96 



Kilaiica and Manna Loa. 



on an ancient flow, one of which was seven feet high, eighteen inches external diame- 

 ter, and with a bore of eight inches. It was brittle, and on breaking it off, I found the 

 hole was six feet deej?, making its whole length thirteen feet. Others of the same 

 height were near b}-, and their sides were always thicker towards the sonrce of the flow, 

 Externally they are rough like aa, but the top was smooth and sometimes proje6led 

 like an umbrella. Wliere several were in close proximity a slab of lava was supported 



Fig. 64. CRATER MAK,\OPUHI IN PUNA. 



like a roof on columns (see Fig. 43). The lining of the tube was smootli and much 

 more compact and vitreous than the exterior. The trees which served as cores, if not 

 burned have entirely decaj-ed, and were mostly tree ferns, although I think that I 

 detected some ohias. I followed the stream down some distance to learn the cause of 

 its subsidence, which must have been rapid, and found that a fissure had opened and 

 swallowed most of the lava. Judging from the great size of the trees growing over its 

 surface, the flow must be quite ancient. The surface of the casts gave fresh indication 

 of the process by which aa is formed and seems to prove that the more refra(5lory par- 

 ticles of the lava cool at an appreciable time before the general molten mass ceases to 



be liquid, much as the less soluble salts in a saline solution cr3-stallize first on cooling; 



L474] 



