madans, mainly petty traders, from the northern oases of the Tarim Basin. Everywhere in the mountains to the north live Mongols who continue in their traditional ways as nomads, known to Muhammadans as Kalmaks and belonging to different Torgut tribes. It is the vicinity of these troublesome nomad neighbours, whose turbulent nature and robbing propensities Chinese reports of the eighteenth century on Kara-shahr duly emphasize,11 which has kept Turki cultivators away from such fertile lands. The same reports also show us that Kara-shahr has suffered from its present conditions of depopulation and neglected agriculture ever since the great inroads of the Dzungars in the last century drove out the old population of Turki Muhammadans.
These observations on the present conditions of Kara-shahr will make it quite clear that, while the territory has been favoured by nature in various ways, its geographical position must at all times have exposed it to a very serious drawback. I mean its close vicinity to, and its easy access from, mountain tracts which, as far back as history takes us, have always had a particular attraction for nomads. It is unnecessary here to explain in detail how the famous grazing uplands of Yulduz have been cherished haunts for all the great nomad nations, from the Wu-sun and Huns downwards, which held sway along the Tien-shan, that natural spina, as it were, in the cycle of Central-Asian migrations. Situated as Kara-shahr is at the very mouth of the big valley leading down from Yulduz, it must have been like a gate specially inviting those who had their favourite summer camps on those grassy plateaus and necessarily looked to the oases on the south as their richest grounds for raids and exactions. Whenever Chinese power was firmly established from Turfan to Kashgar or beyond, the gate might be kept safely closed. The same is likely to have been possible during periods while internal feuds or conflict with nomad aggressors weakened the tribes in the north. But the danger must always have been close at hand, and from time to time Kara-shahr was bound to suffer from its onset. The oases further west would then be exposed, too, to plunder and heavy exactions of tribute. But the additional risk of prolonged occupation would be reserved for Kara-shahr, which alone could offer grazing grounds adequate for the maintenance of large nomad hosts.
The peculiar circumstances just explained as a result of geographical features must be kept in view if we are to understand properly the part played by Kara-shahr in the early history of what is now Chinese Turkestan. It does not appear to have ever been as important as that of Kucha, Kâshgar, Khotan, or Yarkand, either in political respect or with regard to Buddhist culture and all that was connected with it. It is true that Yen-chti figures in the description of the ` Western regions ' given by the Former and Later Annals as a territory with a relatively large population.12 But the records also show that its political fate was always closely bound up with that of its more powerful neighbours on the west and east, Kucha and Turfân. The Later Han Annals' account duly notes that the territory ` on four sides has high mountains which attach themselves to those of Kucha. The roads [leading there] are blocked with obstacles and are easy to defend.' The last remark obviously refers in particular to the routes which gave access to Yen-chi from territories under Chinese control. True also is the statement about ` the water of a lake which spreads in sinuosities within the four mountains ', and what is said about the position of the capital, as we shall see presently. The Chin Annals' notice of Yen-ch`i repeats the essential points of the above description and emphasizes the difficulty of the routes leading to it by adding that ' if a hundred men defend them a thousand could not pass '.13
" Cf. Ritter, Asien, ii. p. 436, where the natural fertility of the land, its former flourishing condition, and the inability of the Mongols to turn its advantages to good account are all quite correctly noted from the Chinese records.
" Cf. Wylie, J. Anlhrop. Inst., xi. pp. 'or sq.; Chavannes, T'ouna pao, 2907, P• 208.
" See Chavannes, Ancient Khotan, i. p. X42. The reference is clearly to the difficult defiles of the Turfân route