The falls of such headwater streams must persist until the plateau is cut away, for the cap rocks over which the streams leap being horizontal cannot be smoothed down till the whole plateau is cut through. They are long-lived features. Moreover every one of the innumerable branch streams must on its way down from the uplands fall over the outcropping edges of all the hard beds. The falls will therefore be common as well as long-lived features. Their frequent occurrence confirms the correctness of this generalization. On the other hand, in regions of tilted rocks, the hard beds are avoided by the streams, which select the softer strata for their valleys. The hard beds soon stand up as ridges or divides, across which only the large streams can maintain their courses, and these are the very ones that soon cut down any fall that may appear in their early stages. Falls on tilted rocks are therefore rare not only because of their brief duration, but also because tilted rocks are crossed by few streams, except the large ones, which soon cut away their falls.
The foregoing considerations show clearly enough that falls like those of northeastern Pennsylvania are rare, and we have now to consider why they should be prevalent in the region in question. The Appalachians contain many water-gaps cut down on tilted beds, every one of which may have been the site of a fall for a relatively brief period of river immaturity, but this brief period is now left far in the past. The streams show many signs of maturity: their slope is gentle and their valleys are wide open from Alabama to Pennsylvania, but in the northeastern corner of the latter State we find a group of streams that leap over high benches into narrow gorges, and the benches are held up by tilted rocks. Manifestly the streams have in some way been lately rejuvenated; they have been, in part of their courses at least, thrown back into a condition of immaturity, at a time not long past, and, as has so well been shown by White, the cause of this is the obstruction of their old channels by irregular deposits of glacial drift. Here first in the whole length of the Allegheny section of the Appalachians we find an exceptional condition of stream life, and here also we come into a region lately glaciated, where heaps of drift have thrown the streams out of their old tracks. The explanation fits perfectly, and if it had not been discovered by inductive observation in the field, the need of it might have been demonstrated deductively. It is a case that has given me much satisfaction from the promise that it holds out
