366 TO TUN-HUANG AND AN-HSI [Chap. X
But when I had a cutting made through the agger the soft reddish soil, quite different from the yellow clay of the natural ground, plainly suggested the presence of rotten or possibly burnt vegetable matter. It was obvious that the construction of the border wall was here different from, and inferior to, that observed along the Limes line farther west.
A little to the east of where we had struck the line, a small semicircular northward bend of it marked the position of a watch-tower, T. xxxvii. a. It had decayed into a little mound of earth about 6 feet high and covered with salt-efflorescence. A hasty search disclosed no ancient refuse near it. But examination of the agger near by furnished interesting confirmation of a surface feature which I had observed on previous occasions, but not quite so clearly. Low scrub was found growing with comparative abundance on the slopes of the mound, whereas its top was léft quite bare. Accordingly, standing on the mound itself, I could make out its direction with ease for a considerable distance by the double and parallel line of vegetation thus produced. Exactly corresponding lines of vegetation had been noticed by me, and Afrdz-gul also, at two different points of the Besh-toghrak valley.10 There the lines had suggested to me the embankments, perhaps, of an ancient canal, and their reappearance here, under conditions excluding all possibility of optical illusion, helped to strengthen my previous conclusion.
No signs of ruined watch-stations could be discovered along the line stretching westwards. I therefore decided to move in the opposite direction, where the line of the agger could be seen to rise gradually. Half a mile farther on it attained a height of about nine feet and disclosed on its surface characteristic layers of brushwood. These were about 3 inches thick and alternated with strata of earth, about 7 inches in thickness. The fact that the reinforcing material consisted not of tightly bound fascines, such as was used for the Limes wall west and north of Tun-huang, but only of thin brushwood loosely laid, was a clear proof of inferior construction and also accounted for the greater width which, it was evident, had originally been given to the agger. The shôr which might have served as a useful binding material appeared to be absent from both the soil and the brushwood that had here been locally collected for the agger ; and this may have accelerated the decay. At one point of this section I found the agger still rising to a height of 12 feet, with five alternate layers of brushwood and earth still clearly distinguishable near the top. The brushwood must have been exposed already in ancient times, for over a distance of about 120 yards I found the edges of its layers charred. As this burning was to be seen in at least three other places, it may be assumed with some probability to mark an attempt to destroy the Limes by fire.
At the distance of about a mile from T. xxxvrr. a, a large refuse heap (marked b) was found near the southern foot of the agger and may indicate the position of a completely decayed post. From the masses of reed-straw, dung, & c., a badly perished wooden slip was extracted of the size usually adopted for ancient Chinese records ; also the pointed wooden stick, T. xxxvir. b. oi, of uncertain use. Half a mile or less farther on there rose by the agger the ruin of a watch-tower built in stamped clay, T. xxxvii. c.1' Most of the northern and eastern faces had fallen, but on the west the original measurement of the square base, 20 feet, could still be ascertained. The extant height was about 14 feet. An extensive refuse heap to the south-east, covering about 3o by 22 yards, could not be thoroughly cleared by the few men who accompanied me. So far as examined, it yielded, apart from stable refuse and chips of wood, only abundance of hard potsherds, mostly mat-marked ; specimens of these are noted in the List below.11a
The line of the agger eastward still rose to a height of 6 to 8 feet and comprised layers of brushwood which in places showed marks of having been fired. We followed it for two miles
10 Cf. above, p. 323.
11 In Map No. 38. D. 4 the symbol of this tower has been shown too far to the east. lla See Chap. XII. sec. iii.