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. iii]   MAIN FEATURES OF THE REMAINS OF THE LIMES   737

mention that, whereas on the section explored to the north-east of Tun-huang there were watch-

towers (T. xxxll-xxxv) to be found within three-quarters of a mile of each other, the distances on

the terminal western stretch of the Limes were generally greater, and in one case as much as

4 miles in length.' On the south-western flank of the Limes, which was well protected by the

great natural defence of extensive marshes, and therefore left without a line of wall, we find, on the

average, intervals of about 5 miles between the watch-stations.° The high commanding positions

on which they were placed here made it easy to secure visibility for signals. It is for the same

reason that also along the proper Limes wall we find high isolated clay ridges, or Mesas, invariably

selected, where available, as the positions for watch-towers.° In a number of instances where

conditions were particularly favourable the building of a proper tower could thus be saved, a mere

guard-room on the top of a high Mesa base equally serving its purpose.10

The watch-towers were always built solid and square, tapering towards the top ; but they Construe-

varied considerably in size, height, and materials used for construction. We find bases from 16 to oif watchures

24 feet square." Owing to the broken condition of the top it is often impossible to determine the towers.

original height. But by the side of towers still reaching 3o feet or so in their actual state (T. v, vi. d)

we have others where the original elevation, as marked by the floor of the guard-room, was not

more than 8-13 feet.12 No doubt, the height adopted for individual towers was largely determined

by the elevation of the place they occupied and by the range of vision which was aimed at for

watching or signalling. The choice of the materials used in construction similarly depended on

local circumstances. In most cases where bricks, always sun-dried, were employed, we find that Bricks used

water was not available on the spot ; over distances or to heights the transport of bricks was, no in masonry.

doubt, easier than that of water in quantities such as would have been needed for the puddling of

stamped clay.13 The size of the bricks shows little variation, a circumstance pointing to approxi-

mately contemporary construction of the towers.'4 The use of stamped clay, usually in layers 3 to

4 inches thick, may safely be taken as an indication that water was at the time of construction

obtainable at no great distance.16 The same observation, with a modification, applies also to

another local material, clods of hard salt-impregnated clay, of which T. x (Fig. 174) furnishes

a striking example.1° Soil producing such clods would ordinarily hold only brackish, undrinkable

water. Whatever the building material used was, we find it always strengthened by the insertion Reed layers

bween

of those layers of reeds, usually after three courses of bricks, stamped clay, or clods, which I have masonry.

come to look for in this region in all structures of Chinese origin belonging to Han times or those immediately following. In addition, the masonry was reinforced by the insertion within the towers

7 See T. XXII. c-d in map PI. 33. In this case the two watch-towers thus separated occupied opposite shores of a southern bay of Lake Khara-nôr. Thus the intervening distance was defended by nature, and guarding and visibility for signals easily assured. Where the foreground was coups and for other reasons also needed a specially careful watching, as between T. xvii-xvii, a, or T. xix-xx, we find the interval reduced to one mile only.

' See above, p. 641.

9 Cf. above, p. 634 (T. iv. a, b), pp. 572, 662 (T. ix, x), p. 666 (T. xi), p. 697 (T. xiv. a), p. 711 (T. xv), p. 712 (T. xvii. a), pp. 717 sq. (T. xix, xx), pp. 718 sq. (T. xxiXXII. a-c), p. 72o (T. xxIii).

10 See above, pp. 712,717 sq. (T. xvii. a, )(ix, xxx).

31 E.g. in T. I, XXIII, xxxiv, the bases are 16' square; T. VIII, xi, xIi. a, XIII, xvi have bases 23-24' square.

1974

12 See pp. 711,721 (T. xvi, XXIII. a).

1s This consideration probably accounts for the use of bricks at T. Iv. b, vI. a-c, Vin, Ix, Ix. a, xIII, inv. a, xv-Xvii, XIX—XXI, XXII. b, XXIII. a, xxvii. At T. XII, XII. a, xv. a, where also bricks were used, water was, however, not far off.

" The prevalent size is 14-15 by 7-8 inches, with a thickness of 4-5 inches; see T. iv. b, vi. a-c, viii, xix, xiv. a, xv. a, Xv-xvii, • XXIII. a. A slightly bigger size, 1718' by 8-9', with a thickness of 44-5', is found at T. Ix. a, xi'. a, xxi, xxvii. It may be noted that the size of bricks, 12 by 6-7 inches, found in the ruined shrine near T. xxix (see above, p. 6o1) is a further proof of its late date.

16 Stamped clay is found used, e. g., at T. I, II, y, vi. d, xviii, XXII. C, XXVI, XXVIII-XXXIV.

36 Hard lumps of salt-impregnated clay are employed in

T. Iv. e, x, XI, xxiI, XXIII.

5B