100 SNAKE KIVER PLAINS OF IDAHO. [bull. 199. 
expected to occur where the gradient is gentle, but still sufficient to 
insure a draining away of the lower portion of a lava stream after a 
crust too thick and strong to be wrinkled or broken by the friction of 
the underflow has formed. The underflow must evidently produce a 
current of restricted width, for if broad, the arch left would fail to 
support its own weight and subsidence would occur, probably resulting 
in the production of a more or less characteristic aa surface. The 
length of the tunnel does not seem to be confined within narrow bonds, 
and it may extend for miles, in the direction in which the lava stream 
flows. The presence of the caves is not suggested at the surface 
unless a sagging of their roofs has occurred. The caves in cross 
section show a flat arched roof resting on vertical side walls and nearly 
flat floors. 
The formation of lava caves of considerable size by the arching of 
a still somewhat plastic lava crust, due to lateral pressure caused by an 
underflow, has already been described. Their presence is indicated at 
the surface by an arched ridge, rising above the adjacent and usually 
flat surface. Within the caves, as a rule perhaps, the arch of the roof 
springs from the flat floor and vertical side walls are absent; there 
are exceptions to this, however, and in such instances there has prob- 
ably been an outflow of material as in the formation of the variety of 
caverns noted above. The roofs are relatively thin from having been 
forced up into true arches, which are more perfect structures from an 
engineering point of view than the flat-topped roofs left by an out- 
flow of lava from beneath a thick crust. These caverns are elongate, 
their length being at right angles to the direction of flow of the lava 
stream which produced them. A characteristic feature of the interiors 
is the roughness of their roofs, which, as already stated, have the 
appearance which is produced by the pulling apart of a highly viscous 
substance. This appearance is more pronounced on the roofs than on 
the floors, owing to the fact that the separation took place when the 
nearly solid crust rested on still plastic material below, which settled to 
an even surface. In addition to the " pulled-dough " forms in the 
roof, there are sometimes pendent stalactite-like masses formed by the 
congealing of lava, which was at first sufficiently fluid or plastic to 
drip. 
There is still another variety of lava caverns due to the blowing out 
by escaping steam of still liquid or highly plastic lava through open- 
ings in a solid crust. The chimney -like openings in the parasitic 
cones described on a previous page are examples of "vertical cav- 
erns" which have recesses opening from them at their bottoms. 
The claim that the horizontal enlargements are due to the cause 
suggested requires confirmation, but the cones built of clots of lava, 
scoriaceous bombs, etc., contain far more material than the vertical 
openings within them could have furnished. This is apparent when 
