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. i] THE EXTENSION OF THE ' GREAT WALL' BEYOND TUN-HUANG 729

meant to protect the great route of commercial and political expansion westwards. It was a step of the kind which in the case of the Roman Limes lines is appropriately described by the term cash-a or praesidia j5yomovere.24 I have already explained in full detail the topographical reasons which caused the Emperor Wu-ti's military engineers to carry the line of their wall to T. iv, on the very edge of the marsh-filled terminal basin of the Su-lo Ho, and to make it finally end there.25 At no other point could they have found a better flanking defence provided by nature itself for their Limes. In this termination of the wall, just as in the clever use made of the line of the Su-lo Ho Iakes and marshes for a ' wet border ',26 they displayed the same clear eye for topography which we have so often had occasion to recogrfize in the old Chinese leaders and organizers. Though among the documents found at the isolated watch-stations T. v—T. vi. d along the eastern edge of the terminal Su-lo Ho basin none date back further than 68 B. C. (Doc. No. 255), yet there is reason to believe that this chain of posts watching the south-west flank was established about the time at which the wall was carried to T. iv. a.27

The same undoubtedly holds good of the short line of watch-towers pushed out beyond the end of the wall and represented by the ruins of T. i and T. ii. It was plainly intended to assure additional security for the exposed western end of the Limes proper. These small advanced posts made it easier to watch the main route coming from the west, the only one really practicable for serious inroads, and to send on warning signals, etc. M. Chavannes has justly pointed out the exact analogy presented by the system of fortified outposts which the Romans employed in their African provinces, notably on the Tripolitan border, where the routes leading through the desert towards the oases of the coast belt required to be guarded. There too, as M. Cagnat's very instructive exposition of the Tripolitan Limes clearly shows, this system served a policy of expansion beyond the actually protected area.23 The same observation applies also to the numerous lines of advanced posts traceable beyond the other desert Limes of the Roman Empire, that of the Province of Arabia.26

We have seen above that the historical record relating to the year z o i B.C. distinctly mentions the establishment of military posts (ring 4) ' from place to place from Tun-huang westwards to

the Salt Marsh (yen-1se   ).' That by the latter term Lop-nor or the dried-up ancient Lop sea
is meant may, on M. Chavannes' authority, be accepted as certain.$0 Yet I must point out that I could not trace any ruins of watch-towers or other remains of structures going back to Han times along the ancient route to the west of T. ii. However, my explorations of 1914 enabled me to trace the ancient route itself to and across the dried-up salt sea of Lop, as briefly mentioned above,3' and the observations then gathered along it about the physical conditions prevailing on this

Limes ends at terminal basin of Su-lo Ho.

Watch-posts advanced beyond Limes wall.

Military posts established to Salt Marsh'.

24 Cf. Kornemann, K&o, 1907, vii. p. 77 : ' Die neuen Einfallslinien wurden durch Kastelle von Auxiliartruppen . . . gesichert, an der Kopfstation durch ein besonders umfangreiches Kastell. . . . Jede neue Eroberung brachte eine Verlängerung des Limes und eine Vorschiebung der Kastelle : castella oder praesidia promovere.'

25 See above, pp. 632 sq.

26 Cf. above, pp. 663 sq.

47 See above, p. 647.

26 Cf. R. Cagnat, La frontière militaire de la Tripolitaine a l'époque romaine, in Ai/moires de I'Académie des Inscriptions, xxxix. pp. loo sqq. (reprint pp. 28 sqq.).

29 Cf. Kornemann, Die neueste Limesforschung, Klio, vii. pp. 112 sq. I may note here that the detailed surveys of the Arabian Limes recorded in Brtinnow and v. Domaszewski's

great work, Die Provincia Arabia, offer an abundance of interesting material for the comparison of its burgi, or watchtowers, fortified camps, etc., with the remains of the ancient Limes of Tun-huang.

6° See Documents, p. vi ; cf. also Wylie, J: Anthrop. Inst., x. p. 22 ; Kingsmill, J.R.A.S., 1882, p. 29. It deserves to be noted that another designation of the Lop-nor marshes

P'u-chang %lfi   , is also to be found in the Former
Han Annals; cf. Chavannes, T`oung pao, 1905, pp. 532, 570. [If the term Yen-tsê were not used in other passages also for Lop-nor, the suggestion might be hazarded that the terminal marsh basin of the Su-lo Ho may be meant by it in the passage quoted in the text.]

31 Cf. above, pp. 341 sq., 553 sq. ; also Third Journey of Exploration, Geogr. Journal, xlviii. pp. 227 sqq.

5A

r-,.   - . ,.   .   •----_.-   ---.. - _.._ ,   - .