ON the morning of December 29, 1906, we left the Lou-lan ruins to regain ground where there was still water and with it life. The journey was to be made in two separate parties. The large band of labourers, whom the constant exertions and privations had severely tried, was to return to Abdal by the route previously followed, together with those camels which showed signs of exhaustion. Rai Ram Singh, the Surveyor, who was still suffering from acute attacks of rheumatism, and was thus unfit at the time for fresh surveys such as I should otherwise have liked him to undertake in the desert eastwards, was put in charge of this party. I myself, taking ten of the fittest camels and a small number of men, set out to the south-west in order to reach the terminal course of the Tarim River across the unknown desert area separating it from the dried-up delta of the Kuruk-darya. The point I was steering for was the small ruined site of Merdek-tim, which Dr. Hedin had visited in 1896 on his journey along the Ilek branch of the Tarim, and which I wished to examine before returning to my tasks at the ruins of Miran.
In Chapter xxxvi of my personal narrative 1 I have given an account of the seven trying marches which brought us safely across a great waste of dunes and high ridges of sand to the line of freshwater lagoons linked up by the flood-bed of the Ilek. As reference may be made to that account and to Maps No. 6o, 56, and 57 for all details, it will suffice here to mention only a few salient facts of topography which have a quasi-antiquarian bearing. For a distance of close on thirty miles in a straight line, covered in the course of the first two and a half marches, the route, which I was steering by the compass south-westwards, led across a succession of dry river-beds, all forming part of the ancient Kuruk-darya delta. Their direction, gradually changing from east and west to north-west and south-east, unmistakably proved that they all branched off from the head of the main Kuruk-daryâ, which our surveys of 1914 have shown to lie approximately west of the Lou-lan Site. Further to the east and south-east these beds linked up, as will be seen from a reference to Map No. 6o, with the ancient river-courses which we had crossed on the march to the Lou-lan Site, and which were traced still more clearly by our surveys of 1914. The first two beds, lying over bare wind-eroded ground, were seen as well-marked, winding depressions. But after we had left them behind, at a distance of only six miles from our Camp 125 at L.B., low dunes kept the ground uniformly covered with sand, and made these depressions more difficult to recognize. Yet there remained the familiar narrow belts of ancient riverine jungle, crossed at intervals by their parallel lines of dead Toghraks and tamarisk-cones, to show plainly the direction of the dried-up river branches once spreading over this deltaic ground.
Nowhere along the route followed onithese first few marches were structural remains or any other signs of settled occupation traceable, and the comparative rarity of bare eroded ground
' See Desert Cathay, i. pp. 415 sqq.