国立情報学研究所 - ディジタル・シルクロード・プロジェクト
『東洋文庫所蔵』貴重書デジタルアーカイブ

Tower on
secondary
wall line.

Wall connecting Yü-mên and Yang barriers.

Ancient track visible.

692   THE JADE GATE BARRIER   [Chap. XIX

throughout. Where this wall approaches the depression north of T. xIv I found a small mound containing the foundations of what may be a completely decayed tower, built of bricks 14" x 7" x 5", with the usual layers of reeds. No refuse or other sign of prolonged occupation could be traced near these scanty remains, which lay to the east of the wall at a distance of about 8o yards. The wall then drops down to scrub-covered ground about 4o feet below the top of the plateau and continues as a low straight mound in the direction of the débris-covered hillock of T. xtv, and to within about a quarter of a mile from it. The wall must have skirted the hillock by the north and west foot ; for, as Plate 40 shows, its line reappears as an earth mound, only 3 to 4 feet high, yet easily distinguished by its straightness, immediately to the west of the fort, and thence runs in the direction of S. 16o° E. across the sandy reed-covered ground close by.14

Looking from the height of the wall of the fort, I could assure myself in i 907 that this wall had its continuation to the south-south-east also beyond the point where it struck the bare gravel ` Sai' to the south. But it was impossible for me then to find time for surveying the line further. On my passing T. xiv again in March, 1914, I could, however, make a reconnaissance to the south, and, though owing to practical difficulties about supplies this had to be short and rapid, actually survey the secondary wall for some distance beyond. I found then that, after emerging from the depression at about a mile from the fort, the wall showed its reed fascines quite clearly again on the bare gravel surface. As the revetting longitudinal fascines were preserved here, I could easily verify my previous observation that the width of the wall was only about 5 feet at its foot. Badly effaced in some parts, in others still up to 3 feet high, the wall continued unbroken for 31 miles to a place where I found the remains of a ruined watch-tower not previously sighted from T. xiv.

It stood about 7o yards to the east of the line of the wall, and was constructed of brickwork similar to that of most of the Limes towers, but decidedly rougher. At intervals of i 8 to 24 inches layers of reeds intervened. The base of the tower measured about i 7 feet square, and its actual height was about 15 feet. A rapid search of a small mound of débris at the south foot of the tower disclosed only fragments of torn fabrics and little pieces of cut wood. For proper clearing we had, unfortunately, neither time nor implements. But what was important to observe is that the wall continued beyond with the same bearing S. i 50 E. which the tower showed in relation to T. xiv. This direction would take it, as a reference to the map in Plate 33 shows, exactly to the ruined tower T. xvin. b, below the northern end of the abandoned cultivation of Nan-hu.

Thus the fact that this secondary wall served to connect the site of T. xiv or, as I take it, of Yü-mên with the Nan-hu oasis containing the site of the ancient ` Yang barrier ' may now be considered as certain. For two miles beyond the tower just described the line of the wall was clearly visible on the gravel ; further on it disappeared in the belt of scrub and Toghraks I had crossed on my march from Nan-hu. To trace it there and beyond towards Nan-hu is a task that I must leave to my hoped-for successor on the Chinese Limes. I may add, as an interesting observation, that along most of the surveyed length of the secondary wall, where it runs over gravel soil, the ancient track previously noticed elsewhere was still clearly visible. It kept always at

Reconnaissance of secondary wall.

" In Map No. 74. D. 3 the fort T. xiv is wrongly shown as lying to the west of the secondary wall, owing to a draughtsman's error already noted above, p. 683, note z. The line of the wall, too, is in need of correction, being erroneously marked there as running south from the fort, whereas the true general bearing is to the south-south-east. These mistakes have been only partially rectified in Pl. 33.

Neither at the time when this map was being prepared in 1913, nor when I was revising the proof of Map No. 74 two }tears earlier, was it possible for me to find leisure for checking such details by careful reference to the records of my diary about the site. Still less was it possible to include in eithermap the result of my supplementary reconnaissance survey made in 1914 to the south.