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

Extension
of Nan-hu
cultivation.

March to edge of tamarisk belt.

Old cart- tracks.

628   THE OASIS OF NAN-HU AND THE YANG BARRIER [Chap. XVI

also, in all probability, were both Upper and Lower Yen-chia. That the abandonment of Kuan-tsou took place just about the time when cultivation in the Domoko area was shifted from Old Domoko to the present main village is certainly curious. But the great distance separating the tracts of Nan-hu and Domoko, and still more the detailed observations recorded above as regards the peculiar conditions affecting the shifts of cultivation in the latter tract,4 must warn us against any hasty conclusions which might be drawn from the synchronism of these local changes.

No traces of cultivation were met with on the bare clay steppe beyond, but a well-marked shallow depression, with living tamarisks and signs of water-erosion of somewhat recent appearance, continued north-westwards and suggested that at rare intervals floods from the mountains, coming down the dry bed of the river of Nan-hu, may penetrate here some way into the desert. At a distance of about two miles from the last farm a much-decayed watch-tower (shown as T. xviii. b in Plate 33) was reached, which rose as a conspicuous object on a slight swelling of the barren plain and seemed to be known to some of the Nan-hu people by a name like Wa-shih-lun.b Measuring about twenty-three feet at the base and still rising to a height of nineteen feet, it was built with solid layers of stamped clay, about two inches thick, just like several towers that I had noted along the ancient Limes. I regret now that I did not examine it more carefully and ascertain whether there were also the usual thin layers of brushwood inserted at regular intervals after a succession of courses. Far away to the north another ruined tower was visible.°

At the time it seemed difficult to account for the position and purpose of these towers. But subsequent discoveries' have suggested the explanation that they may have been connected with the subsidiary Limes which I found running south-south-east from the ruined fort T. xiv, marking the position of the ancient Jade Gate, and which was, no doubt, intended to safeguard the important line of communication between this and the Yang kuan station, i.e. Nan-hu. It is also only in the light of these later discoveries that I could realize fully how helpful it must have been, for those who had to guard the westernmost extension of the Limes in the desert, that the terminal point of the area capable of cultivation from the side of the Yang barrier—assuming that point to have lain near the extreme northern edge of Kuan-tsou—approached within about twenty-two miles of the Jade Gate. This fact must have greatly facilitated the dispatch of supplies and of reliefs to the outlying watch-stations of the Limes.

My original intention had been to move across the desert north-north-west towards the ruined fort T. xiv, which, as I already surmised, might prove to occupy the site of the Jade Gate. But Chiang-huan, our soi-disant guide, on the look-out for an alleged well which he called ` Lao-tsaoching-tzû', took us steadily to the north-west until, after a march of about thirteen miles from the tower T. xvIII. a, we struck the southern edge of a broad belt of tamarisk-cones and scrub. Beyond the tower the bare clay on the surface of the ground had given way, first to patches of gravel and then to continuous stretches of ` Sai '. I was surprised at the time to notice numerous old cart-tracks, rather faint, leading to the north and north-west, and wondered whether they went back to the days when Kuan-tsou was still occupied and its inhabitants were likely to have resorted to the riverine jungle belt for timber and fuel. It was only in the course of my subsequent explorations along the Limes that I became aware how well the absolutely bare gravel soil of this desert can retain such traces of human passage for centuries—and even of tracks followed when there were still patrols and others moving along the Limes wall.

Watch- towers north of Kuan- lsou.

4 Cf. above, pp. 202 sqq.

6 Tun tit seems to be the term generally employed by the people of Tun-huang and the regions further east for all watch-towers, ancient and modern.

6 The position of this tower, T. xvur. a, appears shifted by a slight error to north-north-east in Map No. 78. A. 4 and hence also in Pl. 33.

4 See below, chap. xxx. sec. iii.