russell.] SPRING-FORMED ALCOVES IN CANYONS. 129 
or nearly free from talus, but are so fresh in appearance that the Last 
blocks to fall from them must have been detached during the past few 
years, while the sides of the canyons through which the great springs 
discharge their waters have at their bases talus slopes (PI. XXIII) 
which extend well toward the centers of the gulf between them. The 
Blue Lakes are retained by dams composed of talus blocks. 
In considering the origin of alcoves of the type of Little Canyon, 
etc., the question arises, Why should not landslides be caused by the 
undermining of thick sheets of basalt by springs in the same man- 
ner that a river cuts under the border of a similarly thick sheet of 
resistant rock? The answer to this query is that the fall of individual 
blocks of comparatively small size, so as to form a talus, and the 
breaking away and descent of large masses, as in landslides, are in 
realit}^ but extremes of a single process. Large blocks of such dimen- 
sions as to be termed landslides are not caused to fall by the under- 
mining performed by springs, evidently because the action is too local. 
The fact that large alcoves are not more frequent along the side of 
Snake River Canyon, as at The Thousand Springs, etc., where water is 
pouring out, is because only exceptionally strong springs can under- 
mine thick lava sheets so as to initiate and continue the process 
described when the other essential conditions are most favorable. At 
The Thousand Springs the water emerges from the open basal portion 
of a lava sheet and above a thick layer of compact tuff which stands in 
a vertical cliff, and no approach to the formation of an alcove is to be 
seen. Several of the smaller springs which come to the surface in the 
canyon wall from beneath relatively thin sheets of lava, as in the vicin- 
ity of Hagerman, are situated at the heads of reentrants in the canyon 
walls. These reentrants are of the same general character as the larger 
alcove, being produced by large springs, but the extension head ward 
has been small. In these instances the water-bearing layer is cellular 
basalt and the beds beneath stratified clay. 
The necessary conditions, then, for the production of large alcoves 
which have otherwise a generally straight alignment, and which 
furnish no surface tributaries to the stream in the main canyon, 
are the presence of a surface or superior sheet of compact rock of 
I very considerable thickness, and beneath this a stratum of uncon- 
! solidated fine sand, silt, etc., of fully as great thickness as the overly - 
! ing stratum, and an abundant water supply in the unconsolidated 
stratum. The conditions are still more favorable if the superior sheet 
of compact rock is vertically jointed. The initial step in the process 
is the escape of the water at some locus of discharge forming a spring 
sufficiently powerful to carry away the material composing the porous 
stratum. The necessity for a thick bed of fine sand or other similar 
material beneath the canyon-forming layer is shown by the fact that 
blocks falling from the cliffs in order to form an alcove are not car- 
Bull. 199—02 9 
