128 SNAKE RIVER PLAINS OF IDAHO. [bull. 199. 
reported to be not less than 1,500 eubic feet per second. The canyon 
of the Malade throughout the lower 4 miles of its course is similar to 
the Box and Little Canyon alcoves, but its walls are higher, and dur- 
ing portions of each year, and perhaps continuously during certain 
years, it receives a stream at its head from the plain above. 
The water flowing from each of the four alcoves just described is 
clear, but has a bluish color and a slight opalescence, due to finely 
divided white silt-like material in suspension. It is this fact which 
has suggested the name for the small spring-fed lakes in Blue Lakes 
alcove. These pools and others fed by springs in and near the four 
alcoves referred to have white bottoms composed of clean white sand, 
which is brought to the surface of the springs themselves. 
The four side canyons or alcoves referred to thus have many similar 
features and without question have been produced by similar agencies. 
One conspicuous feature in each is a sheet of basalt from 200 to 300 
feet or more thick. No exposures of the material on which the basalt 
rests are known to me, but as the springs are bringing out white silt 
and fine white quartz sand it is evident that beds of similar material 
exist beneath the basalt. 
The origin of the side canyons which receive no surface streams, 
such as Little and Box canyons and the one in which the Blue Lakes 
are situated, is plainly due to the action of the great springs which 
come out at their heads. The lower portion of the canyon of the 
Malade has a like history, modified by the fact that a surface stream 
is also concerned in the work. The great springs undermine the 
basalt by removing the soft material on which it rests. Thus blocks 
of the usually more or less vertically jointed rock break away and fall 
into the spring, and sooner or later sink into the soft bed beneath, as 
the emerging waters remove the silt and sand from beneath them. 
By this process vertical walls without talus slopes are produced. This 
process continuing, the cliff recedes, leaving a side cut or alcove in 
the wall of the main canyon, which becomes lengthened into a lateral 
canyon. The process would seem to be cumulative, as the farther the 
head of a side canyon receded the greater would be the tendency of 
the escaping waters to converge toward it. The marked tendency to 
an enlargement into an amphitheater, observable especially at the 
head of the Blue Lakes alcove, is apparently due to this cause. 
It is to be expected, under the hypothesis just suggested, that as the 
head of a spring-formed alcove recedes the wall formed would be 
always vertical and kept fresh in appearance by the fall, from time to 
time, of portions of its face; while the sides of the older portion of 
the cut produced would become more or less encumbered with talus. 
The truth of each of these inferences is abundantly sustained by 
observation. The precipices partially encircling the great springs at 
the heads of the alcoves visited are not only vertical and either without 
