Halemaumau Lava La~ke. 343 



has also observed that when, by a combination of circumstances, 

 a downflow of lava is produced at a location where there are 

 generally no large gas bubbles to cause jetting or spattering, 

 the downflow proceeds with perfect quietness. This was nota- 

 bly the case for a short time on the eleventh of August, 1911, 

 when, by some means, a downflow was started along a line 

 stretching entirely across the lake not far from its western end. 

 The crust was sufficiently flexible to be bent downward with- 

 out breaking and the pull of the vertically descending heavy 

 sheets was sufficient to draw the surface layers inward from 

 either side and down into this semi-liquid fissure, the syphon- 

 like action continuing for several hours without any jetting 

 whatever. It is clear that the vertically descending crust must 

 be melted at no great depth, but the fused material also will be 

 denser than the average gas-vesiculated lake substance and the 

 action is the same. This is also seen at some spatter grottoes 

 where the radiant heat is sufficient to partially melt the crust 

 on the in- and down-flowing surface lava even before it has 

 quite entered the grotto, yet the heavy material sinks and 

 maintains the flow, although the crust, as such, no longer exists. 



As regards the ascent of gas in those localities where the 

 lava is going down, it is not necessary to believe that the bub- 

 bles always struggle upward through the descending lava — 

 they may rise to one side of the downflow or between it and 

 the wall of the basin. Large gas bubbles, by their extreme 

 buoyancy, rise very rapidly through the very mobile lake 

 material without carrying the lava with them to any extent 

 and they may even rise through a slowly descending current 

 of the liquid. In this case the opposing forces are often in 

 equilibrium, the surface lava surging toward the grotto, then 

 stopping and then again advancing, like the movement of a 

 cinematograph film. It is the very small gas bubbles which, 

 themselves incapable of rapid ascent, may, by their vesicula- 

 tion of the lava, cause it to rise " en masse " and this mode of 

 action plays its part in the scheme of circulation. 



Before taking a comprehensive view of this we must con- 

 sider a detail which, during the early summer of 1911, formed 

 the most extreme manifestation of relative movement in the 

 lava of the lake and which constituted, while it lasted, the 

 principal element of the circulation. A large cavity had been 

 eroded in the mass of a floating island of rock, extending from 

 the bottom vertically upward to a point under the saddle 

 between the two principal portions of the island. This cavity 

 was open at the lake surface on the north (fig. 3), and also on 

 the east side, in the form of a grotto within which the lava 

 could be seen, free from crust, and in communication with the 

 outside material. Suddenly a fountain would play within the 



