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

Sec. ii]   THE TUN-HUANG LIMES SINCE ITS CONSTRUCTION   735

and also with the regulation length, 7 feet, of the fascines used in its construction.28 The statement of width on the top does not lend itself to such an exact test, as it necessarily varies now with the different state of preservation in the surviving sections of the wall. Judging from the height indicated, only 8 Chinese feet or 6 British, the wall, where that old Chinese antiquary measured it, must have been already badly decayed, and taking into account the extent of abrasion through the action of the wind which I have often observed, the measurement of only 4 Chinese feet may have been perfectly correct as far as the actual width went. Originally, I have reason to believe, it could nowhere have been much less than 6 feet as measured by me east of T. xxxv, while the height was found, even in the present ruined state of the wall, to rise in places to above io feet.

Turning to the measurements of distance which the Sha chou chih records for the ancient wall, it is easy to account for the i 8o li given as its extent eastwards. Reckoned from a point on the line of the Limes where it runs due north of the site of old Sha-chou, this takes us approximately to beyond T. xxxv (Map No. 81. B. 3), where in 1907 I traced the easternmost remains of the wall towards An-hsi. When resuming my exploration of the Limes in 1914 from this point, I found the wall to the east almost completely effaced for a considerable distance, evidently owing to the prevalence of abrading drift-sand in this area. The distance from a point due north of Sha-chou to beyond T. xxxv, where, as I believe, the ` Chick-Ping Signal-station ' may be placed, is approximately 35 miles in a straight line, which agrees very closely with the 18o li mentioned in the text. From the same point the distance of 212 li measured in the opposite direction to the west, if converted into miles at the same approximate value of 5 li to I mile,2$a would bring us near the lake which is overlooked by the conspicuous watch-post T. xx and into which the Su-lo Ho expands after leaving the Khara-nôr (Map No. 78. A. 3). The configuration of this lake and of the wide lagoons adjoining it north of T. xvIII might well account for the designation of the Ch`ü-tsê, or ` Winding Lake', signal-station mentioned by the text, The referénce to the wall ` running out into the desert due west in the direction of the territory of Shih-ch`êng' shows plainly that the local informant was aware of the westernmost extension of the Limes and of the route to Charkhlik which passes along and beyond it. We have seen above that S/zih-ch`eng, or the ` Stone town ', was the name given in Tang times to the site within the present Charkhlik oasis.28

With this late but strikingly accurate local notice we have reached the latest of the Chinese records concerning the Tun-huang Limes and the Jade Gate which I am able to trace at present.3o In late mediaeval times, when China followed once again a policy of strict seclusion towards Central Asia and the West, a ` barrier' maintained much further east took the place of the Jade Gate. But the discussion of it belongs to a subsequent chapter.3'

Distances along Limes noted by Sha chou chih.

Latest Chinese records of Tun-huang Limes.

' To these 7 feet must be added the average thickness, apparently 4-5 inches, of the revetment fascines longitudinally ,fixed on either side of the alternate layers of fascines and stamped clay; cf. above, pp. 568, 570, 606, 678 ; below, p. 736. .8a For other evidence supporting this value of about one-fifth of a mile for the li in Central Asia, see, e.g., above, pp. 559, 649, 716 ; below, chap. xxviii. sec. i, etc.

R° See above, pp. 306, 320.

° Here I may conveniently note the curious fact that in the Chinese map engraved on stone in A.D. 1137, but probably drawn about a century earlier, which M. Chavannes has edited (B.É.F.E.O., 1903, pp. 214 sqq., carte A), the

ancient Wall of Han Wu-ti is still indicated in an approximately correct position. Its western end is marked with the name Yii-min kuan, some distance to the west of Kua[-chou], i.e. An-hsi. Beyond An-hsi the wall is shown as running to the north-east, crossing a great lake or marsh and ending near another which receives a river coming from the south-west. I suspect that by the latter is meant the Su-chou R. branch of the Etsin-gol, to which I traced the Limes in 1914. I regret that I am unable to follow up this interesting cartographical record further at present.

n See below, chap. xxvii. sec. ii, on the ' Gate' of Chiakuan.