Sec. ii] FROM CHIAO-WAN-CH`ÊNG TO SHIH-ÊRH-TUN 383
been taken along the caravan track leading from T. XLI. k to the river. After gaining its right bank we followed it to the point where the river turns almost due south towards Yü-mên-hsien. The tower T. XLI. n which stands here, built of stamped clay and about 12 feet square at its base, marks a usual halting-place for caravans coming from or going to Hâmi. It is known by the name of the village, Ma-ku-t`an, the fields of which approach close to the left bank 12 The tower appeared to be of late date, and the small metal objects picked up near it afford no chronological indication.
The river flows here and for several miles below in a narrow bed cut into steep banks of clay, Volume of 3o to 4o feet in height. Where we measured its volume, it was only about 20 feet wide, and from Su-lo-ho.
3 to 4 feet deep. The volume roughly ascertained was only about 18o cubic feet per second, and obviously represented but a very small proportion of the water carried by the Su-lo-ho at this season, the rest being all absorbed by the branches that irrigate the lands of San-tao-kou and the other oases westwards, and by the several canals of Yü-mên-hsien.
Leaving the route towards Shih-êrh-tun which had already been surveyed in 1907, at a distance Dried-up
of half a mile from T. XLI. n we turned to the north-east in order to reach a conspicuous tower lacustrine
already sighted on the previous journey and evidently lying in the continuation of the line of the Limes. The level plain, which we had found in 1907 distinctly marshy at a point farther to the east, had here the appearance of a marsh bed dried up in a recent period. Where the ground was beginning to rise slightly and the clay bottom of the depression had lain dry longer, wind-erosion had begun to scour patches left unprotected by vegetation. Elsewhere drift-sand had accumulated in low dunes amidst reed growth and scrub.
We came to the tower T. XLI. o (Fig. 212) at the point where this dried-up lacustrine depression Limes line
gives place to gravel Sai and the outermost low stony ridges from the north crop out. It was traced south
of T. XLI. o.
built in stamped clay on a base 32 feet square and like an adjoining square enclosure (Pl. 14) bore no appearance of great age. But about 5o yards to the south of it, closer examination disclosed the line of the Limes agger running for about a furlong along a low ridge of detritus. It was clearly indicated by layers of tamarisk branches cropping out on the sides of a low straight embankment of gravel and rubble. It was aligned north-westwards upon T. XLI. 1, but no intermediate towers could be seen ; nor was the Limes likely to have left any conspicuous traces where it crossed the vegetation belt along the shore of the old marsh bed. Just opposite to T. XLI. o and close to the south of the agger, abundant fragments of Han pottery marked the position of a watch-post that once occupied the stony crest.
Significantly enough not a single potsherd of this type could be found near the walled enclosure Route
outside the Limes line, while pieces of glazed ware and porcelain abounded. A small inscribed guarded
Chinese tablet was found just outside that enclosure, but the writing on it looked late. A well-marked T. XLI. p.
track passes close to the little fortified post ; it comes from Shih-êrh-tun and leads, as subsequently ascertained at the latter place, towards the previously mentioned route of Obrucheff and Grum Grishmailo, which it probably joins near the well shown in Dr. Hassenstein's map accompanying Professor Futterer's paper as ` Ulun-tschuan '. The watch-tower T. XLI. p which was sighted on rising ground to the north-east was obviously meant to guard the approach towards the cultivated area from the same side. As seen through our glasses it had a recent appearance.
We followed the line indicated by the agger segment from T. XLI. o to the south-east along the Ancient foot of the last detritus-covered offshoots of the hill chain. The agger, with its layers of tamarisk agger traced brushwood exposed and still rising to 5 or 6 feet, reappeared in three places, in each for a distance to SE. of about a quarter of a mile. After a tramp of about three miles we reached the remains of a ruined
12 The cultivated area of Ma-ku-t`an, as shown in Map On September 21, 1907, we had passed here in the dusk
No. 4o. c. 5, ought to have been extended farther north. and failed to notice this detail.