Sec. iii] HUA-HAI-TZO AND ITS LIMES REMAINS 397
we found traces of old canals and abandoned cultivation for about two miles beyond the present limit of the oasis. But even thus it seems to me very unlikely that the importance of the oasis could ever have been great enough either to divert to itself traffic from the shorter and more convenient high road passing south of the range or in itself to justify the inclusion of Hua-hai-tzit within the marches protected by the Han Limes. Accordingly, when our search had definitely shown that the line followed by the border wall passed right across the depression from the west to the east, there remained the puzzling question as to why this line had been adopted by Han Wu-ti's ` Wardens of the Marches ' instead of the far more easily guarded natural line of defence which the rugged range to the south appeared so conveniently to offer.18
The answer to this question was found only when subsequent explorations along the Pei-ta-ho revealed the great extension to the north of the territory over which, under the great Han Emperor, a firm Chinese hold was established. But two points of antiquarian interest had already become
clear before that discovery. One was that the alternative name of Ying p`an 2. the
military encampment ', borne by the little oasis, may well be derived from the part which it must have played in the border defence scheme of Han times. Reference to the map will show that it would have been quite impossible, without a garrison placed there, effectively to guard the Limes along a line which here stretched for fully ninety miles across absolute desert. On the other hand, it appears unlikely that this outlying isolated oasis could have been garrisoned with any advantage in later times, when the defensive line on the north-western marches of Kan-su was withdrawn to the vicinity of Su-chou, as it certainly was during the Ming period and right down to the Man-chu conquest of Hsin-chiang. However, the question of the name Yingp`an is one which could definitely be decided only by reference to Chinese historical records inaccessible to me.
Another point of interest is the proof, afforded by the line that the wall follows north of Yingp`an, that this ground was not covered by a large Iake either in Han times or later, as might have been concluded from the representation which the Chinese ` Wu-chiang map ' gives of this area. Professor Futterer had already called attention to the discrepancy between this Chinese cartographical representation and the actual topographical facts,18 and the evidence afforded by the remains of the Limes indicates a similar discrepancy as regards a much earlier period. This discrepancy deserves particular notice in view of the extensive use which has been made of data derived solely from that product of Chinese seventeenth-century cartography for the discussion of the so-called ` Lop-nor problem ' and similar questions.
SECTION IV.—THE LIMES TRACED EAST OF HUA-HAI-TZO
On the morning of April 28th we set out from our camp near the springs of Hsiao-ch`üan-tzü, and after proceeding about three miles across the breadth of the oasis, marched ESE. by the caravan track leading towards Su-chou. Traces of old canals and cultivation were met with over a distance of two miles or so from the edge of the oasis. Beyond this, for about ten miles, the route lay across a flat clayey plain where scrub was abundant and scattered tamarisk-cones rose to a height of eight
18 Cf. Serindia, iii. pp. 1136. It was the great natural strength of the line of defence offered by the high and rugged hill chain flanking the great route west of Chia-yü-kuan that had in 1907 first led me erroneously to look here for the eastern continuation of the Han Limes line coming from Tun-huang and An-hsi ; cf. Desert Cathay, ii. pp. 28o sqq.
18 Cf. Futterer, Whsle Gobi, p. 24. It deserves to be noted that the Wu-ch`ang map, as quoted there, calls the eastern
most portion of the supposed great lake basin by a name rendered as ` sea of flowers '. This is manifestly a transla-
tion of the name Hua-hai-tzti ).4 . , borne by the
Regarding the long-continued reproduction of this apocryphal great lake in European maps, cf. Serindia, iii. p. 1136, note 2.