DAVIS: RIVER TERRACES IN NEW ENGLAND. 289 
north-flowing rivers was turned over the divides into the valleys of 
south-flowing rivers. ‘This may have been the case while the ice still 
covered the northern basins, their waters (as far as they had any) then 
running as subglacial streams, which may have been forced to ascend 
slopes and cross divides. Effective constraint may also have been pro- 
vided after the ice had at least in part withdrawn from the northern 
basins, but when it remained in sufficient force to obstruct their normal 
outlets, thus forming lakes whose overflow ran across a pass in the 
divide to some southern valley. Another cause for increased volume of 
our south-flowing rivers was the importation into their basins of a con- 
siderable snowfall that was received on the ice sheet over some northern 
basin. A fourth cause for increased volume of our rivers lies in a pos- 
sibly greater precipitation during the later stages of the glacial period 
than at present. A fifth cause lies in a relatively rapid melting of the 
retreating ice sheet. It is eminently possible that these various causes 
may have contributed effectively to an increase in river volume while 
the New England valleys were aggrading with drift; but it does not 
follow that volumes decidedly larger than those of to-day were continued 
into the period of terracing. 
Except where direct evidence is given by curvature and are of high- 
level terrace scarps, a formerly greater volume of the terracing streams 
should be regarded only as a possible, not as an actual occurrence. It is 
especially desirable that large bulk and coarse texture of terrace deposits 
should not be too readily accepted as evidence of former greater volume 
of streams; for bulk of deposits is a function of time as well as of 
rate of action, and texture is a function of slope as well as of 
stream volume and velocity. Hence until time and slope are shown to 
have been insufficient to account for bulk and texture of deposits, it is 
not compulsory to account for them by greater stream volume. 
Even if decrease of volume has been of general occurrence during the 
period of terracing, it has nevertheless not been in control of terrace 
development ; for if it had been, stepping terraces should be much more 
abundant than they are to-day. As a matter of fact, the diagrams by 
which terraced valleys are ordinarily represented give an exaggerated 
idea of the prevalence and perfection of these graceful forms. It is 
rare to find a long flight of stepping terraces on both sides of a valley ; 
it is rare to find a flight of terraces continued for any long distance 
along a valley side; when more than three or four low steps are to be 
counted, it is usually only for a moderate fraction of a mile that they 
persist. A large part of the length of our terraced valleys is bordered 
