EELATIONS OK VALLEY DRIFT AND OTHER DEPOSITS. 477 



The phenomena of delta or diverging branches of glacial rivers j^i'ove 

 that from time to time these streams fomid their channels clogged or were 

 for some other reason diverted to new channels. Admitting that these acci- 

 dents were liable to happen at any time, still I can see no especial liabihty 

 of their happening during the very last of the ice at a particular place 

 except on account of the rising of a hill in front. Transverse hills crossed 

 by glacial rivers might often force the streams to escape either east or west 

 after the ice ■ sank to the tops of the hills. In continuous north-and-south 

 valleys containing gravels deposited in ice channels the retreat would cause 

 sediments to be carried beyond the ice front, where they would overlie or 

 be mixed with the previously deposited ice-channel gravels. Cases of this 

 sort of deposition are found in the valley of the Saco River for many miles 

 above Steep Falls, in the upper Kennebec Valley, and elsewhere. 



Where the very latest conditions favored the formation of the broad 

 osar, the channel might often continue to widen till it extended across a 

 whole valley. The marginal part of the plain of sediments that would 

 extend across the valley might be valley drift, and we should hardly be 

 able to distinguish it from the osar terrace proper. But where we find the 

 narrow osars or reticulated ridges we could not fail to distinguish them from 

 a later deposit of overwash matter, which would necessarily border or 

 overlie them. In general, it is astonishing to note how suddenly sedimen- 

 tation ceases. Kettleholes and ridges of coarse matter are found with their 

 shapes clearly defined. Often tliere has been but little postglacial erosion 

 to fill up the bottoms of the kettleholes. We must, therefore, account not 

 only for the valley drift, but also for its absence from long reaches of the 

 osars and reticulated kames right on the lines of glacial rivers, where, on 

 the glacial hypothesis of the valley drift, its presence would be expected. 



In many cases the relief forms of the land would naturally cause the 

 flow of a glacial river to cease at a given place before the ice front had 

 retreated to that point. Thus, where the ice flowed over transverse hills 

 there would be local deflections of ice movement during the last days of 

 the ice. This would make it increasingly easy for the subglacial streams 

 to find new channels east or west along the valley north of the transverse 

 hills, at the same time that the lowering of the level of the ice would make 

 it inci-easingly difiiciilt to maintain the flow south over the tops of the hills. 

 Often we can trace the new channels by transverse series of gravels. Thus, 



