When a little conelet formed on the northern or western floor platform inside Halemaumau, its slope around a splashing and fountaining crack would make a fourth innermost dome a few feet across in the series of progressively smaller cone-in-cone structures from the outer rim of the big mountain inward to the Halemaumau centers of eruption. We saw such a conelet cave in just where I had stood and tested a flame the day before. Quietly the cone collapsed into a fountaining well of boiling lava beneath. The ring-in-ring conception must be held in mind with regard to any volcano, for one thing which we discovered is that cones are not only built up and collapsed but they are also swollen up by internal percolation of cracks and expansion of the hot stuff. This tumefying, or swelling, is concerned with the experiment now to be described. Even after Perret described his “floating island” of 1911 and I saw the triangular islands appearing like shoals in a mud flat and gradually rising into crags in 1916 and 1917, I remained incredulous of the possibility of a basaltic island floating. When solid lava cracked off in pieces from inner cliffs around the lava lakes, the fragments immediately sank. Furthermore, when solid crusts formed on top of the foaming and streaming slag, the shells, when they got thick enough, cracked up, tilted up, and slid down and foundered in the melt beneath. It was obvious that lava rock is heavier than lava foam. Hence as an island is a rock, it would not float. This raised several questions. Where was the bottom of the lava lake on which it rested? Did the lava lake have a bottom, and if so how far down was the bottom when the same lake rose 600 feet in Halemaumau pit between June and December of 1916? In other words, was the lake 600 feet deep in December? What would be the answer at any time if a stiff iron pipe were thrust down vertically into the liquid lake as a sounding rod? No one had ever raised the question. Cross-section drawings had always depicted the liquid as extending downward indefinitely within a vertical tube. When the lake became accessible in 1917, it seemed to me that a long steel pipe might be shoved over the border rampart, end on, and allowed to bend and sink, or to strike bottom. If the pipe could be recovered by dragging it back, fusible samples of known melting point might show the temperature of the depths.