of three and a half miles we arrived at the southern end of a belt of Mesas, projecting into the flat expanse of the valley from the glacis of the hill range and running from NNE. to SSW. Their average height was forty to fifty feet. When looking round from the top of the Mesa on which I had fixed the plane-table, I was struck by the curiously straight line of what looked like a double embankment, stretching away to the north-east for about three-quarters of a mile. The raised edges were covered with reeds, while the space between the two banks, which appeared about twenty feet wide, was bare of vegetation. Where this line passed among the Mesas, just north of the one I stood on, it turned at right angles and took a north-westerly direction. I remembered the difficulty I had experienced in 1907 during my explorations along the Tun-huang Limes, in attempting to follow up on scrub-covered ground the traces of decayed earth walls which I had clearly distinguished before, when looking down from a height.' I therefore did not examine this strange line more closely.
I had occasion the next day to regret this omission when I perused the route report that Afraz-gul had kept while accompanying Lai Singh, and found there that he had quite independently made an exactly corresponding observation some fourteen miles farther west. Marching from Camp 96 and skirting the edge of the gravel Sai, they had come upon a series of old cairns and a strange embanked line running across reed-covered ground lower down. Afraz-gul had been able to trace the line clearly for about one-third of a mile, the bare space in the middle being about twenty-one feet wide and the reed-covered banks raised at least a foot or two above it. The direction of the embanked line where traced was straight and almost due east to west, but a bend at the eastern end of the stretch indicated that it came from the north-east. The cairns, three in all, were found over a distance of three-quarters of a mile and lay in the same direction. They were built of rough stones and looked much decayed. A fourth cairn of the same kind had been met with before about a mile and a half to the south-west.
The impression received by both Afräz-gul and myself was that of a canal rather than of a road. It has consequently occurred to me that the line thus observed at two widely distant points might possibly be the trace left by an attempt to bring water down the valley along the Lou-lan route and thus to facilitate traffic on it. But the evidence gathered is too slight to justify more than a conjecture put forth with all due reserve.
My farther march beyond the Mesa above mentioned lay first to the east-north-east across ground where reeds grew abundantly both on sandy soil and on intervening patches of soft sher. Then low dunes became frequent, completely fixed and overgrown by reeds. After having thus covered about fourteen miles from our last camp I turned to the south-east and reached the two long clay terraces, about a hundred and twenty feet high, between which the caravan track passes before approaching,Bésh-toghrak (Map No. 35. B. 4).8 I well remembered noticing in 1907 the curious gate-like appearance produced here by the narrowing of the valley trough. Ascending the farthest point of the terraces which jut out from the long sand-covered clay ridge on the south, I could clearly make out a corresponding promontory projecting opposite to them from the steep line of clay cliffs which marks the foot of the hill chain overlooking the valley from the north.
From what I could see with my glasses and from the information contained in Lai Singh's plane-table and in Afraz-gul's route report, these spurs facing each other from the south and the north appear to be of exactly the same character and configuration. Considering that the distance dividing their ends is only three miles, the inference seems justified that they represent the remnant of a clay ridge that once stretched ledge-like right across the valley and was cut through by the action of water. The erosive effect of a great volume of water is similarly reflected in the appear-
7 Cf. Serindia, ii. p. 637. 8 See Desert Cathay, i. p. 527.