National Institute of Informatics - Digital Silk Road Project
Digital Archive of Toyo Bunko Rare Books

a decided turn to the east and crossed to the opposite side of the Etsin-gol valley. It was impossible
at the time to follow it there in search of the continuation of the line; for this would have prejudiced
the work that I had planned to carry out lower down in the Etsin-gol delta. But on my return
after accomplishing that task I took care to revisit the ground, with the result to be recorded in
a later chapter.³ For the time being it was a point of antiquarian interest that cultivation ceases
within half a mile, at the foot of this ridge, just as it evidently did in Han times. And character-
istically enough there was found below the tower, towards the left bank of the river, the modest
shrine of Ta-fêng-miao as proof that local worship has not ceased to respect the position once
occupied by a 'gate' of the ancient Han wall.³ᵃ

Ruined fort But it soon became clear that, whether in Han times or later, care had been taken to close the
guarding
approach approach from the north along the Etsin-gol by defensive works more effective than the mere
along river. border wall, and well advanced beyond it. As we moved across the riverine flat of gravel and
coarse sand to the grazing ground, known as Ta-wan, that lies by the westernmost branch of the
river bed (Map No. 42. D. 3), we passed, at about four miles' distance from T. XLVIII. b, a small
fort remarkably massive, T. XLVIII. c (Fig. 233; Pl. 16), which strikingly recalled the fort at the
ancient 'Jade Gate', T. XIV. It measured 32 feet square, inside walls of stamped clay themselves
fully 20 feet thick, which still stood to a height of over 30 feet. Among the potsherds found near
it both ancient and late ware was represented. Accumulations of sheep-dung within the walls
indicated recent occupation as a herdsmen's shelter.

Circum- About a mile to the north of where our Camp 140 stood at Ta-wan I found a large circumvalla-
vallation of
later date. tion, T. XLVIII. e (Fig. 228), quite close to the left bank of the wide river-bed (Map No. 45. A. 3).
Its Chinese name was given as T'ien-tsung-ta-wan, its Mongolian as Arun-tokhai-dürüljin. Its
walls of stamped clay, about 18 feet thick and as much in height, were provided with large square
bastions at the corners and enclosed an area about 220 yards square. The low ruins of three or
four structures inside the circumvallation revealed a masonry of vertically set bricks which was
clearly of late date. The bricks, in what seemed to have been a temple, measured 16″ × 10″ × 6″.
The general impression I received was that of a later stronghold, and the absence of any ancient
pottery debris seemed to confirm this.

Ruined About a mile to the south-east of this ruin and on the right bank of the river I visited on the
stronghold
on right morning of May 17th another stronghold, T. XLVIII. d, which to the Mongols is known as Taralingin-
river bank. dürüljin, the latter part of the designation (also pronounced as durwuljin) meaning 'fort'. It
occupies the western extremity of a low gravel plateau curving round from the desert hill chain that
we subsequently crossed on our way to Kao-t'ai (Map No. 45. A. 4). This extremity of the plateau
is faced on the opposite side of the valley by the foot of the gravel glacis that descends from the
easternmost outlier of the Pei-shan range above the Mao-mei oasis. Between the two the river-
bed contracts to less than half a mile. The stronghold, imposing in its ruin, looked on the whole
older than T. XLVIII. e. Various features pointed to successive structural changes, but there was no
definite evidence to date them.

Ruined fort As the sketch-plan shows, there is an inner fort of irregular oblong shape, enclosing an area
T. XLVIII. d. roughly 250 by 185 feet. The walls of stamped clay, about 12 feet thick at the base and still standing
in places to a height of about 25 feet, are strengthened by two solid bastions, one at the south-
western corner and one on the western face. The gate leading through the eastern curtain is
protected by an oblong outwork with walls of lesser strength, and within, moreover, by massive
flanking towers. The inside faces of the latter show a revetment of manifestly later masonry in
which courses of bricks set flat and upright alternate. A similar revetment covers a great portion