DAVIS: RIVER TERRACES IN NEW ENGLAND. 305 
A terrace whose scarp has been almost evenly trimmed by the small 
vacillations of a down-sweeping meander will face the axis of the valley. 
A terrace whose scarp has been under-cut by the forward half of a down- 
sweeping meander will face obliquely up the valley, as in the foreground 
of Figure 10. Inasmuch as the normal progress of meander-sweeping is 
down the valley, it would seem, at first thought, that no terrace scarps 
could be carved so as to face in that direction ; but on second thought it 
will be seen that the lateral growth of a meander may cause part of the 
curve to grow up-valley faster than the meander is carried in the other 
direction by the normal down-valley sweeping of the meander system ; 
and in this case a terrace scarp facing obliquely down-valley will be 
carved. An example of this kind is shown near the foreground of Figure 
11. It is manifest that the development of terraces facing down-valley 
will be favored wherever the down-valley sweeping of a group of mean- 
ders is for any reason checked while the enlargement of their curves is 
continued. 
There can be little doubt that the height of terraces produced by the 
action of successive meanders would be very small, hardly measuring as 
many inches or quarter inches as actual terraces measure in feet. Let 
it therefore be now supposed that after a series of one-sweep scarps has 
been carved, the river swings away from the western side of its valley 
and for a time occupies itself in carving scarps on the eastern side. 
Many meanders will have swept down the valley during the eastward 
swing of the meander belt, each meander leaving its faint scar on the 
valley floor. When the river swings westward again, it will be working 
at a lower level than before, and as it then once more undercuts the 
high plain, a distinct terrace with a scarp of ten or twenty feet will be 
formed. Terraces of distinctly different levels may therefore usually 
be taken to represent different swings of the meander belt ; terraces 
that represent only the sweeps of successive meanders while the belt 
remains almost stationary must be so faint as to be hardly noticeable. 
A cusp which results from the slightly vacillating forward sweep of a 
single meander, as B6, B7, etc., Figure 9, has already been called a 
one-sweep cusp. The terrace plain extending forward from the base 
of such a cusp will be a smoothly continuous surface on both sides of 
the apex. When the two parts of a terrace plain on either side of a 
cusp differ in height by a foot or two, they are probably the product of 
different (but not necessarily successive) meanders; and such a cusp 
may be called a two-sweep cusp, because the two levels probably repre- 
sent different sweeps in the meander belt. When the difference in 
