1894.] 
267 
[Dodge. 
If we follow the deductive scheine outlined above, river terrace 
may be clussitied as follows. 
Planatioii terract 
not farther considered. 
oj(i aue : 
' norniul . I capture. 
{ <lisc<)very ofliaid bcdt 
I 
Alluvial ! 
terraces 
f simple \ 
subnormal 
, accidental 
composite 
I 
L coiiijilex 
r cliluatic. 
j barrier from alluvial cone, talus, 
^ land slide, etc. 
V reversed gradient. 
r barrier by timber, beaver, lava, 
) or artificial dam. 
^ , . , f obstructive. - 
I olacial < I T 
V » ( overloatlmy. 
! rising of land, 
continuation of any factor to a 
later cju-le. 
\ comi)licated in origin and not 
( capable of classification. 
C<JNIHTIONS OK TkI!KACE FoKMATION. 
Let us now pass from a deduc^tive consiileration of the probable 
cause of alluvial terraces to an examination of the actual coudilions 
as they a[)[)ear iu many terraced river valleys. The top of the 
highest terrace represents the original line of the alius ial plain of 
the stream in which the terrace was formed, and the escarpment 
shows the depth to which the river cut before the next plain was 
formed, and hence it is difficult at first to see why there should be 
more than one terrace in a river valley, if we consider the erosive 
j)rocesses to be uniformly continuous. It is especially hard to 
account for the formation of ten or more terraces in regular order 
on the sides of a valley ; for instance, in certain portions of the Con- 
necticut valley. The terraces above the [>resent level of the river 
are the remnants of the oldest flood plain of the river, and the 
question at once arises why a river should so frequently leave such 
remnants in its down cutting. Why should it not, as soon as it 
has cut its channel as low as it can into the deposits under the 
existing conditions of base-level, swing back and forth across its 
