Under the red cliff of Forked Donkia the valley becomes very broad, bare, and gravelly, with a confusion of moraines, and turns more northwards. At the angle, the present terminal moraine rises like a mountain (I assumed it to be about 800 feet high),* [This is the largest and longest terminal moraine backed by an existing glacier that I examined with care: I doubt its being so high as the moraine of the Allalein glacier below the Mat-maark sea in the Sachs valley (Valais, Switzerland); but it is impossible to compare such objects from memory: the Donkia one was much the most uniform in height.] and crosses the valley from N.N.E. to S.S.W. From the summit, which rises above the level of the glacier, and from which I assume its present retirement, a most striking scene opened. The ice filling an immense basin, several miles broad and long, formed a low dome,* [This convexity of the ice is particularly alluded to by Forbes ("Travels in the Alps," p.386), as the "renflement" of Rendu and "surface bombee" of Agassiz, and is attributed to the effects of hydrostatic pressure tending to press the lower layers of ice upwards to the surface. My own impression at the time was, that the convexity of the surface of the Donkia glacier was due to a subjacent mountain spur running south from Donkia itself. I know, however, far too little of the topography of this glacier to advance such a conjecture with any confidence. In this case, as in all similar ones, broad expanses being covered to an enormous depth with ice, the surface of the latter must in some degree be modified by the ridges and valleys it conceals. The typical "surface bombee," which is conspicuous in the Himalaya glaciers, I was wont (in my ignorance of the mechanical laws of glaciers) to attribute to the more rapid melting of the edges of the glacier by the radiated heat of its lateral moraines and of the flanks of the valley that it occupies.] with Forked Donkia on the west, and a serried range of rusty-red scarped mountains, 20,000 feet high on the north and east, separating large tributary glaciers. Other still loftier tops of Donkia appeared behind these, upwards of 22,000 feet high, but I could not recognise the true summit (23,176 feet). The surface was very rugged, and so deeply honey-combed that the foot often sank from six to eight inches in crisp wet ice. I proceeded a mile on it, with much more difficulty than on any Swiss glacier: this was owing to the elevation, and the corrosion of the surface into pits and pools of water; the crevasses being but few and distant. I saw no dirt-bands on looking down upon it from a point I attained under the red cliff of Forked Donkia, at an elevation of 18,307 feet by barometer, and 18,597 by boiling-point. The weather was very cold, the thermometer fell from 41 degrees to 34 degrees, and it snowed heavily after 3 p.m.