But, more probably, the lower wall leans over more and more towards the top, and so forms as it were a part of the roof.

It is possible that, as the wall has grown, each successive annual layer has projected farther and farther, till at last some year very favourable to the increase of ice has carried the projection for that year nearly to the opposite stones, and then an unfavourable year or two would form the foot of the upper wall.

This seems more probable, from the loose constitution of the floor at the point where it joins the stones, as if it were there only made up of drift and debris, while the part of the floor nearer the foot of the wall is solid ice.

It has been suggested to me that possibly water accumulates in the time of greatest thaw to a very large extent in the lower parts of the cave, and the ice-floor is formed where the frost first takes hold of this water.

But the slope of the ice-floor is against this theory, to a certain extent; and the amount of water necessary to fill the cavity would be so enormous, that it is contrary to all experience to imagine such a collection, especially as the cave showed no signs of present thaw.