274 F, E. WRIGHT OBSIDIAN FROM ICELAND 



much longer period at the center than at the margin ; in short, the center 

 is mucii older than the margin, and the gases active during the alteration 

 were evidently those set free chiefly during the primary crystallization of 

 the spherulites. 



The evidence presented thus far proves that during the crystallization 

 of the spherulites volatile components were active within the cavities; 

 also that at the temperatures, at which the lava was still sufficiently 

 molten to flow into very small cracks, the remarkable deformations de- 

 scribed above were produced. Now the chemical analysis of this obsidian 

 shows that it contains 0.13 CI, 0.07 SO3, and 0.27 HjO— all volatile gases 

 which would be set free, in part at least, on the crystallization of the 

 silicates. It is not probable that either sodalite or noselite would be 

 farmed in the presence of so much quartz, and these are practically the 

 only silicates containing NaCl or Na^SO^ which would be likely to be 

 formed. We have seen, furthermore, that the escape of volatile com- 

 ponents continued as crystallization proceeded and as the cavity was en- 

 larged. The question arises : Did the pressure of this escaping gas force 

 the cavity apart or was the main factor an external uniform tension de- 

 veloped on the shrinking of the magma ? 



It is evident that where gas bubbles are formed in a magma the vapor 

 pressure of the gas has been sufficient to overcome the internal pressure 

 of the magma : also that simple vacua of regular bubble shape in a viscous 

 magma would be difficult to form. Field experience and laboratory prac- 

 tice have shown that, in such instances where the magma is inclosed 

 between frozen walls and shrinkage occurs on cooling, cracks (joints) 

 develop rather than bubbles disseminated evenly through the magma. 

 Reduction of hydrostatic pressure in the liquid favors the formation of 

 gas bubbles just as does the opening of a siphon bottle containing car- 

 bonated water. Bubbles may begin to form, moreover, when the liquid 

 becomes supersaturated with respect to the gas. Increase of uniform pres- 

 sure raises the saturation limit with respect to the dissolved gas; con- 

 versely, reduction of uniform j)ressure loM'ers the saturation limit and 

 favors the escape of the gas. Gravity, furthermore, is a factor which 

 would tend to close any vacuities disseminated through the magma. On 

 cooling, it would seem, then, that a magma inclosed in a solid shell would 

 tend to shrink away from the roof and to leave cracks rather than simple 

 bubbles. The formation of bubbles is facilitated if there be a point of 

 discontinuity in the liquid (differences in potential). Tins is given in 

 the case of gas in the magma reaching supersaturation near some nucleus, 

 such as a minute crystal or spherulite. It is less easy, if not impossible, 

 to explain the formation of a vacuum bubble in a moving viscous liquid. 



