is gathered into such reservoirs in winter. Hence rivers are always
fuller in winter than in summer, and some are perennial, others not.
Rivers are perennial where the reservoir is large and so enough
water has collected in it to last out and not be used up before the
winter rain returns. Where the reservoirs are smaller there is less
water in the rivers, and they are dried up and their vessel empty
before the fresh rain comes on.

  But if any one will picture to himself a reservoir adequate to the
water that is continuously flowing day by day, and consider the amount
of the water, it is obvious that a receptacle that is to contain all
the water that flows in the year would be larger than the earth, or,
at any rate, not much smaller.

  Though it is evident that many reservoirs of this kind do exist in
many parts of the earth, yet it is unreasonable for any one to
refuse to admit that air becomes water in the earth for the same
reason as it does above it. If the cold causes the vaporous air to
condense into water above the earth we must suppose the cold in the
earth to produce this same effect, and recognize that there not only
exists in it and flows out of it actually formed water, but that water
is continually forming in it too.

  Again, even in the case of the water that is not being formed from
day to day but exists as such, we must not suppose as some do that
rivers have their source in definite subterranean lakes. On the
contrary, just as above the earth small drops form and these join
others, till finally the water descends in a body as rain, so too we
must suppose that in the earth the water at first trickles together
little by little, and that the sources of the rivers drip, as it were,
out of the earth and then unite. This is proved by facts. When men
construct an aqueduct they collect the water in pipes and trenches, as
if the earth in the higher ground were sweating the water out.
Hence, too, the head-waters of rivers are found to flow from
mountains, and from the greatest mountains there flow the most
numerous and greatest rivers. Again, most springs are in the
neighbourhood of mountains and of high ground, whereas if we except
rivers, water rarely appears in the plains. For mountains and high
ground, suspended over the country like a saturated sponge, make the
water ooze out and trickle together in minute quantities but in many
places. They receive a great deal of water falling as rain (for it
makes no difference whether a spongy receptacle is concave and
turned up or convex and turned down: in either case it will contain
the same volume of matter) and, they also cool the vapour that rises
and condense it back into water.

  Hence, as we said, we find that the greatest rivers flow from the
greatest mountains. This can be seen by looking at itineraries: what
is recorded in them consists either of things which the writer has
seen himself or of such as he has compiled after inquiry from those
who have seen them.

  In Asia we find that the most numerous and greatest rivers flow from
the mountain called Parnassus, admittedly the greatest of all
mountains towards the south-east. When you have crossed it you see the
outer ocean, the further limit of which is unknown to the dwellers
in our world. Besides other rivers there flow from it the Bactrus, the
Choaspes, the Araxes: from the last a branch separates off and flows
into lake Maeotis as the Tanais. From it, too, flows the Indus, the
volume of whose stream is greatest of all rivers. From the Caucasus
flows the Phasis, and very many other great rivers besides. Now the
Caucasus is the greatest of the mountains that lie to the northeast,
both as regards its extent and its height. A proof of its height is
the fact that it can be seen from the so-called 'deeps' and from the
entrance to the lake. Again, the sun shines on its peaks for a third
part of the night before sunrise and again after sunset. Its extent is
proved by the fact that thought contains many inhabitable regions
which are occupied by many nations and in which there are said to be
great lakes, yet they say that all these regions are visible up to the
last peak. From Pyrene (this is a mountain towards the west in
Celtice) there flow the Istrus and the Tartessus. The latter flows
outside the pillars, while the Istrus flows through all Europe into
the Euxine. Most of the remaining rivers flow northwards from the
Hercynian mountains, which are the greatest in height and extent about
that region. In the extreme north, beyond furthest Scythia, are the
mountains called Rhipae. The stories about their size are altogether
too fabulous: however, they say that the most and (after the Istrus)
the greatest rivers flow from them. So, too, in Libya there flow
from the Aethiopian mountains the Aegon and the Nyses; and from the
so-called Silver Mountain the two greatest of named rivers, the
river called Chremetes that flows into the outer ocean, and the main
source of the Nile. Of the rivers in the Greek world, the Achelous
flows from Pindus, the Inachus from the same mountain; the Strymon,
the Nestus, and the Hebrus all three from Scombrus; many rivers,
too, flow from Rhodope.

  All other rivers would be found to flow in the same way, but we have
mentioned these as examples. Even where rivers flow from marshes,
the marshes in almost every case are found to lie below mountains or
gradually rising ground.

  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