ley is widened. In all those cases the valley-cliffs and slopes on the one side have the same general aspect as those on the other. But when a river cuts its way along the strike of moderately inclined strata its! course assumes a different form. On the one side cliffs, and on the other, where the rocks dip into the valley, slopes tend to be developed. Again, as a river in its journey across a wide tract will necessarily traverse rocks and rock-structures of very different degrees of durability, its valley will widen or contract according as the rocks are more or less readily eroded. In one place the river meanders through a plain bordered by gentle slopes, in another place it hurries through a narrow and sometimes approximately straight or gently winding gorge, the latter often indicating the site of former cascades, waterfalls, and rapids. In a word, every change in the form and character of a valley of erosion is determined by the nature of the rocks and rock-arrangements with which the river and its assistant agents have to deal.

Waterfalls frequently mark the outcrops of relatively hard rock-masses. The Falls of Niagara, for example, owe their origin to the intercalation of a bed of hard limestone amongst more yielding strata, which have a gentle dip upstream. By the constant wash of the water the soft shales underlying the limestone are gradually removed, and the overlying mass, losing its support, breaks away from time to time. along its joint-planes. In this manner the Falls have slowly retreated from Queenstown, and the gorge of

Niagara has been formed. The Falls of Clyde are due to a precisely similar geological structure, and many ravines and gorges in the valleys of our lowlands have originated in the same way as the gorge of Niagara.

The occurrence of great waterfalls in a long-established hydrographic system is somewhat anomalous, and leads at once to the suspicion that the drainagesystem has been interfered with. Waterfalls cannot be of any great age. Sooner or later they must be cut back and replaced by ravines or gorges. Their presence, therefore, shows either that the valleys in which they occur are throughout of recent age, and that the rivers have not yet had time to reduce such irregularities, or that the drainage-system, if long established, has since been disturbed by some other agent than running water. In deformation-mountains of recent age we naturally expect to meet with cascades and waterfalls, for the streams and rivers of such a region are relatively young. They have only, as it were, commenced the work of erosion. But plains and plateaux of erosion which have existed for ages as dry land, and in which a complete hydrographic system has been long established, should show no great waterfalls. Yet we find cascades and waterfalls more or less abundantly developed in all the plains and plateaux of Northern Europe and the corresponding latitudes of North America; and most of these lands are of very great antiquity, their main lines of drainage having been established for a long

time. Obviously the hydrographic systems have been disturbed, and the disturbing element has been glacial action. During the Ice Age the long-established preglacial contours were greatly modified. Frequently, indeed, the minor valleys in plateaux and plains were completely obliterated, while even the main valleys were often choked with débris. When glacial conditions passed away, and streams and rivers again flowed over the land, they could not always follow the old lines of drainage continuously, but were again and again compelled to leave those and to cut out new courses in whole or in part. Hence the frequent occurrence of cascades and waterfalls in formerly glaciated lands.

Another cause for the existence of waterfalls in long-established hydrographic systems must be sought for in crustal disturbances. In general, deformations of the crust would seem to have been very gradually brought about, so gradually, indeed, that they have often had little or no influence upon the courses of great rivers. Anticlines slowly developing across a river-valley have been sawn through by the river as fast as they arose. Dislocations, in like manner, would seem to have been very slowly developed. Frequently these have traversed a river-valley without in any way disturbing the drainage, the rate of erosion having been equal to that of the displacement. On the other hand, we know that faulting or dislocation may sometimes be rather suddenly effected. Thus, a large fault crossing a river-valley

and having its downthrow in the direction in which the river is flowing would certainly produce a waterfall. Such, indeed, would appear to be the origin of the great falls of the Zambesi.

A volume might be written on the many appearances presented by subsequent or erosion valleys, but it is beyond our purpose to enter into further details. It is enough to recognise the fact that the great majority of river-valleys have been excavated by the rivers themselves. Even the most recent tectonic valleys have often been profoundly modified by subsequent erosion. In all regions, whatever their character may be, whether plateaux and plains of erosion or accumulation, or true mountains of elevation, the streams and rivers are constantly striving to reduce the land to their base-level. The main directions or lines of erosion are early established; but in the course of time many modifications arise, owing to the work of the streams and rivers and of epigene agents generally. At first, it may be, the rivers descend by a succession of steps or by alternate steep and more gentle declivities. Cascades, waterfalls, and rapids, and here and there barrier-lakes, may abound. But eventually the irregularities are removed and a true curve of erosion is produced. Each river has then its relatively short torrent-track, and its longer valleyand plain-tracks. As erosion proceeds the plain-track continues to encroach inland upon the valley-track, while the latter eats back into the torrent-track. the same time the entire surface of the land is being

At

continuously reduced, until at last hills and mountains gradually disappear, and the whole region is replaced by a plain. The cycle of erosion, however, is not often allowed to proceed without interruption. Sometimes an upward movement increases the gradients, and so in time the revived rivers deepen their courses, and "valley within valley" appears. Or the whole region may become subject to glaciation, during which the preglacial drainage-system may be considerably modified by erosion here and accumulation there. When at last the ice-covering vanishes, lakes, rapids, cascades, and waterfalls diversify the watercourses. But the removal of these features is only a matter of time. By and by all the direct effects of glaciation must disappear. Again, long before a cycle of aqueous erosion is completed the land may be submerged and more or less deeply covered under new accumulations. Should re-elevation eventually ensue, a new hydrographic system will then come into existence, but this may not coincide in any part with that of the old buried land-surface.

In regions where soluble rocks, such as limestone, abound, the hydrographic system usually presents strong contrasts to those we have just been considering. Much of the rainfall finds its way below ground, where a complex series of channels is gradually licked out, until eventually the whole drainage may become subterranean. Usually, however, the drainage is partly superficial and partly underground, the rivers flowing for longer or shorter distances in ordinary