divides are among the most stable of the geographical features of the land, they are continually changing. This shifting of the position of the lines of parting between opposing drainage slopes is usually exceedingly slow, but under certain conditions may be a comparatively rapid process, as will be seen by reverting to the discussion of the origin of subsequent streams, and the manner in which they extend their channel by headward cutting so as to capture rival streams and divert their waters. This process

leads to great and even rapid changes in the positions of divides.

Another illustration of the migration of a divide may be of interest to the reader. In the Catskill Mountains we have a table-land sloping gently westward, but presenting a bold escarpment about one thousand five hundred feet high, facing the Hudson. A portion of this plateau and its eastward-facing escarpment is shown on the map forming Plate XIII. The rocks forming the plateau are sedimentary beds of hard sandstone and soft shale, which dip gently westward and present their broken edges in the escarpment. As has been shown by Darton,' the regularity of the precipitous eastern border of the plateau was broken by alcoves and recesses, inherited from a preceding geographical cycle, and in these embayments eastward-flowing streams originated. Of these the Kaaters Kill and Plaaters Kill are the best examples. Streams also came into existence on the gentle western slope of the plateau and flowed westward; the head

IN. H. Darton, "Examples of Stream-Robbing in the Catskill Mountains," in Bulletin of the Geological Society of America, vol. vii., pp. 505-507, Plate xxiii., 1896.

branches of one of these, Schoharie Creek, are shown on the accompanying map. The conditions are thus especially favourable for the processes of stream capture and the migration of a divide, already described.

It is evident from an inspection of the map, that the branches of Schoharie Creek were formerly longer than now, and carried away the surplus water from the plateau even to the edge of its eastern escarpment, but the Kaaters Kill and Plaaters Kill have been enabled to extend their head branches so as to capture a considerable portion of the previous western drainage. The divide has migrated westward, and some of the former branches of Schoharie Creek have been diverted. This history is brought out so graphically on the accompanying map that further explanation seems unnecessary.

One result of the process just considered is shown by the direction of flow of the higher branches of the capturing streams. Normally the branches of a stream join the main trunk at an acute angle, the flow in the branch and in the trunk near their place of union being in the same general direction. But in the case of the capturing streams instanced above, their head branches come in at more than a right angle; the captured branches maintain the direction. they had when flowing to Schoharie Creek, and in general flow westward, while the trunk streams to which they are now tributary flow eastward. Such an abnormal arrangement of the branches of a drainage tree in any region should at once suggest that a recent capture has been made, but yet rock texture and other conditions might produce a similar result.

The process of stream capture, so admirably illustrated in the Catskills, furnishes an example of one method by which fishes, mollusks, etc., might be enabled to migrate from one side of a mountain range to the other. The opening of gaps in the crest of a high ridge or mountain range would also facilitate the distribution of plants and animals not dependent directly on streams for their means of travel. Important influences even on the migration of peoples may be traced to the same cause.'

1 Books of reference:

HUMPHREYS and ABBOT. Physics and Hydraulics of the Mississippi. War Department, Washington, D. C., 1861.

THOMAS RUSSELL. Meteorology. Macmillan & Co., 1895. (Chapters IX., "Rivers and Floods," and X., "River-Stage Predictions.")

PARK MORRILL. Floods of the Mississippi River. Weather Bureau, Washington, D. C., 1897.

CHAPTER VIII

SOME OF THE CHARACTERISTICS OF AMERICAN RIVERS

HE many details that have occupied the reader's atten

THE

tion in the preceding chapter have perhaps diverted attention from certain general conclusions pertaining to the lives of streams. A brief review of the leading characteristics of a few of the rivers of America will possibly correct this tendency and at the same time afford an opportunity to apply some of the principles stated, perhaps too empirically, in what has gone before.

The initial slopes of large rivers must evidently be determined by the slope of the land due to upheaval. In many, and probably most, instances, however, the surface slopes that gave direction to the youthful streams have been deformed by movements in the rocks of the nature of a tilting of the land over broad areas; again, the rocks have been folded, or broken and one side of the fracture upraised above the opposite side, so as to affect the surface drainage. While these changes were in progress in many instances, the streams have maintained their positions or right of way by deepening their channels as fast as the rocks were raised, or by filling in the depressions due to subsidence; but in other instances the streams have been reversed or given other di

rections, owing to the modifications in conditions referred to. The present courses of even the larger rivers do not, therefore, in themselves, necessarily record the original slope of the land.

Throughout the lives of streams they have the power of extending their branches in a manner analogous to the growth of a tree by the lengthening of its terminal twigs. This process, as we have seen, leads to rivalry between neighbouring streams, and the shifting or migration of the boundary line between adjacent drainage areas. Climatic changes may also favour the extension of certain drainage areas and the diminution of others. For these and still other reasons, the boundaries of the original slopes which gave the large rivers their general directions have been greatly modified and in some instances rendered indeterminate; yet when the general changes that land areas pass through and the laws of stream development are known, much of the history of a river system can be deciphered.

Some of the modifications that have taken place in the various drainage areas of North America, due to changes in the elevation of the land, variation of climate, normal stream development, etc., can be recognised even in a general view of the present distribution of the streams. Individual rivers furnish too small a unit with which to measure the greater slopes produced in the surface of North America by upheaval, and a better idea of the character the surface of the continent would present, had there been no erosion, can be had by considering the main drainage areas. It must be remembered, however, that the upheavals which established