CAUSES OF CHANGES IN DRAINAGE. 197 



these lakes. This would perhaps at first result in lakes that drained south- 

 westward tlirough the Maumee Basin to the Wabash. But with the advance 

 of the ice field the lake basins would become filled, and ponding would 

 extend southward toward the present Ohio. New lines of discharge would 

 then be opened. This process of shifting to new lines might continue in 

 jDarts of the region down to the culmination of the glaciation, but in other 

 parts the final shifting may have taken place long before the ice field 

 reached its extreme limits. A stream which suffered a late diversion should 

 accomplish correspondingly less work than one that suffered an older diver- 

 sion, so that by the amount of work the relative dates of diversion may be 

 estimated. 



The effect of glaciation in diverting streams would in some cases be 

 restricted to the time when the ice field was present, there being a return 

 to earlier courses upon the withdrawal of the ice. But in other cases the 

 first ice invasion produced a permanent diversion of di-ainage. In each 

 succeeding stage of glaciation the streams would be subject to disturbances 

 similar to those produced by the earliest glaciation. This field is liable, 

 therefore, to contain examples of diversion of various dates, from the first 

 obstruction of the northward drainage by the encroachment of the ice field 

 on the basins of the lower Great Lakes down to the close of the last glacial 

 stage and the final disappearance of the ice from these basins. Instances of 

 diversion and of control by glaciation or glacial features, illustrating the 

 wide differences in date, will appear in the course of the discussion of the 

 glacial features. 



Turning to the Ohio, it will be found that the old Upper Ohio or 

 Monongahela system was diverted to the present course at a date at least 

 as early as the culmination of the earliest glaciation in the Upper Ohio 

 region, a glaciation that was probably Kansan if not pre-Kansan. While 

 a part of the change (in the portion between New Martinsville and Mounds- 

 ville, W. Va.) appears to have been produced through piracy at an earlier 

 date than the first glaciation, there seem to be no grounds for inferring that 

 the great diversion of the old Monongahela from the northward to the 

 southwestward line of discharge took place through pirac}^ On the con- 

 trary, the slope of the old gradation plain toward the Lake Erie Basin is so 

 great that it seems scarcely possible for it to have been disturbed through 

 piracy by the southwest system of drainage. It may be suggested that a 



