valleys no attention has been given to the
phenomena of waterfalls, those accidents of the flow which, as we have
noted, are particularly apt to characterize rivers which have not yet
cut down to near the sea level. Where the normal uniform descent which
is characteristic of a river's bed is interrupted by a sudden steep,
the fact always indicates the occurrence of one of a number of
geological actions. The commonest cause of waterfalls is due to a
sudden change in the character of horizontal or at least nearly level
beds over which the stream may flow. Where after coursing for a
distance over a hard layer the stream comes to its edge and drops on a
soft or easily eroded stratum, it will cut this latter bed away, and
create a more or less characteristic waterfall. Tumbling down the face
of the hard layer, the stream acquires velocity; the _debris_ which it
conveys is hurled against the bottom, and therefore cuts powerfully,
while before, being only rubbed over the stone as it moved along, it
cut but slightly. Masses of ice have the same effect as stones. Bits
dropping from the ledge are often swept round and round by the eddies,
so that they excavate an opening which prevents their chance escape.
In these confined spaces they work like augers, boring a deep,
well-like cavity. As the bits of stone wear out they are replaced by
others, which fall in from above. Working in this way, the fragments
often develop regular well-like depressions, the cavities of which
work back under the cliffs, and by the undermining process deprive the
face of the wall of its support, so that it tumbles in ruin to the
base, there to supply more material for the potholing action.

Waterfalls of the type above described are by far the commonest of
those which occur out of the torrent districts of a great river
system. That of Niagara is an excellent specimen of the type, which,
though rarely manifested in anything like the dignity of the great
fall, is plentifully shown throughout the Mississippi Valley and the
basin of the Great Lakes. Within a hundred miles of Niagara there are
at least a hundred small waterfalls of the same type. Probably three
quarters of all the larger accidents of this nature are due to the
conditions of a hard bed overlying softer strata.

Falls are also produced in very many instances by dikes which cross
the stream. So, too, though rarely, only one striking instance being
known, an ancient coral reef which has become