190 ROCHESTER ACADEMY OF SCIENCE. 



(27) Eskers should be located in large preglacial valleys, in- 

 stead they cross valleys repeatedly. 



(28) Eskers pass along plains where there is little chance for 

 crevassing, zigzag across valleys without regard to the direction that 

 crevasses (29) would naturally follow. There is little or no evidence 

 of erosion of till back of the point of origin of the esker. 



Conchision. Most of these objections have already been 

 answered in the argument given above. Objections 5, 11, 14 are the 

 most vital and important, and apparently can be answered only in 

 part as yet, or in a manner not thoroughly satisfactory. 



Superglacial hypothesis. The superglacial hypothesis of the 

 origin of eskers has been most ably advanced by Crosby (21). Stone 

 was an early exponent but later was converted to the subglacial idea. 

 Hoist has already been mentioned in connection with the early advo- 

 cacy of this theory of esker origin. The hypothesis has found few 

 supporters in this country. 



Statement. Crosby states "this explanation of eskers assumes 

 a stagnant marginal zone of the ice sheet at least 100 miles in maxi- 

 mum width, practically free from crevasses, sufficiently wasted by 

 ablation to be more or less abundantly covered by englacial drift 

 which has become superglacial, with a general southward slope, 

 and toward the southern border at least, thin enough to reflect in its 

 surface contours, in some degree, the underlying topography" (21). 

 Crevasses, if present, would be sealed in the final stages of move- 

 ment, or filled with debris ; it is significant in this connection to note 

 the rarity or lack of crevasses in the Greenland glacier except near 

 the margin. Water divides on the ice must be sufficiently far apart 

 so that the water supply may be great enough to form large rivers, 

 hence over broad valleys and lowlands of the pre-existing topography 

 are places favorable to the formation of eskers (66). 



At the margin of the ice the superglacial stream's elevation is 

 controlled by the level of a glacial lake, bank of till, or detrital cone 

 (21). The stream grades itself with reference to this control. Tt 

 may have tributaries — in fact a river system upon the surface of the 

 ice, comparable to a river system on a land surface. At first the 

 streams of the system will possess such high velocities that no mate- 

 rial will accumulate in their channels, but as grade approaches corra- 



