PHYSICAL GEOGRAPHY OP THE IONIAN ISLANDS, 461 
down from the Mils. They are in fact valleys produced by tlie 
sinking in of the strata into some great natural cavern formed 
or enlarged by the infiltration of the rain-water. They vary 
much in size, but none are too large to be thus explained. 
As might be expected, caverns abound in districts so cal- 
careous, and many of them have been described. They offer 
nothing remarkable, except indeed that in the island of Cerigo, 
which does not belong naturally to the Ionian group, the lime- 
stone fragments have formed a breccia with bones of extinct 
animals ; but all of them illustrate the state of the rock. 
There is, however, a still more curious illustration of this in 
Cephalonia, close to the town of Argostoli. A large and 
broad gulf enters from the south to the western part of Cepha- 
lonia, and from this a small transverse bay opens to the east. 
A spit of comparatively low ground with hills behind marks 
the junction of these two bays. Part of the land here appears 
to have sunk in, and the naked limestone rock, much broken, 
forms a kind of reef, keeping the sea from the narrow low 
tract of sunken ground. At three or four places the reef is 
broken, and the sea enters narrow irregular fractures, and 
is swallowed up in the earth. Advantage has been taken of 
this curious inversion of the usual order of things, and two 
mills are worked by the power of the entering stream of salt 
water made to turn an undershot wheel. The mills worked 
by the rush of sea-water losing itself in the earth have long 
been a subject of great astonishment both to the inhabitants 
and to the few strangers who visit Argostoli. Certainly, that 
there should thus be a permanent rush of water into cavities 
in the rocks close to the sea, and considerably below the sea- 
level, and that this should go on permanently, or for an in- 
definite period, without filling the cavern, is an extraordinary 
phenomenon, and requires explanation. No doubt, the gene- 
rally broken and cavernous state of the limestone, in all 
directions, suggests a clue to the mystery : but the full expla- 
nation is to be found in the climate, which is such that, during 
a great part of the year, the surface is always burnt up, so 
that the evaporation must exceed the natural supply from rain 
if any means exists of supplying the deficiency. The rock 
being split, and full of capillary fissures as well as more open 
cracks, there will always be evaporation from the pools or 
moist rock at whatever depth, so long as the surface is dry 
and heated. Thus, the water entering the broken rock finds 
its level in some subterranean caverns where the surface always 
remains below that of the sea, because more water is carried 
off by evaporation than is received from the sea and the rains 
together. The perfect dryness of the surface, combined with 
the presence of malaria in the Yal di Koppa, and many others 
