behind flower. Row of bead orn. by each long edge, and projecting bars at back across each end, as also in the preceding. o8 broken at edges, 09 perhaps complete. Cf. 06-7. Size complete i II" x 9a", thickness I". PI. L.
Chiao. oio. Fr. of clay tile or hollow brick, moulded
in relief like the preceding. One end only, broken in two ; shows part of curving stem, leaves, and narrow-petalled flower. Raised moulding along each long edge, and return piece of clay along end. 6" (incomplete) x 71" x (return end) z".
SECTION III.—HUA-HAI-TZÛ AND ITS LIMES REMAINS
Having already given an account of the physical features of the ground over which our march of April 24th east of Shih-êrh-tun took us, I may now describe the remains of the Limes traced on that march. Within less than half a mile to the east of T. XLII. d the line of the wall was lost on low ground covered with abundant scrub and reeds. We followed its direction towards T. XLII. e, a tower well in sight from Shih-êrh-tun, and came again upon the wall extending in a straight line marked by reed fascines cropping out on its sides, as soon as we struck gravel soil at the foot of a low ridge of detritus. Near the tower T. XLII. e, which stood at a distance of about two miles, the agger still rose to a height of about 6 feet. From here onwards it remained clearly traceable for close on twelve miles eastwards, except for short breaks where the foot of the gravel Sai gave way to clayey patches with vegetation. For the whole of this distance, apparently, the wall had been built of alternate layers of reed fascines and stamped clay, after the fashion observed along the Tun-huang Limes. It may safely be concluded from this use of reed fascines that the depression which stretches to the south of the line and is drained by the Hsi-wan-ho presented much the same marshy appearance in Han times as at present.
The ruined watch-stations found along this stretch of the Limes wall, T. XLII. e–j, showed much similarity in construction and had evidently been occupied and repaired down to later times. This latter circumstance is easily accounted for by the fact that the route leading to Hua-hai-tzû and Su-chou here passes parallel to the line and within a mile or so to the south of it. The ancient towers could, it is evident, be conveniently employed for the protection of this route and of the scattered settlements to the south, and at one time they all seem to have been enlarged into small shelter-posts on much the same plan.
The towers are built of stamped clay and measure to-day from 22 to 28 feet square at the base. Their height varies between 18 and 25 feet, including brick parapets which manifestly are of later date. The towers now occupy the north-western or north-eastern corner of walled enclosures, as seen in Fig. 214 representing T. XLII. f. These enclosures, also built of stamped clay but of less solid construction, have the appearance of later additions and form squares of 6o to 62 feet inside the walls.' These walls, 3/ to 4 feet in thickness, have been breached by wind-erosion usually on the west face, an observation which suggests the prevalence here of winds blowing from the Su-lo-ho valley down into the depression of Hua-hai-tzû.
No large accumulation of refuse was found within any of the enclosures. But the presence of porcelain fragments by the side of potsherds of Han type sufficed to prove that the posts had been occupied down to later times. Whether the ornamented shoe with string sole, T. XLII. e. of (Pl. XLVI), is of any antiquity is doubtful. On the other hand, we may recognize a relic of the period when the Han Limes was regularly guarded along its whole length, in two stacks of half-petrified reed fascines, 4 to 5 feet high, which I discovered about 3o yards to the south-east of T. xLII. i on a slight swelling of detritus-covered ground. These stacks, like those found near watch-towers of the Limes west of Tun-huang,2 served the purpose of providing material for signal fires ready
I For the plans of T. xLII. e, f, i, see PI. 14, 16. 2 Cf. Serindia, ii. p. 754 ; above, p. 384.