lift SNAKE RIVER PLAINS OF IDAHO. [bull.199. 
have progressed far are normally compact. Either lapilli or the frag- 
ments produced by contact of hot lava with water may be glassy, but 
such fragments, at least in the tuff cones of Idaho that have been 
examined, are rare, while in the breccias produced by hot lava entering 
water they are common. 
In association with the lava sheets cut through by Snake River there 
arc immense beds of f ragmental material sometimes 100 to 160 or more 
feet thick, and traceable for several miles, which resemble closely the 
tuff about the Market Lake and other craters. In fact, they seem to 
have the characteristics of compacted lapilli or tuff, but again simu- 
late the fragments produced by contact of hot lava with water. In 
which of these two ways the material referred to was principally pro- 
duced I am not sure, but incline to the opinion that in the main it 
resulted from volcanic explosions and is a true tuff. 
When a breccia is produced, including at times cellular lava and the 
"lava balls" referred to, the interspaces are completely filled with 
what was once sand or mud, and the mass is, as a whole, compact and 
essentially impervious to water; but when the cellular and steam-torn 
rock occurs, it is open in texture and in a condition to permit of the 
flow of water through it. These differences are significant in connec- 
tion with the occurrence of springs and in reference to the possibility 
of obtaining artesian water. The truly wonderful springs at what is 
termed The Thousand Springs (Pi. Ill), for example, pour out of a 
cellular lava of the nature described above. Near Hagerman, at two 
localities, tunnels or horizontal wells have been excavated in the basal 
portion of the lava flow and a strong outflow of water has been obtained. 
The facts presented above seem to indicate that whenever a lava 
stream flows into water there will be steam explosions and a great dis- 
turbance of the bottom over which the lava flows. Remarkable exam- 
ples of results of this nature may be seen in the northern wall of Snake 
River Canyon, about a mile upstream from the mouth of Salmon River, 
and near the Riverside ranch, where the involved and steam-torn lava 
and the associated breccia contain fragments of clearly recognizable 
stratified ehrv and masses of sand of all sizes up to 5 or 6 feet in diam- 
eter. Other exposures of the bases of lava streams, however, seem to 
show that lava may flow over a lake bottom without disturbing it or 
producing scarcely any discernible change in it. 
An instance of the nature just referred to occurs in the east wall of 
the canyon of Canyon Creek, about a mile above its junction with 
Snake River. At this locality the rim rock of the canyon is formed by 
a lava sheet between 30 and 40 feet thick, which rests on fine, white, 
thinly laminated lake beds, of which about 12 feet in thickness are ex- 
posed. The lava, owing to wind erosion, projects about 10 feet beyond 
the soft beds beneath, and its bottom is well exposed. The base of 
the lava is a flat, but moderately rough, surface, composed of cellular, 
