Sec. ü) OLD REMAINS NEAR AN-HSI & HSÜAN-TSANGS YÜ-MÊN KUAN 1099
the above-mentioned old bed and the first halting-place, Pei-tan-tzû, with its spring, on the present caravan road.
As to the exact position of the Yü-mên barrier, as established at the time of Hsüan-tsang's departure, I am unable to state anything definite. Whether traceable remains of such a watch-station could survive in the immediate vicinity of a river-bed subject to changes is doubtful. At any rate my Limes explorations have solved the question as to the original position of this famous frontier station, once far away to the west of Tun-huang, and there is reason to believe that even in Hsüan-tsang's time its transfer to the north of Kua-chou could not have been of old date. A passage of the Tang Annals referring to the dispatch in A. D. 610 of the famous Chinese commissioner P`ei Chit to Yü-mên kuan distinctly places this frontier station at the town of
Chin-ch`ang g.19 Chinese antiquarians seem to agree in considering Chin-ch`ang as a sub-
prefecture dependent on Kua-chou and situated to the east of the present An-hsi.20 But its exact position still remains to be determined. In any case it is clear that the Yü-mên barrier in A. D. 610 was not where Hsüan-tsang found it twenty years later. In the interval it may have been advanced to the west, in conformity with the resumption of Chinese political activity in Central Asia, which commenced soon after the accession of the Tang dynasty (A.D. 618). How long it remained near Kua-chou, and when and how the present Yü-mên-hsien, between An-hsi and Su-chou, came by its name, derived from the ancient frontier station, are questions which must be left for future inquiry.
SECTION III.—THE RUINED SITE NEAR CHIAO-TZÛ
On June 24, 1907, I was able to leave desolate wind-swept An-hsi for the high mountains - south-eastwards. Before beginning there what was to be my geographical work of the summer, I arranged to visit en roule two sites in the outer hills of the Nan-shan which held out promise of archaeological interest. My immediate goal was a ruined town to the south of the village of Chiao-tzû about which a Turki Muhammadan trader at An-hsi had first given me information, necessarily of the vaguest character. The position of Ch`iao-tzû itself was quite uncertain, that part of the lower hills within the great bend of the Su-lo Ho not having been visited by any European traveller. For an account of the long and, owing to fairly heavy rainfall, somewhat trying march which brought me to Ch`iao-tzû, I may refer to my Personal Narrative.' This march took us across the outermost and absolutely barren range of the Nan-shan. It was on approaching its foot, south of the hamlet of Huang-ch`u-k`ou, that I first discovered that short but well-defined stretch of the ancient Limes which assured me of the extension of its wall-line beyond An-hsi, and to which I have already referred in the preceding section. As I could not examine it closely until my return journey from Su-chou to An-hsi, I leave its description for the next chapter.2
The village of Ch`iao-tzû proved to be the chief place of a small but very fertile oasis situated within a wide grassy plain which fills here the bottom of the broad valley dividing the two outermost
See Chavannes, Turcs occid., p. r8.
20 Cf. the references given by M. Chavannes, Turcs occid., p. 18, note 3 ; Dix inscriptions, p. 67, note 2 ; p. 93, note 8.
Learned local opinion, as communicated to me by An-hsi officials, placed Chin-ch'ang to the west of Yii-men-hsien (Map No. 85. A. 3). I regret that I did not ascertain the exact position meant, and still more that I did not follow up the point locally on my way from Yii-mên-hsien to An-hsi.
The Chien-fo-tung inscription of A.D. 894 speaks of Chin-ch'ang as a ' strategic point' (Chavannes, Dix inscriptions, p. 93) where a prefect of Kua-chou earned distinction
by his bravery. This has suggested to me that some place near the foot of the Wan-shan-tzü spur, where the high road from An-hsi to Su-chou passes a defile of the Su-lo Ho (Map No. 83. B. 2), might be intended. Regarding the suitability of this position for a frontier watch-station guarding the great route, see above, p. 727. The large fortified station at Bulungir, to the east of it, now mostly in ruins, served this purpose until relatively recent times.
' See Desert Cathay, ii. pp. 242 sqq.
See below, pp. 1139 sqq.