Febbcary 14, 1913] 



SCIENCE 



247 



was northward, but the tributaries had 

 various directions. The glacial drainage 

 took advantage of the favoring valleys and 

 connected them into sequence of south- 

 ward flow. Tarr thought that the work of 

 stream diversion and of channel erosion 

 was mainly Prewisconsin, for the district 

 described in the Folio 169 (page 30). He 

 specially cites the outlet of Cayuga Lake, 

 the gorge of Tioughnioga creek and the 

 gorge of Chemung River behind Hawes 

 hill, west of Blmira. The copious waters 

 from the waning Laurentian ice sheet were 

 supplied with such volume of detritus that 

 they were largely aggrading agents. It is 

 possible that the south-leading valleys were 

 mostly established by Prewisconsin glacial 

 drainage and that the work of the latest 

 glacial floods was chiefly transportative. 

 In the eastern half of the state the glacial 

 outflow was freely into the Susquehanna 

 and Delaware escape or into the Hudson- 

 Champlain marine inlet, so there was no 

 necessity for cutting new channels. 



The heaviest normal drainage was that 

 in south-central New York, concentrated 

 in the Susquehanna, which cut the gorge 

 south of Say re. Pa., and the river which 

 drained Lake Iroquois through the ilohawk 

 Valley, the Iromohawk. This great river 

 was the predecessor of the St. Lawrence, 

 which it probably exceeded in volume, as it 

 carried not only the outflow of the glacial 

 Great Lakes, but the copious waters from 

 the glacial melting. 



Subglacial Drainage. — This class of gla- 

 cial streams has been noted chiefly in rela- 

 tion to eskers, which fall under another 

 head in this writing. It is not likely that 

 all eskers were laid down in the beds of 

 streams actually beneath, or in tunnels 

 under, the ice sheet, though some probably 

 were. Probably most subglacial or engla- 

 eial streams were full loaded with detritus, 

 and it is not likely that many streams be- 



neath the ice margin were so free of drift 

 or under such hydraulic pressure as to 

 seriously erode their beds. However, a few 

 peculiar channels, or "dead" creeks, have 

 been noted which have such form and rela- 

 tions as to suggest erosional flow beneath 

 the stagnant margin of the ice. One of 

 these bayou-like channels is that of Dead 

 creek, a tributary of Seneca River, lying 

 southwest of Baldwinsville, and mapped 

 on the Baldwinsville sheet. 



Marginal Drainage. — This class of drain- 

 age phenomena included many of the most 

 conspicuous and interesting features con- 

 nected with the disappearance of the ice 

 sheet, and they have been the subject of 

 much work by the writer. The ice-border 

 drainage channels are important as they lo- 

 cate ice-front positions and determine the 

 altitude of the glacial lakes which they 

 drained. They are humanly, or econom- 

 ically, important since they have graded 

 the ways for many lines of communica- 

 tion or transportation. And they are spe- 

 cially valuable for geologic instruction 

 since they are widely distributed and easily 

 recognized products of long extinct agen- 

 cies. 



It is evident that stream flow along the 

 ice margin could occur only where the 

 land surface sloped toward the ice, and 

 consequently only north of the divide. The 

 remarkable physiography of the western 

 half of the state favored the production of 

 glacial lakes, which required outlet chan- 

 nels for the imprisoned waters. 



The most notable series of ice-border 

 drainage channels occur in five districts. 

 (1) On the south slope of the Erie basin, 

 where the ice-impounded waters in the 

 north-sloping valleys escaped westward into 

 the Brian glacial lakes. (2) Along the 

 south slope of the Ontario basin the glacial 

 waters found eastward escape toward the 

 Mohawk-Hudson depression. (3) On the 



