J. D. Dana — History of the Changes in Kilauea. 287 



The intervals of cooling and refusion vary in length from a 

 few seconds to an hour and longer. The crust of a lava-lake 

 is often only the thin, easily fusible glassy scum, but thicker 

 crusts also yielded to the heat. 



In the case of rapid transitions, the cooling and refusion 

 may be due to the loss of heat by the expansion in the process 

 of vesiculation ; and if the vesiculation takes place intermit- 

 tently for any reason (as from oscillating movement in the lava 

 column, or other condition) it would occasion the alternations 

 between the fused and crusted state. But for the crustings at 

 longer intervals deeper movements may be concerned, and 

 more study is needed before they are fully understood. 



(2.) The disappearance of floating islands is another effect 

 of heat and chiefly of refusion. For in some cases the islands 

 have after a while — a year or more — disappeared. 



(3.) The destruction of debris-cones in Halema'uma'u is de- 

 pendent on undermining by the active lavas and vapors be- 

 neath ; and, in one case, the destruction was probably com- 

 pleted before a period of eruption (xxxiv, 88). 



The debris-cone, 1500 to 2000 feet broad at base, which now 

 occupies the center of the Halema'uma'u basin is already in 

 process of dissolution. It made no increase in height during 

 the summer of 1887, but, instead, rather lost ground through 

 the changes going on. This fact was obvious in August last ; 

 for the east wall, a single continuous ridge in October, 1886, 

 had become divided into two ridges, and dense vapors rose 

 from the interval between, with sounds of splashing lavas that 

 were evidence of an active lava-lake below A photograph taken 

 in October, 1886, copied on Plate IV, shows the condition of this 

 side of the cone at that time when it had reached its maximum 

 height ; and Plate V, from a photograph taken nearly a year 

 later, in September last (a month after I left the crater), ex- 

 hibits the condition above described. Both views were taken 

 looking westward, and the foreground in each is the bottom of 

 Halema'uma'u on the east side of the cone. In Plate IY dense 

 vapors may be seen issuing from a large aperture near its mid- 

 dle, and other vapors from a lower place to the right (north). 

 In Plate Y, the vapors escape copiously all the way between 

 these two places and far southward, showing the subdivision 

 nearly completed. A photograph taken in the spring of 1887 

 exhibits an intermediate stage in the process of division. A 

 letter of Jan. 2 of the current year, from Mr. J. H. Maby, 

 proprietor of the Yolcano House, says that the bottom of the 

 Halema'uma'u basin on the east side (that shown in the plates) 

 has risen much and is now nearly on a level with the upper 

 surface between it and " New Lake" and a lava-lake has opened 

 in it, so that the fires and the overflows are visible from the 



