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

694   THE JADE GATE BARRIER   [Chap. XIX

distinctly seem to support it. If the cross wall is of later date, as we have shown reason to believe, it may well have been constructed with the purpose of reducing the length of the border-line that had to be guarded across desert ground. A glance at the map is sufficient to demonstrate the advantage of such a reduction.2

Limes   Now it is certainly important to note that, while east of the transverse line the dated records
retrenched brought to light on the Limes prove occupation of its watch-stations from the beginning of the first

cent. A. first century B.C. down to A.D. 137, the still more numerous datable records found west of it, with two

cent. A. D.   ry   3 7>   >

apparent exceptions, stop short with the time of Wang Mang, and in the case of the outlying stations on the south-west flank, T. iv. b-vi. d, with dates considerably earlier.3 This fact seems strongly to favour the presumption that a gradual retrenchment of the border-line lying westwards of T. my may have set in during the troubled times of Wang Mang's usurpation or soon after. The abandonment of the outlying westernmost portion of the Limes was bound to effect a considerable reduction in the difficulties about victualling, garrisoning, etc., which must always have been felt most in the case of posts pushed out into the desert far away from the inhabited bases. If due attention is paid to the topographical factors, it appears probable that such a retrenchment would necessarily have commenced from the side of the outlying line of watch-stations on the south-western flank, away from the Lou-lan route. The posts along this route may have still been occupied during a portion at least of Wang Mang's reign, as proved by T. viii. ii. 2, Doc., No. 585, dated A. D. 8, and No. 586. In the case of T. xii, xi'. a, which, as has been shown above, formed very convenient advanced watch-posts for the Jade Gate ',4 occupation is likely to have been continued even somewhat later.

As soon as the line of watch-stations guarding the south-western flank was abandoned, the need must have made itself strongly felt for another line, also drawn across the desert but further back, which would render the crossing of the administrative frontier' of the empire—to use the term of Anglo-Indian official language—impossible, whether for marauding bands or for unauthorized persons in general. It is at that time and in the circumstances just mentioned that I think the construction of the transverse wall from T. xiv to Nan-hu or Yang kuan would best be accounted for. Without it, access to Tun-huang would have lain open for any party which might have avoided the stations still guarded along the Lou-lan route by moving round the terminal marshes of the

Su-lo Ho, or have crossed the route beyond the last station still occupied. But what was of even

more immediate moment is the obvious fact that without such protection the important line of

Transverse wall needed after retrenchment.

2 The distance from the northern end of the tranverse wall to the northernmost point of modern (though now abandoned) cultivation in the Nan-hu area is about 24 miles by the map. From the same point the distance along the line of watch-stations as far as T. vi. d is more than twice as long.

3 The total number of exactly dated documents, or of those for which an approximate chronological limit can be fixed, is thirty-two in the case of the stations from T. xiv to T. xxix. c. The earliest among them is T. xxix. C, Doc., No. 271, of 98 s.c.; the latest, T. xv. a. i. 6, Doc., No. 536, dated A. D. 137.

Taking the stations from T. vr. c to T. xIII, we have a total of forty-four datable documents, counting Doc. Nos. 9-24 and 25-35, which contain portions of calendars for the years 63 and 59 s.c. respectively, as one document in each case. The earliest in the series is T. iv. b. ii. 1, Doc., No. 430, of 94 B.c., or, if another possible reading were adopted,

T. V1. b. ii. 6, No. 255, of 68 B.c.; the latest, T. xII. a. 3, No. 592, dated A.B. 21.

Of the two apparent exceptions referred to in the text above, T. xiv. a. i. 1, No. 390, bearing the date of A.D. 87, comes from a watch-tower which closely adjoins and overlooks T. xv. a. Though slightly west of the transverse line, it certainly had to be guarded as long as the latter important station was occupied ; see below, pp. 697, 708 sq. In the case of T. xi. ii. 6, No. 68o, we have the fragment of a calendar the date of which, A.D. 153, has been inferentially calculated by M. Chavannes on the conjectural assumption that another slip, T. xi. ii. 8, No. 8, found in the same place, belongs to a supplement to a lexicographical text composed in 48-33 B.C. For our purposes it is enough to remember that T. xi was a roadside station which is proved by archaeological evidence to have been used as a halting-place by passing travellers down to Sung times, if not later; see above, p. 667.

4 Cf. above, pp. 670, 679 sq.