F. H. Lahee — Crescentic Fractures of Glacial Origin. 43 



but when traced outward, away from the axis, or downward, 

 into the rock (as seen in microscopic slides), it often wanders 

 between the grains and may jump small intervals and become 

 echelon in habit, as in the case of echelon jointing. Within 

 the rock the fracture surface dips toward the axis, or forward, 



Fig. 2. 



Fig. 2. Plan of a series of crescentic fractures. X-Y is the direction in 

 which the ice advanced, a-b is referred to as the 'length' of the longer 

 cracks. 



at an angle varying from 60° to 80° (fig. 1). As for their size, 

 individual cracks range up to seven inches in length (a-b, fig. 2), 

 although commonly this dimension is between 1/4" and 2" ; 

 and their depth is rarely over 1/4". 



These crescentic fractures, then, are of glacial origin. Gil- 

 bert has suggested that they may be due to frictional resistance 

 between the ice and the bed-rock, locally increased by the 

 presence of sand-pockets in the ice-base.* In view of this 

 interpretation it is interesting to observe that the long north- 

 ern slope of Northey Hill rests upon mica and staurolite 

 schists which are very friable and easily pulverized. We are 

 inclined, however, to attribute the cause to more concentrated 

 action than could result from fine, disintegrated materials, 

 even though they were irregularly scattered. Thus, as evi- 

 dence for the operation of a pointed or edged tool are (1) the 

 increase in the number of fissures toward the axes of the sets ; 

 (2) the dying out of these fractures, not only downward, but 

 also laterally, away from the axes ; (3) the frequent association, 

 in a particular manner, of striae and grooves with many of the 

 sets ; and (4) the fact that the curves are hyperbolic. Accord- 

 ing to this view, the tool, in each case, may have been one of 

 the rock fragments witli which the ice must have been loaded. 



It is worthy of remark that actual contact of this tool with 

 the bed-rock does not seem to have been necessary for the pro- 



*Op. cit., p. 304. 



