time they must collapse; yet no change is ever visible, no signs of recent action can at any point be seen. The rarely occurring great flood serves but to remove one film of sand from the floor and to leave another in its place as the waters subside. So slow is the undermining action that no eye can ever detect a change though the waters rise ever so high. Yet this channel is but one of five distinct levels which the river has occupied from time to time, until it has found in turn a lower course, leaving its sands as a record upon each, here and there sealed down beneath a mass of stalagmite. What untold ages have elapsed since first the river flowed through these upper channels! [Illustration: PLAN AND SECTION OF WOOKEY HOLE CAVERN.] But an examination of the top of the Mendips points to a vaster antiquity still. The published horizontal section No. 17 of the Geological Survey gives an excellent idea of the plateau of Mendip, which stretches from immediately north of Wells to the neighbourhood of Compton Martin. This plane of denudation would never have been reached save by the long-continued action of subterraneous streams, an assumption supported by the existence of the great depression crossed by the road from Wells to Priddy. That depression of nearly 100 feet in depth and several miles in length, hollowed in the hard Carboniferous Limestone, here dotted with every known type of swallet or swallow hole, has been obviously formed by the slow action of swallet streams prolonged through vast periods of time. Every atom of the millions of tons of solid rock represented by this depression has been borne down the course of the subterranean Axe. Tributary to this depression a little valley has been eroded across the Old Red Sandstone anticline immediately to the north, and in it are deposited masses of Dolomitic Conglomerate, the component pebbles of which were derived from the surrounding rocks. The same valley existed, therefore, in pre-Triassic time, and as there was obviously no other outlet for its water, the cavities into which it flowed--that is to say, the swallets and subterranean channels--must have existed also, and are therefore pre-Triassic in date. Though at first sight this appears impossible, inasmuch as the known course of the resulting Axe River is through Triassic Conglomerate, I propose to show that such a conclusion is necessary and inevitable. Long ago I was struck with the fact that at Wookey Hole the Triassic Conglomerate attains an abnormal thickness, and measurements have shown that at the far end of the cavern there is certainly a thickness of over 350 feet of this rock. As there is no sign of any approach to the Limestone against which it must abut, nor any change in the character of the Conglomerate itself at this point, I think that we may fairly conclude that the total thickness of it must be at least 500 feet. Now this is a vast deposit, far exceeding