382 IN SEARCH OF THE LIMES TO SU-CHOU XI ticable for nomads by which the easternmost Pei-shan can be crossed from the side of Mongolia and Dzungaria.
Finds at By sunrise of April 22nd I resumed the survey of the Limes agger eastwards. It was followed
tower with ease for a mile over the gravel soil, to where the much-decayed remains could be traced of
a watch-post with walls 3 feet thick built of bricks measuring 14" x 9" x 4". Among the miscellaneous small objects found on clearing the refuse and described in the List below may be mentioned the part of game-trap, T. xi.i. i. o6 (P1. XLVI), interesting by its type still known in widely distant parts of Africa and Asia.9a Besides pieces of mat-marked grey pottery of Han type, fragments of porcelain were also picked up here, suggesting occupation of the post in later times also.
Positions For about two miles farther the low straight mound marking the line of the Limes remained of watch-
posts T. clearly traceable. But the position once occupied by a watch-station, T. xLI. j, at a distance of
j, k, one mile, and by another, T. XLI. jj,10 at a similar interval, could be recognized only by low gravel
mounds thickly strewn with potsherds of Han type. The effects of wind-erosion became still more noticeable beyond, where the ground turned to bare clay, with the line of the agger indicated only by a perfectly straight line of dark gravel running across it on the surface. Having covered about four miles from Ch`iao-wan-ch`êng, we crossed the well-marked track by which the route from Hâmi, above mentioned as having been followed by Professor Futterer, reaches the right river bank in present times. Close to it a clay terrace, about 3 feet high and about 75 feet in diameter, thickly strewn with Han pottery fragments and stones, marked the spot once occupied by a watch-post, T. xLI. k. A large Wu-chu coin was found near it.
Remains at Gravel-covered patches of ground allowed the line of the agger to be followed for about a mile
T. XLI.1. farther to T. XLI. 1. There a couple of small Stüpas, built of bricks on end and obviously of late
construction, were found close to the line of the Limes. A clay terrace about 18o yards to the south of the latter bears the remains of a small domed structure, partly cut into the clay. The pottery debris found near it seemed mostly of later date, and so also the ornamented bronze stud, T. XLI. 1. 01. A curious feature of the site was a canal, about io feet wide, traceable for about a quarter of a mile and running from east to west at about 6o yards distance within the line of the agger. It may have once received water from a shallow Nullah upon which we came on our way to T. XLI. m.
Beyond T. xLI. 1. the trace of the agger was soon lost on ground which had suffered much from erosion and which gradually passed into a wide sandy expanse abundantly covered with reeds and scrub. The tower, T. xLI. m, drew us away to the south-east, but when reached after we had proceeded about 2 miles from T. XLI. 1, proved to be of later origin, with a miniature shrine on its top.
Depression We had now entered a depression three or four miles wide, extending along the right bank of
along Su-lo- the Su-to-ho, where the river curves round in its bend to the west. I had passed across the south-ho bend.
easterly portion of this depression in September, 1907, on the reconnaissance which took me from Yü-mên-hsien to the remains of the Limes near Shih-êrh-tun (Map No. 4o. c. 5) and thence to San-tao-kou on the An-hsi high road.'1 I had then ascertained that this area, extending between the gravel glacis of the Pei-shan and the river, probably represents an old lacustrine bed. It still receives abundance of moisture by percolation and in parts by seasonal inundation from river-beds branching off below Yü-mên-hsien. I was therefore aware that on ground such as this the chance of any remains of the Limes surviving was slight. On the other hand, I knew with certainty where I should find the continuation of its line to the south-east.
Track This made me regret less the necessity of leaving the probable direction of the line at this
followed to point, and of turning south in order to resume touch with our baggage train, which had wrongly river bank.
9a Regarding the use of this trap, see Mr. H. Balfour's 10 Not marked on Map No. 40. B. 4.
note in Add. & Corr. to p. 421. 11 Cf. Serindia, iii. pp. 1136 sqq.