Chinese wayfarers must have held these repellent belts of salt-coated terraces and the equally trying sea-bed between them.
That the name Po-lung-tui was applied also to an area much wider than that actually covered by those ` dragon-shaped ' white terraces was rightly recognized by M. Chavannes when discussing the mention made of them in the Wei lio, though the absence of adequate geographical data would not allow him to locate them correctly. This wider application is proved by the passage quoted by him from the Former Han Annals, which says : ` straight to the west of Tun-huang, outside the barriers (kuan) [of Yü-mên and Yang] there is the Po-lung-tui desert it! and the lake P`u-ch`ang g A ! '.24 The reference shows that the geographical knowledge of Han times placed side by side, as they are in reality, the two areas, corresponding to the dried-up portion of the old sea-bed and to that still containing marshes. But by itself the passage would not help us to locate the ` White Dragon Mounds ', and as it is the only other reference to them that I am able to trace in the texts accessible to me, we may turn now to the antiquarian question still left open, viz. as to where exactly the ` route of the centre ' is likely to have passed through the
This question, which the textual references do not help to settle, is equally incapable at present of definite solution by the available archaeological evidence. But if we carefully compare the latter with the topographical facts as our surveys recorded in Map No. 32 show them, we may, I think, arrive at certain conclusions considerably restricting the limits within which the line of the ancient route is to be looked for. Good fortune—or was it, perhaps, more than that ?—had made us come upon unmistakable relics of ancient traffic just at those points where the route coming from Lou-lan entered the western belt of salt-coated terraces and again where it passed out of the eastern belt. I refer to the finds of Han coins, the dagger and other small objects at the Mesa to the east of Camp ci, and to our discoveries where we emerged from the Yârdang belt on the eastern side of the dried-up sea and approached Camp civ.
Looking at the configuration of both Yârdang belts on either side of the sea-bed, as our survey shows them in Map No. 32. B, C. 3, it is easy to realize that if we had continued on our eastern course from the above-mentioned Mesa to the edge of the Yârdang belt and had thence struck across the sea-bed with a bearing approximately south-by-east, we should have been moving on a line which would have brought us straight to the place of the first coin find on the eastern side. On this line, the edge of the opposite Yârdang-belt with its far easier going could be reached by a crossing of the difficult expanse of hard salt only a couple of miles longer than our actual crossing between Camps cii and ciii. At the same time the total marching distance would be greatly reduced as compared with our circuitous route farther north. This saving is obvious from the map and could scarcely be less than fifteen miles. On the other hand, a course from the same assumed starting-point with a more southerly bearing would certainly have considerably lengthened the extent of that trying surface of salt crust which travellers by the ancient route had to cover. For our survey shows that the eastern belt of ` Dragon Mounds ', with the far softer shôr between them, ends close to the south of the line previously indicated, and that beyond it the salt-encrusted sea-bed steadily widens.
It would have been quite impossible for me, for obvious practical reasons, to turn back from Camp civ in order to search for traces of the ancient route where it was likely to have entered the Yârdang belt to the west or north-west of our last coin finds. But when a year later the opportunity offered of letting Afrâz-gul carry out supplementary surveys to the east and south-east of the once
M Cf. T`oung-pao, 1905, p. 531, end of note 7.