and engaged in a battle which lasted from the ch`ên period (7-9 a.m.) to the szzi period (9-I i a.m.). He inflicted a great defeat upon the barbarians, who fled when the night came. He pursued them, killed 5,000 men and made i,000 prisoners ; all the rest dispersed. He took more than i,000 horses, and warlike stores and arms beyond counting.'
The analysis given above of the routes followed by the Chinese columns and of Kao Hsienchih's three days' march to Mount Tan-chü, or the Darkôt, confirms M. Chavannes in locating the
Tibetan stronghold of Lien-yün * near the present Sarhad." It is equally clear from the
description of the river crossing that the Chinese concentration must have taken place on the right or northern bank of the Ab-i-Panja, where the hamlets constituting the present Sarhad are situated, while the stronghold of Lien-yün lay on the opposite left bank. Already, when briefly discussing the record of the expedition in Ancient Khotan, I had expressed the belief that the position taken up by the Tibetan main force, fifteen li to the south of Lien-yün, must be looked for in the valley which debouches on the Ab-i-Panja opposite to Sarhad and leads due south up to the Barôghil and Shawitakh Passes.' I also surmised that the Chinese general, apart from the confidence aroused by the successful river crossing, owed his victory mainly to a flanking movement by which his troops gained the heights and thus successfully turned the fortified line behind which the Tibetans were awaiting them.
This opinion was confirmed by what I saw of the valley leading to the Oxus on my descent from the Barôghil on May 19, and by the examination I was able to make two days later of the mountain-side flanking its debouchure from the west.' The valley into which the route leads down from the Barôghil is quite open and easy about Zartighar, the southernmost hamlet. There a ruined watch-tower shows that defence of the route had been a concern also in modern times. Further down the valley-bottom gradually contracts, though still offering easy going, until, from a point about two miles below Zartighar to beyond the scattered homesteads of Pitkhar,' its width is reduced to between one-half and one-third of a mile. On both sides this defile is flanked by high and very precipitous rocky ridges, the last offshoots of spurs which descend from the main H indukush watershed. These natural defences seemed to provide just the kind of position which would recommend itself to the Tibetans wishing to bar approach to the Barôghil, and thus to safeguard their sole line of communication with the Indus Valley. The width of the defile would account for the relatively large number of defenders recorded by the Chinese Annals for the enemy's main line ; the softness of the ground at its bottom, which is almost perfectly level, covered with fine grass in the summer, and distinctly swampy in the spring owing to imperfect drainage, would explain the use of palisades, at first sight a rather strange method of fortification in these mountains.10 Finally the position seemed to agree curiously well with what two historical instances of modern times,
See Chavannes, Turcs occid., p. 154, noted; also above,
pp. 53 sqq.
8 Cf. the description of these marches in Desert Cathay, i. pp. 64 sq., 7o sqq.
'° In my note of Ancient Khotan, p. 9, I ventured to suggest that, considering how scanty timber must at all times have been about Sarhad, there was some probability that walls or Sangars constructed of loose stones were really meant by the palisades' mentioned in the translation of the passage from the Tang Annals.
This suggestion illustrates afresh the risk run in doubting the accuracy of Chinese records on quasi-topographical points without adequate local knowledge. On the one hand I found that the peculiar nature of the soil in the defile would make the construction of heavy stone walls inadvisable if not distinctly difficult. On the other, my subsequent march up the Ab-i-Panja showed that, though timber was as scarce about Sarhad ifself as I had been led to assume, yet there was abundance of willow and other jungle in parts of the narrow river gorge one march higher up near the debouchure of the Shaor and Bahârak streams. This could well have been used for palisades after being floated down by the river.