Jaggar — YolcavologiG Investigations at Kilauea. 169 



Evidences of Convection ; the Duplex Lava Column. 



Dalj terms "two-phase convection '' the meclianism whereby 

 deep gases rise as bubbles witli increased rapidity toward tli'e 

 surface of a lava column and the liquid, released by surface 

 collapse of the bubbles and loss of the gas, tends"^ to sink 

 because it is heavier than the rising froth. 



Both he and Ferret made diagram sketches^ shoAving the lava 

 basin in the Kilauea pit to be shallow and saucer-shaped with 

 one conduit beneath smaller than the visible pool. If conduit 

 remains small and saucer remains shallow, the pool being 

 bordered by a bench of overflow witliin the pit, how is it that 

 the bottom of the saucer is built up while the lava surface rises 

 600 feet in six months ? (See section, fig. 2.) 



As will be shown below, there is every evidence that just 

 this relation, but with several tubular conduits, existed from 

 June to December, 1916, and that the lava lake remained con- 

 tinuously shallow. The bottom of the lake builds up by 

 accretion of the relatively cooled and denser sinking surface 

 layers of the convection, step by step with the building up of 

 the shore bench or " floor " by the spasmodic overflow which 

 takes place from time to time owing to the inflation of the 

 liquid part by the rising gas-bubble currents of the convection. 

 The obvious surface evidence of such convection lies in the 

 persistence of deep wells at certain fixed sites from which 

 vesiculated lava springs up and streams outward across the 

 surface of the lake, while on or near the opposite shores the 

 solidified crusts and skins founder in the grottoes and the 

 fountains. There have been such source wells or spring holes 

 at the west and north sides of Halemaumau for many years 

 and the dominant convectional streaming has been away from 

 them. That they are definite wells in the fresh bench lava has 

 been repeatedly demonstrated at times of faster lowering of the 

 level of the lake, when they w^ere revealed as local small pits. 

 The inflow pits frequently persist for long periods as separate 

 ponds of lava (fig. ^d). 



*Loc. cit., Daly, p. 77 ; Ferret, p. 345. 



Fig. 2. Map and diagrammatic section of Halemaumau, Jan. 12, 1917. 

 Lava lake in black, crusted conduit ponds shaded, overflow benches diag- 

 onal lines, raised crags contoured. Coarse dotted outline lava lake of Feb. 

 18, 1912. Fine dotted outline June 23, 1916. Rectangle (5) site of lava 

 spring of June 5, 1916. Rectangle (6) west corner of pool June 6, 1916. 

 Note that N.W. corner has been conduit source on all these dates. Slight 

 slope lake surface from conduits W. to overflow bench E. Bench magma 

 elevated on conduit side W.S.W., subsided on sinkhole side E.N. E. Sec- 

 tion without vertical exaggeration, lower profile shows simple rising pool of 

 June 23, 1916. Shoal shown in lake bottom, upper profile, was revealed by 

 subsidence February, 1917. Depths from soundings and subsidence records. 

 Note progressive shoaling from W. to E. Diagrammatic sinkhole E. shows 

 ridge of accretion on lake bottom margin which produces cascade ledge when 

 subsidence takes place. Surveys with transit by T, A. Jaggar. Bench marks 

 (B.M.) U. S. Geological Survey, trig, stations Hawaiian Volcano Obser\-atory. 

 Meridian approximately 155° 17' 8" W., lat. 19° 24' 33" N.; 10-foot contours 

 above lake as datum plane. 



