—316 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY. 
side. The short curves and blunt cusps may be blended so as to pro- 
duce a convex terrace front for a little distance down-valley from the 
ledge ; and this convex front will be separated from the long concave 
re-entrant by a more or less pronounced angle at the defended cusp. 
The reason of this may be easily understood from Figure 32. The ledge 
was here first discovered by a down-sweeping meander, of whose first 
work only the short scarp A remains. The meander was then com- 
pressed so as to scour out a large concave re-entrant, and was with- 
drawn from the channel, B, at its base by a short-cut (outside of the 
diagram). A meander on the new course of the stream, swinging 
westward, trimmed off the terrace front in successive lines, C, D, E; 
but as this meander had no definite relation to the ledge, and as the 
general sweeping of its curve was down-valley, it did not, trim the ter- 
race front with any special regard to the down-valley side of the de- 
fending ledge. In brief, the tendency of a stream to sweep its meanders 
down-valley commonly results in trimming off the terrace front close to 
the up-valley side of a defending ledge, but it is only by chance that the 
terrace is worn away close on the down-valley side of the ledge. The 
pattern here deduced may be matched in many actual examples. 
Derenpep Cusps: Larter Stace. After a ledge has once been dis- 
covered by the swinging river, there is much probability that the for- 
ward reach of its under slope will present an obstruction to the stream 
every time it swings again towards the valley side. For example, let 
