FISHER: RIVER TERRACES. 
35 
happens that ledges which have been of greatest importance in con¬ 
trolling the deflection of streams are not visible because of wash 
deposits which could most easily cover and apparently obscure them. 
Yet the terrace patterns which appear, make it absolutely evident 
that a ledge existed there. Otherwise the left-hand, 280-foot alluvial 
spur would have been swept away by any river swinging strongly 
against its northern up-stream side. The positions of the rock ledges 
on the map give many other illustrations of rock control, though possi¬ 
bly not so striking as these. For instance, as the West River swung 
to the left at the back of the 220-foot level, its meanders impinged 
against the alluvial material and would probably have eroded more to 
the left had it not been for the existence of the rock ledges. As the 
West River swung southward across the meadow, its passage to the 
right was definitely limited by rock ledges. 
Diminished swinging of the meander belt .— While rivers are flowing 
at high levels in their drift valleys, they may not be expected to en¬ 
counter ledges. As they cut deeper, the possibility of encountering 
rock grows greater, and rock ledges are constantly nearer and nearer 
the axis of the valley and so tend to narrow the meander belt. As the 
valley deepens, the number of defended cusps tends to increase; and 
as the meandering belt is narrowing because of the rock barriers, the 
horizontal distance between low-lying scarps must be less than between 
the high-level scarps,— a principle which is of great interest in this 
connection because it shows that the narrowing of the valley must be a 
normal result due to the discovery of rock ledges and not referable to 
diminution in volume as according to the earlier but discarded theories. 
Since there is no normal distribution of reefs of rock upon the valley 
floor, their discovery is not only accidental but without systematic 
arrangement. Consequently the rock reefs which appear on both 
sides may chance to be near the axis, and so narrow the valley, or may 
be widely separated, and so enable the meander belt to cover a broad 
area. There may be stretches along any river comparatively free 
from ledges, or if not free, in which the ledges lie so low as to have little 
influence and to allow a broadly open valley. Under such circum¬ 
stances, few of the middle terraces will be preserved, and on either 
side will be a high, strong escarpment with a few low-lying escarp¬ 
ments protected by ledges. The Holyoke-Springfield district may 
again be mentioned as an example of this type. Such types, however, 
attract comparatively slight attention, since it is the existence of step- 
