284 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY. 
behavior of a meandering and swinging stream, slowly degrading a pre- 
viously aggraded valley without change in volume ; and, second, by the 
control exerted here and there over the lateral swinging of the stream 
through the discovery of rock ledges, as suggested by Miller. The 
following pages are devoted to a fuller consideration of this conclusion. 
II. Preliminary Inquiry. 
Various Kinps oF Terraces. For the sake of clearness it is desirable 
to exclude at the outset all kinds of terraces other than those here 
studied. The terraces that occupy so many of our valleys are known as 
river terraces, drift terraces, or alluvial terraces. They have as to 
origin nothing in common with the terraces of sea-shores, such as occur 
on the coastal slopes of Cuba; or with the lake-shore terraces so well 
developed in the basins of Bonneville and Lahontan. They bear little 
resemblance to structural rock benches, such as break the slopes of 
valley sides in dissected plateaus, as in West Virginia or on a still 
larger scale in the Colorado canyon. They have little likeness to the 
silt and gravel-covered rock terraces formed when a graded river, 
revived by uplift, cuts a new valley in its former valley floor, as along 
the gorge of the Rhine on its way through the Schiefergebirge of 
western Germany. 
Our New England drift terraces have a flat and nearly level upper 
surface or plain, limited backwards by rising ground and forwards by 
falling ground, and to that extent they resemble the terraces of all the 
classes above mentioned ; but they have certain well-marked features of 
their own. They are evidently the river-carved remnants of a body 
of stratified clays, sands, or gravels that once occupied in larger volume 
than to-day the rock-floored valleys of still earlier origin. Their upper 
surface, the terrace plain or floor, slopes with the fall of the stream by 
which their scarped face or front has been eroded; and in this they 
differ from sea and lake shore terraces and from structural rock benches, 
none of which have any particular relation to the slope of neighboring 
streams. They consist of unconsolidated, stratified drift; if a ledge 
appears in any part of a drift terrace it is manifestly an accidental 
element, although as will be shown it may exert a controlling influence 
on the pattern of the terrace front ; in this our drift terraces differ from 
the structural rock benches of valley sides in dissected plateaus, and 
from the rock terraces that represent the former valley floors of 
revived rivers, both of which consist essentially of rock, even though the 
