390 IN SEARCH OF THE LIMES TO SU-CHOU [Chap. XI
to hand. I may also mention that near this post and the next one eastwards, T. XLII. j, short stretches of an old canal could be traced, evidently connected with the Hsi-wan-ho.
Tati' with Beyond T. XLII. j the line marking the wall could be made out for about another mile, after
Han pot- which it disappeared amongst tamarisk-cones in the depression eastwards. Here we had to leave sherds.
it in order to gain the well which serves as the usual halting-place on the caravan route to Ying-p`an. Striking to the south-east from T. XLII. j, we passed, at a distance of a mile, a small ` Tati ' covered with potsherds of Han type this may well mark the position of a settlement extant during the early occupation of the Limes. Another five miles across the basin previously referred to brought us to the bare slope of a conspicuous rocky ridge. It is one of those which, as mentioned before, crop out along the northern edge of the big peneplain dividing the valleys of the Su-lo-ho and Ch`ihchin rivers.3
A knoll near the north-western end of this ridge, rising about 200 feet above the brackish well where our Camp 126 stood, was found to carry the ruin of a massive tower of conical shape, which, though standing away from the Limes, had a distinctly ancient look. It was built of layers of clay reinforced with large trunks of wild poplar and measured about 33 feet square at the base. Its top, at a height of some 12 feet, supported a small look-out platform constructed of layers of reed fascines. This tower commands a very extensive view to the north and north-west, whereas the view from the watch-posts of the Limes which we traced on the following day across the gravel plateau beyond the ridge, is masked in those directions by rising ground. I t is therefore likely enough that the tower that crowns this knoll dates from ancient times and enabled the defenders to keep a more efficient watch across the border line.
On the morning of April 25th I sent off the baggage to Ying-p`an by the caravan track and set out myself with Lai Singh and the few available mounted men northward in search of the line of the Limes. It proved an interesting day's work, disclosing remains so abundant that two more days were occupied in clearing and examining these. This work was carried out from a base established at the oasis and with the help of Chinese labour secured there. Having skirted the western foot of the rocky ridge and then covered about two and a half miles to the north, we struck the line of the wall stretching straight across a wide plateau from west-north-west to east-south-east.' The character of the ground, either bare clay or gravel, had helped to preserve the wall in a better state than we had found it since leaving An-hsi.
In many places, indeed, wind-erosion had reduced it to the condition of a low gravel-covered embankment. But there remained numerous segments where it still rose practically intact to a height of 6 or 7 feet and clearly showed its construction of alternate layers of stamped clay and fascines, each from 8 to Io inches in thickness. The fascines along this section of the Limes were composed mainly of tamarisk branches, easily gathered, no doubt, in the depression northward, where, as the map shows, tamarisk growth still abounds at the present day. The clay layers were extremely hard, though the water for compacting them must have been carried from a considerable distance.. The fact that the wall, in the well-preserved segments, was about 5 feet wide at the top makes it appear likely that along this portion of the line the original thickness of the wall conformed to the regular measurement found on the Tun-huang Limes, viz. 8 feet at the base.5
3 There appears to me some reason to doubt whether the aneroid reading which gave an elevation of 5,420 feet for Camp 120 can be safely accepted. My recollection of the ground towards Shih-êrh-tun and Ying-p`an suggests a lesser height. Professor Futterer's reading for the same point was I,35o metres (4,429 feet). In Map No. 40. D. 5 the
clinometrical heights for several points observed from Camp 126 are rightly marked with a query.
4 In Map No. 40. D. 5 the line of towers and wall, from T. XLIII. a to T. XLIII. f, has been shown by a draughtsman's error in the direction from NW. to SE.
5 See Serindia, ii. p. 736.