Sec. ii] RUINED SITES WEST OF THE MUZ-ART RIVER 817
be difficult, owing to thick tamarisk growth, and the search was fruitless. The Surveyor observed no soil subject to wind-erosion. Hence the existence of any ruined site of the ` Tati ' type in this region appears very doubtful. Nor did he discover any ancient remains on the long and trying march to Kara-yulghun which occupied the next three days ; much trouble was caused by the scarcity of wells, and by their brackish water. The experiences recorded by the Surveyor explain why the desert route between Kuchâ and Ak-su, although so direct, is seldom used nowadays, except by men anxious to escape observation.
This route is probably that described in an itinerary of the Tang shu, translated by
M. Chavannes,6 as connecting An-hsi it:4 or Kuchâ with the town of Po-huan 4 which
corresponds to the present Ak-su.7 It is apparently also the route that Hsüan-tsang followed
when, starting from Kuchâ, he reached ` the little kingdom of Po-lu-chia ' g, after
crossing a small sandy desert for 600 li westwards ; for a passage of the Tang shu definitely identifies Po-lu-chia with Po-huan.8 Hsüan-tsang's Memoirs furnish no details about the route ; but the bearing and distance agree with the direct desert route, for the distance between the extreme western edge of Kuchâ. cultivation and the eastern edge of cultivation on the side of Ak-su, marked by the village lands of Jam (Map No. 12. A. 2), is approximately 120 miles. Neither the distance, nor the mention of a sandy desert, can be reconciled with the only alternative route, that followed by the present high road through Sairam and Bai. This is considerably longer and nowhere crosses a sandy desert, as reference to Maps Nos. 17. A, B. I ; 12. A, B. 2 ; C, D. I will show.
In the itinerary of the Tang shu we find a similar correspondence as regards the bearing and the character of the district, though the particular stages mentioned cannot be located with certainty. We are told that starting ` from An-hsi or Kuchâ, westwards one passes through the barrier (kuan) of Chê-chiieh fi gl and crosses the Po-ma-ho È-j ,f6 hpJ or River of the White Horse ', obviously the Muz-art-daryâ ; ` i8o li farther one passes westwards into the stony plain of Chii p`i-lo One passes the salt wells and at the end of 120 li arrives at the town of Chii p`i-lo. 6o li farther
on one arrives at the town of A-ksi-yen I~p7~ . 6o li farther on one arrives at the town of
Po-huan.' Here it must be noticed, in the first place, that the total distance indicated, 420 li, is considerably less than the 600 li recorded in the Hsi yü-chi. Apparently the distance between the entrance into the desert and the `salt wells' has been omitted.
Allowing for this omission, one can account for the other measurements by the following conjectural locations. The distance of i8o li, if reckoned from Kuchâ town, would bring us approximately to the westernmost edge of Yulduz-bâgh cultivation, a distance of about 32 miles as the crow flies. The ` salt wells ' might well be placed near Afraz-gul's camp at Shôr-yâr (Map No. 12. C. 2), where a spring supplies brackish but drinkable water close to a small stream wholly salt. The 120 li thence reckoned to Chü-p`i-lo town would take us to cultivable ground where the map (No. 12. B. 2) shows the small detached oases of Ulugh-yâr, about 25 miles to the west in a straight line. From there to the present `old town' of Ak-su is about 32 miles in a straight line, a distance somewhat in excess of the aggregate of 120 li indicated by the Tang itinerary to A-hsi-yen and thence to Po-huan. We cannot, however, be sure that the chief place of the Ak-su district, which the ` town of Po-huan ' undoubtedly represents, occupied the same position as the present ` kôneshahr ' of Ak-su ; and as cultivation is practically continuous from Jam village westwards, a nearer position of Po-huan town, more in keeping with the road measurement given by the itinerary, is conceivable. But in the absence of definite indications the exact location of the town of Po-huan and of the intermediate stage of A-hsi-yen must evidently remain doubtful.
6 See Chavannes, Turcs occid., p. 8. 8 Cf. Chavannes, Turcs occid., pp. 8, note ; 120.
7 For this identification, cf. Serindia, iii. p. 1297.