126 NATURAL SCIENCE. FEB., 
gravelly and sandy tracts, the rainfall formed subterranean courses, 
flowing, in some instances, even out to sea in that way, as we now 
witness in the case of numerous springs issuing from our cliffs on the 
coast. There deep channels or ‘‘chines”’ are in time formed, and the 
gravelly accumulations being ultimately cleared out, the clayey strata 
are exposed. In one case, I noticed a narrow gorge in the Contorted 
Drift, 9 feet deep and 23 to 3 feet wide—like a miniature canon—at 
the bottom of one of these chines. Thus streams flowing for a time 
underground, and carrying away material from below, will cause the 
surface to sink. 
From these and other considerations we may be able to under- 
stand how some of our great sheets of gravel and sand, like those of 
the neighbourhood of Holt and Cromer, have become isolated hills, 
with ramifying spurs ; for although some of the larger sheets of gravel 
may have been deposited in patches, yet the surrounding strata have 
been denuded, and they themselves have been broken up into 
smaller patches or outliers. When once this has taken place, the 
surface-features of these sandy and gravelly tracts may remain for 
long periods much the same, although the level of the whole may be 
slowly reduced by springs carrying away material from the lower 
portions of the strata at their junction with clayey beds beneath (3). 
Subterranean erosion is suggested by the irregular channels that 
are often met with on the surfaces of clays that lie beneath Valley 
Gravel. On the Lias Clay at Bath, on the Oxford Clay at Oxford, 
and on the London Clay in the Thames Valley, we find such 
channels beneath the gravel; and in many instances the stones 
filling the channels lie with their longer axes more or less upright. 
These appearances have sometimes been attributed to the action of 
land-ice, or to the movements of thawing and slipping soil.t_ Without 
questioning that these explanations may be true in certain cases, I 
think the possibility of erosion by streams has often been over- 
looked. 
On the Dorsetshire coast, between Lyme Regis and Bridport, 
cullies similar to those noted on the Norfolk coast may be seen in 
the Lias clays, and these are formed by springs that issue from the 
porous Cretaceous Beds above. Conybeare, in 1840, in his explana- 
tion of the great Landslip of Dowlands and Bindon, remarked that 
where the loose sands of the Upper Greensand rest on the Lias 
clays, and are exposed in the cliffs, ‘copious land springs will gush 
forth, and carry away in different seasons greater or less quantities of 
the loose material through which they flow; and thus, in process 
of time, the superincumbent rock will become partially undermined.” 
Some of the landslips on Portland, and the great gullies or 
fissures that affect the limestone-rocks, may probably be attributed 
1 See Letter of Darwin to Professor James Geikie (1876), Life and Letters of 
Darwin, ed. 2, 1887, vol. iii., p. 214. 
