Sec. üi] ON HSÜAN-TSANG'S ROUTE TO KASHGAR 75
greatly increased the trouble of construction. Both here and at Kansir the conjecture suggested itself that the insertion of thin layers of twigs and brushwood (here from the juniper growth which is still to be met with in some of the neighbouring side-valleys 7) was primarily intended as a substitute for an adequate supply of wet plaster to set the bricks ; want of water at such an inaccessible height must have rendered this very difficult to prepare. This opinion has been greatly strengthened by the observations subsequently gathered along the ancient Chinese border wall in the desert west and north of Tun-huang, where the difficulties about the carriage of water for building purposes must have been equally great in most places, and where the same expedient was probably first resorted to as a regular constructive feature.
But whatever the origin of this method of strengthening the brickwork may be, I felt certain that the old mountain fastness was the same which Hstian-tsang had seen or heard of. The way in which he records the ancient legend then clinging to it leaves no doubt that it had become ruined long before his own time. The local tradition he had heard ascribed the stronghold to the Han times, the earliest period of Chinese influence in the Tarim Basin ; and it was no small satisfaction to me to see that here, at the very first point where I had touched again his Central Asian route, definite archaeological evidence on the spot confirmed afresh the trustworthiness of the great Chinese traveller. But striking,-too, was the evidence afforded by the ruins for the dryness of the climate which prevails in these mountains, and which alone could account for their survival in so exposed a position from so early a date. If Sarikol had ever in historical times enjoyed much heavier snow and rainfall than it now receives, these ancient walls, perched above precipitous slopes and at an elevation probably close on i 3,000 feet above sea level,' would have long ago disappeared.
Short as my stay at the site had to be under the conditions described in my personal narrative, it was quite sufficient to convince me of the exceptional natural strength of the position, amounting almost to impregnability in days which knew not gunpowder. Though overlooked by higher spurs both on the north and west, the terraces of the hill-top were quite beyond bowshot. In various ways the position, though much stronger, recalled that of Adh-i-Samûdh, the old hill stronghold I had examined in 1904 above the Kohat Valley on the North-west Frontier? The absence of all pottery débris suggested that, just as the latter site, Kiz-kurghan could have served only as a temporary refuge in case of danger, not as a place of permanent occupation. The great natural strength of the position impressed me more than ever when we had safely scrambled down over the slopes of rock and detritus on the south-west and resumed our march to Pisling in the failing light. The narrow track leading by the left bank of the river was completely commanded by the rock walls of Kizkurghan. These towered so sheer above it that some contrivance of ropes would have enabled the defenders to gain direct access to the river water, an important consideration of defence. Close investment of the fastness was impossible either from the river or the gloomy winding gorge of the Kiz-kurghan Jilga, which on the north and north-west forms a huge natural fosse with precipitous rock scarps on both sides hundreds of feet high.
Proceeding down the gradually widening defile of the Taghdumbash River I passed, some four miles below Kiz-kurghan, old terraced fields with traces of canals from the side-valley of Karajilga and reached the Sarikoli hamlet of Pisling, where present cultivation commences, after another five miles. Thence a long day's march of some forty miles brought me down on May 31 to Tash-
1 When descending by the left river bank to Pisling after my survey of Kiz-kurghan I heard the name Archalik, meaning ' place of juniper growth', actually applied to the slopes above the debouchure of the Kara-jilga some five miles below Kiz-kurghan.
6 The map based on the Pamir Boundary Commission's surveys shows the elevation of the Ghujak-bai (Ujadbai) post on the opposite river bank as 11,95 r feet.
9 Cf. Stein, Archaeol. Survey Report, N.W.F.P., 1905, pp. 2 sqq.