be impossible; for it cannot be done even after the whole of the country has been mapped. Yet it is at all events allowable to believe, that their migration tracks bear some sort of fixed relation to the relief of the highlands: that is to say, that the lines which would, on the map imagined, indicate the paths taken by the wild-geese in their periodical migrations would also coincide with those parts of the country in which the isohypsometrical curves would show that the lowest elevations run.
I have briefly alluded to the strand-terraces which we saw beside the southern shore of the Perutse-tso, and which in point of age correspond to the lower beach-lines around the Lakor-tso. They show more distinctly than anywhere else beside the little dry watercourse or glen that winds down to the lake from the south-southeast. From our route we saw only three terraces. On the 3oth October, starting from near the outlet of the glen at the lake-side, I took levellings along the east bank of the watercourse, that is towards the south-south-east, with the object of ascertaining exactly the altitudes of the terraces above the lake, and their respective distances from the water's edge. The levelling-mirror was supported, as before, 1.50 m. above the ground. The first terrace, which is quite low and insignificant — in fact it was not visible at all from our route — is situated at a distance of 297 m. from the lake. It can hardly be called a terrace, but is rather a rampart, frequently broken, which appears to be finally lost amongst the littoral marshes to the west. The second old beach-line or strand-terrace is likewise low, but distinct. It is situated 1046 m. from the lake, and appeared to be continued towards the S. 73° W. It is between these two terraces, which thus lie a considerable distance apart, that the eroded watercourse becomes lost in the schor, growing shallower and less distinct, and finally giving out altogether; and there is nothing to indicate that its temporary torrents ever succeed in getting down as far as the lake. Probably, when the stream is bigger than usual, a temporary marsh or shallow marginal lake will be formed between the two lowest terraces. It is indeed possible that there will then exist an outflow from this marsh into the lake through one of the gaps in the lowest rampart. The third terrace is situated at a distance of 1076 m. from the shore; thus the distance between it and the second terrace is only 3o m. It appeared to continue at first towards the S. 7o° W. and afterwards, like the second terrace, to incline more and more towards the north-west; but very soon it disappears altogether. We found the fourth terrace at 1279 m. from the shore, and it continued towards the S. 8o° W. And lastly there was a fifth, 135o m. from the shore, which proceeded due west. All the four upper terraces are pierced by rainwater channels, and these grow increasingly deeper in proportion to their distance from the lake. Above the terrace which I have here called the fifth, but which, strictly speaking, ought of course to be designated the first, we saw no further signs of the Perutsetso having reached a higher level. The lowest terrace, no. 1, lies 4.5o m. above the existing lake; no. 2 lies 24 m. above it; no. 3, 25.50 m. ; no. 4, 31.50 m. ; and no. 5, 39 m. Thus the rise along our measured line was 39 m. in 1350 m., and consequently not altogether inconsiderable, seeing that the outermost, flat spurs of the southern range are still in evidence there. But quite different are the relations at the point where Camp CXVII stood. That was pitched, as I have said, at the edge of the extensive marsh, and immediately above it ran the lowest of the five ter-
IIIF Color HighRes
