Below the loess of Canstadt, in which bones of the mammoth are so abundant, is a bed of gravel evidently an old river channel now many feet above the level of the Neckar, the valley having there been excavated to some depth below its ancient channel so as to lie in the underlying red sandstone of Keuper. Although the loess, when traced from the valley of the Rhine into that of the Neckar, or into any other of its tributaries, often undergoes some slight alteration in its character, yet there is so much identity of composition as to suggest the idea that the mud of the main river passed far up the tributary valleys, just as that of the Mississippi during floods flows far up the Ohio, carrying its mud with it into the basin of that river. But the uniformity of colour and mineral composition does not extend indefinitely into the higher parts of every basin. In that of the Neckar, for example, near Tubingen, I found the fluviatile loam or brick-earth, enclosing the usual Helices and Succineae, together with the bones of the mammoth, very distinct in colour and composition from ordinary Rhenish loess, and such as no one could confound with Alpine mud. It is mottled with red and green, like the New Red Sandstone or Keuper, from which it has clearly been derived. Such examples, however, merely show that where a basin is so limited in size that the detritus is derived chiefly or exclusively from one formation, the prevailing rock will impart its colour and composition in a very decided manner to the loam; whereas, in the basin of a great river which has many tributaries, the loam will consist of a mixture of almost every variety of rock, and will therefore exhibit an average result nearly the same in all countries. Thus, the loam which fills to a great depth the wide valley of the Saone, which is bounded on the west side by an escarpment of Inferior Oolite, and by the chain of the Jura on the east, is very like the loess found in the continuation of the same great basin after the junction of the Rhone, by which a large supply of Alpine mud has been added and intermixed. In the higher parts of the basin of the Danube, loess of the same character as that of the Rhine, and which I believe to be chiefly of Alpine origin, attains a far greater elevation above the sea than any deposits of Rhenish loess; but the loam which, according to M. Stur, fills valleys on the north slope of the Carpathians almost up to the watershed between Galicia and Hungary, may be derived from a distinct source. OSCILLATIONS OF LEVEL REQUIRED TO EXPLAIN THE ACCUMULATION AND DENUDATION OF THE LOESS.