required to form a regular system of deep and wide valleys; for time, which they are so unwilling to assume, is essential to the operation. Time must be allowed in the intervals between distinct convulsions for running water to clear away the ruins caused by landslips, otherwise the fallen masses will serve as buttresses, and prevent the succeeding earthquake from exerting its full power. The sides of the valley must be again cut away by the stream and made to form precipices and overhanging cliffs before the next shock can take effect.

If a single convulsion of extreme violence should agitate at once an entire hydrographical basin, or if the shocks should follow each other too rapidly, the previously existing valleys would be annihilated instead of being modified and enlarged. Every stream might in that case be compelled to begin its operations anew, and to shape out new channels, instead of continuing to deepen and widen those already excavated. But if the subterranean movements have been intermittent, and if sufficient periods have always intervened between the severer shocks to allow the drainage of the country to be nearly restored to its original state, then are both the kind and degree of force supplied by which running water may hollow out valleys of any depth or

size consistent with the elevation above the sea which the districts drained by them may have attained.'

Here, in direct contradiction to his own submarine system of forming valleys, Lyell accords that function to rivers. But it appears that the operation is much facilitated if from time to time the valleys are choked up by earthquakes; or if the previously existing valleys should be annihilated, and every stream be compelled to begin its operations anew, and to shape out new channels, instead of continuing to deepen and widen those already excavated.' Also that the action must of necessity be most powerful while the land is rising or sinking;' why, he neglects to inform us. Rivers can only cut valleys down to the level of the sea, and if the land is then raised the rivers may then deepen the valleys. This would give the rivers more to do, if that is to make them more powerful.' But how would a sinking make them more powerful? With submission, neither Lyell nor earthquakes are likely to throw light on the subject; and 'if we consider the question' as dependent on 'the agency of earthquakes,' the formation of valleys by running water can never be understood.' However, in the proper districts of earthquakes, in volcanic regions, Lyell is right

Lyell's sub- to call in running water, or any agent to form marine theory his valleys. His own submarine theory will not

does not apply to volcanic islands.

apply to volcanic districts or islands, for their surfaces have been ejected over them when they were already above the sea; and they have been coated and recoated thousands of times, by floods of melted rock when they had long been sub dio. Yet the gradients of the river valleys and dry valleys, and the whole form of the ridges and furrows of the entire surface drainage of a volcanic region (say Madeira), are so precisely the same as those of any other mountainous district, that no eye can glance over the two, and doubt for an instant that the same cause caused the form of the drainage of both.

In Tenerife I have seen numberless cases of volcanic dikes which projected from the opposite sides of valleys, and ran across the streams in their beds at all angles; and sometimes four or five dikes close to each other, and not parallel. The direction of the water-courses is never altered by these dikes, as is asserted by Keith Johnston. A dike simply retards the deepening of the ravine where the dike is, and that part of the ravine which is above it. The part below it will be deepened precisely as if the dike did not exist. But the part above must wait for the lowering of the dike, and the deepening will then

6

go on backward; that is, from the dike towards the hill. When a dike runs diagonally across the main ravine, it of course does the same with the minor ridges and furrows, with which rain has scored the entire surface of the sides of each ravine. These minor ravines, however, run right up and straight down,' regardless of the dike or Keith Johnston. This may be well seen from the acqueduct in the ravine north of Santa Cruz. We have only to look at a geological map, to see the incorrectness of Keith Johnston's remark, that rivers change the direction of their courses with every change of geological strata. These valleys of Tenerife are almost all dry, except in rain. The beds of the upper parts are sheer rock, the middle parts wear the appearance of torrents of stones, the lower beds are alluvial plains of stones, and opposite the mouths of these valleys are very commonly deltas and bars of stones. Behind these bars, after each rain, large deposits of earth and sand are formed, which the people diligently collect. Where permanent streams exist, they are usually lost at a considerable distance above the mouths of the valleys; that is, except in rains, they percolate to the sea, beneath the alluvial plains, deltas, and bars. From the sides, hundreds or thousands of torrents of stones fall into these rivers of stones. Some

burst.

times these lateral shoots have formed barriers across the valley, behind which large beds of boulders and earth have accumulated, again to be thrust out by heavy longitudinal floods. In this case, I imagine that the overwhelming force Flood of out of what the French call the débâcle comes into play. I shall translate it by the flood of outburst.' This takes place usually on the breaking up of the ice in all northern rivers or of dams formed by avalanches; and in such cases credidimus Jovem regnare. But without frost or snow, in all steep hollows and mountain torrents, the flood of outburst' is of constant occurrence. The paving blocks of the beds of torrents catch floating rubbish, large and small, and form temporary dams and lakes. Away goes a block, which deepens the outlet some feet. Rock after rock follows, dashing itself and its neighbours below into pieces a tenth part of their former size. A day or two after, we decide that the brawling streamlet, which seems to ask permission to steal through these fragments, never did, nor ever will, have power to move one of them. This flood of outburst' is sometimes taken advantage of by man. On precipitous mountain sides, where whole trees could by no means be got out, they are cut and split into logs for firewood, and pitched into the streamlet

6