(Ieien shên , which probably means Manichaeism, while at the same time believing in the
Buddhist doctrine. The statement that sheep and horses were kept in distant little-known localities is probably to be explained by the fact that grazing grounds are to be found only in some of the highest side valleys south of the main range and are difficult of access.29a Another portion of the account, relating to the period (557-618) of the Chou and Sui dynasties, which preceded the Tang, mentions sixteen towns in Kao-ch`ang, later on increased to eighteen, and gives details of the administrative system organized after Chinese models. While the men dress as is the custom of
the barbarians (Hie), the women in costume and hairdressing follow Chinese fashions '. Writing was the same as in China, but the scripts of the Hu were also in use. Laws, customs and ceremonies were in essentials those of China.
It is of interest to note that the account given by the Pei shih concludes with a reference to the great desert stretching between Kao-ch`ang and Tun-huang ` where there is no road and travellers have to seek their way by the skeletons of men and animals. On the way one hears sounds of singing or wailing, and if people follow these they usually come to their end. Hence travelling traders ordinarily follow the route via I-wu (Hâmi).' 3° I think that we can safely conclude from this record that a direct route from Turfan to Tun-huang, leading perhaps past those easternmost springs of the Kuruk-tagh which Lai Singh explored in January, 1915, to the Besh-toghrak valley, was still occasionally followed by adventurous wayfarers in the seventh century.3'
The Tang shu's notice of Kao-ch`ang takes up the story with the death of Ch`ü Po-ya and the
accession of his son Ch`ü Wên-t`ai 4, which occurred in A.D. 619, within a year of the
establishment of the Tang dynasty. The account of his reign throws a characteristic light on the position in which Turfan was necessarily placed when plans of Central-Asian expansion had once again been resumed by China. Embassies of homage from Kao-ch`ang are recorded in the years 619 and 620.32 Among presents offered by its king to the imperial court in the years 624 and 627 are mentioned two performing lap-dogs said to have come from Fu-lin or Syria. This is of interest as pointing to trade connexions with the distant Byzantine Empire, confirmation of which is afforded for this period by archaeological finds in Turfân.33 In 63o Ch`ü Wên-t`ai personally paid a visit of homage to the Emperor T`ai-tsung. But some time after his return, he helped the Kagan of the Western Turks to plunder missions that were proceeding to the imperial court and to attack Hâmi, which in A.D. 63o had come under Chinese control. The remonstrances made thereupon by the Emperor produced no result. Ch`ü Wên-t`ai did not proceed in person to the court, as invited ; nor did he send his commander-in-chief, who had previously been summoned
there to account for the attack upon Hâmi. The family name A-shih-na [4. ms borne by
this personage, as M. Chavannes points out, proves his Turkish origin and by itself serves to indicate the influence then wielded by the Turks in the administration of Turfân.34
Thereupon a large force was organized for the conquest of Turfan. This was to open the way to the establishment of Chinese supremacy over the Western countries. Ch`ü Wên-t`ai appears to have relied on the protection afforded by the difficulties of the desert crossing, and died from terror in A. D. 640, when the Chinese army had effected its passage through the desert. The inscription of that year set up on the Barkul pass, to which reference has been made above, shows the care with which preparations had been made by the Chinese commanders to assure the
288 See above, p. 562.
30 It is this passage that Sir Henry Yule, Marco Polo,
i. p. 210, quotes from Ma Tuan-lin (in Visdelou's translation) in illustration of similar folk-lore beliefs about the Lop Desert; cf. Serindia, ii. pp. 562 sq.
31 As to traditional recollections of such a direct route, cf. above, pp. 273, 319.
32 See Chavannes, Turcs occid., p. 24, note 3.
33 See below, Chap. xix. sec. i, v.
34 See Chavannes, Turcs occid., p. 104, note 2.