Sec. iv] THE ' WHITE DRAGON MOUNDS' 309
extremity of the hill range which overlooks the valley and inlet from the north (Map No. 32. c. 4).
When describing below our actual progress eastwards along the foot of that range, I shall have occasion to indicate the direct archaeological evidence that the ancient route kept close to it. On
the gradually widening strip of ground which separates the edge of the dried-up bay with its hard
salt crust from the clay cliffs and plateaus marking a geologically still more ancient shore line, we came first upon living vegetation on sandy soil near Camp cvi, and this became plentiful some
miles beyond it eastwards. Water, too, could be found there at no great depth from the surface,
though it proved salt. It is in this vicinity that we have in all probability to look for the position of the Sha-hsi well, where travellers from the side of Tun-huang would find grazing and, at a time
when ` desiccation ' was less marked, even drinkable water, before having to face the absolute waste of salt and bare clay towards Lou-lan. It is only up to this point that drift-sand would be found by them along the foot of the hill range, and I have pointed out in Serindia that this topographical
feature may well have suggested the designation Sha-hsi f§ 4, literally meaning ` the well
west of the sand '.17
Beyond this stage the Wei lio's itinerary mentions only a single locality on the way to Lou-lan, but one of special interest to us. After turning to the north-west the route, we are told, ` passes
through the Lung-tui (Dragon Mounds ), arrives at the ancient Lou-lan '. In Serindia I have
already expressed the belief that in these ` Dragon Mounds ' we must recognize those arrays of salt-impregnated Yardang ridges which, as our surveys between Camps ci and civ have shown,
extend on either side of the north-eastern portion of the dried-up sea-bed.18 They certainly form the most striking feature of the great dismal waste which separates the foot of the hills bounding the eastern rim of the Lop basin from the nearest confines of what was once the habitable territory of Lou-lan. Their identification with the ` Dragon Mounds ' of the Wei lio is directly supported by topographical evidence contained in an interesting passage of the Former Han Annals.
In the account which the chapter of the Annals dealing with the ` Western regions ' devotes to the relations of China with the territory of Shan-shan or Lou-lan, there is an instructive reference,
in connexion with events following the year 92 B. c., to the desert route leading from it to China's
` Now the extreme eastern border of the kingdom of Lou-lan where it approached nearest to China,
was opposite to the Po-lung-tui j J (` White Dragon Mounds '), where there was a scarcity
of water and pasture ; and it always fell to its share to provide guides, to carry water and to forward provisions to meet the Chinese envoys ; but being frequently exposed to the oppressive raids of the soldiery, they at last resolved that it was inconvenient to hold intercourse with China.'
This passage makes it perfectly clear that the name Po-lung-tui, ` White Dragon Mounds ', the identity of which with the Wei lio's Lung-tui, ` Dragon Mounds ', was duly recognized by
M. Chavannes, was applied by the Chinese to a desert tract facing the extreme eastern confines
of Lou-lan territory and situated on the direct route leading to the latter from China. We have been able in the light of direct archaeological evidence to trace that route from the easternmost
habitable ground of Lou-lan to the belt of salt-coated ridges which lines the north-western shore
of the sea-bed. We can now fully comprehend why it was necessary for the Chinese missions arriving on that ground, then as now utterly devoid of means of sustenance, to be met there by
guides and to be furnished with supplies and above all with water. We can appreciate also what a tax the provision of all these necessities for safe transit must have thrown upon the scanty population of Lou-lan.