39. _Migration of the Atlantic-Ohio divide_.--There are certain shifted courses which cannot be definitely referred to any particular cycle, and which may therefore be mentioned now. Among the greatest are those by which the divide between the Atlantic and the Ohio streams has been changed from its initial position on the great constructional Nittany highland and Bedford range. There was probably no significant change until after Newark depression, for the branches of the Anthracite river could not have begun to push the divide westward till after the eastward flow of the river was determined; until then, there does not seem to have been any marked advantage possessed by the eastward streams over the westward. But with the eastward escape of the Anthracite, it probably found a shorter course to the sea and one that led it over alternately soft and hard rocks, instead of the longer course followed by the Ohio streams over continuous sandstones. The advantage given by the greater extent of soft beds is indicated by the great breadth of the existing valleys in the central district compared with the less breadth of those in the plateau to the west. Consider the effect of this advantage at the time of the Jurassic elevation. As the streams on the eastern slope of the Nittany divide had the shortest and steepest courses to the sea, they deepened their valleys faster than those on the west and acquired drainage area from them; hence we find reason for the drainage of the entire Nittany and Bedford district by the Atlantic streams at present. Various branches of what are now the Alleghany and Monongahela originally rose on the western slope of the dividing range. These probably reached much farther east in pre-Permian time, but had their headwaters turned another way by the growth of the great anticlinal divide; but the smaller anticlines of Laurel ridge and Negro mountain farther west do not seem to have been strong enough to form a divide, for the rivers still traverse them. Now as the headwaters of the Juniata breached the eastern slope of the Nittany-Bedford range and pushed the divide westward, they at last gained possession of the Siluro-Devonian monocline on its western slope; but beyond this it has not been possible for them yet to go. As the streams cut down deeper and encountered the Medina anticline near the core of the ridge, they sawed a passage through it; the Cambrian beds were discovered below and a valley was opened on them as the Medina cover wore away. The most important point about this is that we find in it an adequate explanation of the opposite location of water-gaps in pairs, such as characterize the branches of the Juniata below Tyrone and again below Bedford. This opposite location has been held to indicate an antecedent origin of the river that passes through the gaps, while gaps formed by self-developed streams are not thought to present such correspondence (Hilber). Yet this special case of