Village of Shih-érhtun.
T. xLII. a.
Mound marking post
T. XLII. b.
Rampart traced eastwards.
384 IN SEARCH OF THE LIMES TO SU-CHOU [Chap. XI
tower, manifestly old, T. XLI. r, built of bricks measuring 14 x 9 x 7 inches. Here a Wu-chu coin of the large type was picked up on the surface. Close to the tower, the walls of a small guard-room, only 6 feet square inside, still stood to a height of about 4 feet. At the entrance the thick wall still showed the socket meant to hold a heavy bar.13 A small mound which we had passed about half-way, T. XLI. q, consisted of decayed masonry with bricks set on end. It probably marked the remains of a later tower erected where a post of the ancient Limes had once stood.
From the vicinity of this mound onwards soft salt-encrusted ground, evidently liable to be flooded at times, extended to the south. Skirting this we reached, after covering another mile and a half, the western end of the low rocky ridge which bears the Limes towers, T. XLII. a—d, above the small village of Shih-êrh-tun. A rapid reconnaissance made to this ground from Yü-mên-hsien on September 21, 1907, had enabled me, as already recorded in Serindia,14 to locate here definitely the line of the Limes. In view of the geographical interest which, as we shall see presently, this ground offers, I was doubly glad to make a day's halt by the lively little stream which passes north of the quiet tree-girt farms of Shih-êrh-tun, and to use it for the closer examination of the remains of the Limes and for survey work around it.
These remains extend along the outermost of a succession of low narrow ridges, which mark here the last offshoot of the southernmost Pei-shan range overlooking the Su-lo-ho valley at its westward bend. These dark ridges, all uniformly bare of the slightest vegetation, are composed of much-fissured rock, which to me looked like granite, but for the most part are thickly covered with stony detritus. Where the outermost ridge sinks away to the salt-encrusted level plain extending westwards, I had in 1907 come upon a definite trace of the agger, marked by small half-petrified branches of tamarisks and Toghraks strewing the slope. From this point the line of the agger, growing gradually more defined with twigs and branches embedded between layers of detritus, brings us, as the plan in PI. 15 shows, at a distance of about 30o yards, to the watch-tower T. XLII. a. It stands on a small rocky hillock about a furlong south of the rampart and at a height of about 5o feet above the foot of the ridge. It commands, like the towers T. XLII. b—d eastwards, a complete view of the line to be guarded in front as well as of the flat scrub-covered ground behind. The tower is built of stamped clay with thin layers of reeds dividing successive layers. It had suffered much decay, and a fissure due to wind-erosion cuts through the portion still standing, which reaches a height of about Io feet.
For three-quarters of a mile the agger runs on towards another small rocky eminence, where a completely decayed clay mound about 12 feet high is all that remains of the tower T. XLII. b. Plentiful fragments of Han pottery indicate that it was occupied contemporaneously with the wall. From some refuse close to the mound we recovered a roughly carved piece of Toghrak wood, perhaps the lintel of a door, and the fragment of a Wu-chu coin. At a point of the crest of the ridge between the two watch-posts just named, two or three heaps of reed-straw permeated with sand looked as if they had been deposited, like the stacks of fascines found at watch-stations of the westernmost Tun-huang Limes, for the purpose of being lit as fire signals.'5
From T. XLII. b, which lies immediately behind the agger, the line of the latter can be traced quite clearly for a mile, on the stony ground of a little plateau that extends by the side of the ridge. Then the line is lost eastwards in a belt of scrub-covered sandy soil. Layers of brushwood are everywhere exposed along the sides of the agger or can be quickly found by digging. The rampart, nowhere more than 4 or 5 feet in height, has a width of about 14 feet on the top. In many places it has a curious depression along the centre line, about 6 feet wide, recalling a similar feature which
13 Cf. Serindia, ii. pp. 645, 658. 15 See ibid., ii. pp. 650, 677 sq., 71I, 7$4 ; above, p. 344 ;
14 Cf. ibid., ii. p. 1137. below, pp. 389, 395.