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 wood. The outside wall on the south-west of iv had completely disappeared, except for remains
of layers of Toghrak branches at the south corner. Fallen timber strewed the slope some 150 feet
lower down in plenty. The fall of the wall was evidently due to the foundations here having slid
down the precipitous rock ledge. It was through a gate on this side that the keep-like main struc-
ture within the fort i (Fig. 330) was entered. This appeared to have been originally built with walls
of coarse brickwork, 4 feet thick, which were subsequently strengthened to a thickness of 8 feet on
the north, east, and south. The fact that the main west wall of the fort adjoining i was built of a
different material seems to indicate that i was the earliest structure occupying the site. It had con-
tained more than one story ; for the square holes found in the walls, well below their top level in their
present broken condition, were obviously intended to carry a flooring. The interior space, 20 feet
square, was found filled with débris including burned timber to a height of some 7 feet.

Walled On the north-east a narrow apartment, ii, 20 feet long and only 6 feet wide, had been spared
outer court. between the 'keep' and the fort wall (Fig. 331). The latter was found here much broken, and
through it ii appeared to have communicated with the outer court v by means of a flight of
stairs. This court v, situated on a level more than 20 feet lower than the area of the fort
proper, was clearly marked as a later addition. Its walls, though also of considerable thickness and
built with sun-dried bricks of the same size as those of the fort, lacked the strengthening layers of
tamarisk brushwood and the timber reinforcing the latter, and consequently had suffered more
damage, especially at the north corner. The space enclosed by them, about 90 feet by 29, had
probably been roofed over ; for charred timber was abundant among the débris and refuse, and all
the brickwork was reddened by fire. The masses of horse-dung found on clearing the space along
the north-east wall and elsewhere showed plainly that this outer court, through which the fort was
entered, had served mainly for stabling. The gate led through the 10-feet thick wall on the south-
east.

Tower The tower (Fig. 329) which crowned the crest, at an elevation of 225 feet above the top of the
crowning clay banks by the river-bed and at a distance of over 60 yards from the west bastion of the fort, was
hill-crest. of remarkable solidity. In construction, size, and conical shape it curiously recalled the familiar
watch-towers of the Tun-huang Limes. It was built of flat and fairly regular pieces of hard clay, no
doubt brought from the banks lining the foot of the hill above the river, with layers of tamarisk
branches at intervals of 10 inches and Toghrak posts and beams inserted in the masonry. Its base
measured 25 feet on the south-west and north-east, and apparently 22 feet on the two other sides.
But, as the south-east face was badly broken, the apparent greater length of two sides cannot
be depended upon. The extant height was over 20 feet. Even without ascending to its top, the
view ranged far up and down the broad bed of the river and its jungle belt, and beyond to the
distant high ridges of bare sands.

Excavation The excavation of the remains within the fort was but the beginning of labours which were to
of 'keep'. keep us busy for three long and hot days. Marks of recent burrowing within the 'keep' i clearly
showed where the Tibetan wooden slips brought to me at Islāmābād had been obtained. But these
diggings had left the heavy débris lower down undisturbed, and on clearing this a number of Tibetan
documents on small tablets, exactly like those found in the Mirān fort, were recovered, besides rarer
fragments of Tibetan records on paper. More of such relics, evidently left behind by the last occu-
pants of the fort, turned up among débris thrown into the recess iii outside the entrance of the
room i. Of miscellaneous small objects found in the latter I may mention a reed pen, M. Tagh. i.
006, cut with nib exactly as a pen found at Mirān ; a wooden die, M. Tagh. i. 007, with numbers
arranged as in its ivory pendant from Mirān ; a wooden key of the Khādalik type, M. Tagh. i. 0011 ;
and the well-made terra-cotta saucer in the form of a tortoise, M. Tagh. i. 0029 (Plate LI).