to mere fragments of fantastic shapes. Between them reeds and scrub reappeared in patches. On digging through the sand for a well, water oozed out slowly from a layer of hard red clay at a depth of three feet ; it tasted brackish, but was drinkable for animals. As we proceeded southeastwards across this flat plain, the sand on the surface'soon became permeated with moisture and remained so until we arrived, after a march of about three miles, at a broad ridge of dunes, forty to fifty feet in height, near the point where the caravan track crosses it.
The surface of this plain, like that of the sandy plain crossed on the previous day, was clear of reed growth ; it manifestly marked a depression of the basin, which was periodically covered with water. The total absence here, too, of any salt efflorescence supported the belief that the disappearance of this water was due not to evaporation but to subterranean drainage. The view obtained from the ridge (Fig. 188) was almost as commanding as that from the Mesa near Camp cxii. It confirmed the observation made from that point that two more belts of wind-eroded terraces stretched down into the eastern part of the basin, and that there was an absence of features suggesting that surface drainage reached it to any appreciable extent from the unbroken gravel slope to the north-east.
There remained an important point to be settled in the course of our survey of this ancient lacustrine basin. It was whether it was possible for floods from the actual terminal course of the Su-to-ho, as seen by us in the spring of 1907, to reach its southern edge. Below Toghrak-bulak, where the caravan route crosses the river (Map No. 35. c. 4), the survey of this terminal course had to be entrusted in April, 1907, to Ram Singh, as excavation work along the watch-stations of the Limes kept me busy farther away to the east. He had descended the river-bed to a point marked by his Camp 174, and sketching the river's farther course from there, had shown it on his plane-table with a distinct north-westerly bearing from about three miles lower down.
This, together with certain information supplied later by the Surveyor, had led me to show the termination of the river in Map No. 74. A. 3 of Serindia as immediately adjacent to the old lacustrine basin passed on our way from Besh-toghrak. When preparing the Personal Narrative of my second journey I had, rather rashly, as subsequent experience has shown, expressed the conjectural view that water from the Su-lo-ho still reached this ancient lake-bed in its southern portion 10 Subsequent considerations had led me to doubt the correctness of this interpretation, and suggested that the true termination of the Su-lo-ho would have to be looked for farther west. Not content therefore with having instructed. Lal Singh to survey the river's course to its end, I was anxious now to examine the south-eastern portion of the old lacustrine basin myself.
For this purpose I moved first to the south-south-west along and across three branches of the big ridge of dunes which, as the sketch of the ground in Map No. 35. c. 4 shows, projects from that direction into the basin. These branches of the sand ridge rose everywhere 4o to 5o feet above the reed-covered narrow valleys between them. The view obtained from them allowed me, after we had covered about three and a half miles in this way, to make sure that to the south there extended an unbroken line of gravel Sai, sloping up to a far-stretching plateau running approximately from east to west. In order to make certain that there was no gap in this barrier through which water from the terminal Su-lo-ho course, as sketched in the former map, might yet make its way into the basin, I then changed our direction to the south-east until we actually struck the Sai edge rising about eighty feet above the scrub-covered ground immediately below.i1 This gravel plateau extended without a break to the east, and we followed its edge in that direction for close on three
to Cf. Desert Cathay, i. p. 535. passing on either side of Lai Singh's Camp zoo, is wrong and
11 The stipple marking drift-sand which Map No. 35. c. 4 ought to be removed. The ground is a bare gravel slope.
shows between this point of my route and the cliff symbols