THE WISCONSIN GLACIATION 607 



diversion, as that at Adrian, where Crab Creek was completely blocked by 

 Wisconsin gravel. The basin fill is not dissected, however, as are Spokane 

 gravel terraces in the npper Crab Creek drainage. This may be because 

 it does not lie in a narrow valley and because it is very porons, absorbing 

 all rainfall and allowing no surface streams to form. The dissection 

 takes the form of three large meridional channels converging to the 

 Drumheller plexus. Two of these lead from the mouth of Grand Coulee ; 

 the third and^ easternmost leads from upper Crab Creek near the mouth 

 of Dry Coulee. It seems probable that, though some intergiacial trench- 

 ing by Crab Creek occurred, these channels were eroded largely by the 

 diverted Columbia during the Wisconsin epoch, the erosion being ren- 

 dered possible because of the contemporaneous deepening of the two main 

 Drumheller channels. The western channel is broad but much shallower 

 than the other two. Its proportions indicate that it is the channel of a 

 large stream, not the valley of a small one, and its shallowness indicates 

 that it was abandoned early in the Wisconsin dissection of the fill. 



The mouth of Grand Coulee is an undrained depression containing 

 Soap Lake. It is dammed by the gravel deposits in Quincy Valley. All 

 drainage of the lower canyon of Grand Coulee comes to it, upper Crab 

 Creek flows to it (when it flows at all), and the neighboring gravel plain, 

 through an arc of 180 degrees, from east through south to west, slopes 

 back toward it. The slope in this arc is gentle and clearly constructional. 

 Whether this back slope is wholly of Wisconsin age or dates in part from 

 the Spokane epoch, it seems clear that it is a graded subfluvial slope, 

 adjusted to the traction load and the velocity of a current emerging 

 from Grand Coulee. The velocity here was greatest within the rock walls 

 6t Soap Lake and decreased rapidly as the waters spread out in the 

 Quincy basin. The depth of that glacial stream can not be measured, 

 however, by the difference in altitude of lake floor and gravel rim. Dur- 

 ing the later stages of Wisconsin discharge, as the Drumheller channels 

 were gashed more deeply and the three channels eroded across Quincy 

 basin fill, notches appeared in the rim of the gravel barrier and the con- 

 stricted portion containing Soap Lake probably was then deepened. 



In all gravels in glacial spillways across the Columbia plateau, basalt 

 is by far the most important constituent. Only a small fraction of 1 

 Ijei cent is of other material. It therefore is not to be considered as 

 glacial outwash in the ordinary sense, for it has not come from the 

 Cordilleran glacial drift. It represents basalt eroded by the high-gra- 

 dient glacial streams in producing scablands and canyoned coulees. 



Thus a brief episode in the latter half of the Pleistocene (the maxi- 

 mum of the Spokane glaciation) introduced conditions under wiiicli ilio 



