When subsidence again set in the strip of alluvium was soon submerged and the two seas again met; but in all probability for a long time the Strait was only a narrow one, over which animals could easily swim. Then tidal scour, deeper submergence, and the action of the waves did the rest, so that ever since that time the Strait of Dover has been getting steadily wider and wider, and also deeper. Its bottom is to a large extent composed of bare chalk with patches of gravel; and the movement of this gravel during storms, combined with the action of boring molluscs must slowly eat away the chalk far below ordinary wave-action. The above explanation is needed, for it will not do to take existing soundings, and say that all the sea-bottom below a certain level, corresponding with a particular submerged forest, was then sea and all above was then land. This is an easy way of reconstructing the physical geography; but it may be a very misleading one. A little consideration will show that whilst in large areas sand-banks have accumulated to a great thickness, in other areas, of which we know the Strait of Dover is one and the Dogger Bank a second, there has been much submarine erosion, which is still going on. In neither case is it safe entirely to reconstruct the ancient contours from the present-day soundings. Even such a gigantic feature as the continental platform, which ceases suddenly at a depth of 100 fathoms, is in all probability in the main a feature formed by the deposition of sediment during long ages. Its outer edge marks, not the limit of some ancient continent, but the limiting depth at which gentle wave-action has been felt, and beyond which the sediment cannot be carried. After this necessary digression we must return to our study of the actual evidence for such changes of sea-level in the English Channel. It has been pointed out already that for this purpose the present depth below sea-level of the rocky floor of the Strait cannot in itself be accepted as sufficient evidence. Nor can the depth at which rock was met with under the Goodwin Sands; though here a cylinder was sunk 75 feet before it reached the chalk. Unfortunately no record of the strata passed through seems to have been preserved, though it is perhaps implied that nothing but sea-sand was penetrated. Romney Marsh is a wide alluvial flat occupying a silted-up bay, the floor of which in places lies at least 70 feet below sea-level. There are here unfortunately no extensive excavations for docks, and all we can say is that the few borings which have penetrated the alluvial strata prove the existence of a slightly undulating rock-surface below. In short Romney Marsh appears to be a submerged flat-bottomed open valley, like that which we have already seen underlies the marsh deposits of the Fenland.