1122 THE NORTH-WEST FRONTIER OF KAN-SU [Chap. XXVII
towards Chia-yii kuan and Su-chou, as I had been led to assume before.' In reality a practically unbroken chain of remains shows that it at first continued eastwards past the outlying oasis of Hu-haitzû or Ying-Van, well to the north of the,An-hsi—Su-chou high road, whence it led far away towards the north-east to beyond the Mao-mei tract on the united course of the Su-chou and Kan-chou rivers.$ This discovery has made it perfectly certain that the well-known line.of wall following more or less the great route from Kan-chou to Su-chou can have nothing to do with the border line drawn on the first Chinese conquest of westernmost Kan-su and with Han Wu-ti's extension of the Great Wall. It must be of much later origin, and everything in the way of archaeological and historical indications combines to support the view expressed to me by local informants that it dates from Ming times.
It was then that the Chinese Empire assumed that policy of strict seclusion towards Central Asia the application of which on this very frontier is so clearly proved by the Western notices to be presently mentioned. A purely defensive line drawn so closely to the limits of the cultivated areas along the foot of the Nan-shan and the high road connecting them would have had no raison d'être in Tang times while China was asserting political. and military control over vast portions of Central Asia. When the hold of Chinese imperial power over Kan-su had been brought to an end in most parts by Turkish and Tibetan invasion, a work of this kind manifestly could not be undertaken. While Uigurs, and then Tanguts, held sway over ` Ho-hsi ', their power extended not merely over the cultivated tracts which this wall was meant to protect, but also over areas extending far beyond them northward. At that period the construction of the wall would have served no purpose, even if such a scheme of defence could ever have recommended itself to semi-nomadic rulers. The same observation obviously applies also to the period of Mongol dominion over the Empire, when the frontiers of China were kept open more than ever before to intercourse with Central and Western Asia.
Historical records abundantly attest the complete change of policy which took place on the advent of the Ming dynasty (A. D. t 368) and the rigid system then adopted of closing China's western frontiers to foreign intercourse. This system of seclusion alone can account for the construction of a wall which, unlike the ancient Han Limes, ignores all strategic advantages of the ground and clings closely to the line where a mere policing of the border was easiest. Here we find the explanation at once of the wall being carried always close to the occupied area or else to the high road, even where military considerations would manifestly have recommended a different position,9 and of the abandonment of oases like Chin-ea and Mao-mei which could not have been brought within the Wall without making police control over it more difficult. But the most significant indication lies in the fact that this wall ends exactly at Chia-yü kuan. This was the very place during Ming times where, as the concordant accounts of Western travellers prove, the rare missions and caravans annually admitted ' within the Wall ' from Central Asia were subjected to close examination before being allowed to enter Chinese territory.
' Cf. Desert Cathay, ii. p. 282. The erroneous view there expressed was influenced largely by my former interpretation of the double wall closing the Hao-Shan-k'ou gorge in inverse directions. The fact that on my reconnaissance of September, 3907, to the north of Chin-t'a I failed to trace the remains of the Han Limes proved also misleading—as negative evidence Often does.
See Third Journey of Exploration, Geogr. Journal, xlviii.
pp. 195 sq.
9 As a striking illustration of this indifference to military
advantages, I may refer to the line of wall from Yeh-ma-wan eastwards, as shown by Map No. 88. B-D. I. The commanding ridges stretching from that point east to the Hsi-mên-k'an defile of the Pei-ta Ho offer an excellent line of defence, as I had occasion to realize when crossing them in 1914. The engineers of the Han Limes would certainly have turned it to good account by taking their border-line along it. The Ming wall instead clings closely to the cultivation edge, where it was easier to maintain police guards for keeping-out unauthorized persons, etc.