
larger type, best illustrated by ruin xiii (Plate 53), has a cella with a vaulted narrow chamber behind
the wall facing the entrance. Low vaulted openings adjoining the side walls give access to this
chamber or passage, thus permitting circumambulation of the principal image, which once must have
1274 7 M