SPHERULITES AND LITHOPHYSvE 263 



adjacent bands are not spherical l)ut tiiinilar in shape, the direction of 

 elongation being that of the lines of flow of the lava. This indicates 

 that the lava during the period of its final flow was sufficiently viscous 

 to prevent the escape of the free gas bubbles which it inclosed. The 

 restriction of the gas bubbles to definite bands and parts of the mass 

 might be considered to indicate that during the period of its flow^ the 

 lava encountered physical conditions of such nature (especially release 

 of pressure) that certain bands became supersaturated with volatile com- 

 ponents, which were then released and formed bubbles. These could not 

 migrate through the lava to any great extent because of its high viscosity. 

 It is, however, conceivable and a priori more probable that the appearance 

 of a pumiceous band is not the result of direct evolution of gas from that 

 band alone, but that either before or during the eruption of the molten 

 obsidian there was an accumulation of gas bubbles at certain points 

 (magma hotter and less viscous, thus allowing freer circulation and con- 

 centration of evolved gas bubbles at favorable pockets near the margin 

 of the magma chamber), and that on final outflow of the obsidian these 

 foamy portions of the lava were drawn out and appear now as vesicular 

 streaks, which serve to emphasize the lines of stiff viscous flow of tlie 

 lava. On this hypothesis the amount of pumice accompanying a rhyolitic 

 flow can not be considered to be indicative of the amount of gas given 

 off by the obsidian. Most of the gas thus liberated from the solution no 

 doubt escaped, and that which produced the pumice represents only a 

 small part of the total amount contained originally in the molten ob- 

 sidian. 



Passing now to the spherulites, we find that they occur in typical de- 

 velopment in several of the specimens. They range in size from a few 

 tenths of a millimeter to over 5 millimeters in diameter. In the outcrop 

 they are not evenly distributed through the rock mass, but are confined 

 to certain bands or layers, thus indicating that in these bands crystalliza- 

 tion took place more rapidly than in others. In the case of finely lami- 

 nated floM' banding, however, the spherulites cut across the banding. In 

 no case was a suggestion of flow banding around a spherulite observed : 

 this would occur were the spherulites older than the banding. In one 

 instance the wall of a hollow spherulite was serrated as the result of a 

 difference in behavior of the different bands with respect to crystallization 

 and to attack by the volatile components released during the crystalliza- 

 tion of the spherulite. 



The relations outlined above suggest that the determinative factors in 

 the development of the dift'erent structures which are now found in the 

 obsidian — ])umiceous, spherulitic. and litliophysal — were the physical con- 



