Sec. ii] THROUGH AK-SU AND UCH-TURFÂN 1299
The present name Ak-su is obviously of Turkish derivation, and the earliest mention of it that I can trace does not go back beyond the fourteenth century.14
On May 6 I left the ` New Town ' of Ak-su in order to visit Uch-Turfan and the little-known tract of Kelpin in the outer hills of the Tien-shan to the south: - I knew that the Kelpin region had been but imperfectly surveyed, and information secured at Ak-su with Fan Ta jên's kind help pointed to the existence of ruined sites near its small oasis and towards Maral-bashi. By visits to these I wished to conclude my archaeological labours in the field before returning via Yarkand to Khotan, where many tasks urgently called me. The seven-miles ride from the ` New ' to the ` Old Town ' of Ak-su took me along the edge of the .wide barren plateau which rises with precipitous cliffs of loess from 6o to loo feet above the two towns and the irrigated area adjoining them. Its top is covered with large cemeteries, Mazars, and crumbling forts or ' Karauls ', manifestly of late Muhammadan times. Of earlier remains, such as the presence of all the numerous modern shrines might have suggested, I saw none. Yet plenty of quasi-troglodyte dwellings and tombs showed what excellent sites for cave-temples these steep and easily worked walls of hard clay might have furnished.
The two easy marches which brought me to the easternmost edge of the Uch-Turfan district lay across fertile and easily irrigated ground showing the same neglected cultivation I had observed on my way from the Tarim. It was ascribed here to scarcity of population and inadequate labour. Both the river of Ak-su, or Kum-arik-darya as it is known here, and the Tushkan-darya which had to be crossed en route, carried a considerable volume of water, far larger than is to be found in the Khotan rivers at that season. The second, though filling then only a small portion of its flood-bed, fully three-quarters of a mile wide, was barely fordable.15 MM. Grenard and Chavannes have already pointed out that these two rivers are correctly mentioned in the itinerary of the Tang Annals where it describes the route from the town of Po-huan, i, e. Ak-su, to the town of Ta-shih
A, or Wên-su S, i. e. Uch-Turfan.'s The first is named as the ` river of Po-huan ', and the
second as the ` river Hu-lu ' J 1.17 The distance of the first river from the town of Po-huan is not stated, but a distance of 4o li is indicated between it and the second river. This corresponds fairly closely to the 12 miles or so passed by the present road between the marsh-bed, marking an old western bed of the Kum-arik-darya (Map No. 19. D. 4), and the village of Terek-bagh at the crossing of the Tushkan-darya (Map No. 19. c. 4). The ` town of Hsiao-shih ' iJ\ , which the itinerary places half-way between the two rivers, may be looked for approximately near the present villages of Barun and Chawarik.
A pleasant march on May 8 carried me from near the west bank of the Tushkan-darya 18 to
Tang itinerary from Ak-su to Uch-Tur- fan.
14 Cf. Elias-Ross, Tdrikh-i-Rashidi, pp. 7 sqq. Haji Muhammad, whose account Ramusio took down about 155o, also mentions Ak-su ; see Yule-Cordier, Cathay, i. p. 293. His reference to the journey thence to Kashgar lying through the wildest desert ' is significant : it shows that either UchTurfàn or Marâl-bâshi or possibly both were uninhabited at the time.
15 In the Kum-arik-darya, which feeds all the canals of the main oasis of Ak-su to the east of the Tushkan-darya,
I measured a volume of about 1,640 cubic feet per second, apart from the water caught in canals higher up. The Tushkan-darya, then flowing in three main and three shallow channels, had a total volume of over 6,800 cubic feet per second. Later on in the season, when the melting of the great glaciers lining the western slopes of the Khan-tengri
massif has begun, the volume of the Ak-su River, which they feed, is far greater than that of the Tushkan-darya; see Hedin, Reisen in Z.-A., p. 253.
16 Cf. Chavannes, Turcs occid., p. 9; T`oung.pao, 1905, p. 553, note r.
17 The itinerary mentions the ` river Hu-lu ' as of the Yltt`ien J gg territory ; for Yill'ien must be read Yrl-chu
j p the name of Uch-Turfân ; see T`oung-pao, r 905, p. 553, note i.
18 This form of the river name is the one I heard from my local guides belonging to the settled population. The form Taushkan-darya found in Russian maps and elsewhere may be the one used by the Kirghiz occupying the valley above Uch-Turfân.