DAVIS: THE GRAND CANYON OF THE COLORADO. 135 
tinctly towards the base of the scarp ; several upper branches may unite, 
in a single lower trunk, and thus the lower ravine receives the waste 
from a considerable length of cliff face; at the same time, many spurs 
fail to reach the base of the scarp, but terminate acutely between the 
forking ravines, as in Figure 9. On the other hand, where a projecting 
scarp or promontory stands forth, the ravines diverge, and a single spur 
at the top of the scarp may broaden and repeatedly divide in its descent, 
assuming a sprawling paw-like form, and including many short ravines 
that begin within it. Variations of this kind are repeated systematically 
over and over again in the Permian scarps. When the entire height of 
the spurs is only some four hundred feet, it may be hardly worth while 
to develop any special terminology for their different forms; but when 
Figure 9. 
Diagram of Permian spurs under Shinarump cliff. Tapering spurs on the left, split spurs 
on the right. Constructed from rough sketches. 
one finds that very similar spurs with heights to be measured in thousands 
of feet occur on the dissected slopes of Mt. San Francisco, the value of 
a simple terminology becomes more apparent. 
There seems to be a curious alternation in the generation of spurs and 
ravines where a scanty or insufficient supply of resistant talus fragments 
from the cliff-maker is associated with an active erosion of the strata in 
the under-slope. As the coarse waste weeps and washes down from the 
cliff, it tends to avoid the spurs and to accumulate in the ravines. The 
bare spurs are then eroded more rapidly than the protected ravines, and 
thus a new ravine may come to be worn down the axis of what was a 
spur, while a spur stands forth where the ravine was before. A new 
supply of waste then gathers in the new ravine, while the older waste is 
