37G GEOLOGY. 



their surfaces, showing the presence of the ice during their building. 

 The bowlders might have been crowded in from the sides, or let down 

 from the ice above. As in kames, the gravel is often not well rounded. 

 Eskers often end in kames, and where they are interrupted, the inter- 

 val is often occupied by kames. Occasionally they end in deltas, 

 where the constructing stream issued from the ice into a lake, or in 

 alluvial fans, where the stream issued upon a plain. 



Most existing eskers were probably made just before the disappear- 

 ance of the ice from the region where they occur. Eskers made during 

 the advance of an ice-sheet were likely to be destroyed at a later time. 

 Probably most eskers were made by streams flowing essentially parallel 

 to the direction of the ice movement. The deposits of streams in other 

 positions would stand much less chance of developing distinct ridges 

 before being destroyed by the movement of the ice. 



It is probable that kames are sometimes developed beneath the 

 ice. It has been noted that eskers are occasionally interrupted, prob- 

 ably both where the channels of the subglacial streams suffered con- 

 striction, and great leakage. It is now to be added that kames are 

 sometimes developed at the point of interruption. Irregular and 

 ill-defined patches of sand and gravel, instead of kames, often occur 

 where the eskers are broken. 



(4) Deposits of superglacial and englacial streams. — Superficial and 

 englacial streams have been supposed to make deposits in their channels. 

 It has even been conceived that this was the principal mode of origin 

 of eskers. Against this view, and against the view that superglacial 

 stream deposits are of consequence quantitatively, stand two facts. 

 (1) So far as known, the surfaces of ice-sheets are free from drift (apart 

 from wind-blown dust) except for a fraction (and generally a small 

 one) of a mile from their edges; 1 and (2) superficial streams are, in 

 general, much too swift to allow of the accumulation of drift in their 

 channels. The channels of most superficial streams in North Green- 

 land, even near the edge of the ice where surface debris is abundant, are 

 free from drift. Judging from the force with which they issue from 

 the ice, englacial streams are likewise much too swift to allow of depo- 

 sition along their channels, as a general rule. 



Such trivial accumulations of drift as may be made in superglacial 



1 Jour, of Geol., Vol. IV, p. 804. 



