Jaggar — Yolcanologic Investigations at Kilauea. 201 



remarkable. Moreover the linings and stalactites of such 

 grottoes indicate refusion, and oxidation of the iron to the 

 ferric condition. The variability in fountain action at grottoes 

 may be from zero, when all is quiet and crnsted, to tremendous 

 spraying and blowing Avith towering banners of flame (tig. 20) 

 and an inrnsh which amounts to a cascade from the lake to a 

 cauldron within the cavernous dome. Times of extreme quiet 

 are characteristic of rising, times of extreme activity are char- 

 acteristic of sinking. This last antithesis is true also of all 

 other fountains. 



The grotto fountain appears to be a true case of a stack 

 or furnace built at the margin of the lava lake remote from 

 inlet conduits. The furnace is fed by relatively continuous 

 downflow of lava vesiculated with combustible gases, and 

 supplied with oxygen by continuous downsucking with this 

 magma of large quantities of air. The reactions are quite 

 as in the other cases cited, but the recess constructed by spatter 

 around the margins permits a concentration of the blast at one 

 point. It is quite probable also that chemical activity of gas 

 with lava, reheating the substance of the grotto walls, adds to 

 the heat supply in these specialized edifices. The opening of 

 a confined glowing cavern toward the air over the lake, with 

 continuous indraught of that air induced by the downflow, 

 brings about very complete combustion of the less active gases 

 such as sulphur vapor at the orifice of the stack, and prevents 

 the condensation of free sulphur which elsewhere appears as a 

 white smoke or kind of gaseous emulsion when the vapor rises 

 gradually cooled through border cracks in the bench lava. 



Construction upon Lake Bottom. 



The great quantities of air-filled crust sucked under the lake 

 mai-gins at the border grottoes probably descend to the bottom 

 with much of their oxygen unconsumed and with their relatively 

 cool and heavy substance incessantly adding, by a backward or 

 eddying subsurface circulation, to layers of more or less com- 

 minuted material on the bottom of the lake increasingly viscous 

 downward (fig. 2). This subsurface crustal material is what 

 builds up shoals in those portions of the shallow lake least 

 eroded by streaming and least heated by oxidation. Com- 

 minution, by destroying the buoying effect of vesicles, accelerates 

 sinking. 



We should thus expect shoals to be built out at the sides of 

 inflowing fresh lava from the conduits, where fountains are 

 absent, and where quiet eddies would permit construction 

 upward from the bottom to join with the thickening crusts at 

 the surface. There has been repeatedly demonstrated a ten- 

 dency for the lake to grow from the western conduits eastward 



