92 bowen: significance of glass-making processes 



rangement was due to differential melting with the sinking of 

 heavy liquid at an early stage and the rising of silica grains at a 

 later stage, though I had not at that time seen or studied the 

 phenomenon. A careful study of the behavior of the ingred- 

 ients of a glass batch leaves no question as to the correctness of 

 this interpretation. There is, then, no experimental basis for 

 the belief in an appreciable result from the Gouy-Chaperon 

 action in a small pot and, therefore, no present reason for as- 

 signing to it any greater importance in rock magmas than that 

 which theory would indicate. 



There are two other factors making for inhomogeneity in 

 glass: the solution of the pot, and the volatilization of certain 

 ingredients from the surface of the liquid. No doubt the cor- 

 responding processes, namely, solution of the surrounding rocks 

 and escape of material into them, have their place in magmatic 

 differentiation, but if their quantitative effect in the glass-pot is 

 any criterion they cannot be regarded as approaching in impor- 

 tance the two processes (sinking of liquid and floating of silica) 

 that have been described above. However, the conditions are so 

 different that one should be careful not to push the analogy too far 

 in these cases. It may be safely stated, however, that, contrary 

 to certain claims that have been made, glass-making processes 

 offer no support for the belief in liquid immiscibility among 

 silicates, nor for the belief in a significant density stratification 

 in a mass wholly liquid. They do, however, suggest the impor- 

 tance of gravity acting on a mass partly liquid and partly solid, 

 and emphasize two stages, (1) that at which there is much 

 liquid and little solid, and (2) that at which there is little liquid 

 and much solid. The effects of these processes in magmas, — ■ 

 sinking of crystals at an early stage of crystallization and squeez- 

 ing out of residual liquid at a late stage, — have been discussed in 

 some detail elsewhere.^ 



The association of gabbro with granite or of basalt with rhyo- 

 lite, and the complete absence of intermediate types that is 

 often noted, have been held by some to necessitate some sort 

 of discontinuous differentiation, whereas crystallization-differen- 

 tiation should, for the most part, be continuous. Evans has 

 offered the suggestion that in aqueous magmas there may be a 



2 Op. cit. * 



