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

Records
found at
T. xmv. a.

Discovery of station T. xv. a.

Transverse wall aligned upon

T. xv. a.

698   THE JADE GATE BARRIER   [Chap. XIX

Tung frontier district '. The locality named here still awaits identification. The preservation of this clay sealing is obviously due to the burning of the wooden seal-case which once had contained it. The remains of the wall near which it was discovered also bore marks of having been exposed to fire. Among the objects recovered from the refuse close to the quarters, a well-preserved string

shoe, T. xtv. a. oo1, and a woven string sandal, 002 (Plate LIV), of special make may   mentioned.

One of the two wooden slips found in the room T. xiv. a. i, Doc., No. 390 (Plate XII), bears evidence of the date A. D. 87. It furnishes a receipt for a certain payment made to the commander of a post belonging to the western section of Ping Wang, and thus seems to indicate that this watchtower T. xiv. a was included in Ping-wang. I have already had occasion to note that, judging from the position of the several stations where documents mentioning this local name were found, it appears to have been applied to that part of the Limes which extended from T. xxii. c on the shore of Khara-nor to the neighbourhood of T. xiv. a, the latter probably marking its westernmost point.' Of the wooden records found in the watch-place on the top of the tower, Doc. Nos. J43-6, 348, 349, the first preserves a mention of Yin-mên, while among the rest are fragments of private correspondence apparently exchanged between exiles posted on this border.2

Beyond the winding marsh-basin to the east of T. xiv. a the remains of the wall were at first difficult to trace. It ran here for about 1-2i- miles across scrub-covered low ground broken by numerous small clay terraces, of the usual lacustrine origin, and in most places was badly decayed. The alignment of the wall was, however, here clearly marked by the towers T. xiv. a and T. xv at its ends, both easy to sight owing to the high ground they occupied. No ruins of a tower intermediate between the marsh and T. xv could be traced, and, but for the keen eyes and intelligence of my Yârkandi follower. Tila Bai, the scanty surface remains of an important station, T. xv. a, once established near by, might easily have escaped me. They proved to be extensive refuse-heaps that covered the gravel slopes of a narrow terrace situated about a quarter of a mile to the south of the line of the wall and close to where it abutted on the spring-fed marsh north-east of T. xiv. a. The terrace, stretching for about 200 yards from east to west (see Plate 40), showed no structural remains whatsoever. But at the south-west end of its flat top (marked i in plan) there was seen a slight swelling, only about 2 feet high and about 40 feet across. On excavation it was found to consist of refuse and completely decayed brickwork, hiding the remnant of a wall about 5 feet long and 2 feet 8 inches thick. The size of the sun-dried bricks was 14" x 7" x 4", approximately the same as that usual in the towers of the Limes.

Before describing the abundant finds which two clays' clearing brought to light here and from the extensive rubbish layers discovered lower down, especially on the slopes to the west, I may call attention to the fact that the transverse wall discussed in the preceding sections, coming from T. xiv, is aligned upon the ruined station T. xv. a marked by these refuse accumulations. The point which it reaches on the south-west shore of the marsh was found to lie about a mile due south of that occupied by T. xv. a on the north-east shore. But, owing to the depth of open water and reedy marsh intervening, a detour of nearly 3 miles had to be made in order to get from one point to the other. This alignment of the transverse wall creates the presumption that, at the later period when it was built, there was still an intention of retaining the position of T. xv. a within the guarded line. We shall see that the prolonged period of occupation proved for T. xv. a by its dated records is in full agreement with this.

But even without such documentary evidence it would have been easy to realize that the Limes

' Cf. above, p. 691 ; also Chavannes, Documents, Nos.   been classed, Documents, pp. 78 sqq., with finds from

271, 274, 275, 313, 314, 377, 484, 592, 66z, 693.   T. XIS'.

2 The documents above named have by an oversight