OLD UPPER ALLEGHENY DRAINAGE SYSTEM. 131 



ence to the map of the restored drainage (fig. 1, p. 89). The naturahiess of 

 this restored system lends support to the more positive evidence just cited. 



It should be noted that the old divide between the small streams which 

 flowed south into the Upper Allegheny and those which flowed north into' 

 Cattaraugus Creek was farther south than the present divide. Above the 

 village of Ischua the headwater portion of Ischua Creek appears to have 

 discharged northward past Machias to Cattaraugus Creek. The changes on 

 the headwaters of other tributaries are of less consequence. Cattaraugus 

 Creek departs somewhat from the ancient line of drainage, as noted in the 

 discussion of that stream. 



It is somewhat difficult to decide upon the extent of the basin drained 

 by the western tributaries of the old Upper Allegheny, for the valleys and 

 lowland tracts have been so greatly filled b)^ glacial deposits that the old 

 divides are concealed. Probably much of the drainage areas of Cassadaga 

 Creek, Chautauqua Lake, and the lower portion of Conewango Creek were 

 tributary to the Upper Allegheny along the valley leading from Jamestown 

 eastward to Randolph, N. Y. 



It is possible that the Conewango Valley was a line of northward dis- 

 cliarge for a small section of the present Allegheny between the old divide 

 near Kinzua and a divide near Thompson station, about 12 miles below 

 Warren, Pa., and also for the part of the Tionesta drainage basin above 

 Barnesville, Pa. It is certain that the upper portion of the Tionesta di.s- 

 charged northward through Glade Run to the present Allegheny at Warren, 

 as pointed out by Carll. The old divide where reversal took place is readily 

 located near Barnesville, where, as noted by Carll, the stream enters a 

 narrow gorge scarcely one-fifth the width of the abandoned channel. It also 

 seems evident, from a constriction of the Allegheny Valley that sets in 

 near Thompson and from other features discussed below, that the discharge 

 could not have been down the present Allegheny. The only element 

 of uncertainty is the course of the drainage — whether it was northward 

 through the Conewango reversed, or westward through the lower course 

 of Brokenstraw Creek and a deeply filled broad valley connecting Broken- 

 straw Creek with Oil Creek along the line of the Dunkirk, Allegheny 

 Valley and Pittsburg Railway. The rock floor in the lower course of 

 Conewango Creek and on the Allegheny between Warren and the mouth 

 of Brokenstraw Creek is shown by numerous oil borings to be nearly level, 



