Sec. iii] BY THE HAN LIMES TO AN-HSI 365
perfectly straight across the gravel-covered flat and rising from three to four feet above it.6 It exactly resembled the remains of the Limes wall where traced in its completely decayed state to the east of T. xxvi.7 In the light of the setting sun, I thought I could also recognize the slight depression left in the gravel by an ancient track keeping parallel to the wall line and at a distance of about nine yards south of it. I had previously noticed a similar depression along more than one stretch of the Limes west of Tun-huang.8
The necessity of reaching water before nightfall prevented any search along the line of the wall and obliged us to hurry on towards the Su-lo-ho across the bare Sai. At a distance of about a mile and a half we came upon a clearly marked dry river-bed, about 20 yards wide and lined with Toghraks, most of them alive. Its appearance and direction left no doubt that it was the same that I remembered having crossed when making my way seven years before to the Su-lo-ho from the easternmost portion then traced of the Tun-huang Limes.9 Beyond it came a narrow strip of ground where the exposed clay was cut up into regular small Yârdangs, 4 to 5 feet in height ; and after this again, level gravel-covered Sai, with only scattered tufts of thorny scrub and here and there a few old Toghraks surviving. There was nothing to assure us that we were approaching the eagerly expected river, until we almost stumbled upon its deeply sunk bed, fringed by a narrow belt of reeds and young Toghraks.
The place we had reached for the night's camp showed signs of having been frequented from time to time by men grazing camels or collecting fuel. A rough cart-track was found in the morning leading from it in the direction of An-hsi. This we followed for close on eight miles eastward to where a large refuse mound, rising to a height of about 8 feet, indicated the former existence of some regular halting-place. Experimental digging disclosed only layers of stable refuse and left it doubtful how long it had been abandoned. From this point the baggage was sent on ahead, with instructions to camp by the river, while Afrâz-gul and I, with a few men, set out to the south-south-west to search for the line of the Limes. At a distance of two miles from where we had left the cart-track, we again crossed the winding river-bed previously mentioned. Its general course lay here from the south-east, and this direction was confirmed by subsequent observations, as Map No. 38. D. 4 shows. I am thus led to conclude that this dry bed probably represents the continuation of the course followed by the T`a-shih river where it turns sharply to the north to lose itself on the flat of the Su-lo-ho valley.
On both sides of this old flood-bed the alluvium once deposited by it was furrowed by wind-erosion into 'Val-clangs. Here they were only 2 to 3 feet high ; but at points farther to the east, where we subsequently had occasion to cross this eroded belt along the dry river-bed, the Yârdangs rose higher—up to 8 to io feet. As we continued our march to the south, the bare Sai of gravel crossed on the previous evening stretched again before us. But in spite of remarkably clear atmospheric conditions, which allowed us to sight the successive ranges of the Nan-shan right away to the snowy chain south of Shih-pao-ch`êng (Map No. 39. D. I ; 41. A, B. I), no ruined towers of the Limes could be seen. Nor did we, owing to the sun being high and in our face, notice the straight line marking its agger until we were quite close upon it.
We there found that the low gravel mound into which it had decayed ran straight like a railway alignment, with the approximate bearing from east to west. Where we first struck it, the mound rose to a height of only 4 feet above the level of the bare gravel, but showed a width of about 32 feet at its foot. No remains of fascines or other reinforcing material were traceable on the surface.