254 Chamberhn and Leverett — Studies of the 



recession were interrupted by a later filling of glacial gravel. 

 An apparent objection to this is found in the fact that there is 

 no appreciable decline in the rock floor from Warren to the 

 Thompson col. This level condition, however, may perhaps 

 find its explanation in northward differential depression of a 

 floor once sloping southward with a low gradient. In profile 

 A, a broken line is introduced which has, at Warren, the alti- 

 tude of the lowest rock shelf covered by early gravel and 

 which perhaps represents the limit of preglacial erosion. This 

 line would give to the rock floor a continuous northward 

 descent. 



The course of discharge was probably through* the Cone- 

 wango-Cattaraugus outlet of the upper section of the Alle- 

 gheny, for there seems to be no direct northward outlet to 

 Lake Erie from either the Chautauqua or Cassadaga valleys so 

 deep as the Conewango-Cattaraugus outlet. 



The Oil Creek Basin. — We need refer but briefly to this 

 basin since it does not involve a section of the Allegheny but 

 only a tributary, and since the evidence of reversal has been 

 fully presented by Mr. Carll. The greater part of this basin 

 drained northwestward through Muddy creek to the French 

 creek valley near Cambridge, Pennsylvania. There is some 

 uncertainty as to the course of the outlet from Cambridge, the 

 data being insufficient to show whether it was direct to the 

 Lake Erie basin along the deeply filled Conneautee creek 

 valley or indirect along French creek valley past Mead- 

 ville to the Conneaut outlet and thence northwest past Con- 

 neaut lake and through Conneaut creek valley to the lake 

 basin. Mr. Carll thought it had the latter course because of 

 doubt as to the existence of a deep channel connecting Con- 

 neautee and Elk creek valleys. That region is, however^ so 

 thickly covered with drift that there is no certainty as yet 

 that a deep channel does not cross it. Furthermore there 

 seem to be objections to the view that the old stream followed 

 down French creek, there being a series of island-like hills 

 in the midst of the valley below Venango, midway between 

 Cambridge and Meadville. These suggest the possibility of 

 an old col there. A decision cannot be reached until each val- 

 ley is more thoroughly tested by borings. 



The Middle Allegheny Basin. — This basin includes the 

 lower portion of the Tionesta (below the old col near Barns- 

 ville P. O.), the Allegheny from the old col at Thompson's to near 

 the mouth of the Clarion, and the lower part of French creek. 

 Several lines of evidence unite in indicating that this district 

 formerly discharged northwestward past Conneaut lake to the 

 Lake Erie basin. Evidence in favor of the reversal is found 

 in the narrowness of the Allegheny valley above the mouth of 



