1 72 RALPH S. TARR 



products of this decay is proved by the fact that fresh diabase was 

 collected in the bottom of the valley. 



This dike chasm, produced before the last ice advance, was not 

 entirely erased by glacial erosion, proving that at this point the ice 

 did not accomplish much work. It would be interesting to know 

 how widespread this weakness of ice-erosion was along the Labrador 



Fig. 9. — Dike valley, formed before last ice advance, on Turnavik Island, Lab 

 rador. 



coast, and whether this particular instance is due to some local re- 

 tardation of the ice; but facts are not available for answer of these 

 questions. 



These observations, in widely scattered localities, do not prove 

 anything beyond the regions in question. It is a well-known fact 

 that the erosive power of ice varies with the conditions, and it is 

 possible that close by some of these localities of moderate ice-erosion 

 there may have been profound changes caused by the ice. Never- 

 theless, such pronounced cases of moderate ice-erosion, in such widely 

 different situations, including surroundings apparently favorable to 

 erosion, prove that under some conditions the erosive power of the 



