Meteorology By Aristotle

It is clear then that we must not suppose rivers to originate from
definite reservoirs: for the whole earth, we might almost say, would
not be sufficient (any more than the region of the clouds would be)
if we were to suppose that they were fed by actually existing water
only and it were not the case that as some water passed out of existence
some more came into existence, but rivers always drew their stream
from an existing store. Secondly, the fact that rivers rise at the
foot of mountains proves that a place transmits the water it contains
by gradual percolation of many drops, little by little, and that this
is how the sources of rivers originate. However, there is nothing
impossible about the existence of such places containing a quantity
of water like lakes: only they cannot be big enough to produce the
supposed effect. To think that they are is just as absurd as if one
were to suppose that rivers drew all their water from the sources
we see (for most rivers do flow from springs). So it is no more reasonable
to suppose those lakes to contain the whole volume of water than these
springs.

That there exist such chasms and cavities in the earth we are taught
by the rivers that are swallowed up. They are found in many parts
of the earth: in the Peloponnesus, for instance, there are many such
rivers in Arcadia. The reason is that Arcadia is mountainous and there
are no channels from its valleys to the sea. So these places get full
of water, and this, having no outlet, under the pressure of the water
that is added above, finds a way out for itself underground. In Greece
this kind of thing happens on quite a small scale, but the lake at
the foot of the Caucasus, which the inhabitants of these parts call
a sea, is considerable. Many great rivers fall into it and it has
no visible outlet but issues below the earth off the land of the Coraxi
about the so-called 'deeps of Pontus'. This is a place of unfathomable
depth in the sea: at any rate no one has yet been able to find bottom
there by sounding. At this spot, about three hundred stadia from land,
there comes up sweet water over a large area, not all of it together
but in three places. And in Liguria a river equal in size to the Rhodanus
is swallowed up and appears again elsewhere: the Rhodanus being a
navigable river.