are composed, we perceive that these latter accumulations are very
important features. Beginning at first with small and imperfect
alluvial plains, the river, as it descends toward the sea, gaining in
store of water and in the amount of _debris_ which comes with that
water from the hills, while the rate of fall and consequent speed of
the current are diminished, soon comes to a stage where it is engaged
in an endless struggle with the terrace materials. In times of flood,
the walls of the terraces compel the tide to flow over the tops of
these accumulations. Owing to the relative thinness of the water
beyond the bed, and to the growth of vegetation there, the current
moves more slowly, and therefore lays down a considerable deposit of
the silt and sand which it contains. This may result during a single
flood in lifting the level of the terrace by some inches in height,
still further serving to restrict the channel. Along the banks of the
Mississippi and other large rivers the most of this detritus falls
near the stream; a little of it penetrates to the farther side of the
plains, which often have a width of ten miles or more. The result is
that a broad elevation is constructed, a sort of natural mole or
levee, in a measure damming the flood waters, which can now only enter
the "back swamps" through the channels of the tributary streams. Each
of these back swamps normally discharges into the main stream through
a little river of its own, along the banks of which the natural levees
do not develop.

We have now to note a curious swinging movement of rivers which was
first well observed by the skilful engineers of British India. This
movement can best be illustrated by its effects. If on any river which
winds through alluvial plains a jetty is so constructed as to deflect
the stream at any point, the course which it follows will be altered
during its subsequent flow, it may be, for the distance of hundreds of
miles. It will be perceived that in its movements a river normally
strikes first against one shore and then against the other. Its water
in a general way moves as does a billiard ball when it flies from one
cushion to another. It is true that in a torrent we have the same
conditions of motion; but there the banks are either of hard rock or,
if of detritus, they are continually moving into the stream in the
manner before described. In the case of the river, however, its points
of collision are often on soft banks, whi