CONTAINING FLINT IMPLEMENTS, AND ON THE LOESS. 275 
of the lowest valley-gravel ; that the higher beds ( b ) were formed at the time the higher- 
level gravels were being accumulated in the bed of the old river ; the beds (b 1 ) after 
the gravel ( d ) was left dry ; whilst the lower beds ( b ") result from inundations of the 
river after the excavation of the valley C, and when the higher levels were beyond the 
reach of floods. The same characters prevail throughout all the levels, except that the 
lower masses usually contain more shells, and show more distinct traces of bedding and 
lamination. That the deposit of Loess was out of the reach of the ordinary current of 
the river is evident from the circumstance of its containing no rolled pebbles or fluviatile 
shells — except a few of either sometimes swept in from the ordinary river-channel, and 
serving to show the connexion of the two deposits, — from the uniformity of its compo- 
sition, and from the local nature and angularity of its debris. It presents precisely 
those characters which would result from fine deposition from turbid waters, with the 
occasional presence of a few angular fragments of the adjacent rocks. As the waters 
subsided, this silt would be left in sheltered positions, would be added to from year to 
year, until, as the river wore a deeper channel, the first-formed sediment would be left 
permanently dry, only to be disturbed by meteorological agencies, which would from 
time to time carry down portions of it into the river, now at a lower level, there to be 
re-deposited at such lower levels. 
Another proof of its limited extent and of its not having at any time filled the valleys, 
is that in the lower levels, when it abuts against the hill-sides, it always slopes up 
towards the hill, following in fact nearly the present shape of the ground, and does 
not form horizontal beds cut off and truncated in the way which so generally occurs 
with the higher-level gravels, or when sedimentary strata have been excavated. 
On this view of the subject the difficulty otherwise experienced in conceiving floods 
of a magnitude such that the waters would rise to the brim of these deep valleys, or in 
supposing these valleys to be filled up with silt and afterwards re-excavated, is avoided, 
for it would reduce the rise to be accounted for to limits more compatible with our 
experience of certain existing rivers. It is possible also that variations in the extent of 
elevation of the land may even cause the difference often to appear greater than it is. 
There is one seeming difficulty to this hypothesis, which must occur to every field 
geologist: it is, the apparently distinct separation of the Loess and the underlying 
gravels wherever they occur in association. But a close examination will sometimes 
show thin seams of loam, not to be distinguished from the Loess, intercalated in the 
gravels, especially in the higher levels, though as a rule the masses of this material 
invariably overlie the gravels. The explanation of this fact is, I apprehend, to be found 
in the circumstance that it is only when the shingle or gravel ceased to be deposited, 
owing to the changes in the level or course of the river, that the Loess could accumulate. 
So long as the place was in the direct line and on the level of the stream, and subject 
to its scouring action, no such deposit could take place. Consequently, so long as the 
bed of the old river was on the level of d, diag. fig. 17, so long would the accumulation 
of the Loess from the rise of the flood-waters to o be restricted to b. When the valley 
