nce. Even now the belief still lingers; and
we continually hear of old Atlantic or Pacific continents, of "Atlantis" or
"Lemuria," of which hypothetical lands many existing islands, although
wholly volcanic, are thought to be the remnants. We have already seen that
Darwin connected the peculiar geological structure of oceanic islands with
the permanence of the great oceans which contain them, and we have shown
that several distinct lines of evidence all point to the same conclusion.
We may therefore define oceanic islands, as follows:--Islands of volcanic
or coralline formation, usually far from continents and always separated
from them by very deep sea, entirely without indigenous land mammalia or
amphibia, but with a fair number of birds and insects, and usually with
some reptiles. This definition will exclude only two islands which have
been sometimes classed as oceanic--New Zealand and the Seychelles.
Rodriguez, which was once thought to be another exception, has been shown
by the explorations during the Transit of Venus Expedition to be
essentially volcanic, with some upraised coralline limestone.

_Continental Islands._--Continental islands are always more varied in their
geological formation, containing both ancient and recent stratified rocks.
They are rarely very remote from a continent, and they always contain some
land mammals and amphibia, as well as representatives of the other classes
and orders in considerable variety. They may, however, be divided into two
well-marked groups--ancient and recent continental islands--the characters
of which may be easily defined.

Recent continental islands are always situated on submerged banks
connecting them with a continent, and the depth of the intervening sea
rarely exceeds 100 fathoms. They resemble the continent in their geological
structure, while their animal and vegetable productions are either almost
identical with those of the continent, or if {244} otherwise, the
difference consists in the presence of closely allied species of the same
types, with occasionally a very few peculiar genera. They possess in fact
all the characteristics of a portion of the continent, separated from it at
a recent geological period.

Ancient continental islands differ greatly from the preceding in many
respects. They are not united to the adjacent continent by a shallow bank,
but are usually separated from it by a depth of sea of several hundreds to
more than a thousand fathoms. I