446 GLACIAL GKAVBLS OF MAINE. 



can assume that during the decay of the ice-slieet this enlai'gement went 

 on at a somewhat uniform rate, so that at last they attained the dimensions 

 of the broad osar channels. But if so, how can we, on the subglacial 

 hypothesis, account for the discontinuous gravels, whei'e the channels con- 

 necting the successive lake-like enlargements were so narrow and the 

 resulting velocity was so great that for long distances no sediment was 

 deposited ? Besides, wear of surface streams ought to enlarge the chaimels 

 somewhat uniformly — that is, produce ordinarj^ osar channels ; but I see 

 no method of wear by which these extraordinary local enlargements would 

 be produced. 



On the other hand, if we postulate a body of water open to the sun- 

 light, we at once find a sufficient local supply of energy to produce these 

 local enlargements, in the heat absorbed directly from the sun by the water 

 of the channel, pool, or lake. We are also saved from a self-destructive 

 assumption of so great power of the ordinary superficial waters of the 

 glacier — such as are exposed for only a short time to the sun and then 

 plunge beneath the ice — in enlarging their channels, as would make it 

 impossible to account for the narrow tunnels. 



On the subglacial hypothesis the broad osar channels originated as 

 ordinary narrow osar rivers, the roofs of whose tunnels subsequently dis- 

 appeared. Were these, then, superficial streams? In my earlier writings 

 they were so interpreted, and formed one of the principal arguments for 

 the belief that superficial streams were able to cut canyons down to the 

 bottom of the ice and deposit stratified sediments Avithin them resting on 

 the till or rock. Professor Chamberlin suggests that they were neither 

 subglacial nor superficial. It is probable the water that flowed in them was 

 in other portions of the glacier a part of the subglacial drainage. They are 

 in general equally consistent with either the subglacial or the superglacial 

 hypothesis, and therefore must certainly be withdrawn as evidence of 

 superglacial streams. On the subglacial hypothesis all tunnels at some 

 time lost their roofs, but these are supposed to have lost theirs before osar 

 deposition was completed. 



2. Another hypothesis would be about as follows: ' 



As the subglacial tunnels attained considerable breadth, and the ice 

 became thin, sagging or collapse of the roofs became more rapid and the 

 cross section of tlie tunnel became a more and more flattened arch. In 



