RIVER TERRACES. 513 
on either side, and this is why river banks are often higher than the 
flood plain further away. 
In addition to the finer material held in suspension, a river | 
pushes along its bottom a quantity of heavier sand and shingle; the 
amount and size of the material moved depending of course upon the 
velocity and volume of the river. This heavier material always 
remains in the river channel, it cannot be spread over the flood plain. 
So that the river channel and flood plain are generally easily distin- 
guished by the size of the material spread upon them. Theoretically 
a river forms a parabolic curve from its source to its mouth, measured 
along all its windings. Of course the curve is not a true parabola for 
it is affected by the varying hardness of the rocks in its course; hard 
rocks forming rapids or cascades long after the softer ones have been 
worn away. In this case the river forms a series of parabolas between 
each rapid. The upper part of the river is called the ‘torrent 
portion,’ the lower the ‘river portion.’ The velocity varies not only 
with the amount of water but also with the slope, so that it is greatest 
in the upper part of the torrent portion and least in the lower 
part of the river portion. When a river, running between high 
banks, suddenly debouches on to a sloping plain, or into the 
sea, the heavy materials pushed along its bed can be carried 
no further but are spread out into what is called a fan, the river 
running sometimes on one side sometimes on the other. But it is not 
only on fans that the position of the river channel changes. In every 
part of its course the current is liable to be deflected by some hard 
substance and is thus thrown against one of its banks, undermining it, 
and causing the river to curve, the old channel being left and a new 
one cut out. Now the ease with which the current can be deflected 
by an obstacle depends upon the velocity. If the velocity is small it 
is easily turned on one side and vice versd. This is why the river 
portion is more winding than the torrent portion, and why rivers 
flowing with small velocities over flat plains meander in a most 
extraordinary way. 
I have said that a river tends to reduce its bed to a parabola. If 
any part is below the curve it fills up; if any part is too high it cuts 
if down until what is called the non-erosion point is reached, below 
which it can cut no further. When a river has reduced its bed to a 
parabola no further change would take place if the position of the 
channel did not alter. But the river constantly sways from side to 
side owing to deflections of the current and this keeps altering its 
fength. If the length increases the parabola becomes flatter, the 
velocity is decreased and deposition takes place. If the course 
becomes more direct the velocity is increased and erosion takes place 
until the point of non-erosion is once more reached. So that a river 
is constantly cutting away in one place and filling up in another. 
But this is only on a very limited scale as the length cannot vary very 
much. Again, the source of the river is always working backwards, 
and thus lowering the position of the parabola, and giving a tendency 
to the river to erode; but this may be left out of account as the cut- 
ting back of the river for many yards would but slightly affect the 
curve of a river even in its torrent portion and would be quite 
inappreciable in the river portion. 
This effect of the lengthening and shortening of a river course 
between any two points will help us to understand the remarkable 
