Jaggar — Yolcanologic Investigations at Kilanea. 171 



islands, mere crusted summits of irregularities on the main 

 lava column, were eventually buried under the talus. The 

 foaming gaseous member of the column appeared towards the 

 end as a lava spring, trickling down the western talus like a 

 mountain torrent (tig. 2, loc. numbered o), and within a few 

 days this and other springs cascading into the lake restored to 

 it its central position in the pit (iig. 2, loc. numbered 6 west 

 corner lake June 6) surrounded by debris slopes (fig. 4<2). 



There is reason to believe that during the progress of a com- 

 plete eruptive cycle of Kilauea between two repose periods, 

 like the interval 1907-1913, the adjustment of rising gaseous 

 lava to sinking viscous lava becomes most perfect at the cul- 

 mination of the eruption, when effervescence is general and 

 uniform, surface heat is distributed and at a maximum, and 

 border benches are absent. Such a culmination was reached 

 in December-January, 1911-12, when the lava lake extended 

 from wall to wall of the pit, and the subsidence thereafter 

 for eighteen months involved a very rapid dwindling in the 

 size of the liquid phase (fig. 2, heavy dotted outline) and a 

 final retirement of the entire column to the depths amid the 

 wreckage of the solidified portion. 



Evidences of Shallowness of Liquid Lava Lake; the 

 Bench Magma, 

 Lava Islands. 



The writer has studied the formation of islands in the Tlale- 

 maumau lava lake repeatedly during the last five years, and has 

 determined that they are capable of extraordinary shifting of 

 position both vertically and laterally in the course of twenty- 

 four hours. He has never, however, seen any evidence that 

 they floated as independently buoyed objects in the liquid part 

 of the lava column, after the fashion of an iceberg in water. 

 That they could not be so buoyed is shown by the facts that 

 they do not equally rise and fall with the risings and fallings 

 of the lake, that they tend to acquire and retain fixed positions, 

 by survey, while the convection currents stream about them, 

 and that movements of tilt in the island escarpments have been 

 repeatedly diagnosed, the tilt according in direction with a 

 place of overloading of the border bench with lava flows. 

 They are thus integral with the shore and bottom of the lake. 

 Their horizontal shiftings are rare, when they njove pivotally 

 as crust blocks of the bench magma. 



Also these islands pass by gradations into craggy scarps in 

 or partly in the border bench where all stages of their forma- 

 tion have been photographed in sequence. Here the relation 

 to weighting by overflow has been repeatedly demonstrated. 

 The effect is stnkingly like the supposed relation of sedimenta- 

 tion to isostatic uplift. On this small scale, moreover, the 



