even chalk. Thus plains of Tertiary deposits capped by gravel, under the action of rain or rivers develop into gravel-capped plateaus and hills, which fall abruptly into open flat-bottomed valleys. The denudation takes place at the edge, where the gravel rests on the Tertiary strata and numerous springs are given out; there is scarcely any denudation in the gravel flat, and unless the height of the land is considerable there is no great amount of denudation in the flat bottom of the valley.
Thus there is a tendency for the valley to widen out on every side, wherever the gravel rests on impervious or soft strata. But where the gravel plunges below the water-level, as it did at the entrance to each of these harbours, the valley narrowed, for there were no landslips, the drainage was subterranean, and the stream could not readily remove the large flints.
The widening of the valleys, where they were cut in soft strata, led to the development of small lateral valleys to the right and left, leaving only narrow divides between their head waters and those of the next valley. When the land sank these divides were flooded, and so were developed the shallow cross connexions, much as we now see them.
In order that it may not be imagined that this reconstruction is merely hypothetical, it will be as well to give some evidence that such an elevation
