the reasons are sufficient which lead us to believe that it accounts
for the main features which we observe in this class of explosions--in
other words, it is a good working hypothesis. We shall now proceed in
the manner which should be followed in all natural inquiry to see if
the facts shown in the distribution of volcanoes in space and time
confirm or deny the view.

The most noteworthy feature in the distribution of volcanoes is that,
at the present time at least, all active vents are limited to the sea
floors or to the shore lands within the narrow range of three hundred
miles from the coast. Wherever we find a coast line destitute of
volcanoes, as is the case with the eastern coast of North and South
America, it appears that the shore has recently been carried into the
land for a considerable distance--in other words, old coast lines are
normally volcanic; that is, here and there have vents of this nature.
Thus the North Atlantic, the coasts of which appear to have gone
inland for a great distance in geologically recent times, is
non-volcanic; while the Pacific coast, which for a long time has
remained in its present position, has a singularly continuous line of
craters near the shore extending from Alaska to Tierra del Fuego. So
uninterrupted is this line of volcanoes that if they were all in
eruption it would very likely be possible to journey down the coast
without ever being out of sight of the columns of vapour which they
would send forth. On the floor of the sea volcanic peaks appear to be
very widely distributed; only a few of them--those which attain the
surface of the water--are really known, but soundings show long lines
of elevations which doubtless represent cones distributed along fault
lines, none of the peaks of sufficient height to break the surface of
the sea. It is likely, indeed, that for one marine volcano which
appears as an island there are scores which do not attain the surface.