LATER STAGES OF EVOLUTION OF IGNEOUS ROCKS ii 



liquid portions should be. If they are ever due to a separation in 

 the liquid state, it is remarkable that sudden extrusion has not 

 happened to give plain evidence of the fact in the occurrence of 

 similar orbules of undercooled liquid in the extrusive tjrpes. In the 

 absence of evidence of this kind, it is necessary to reject liquid 

 immiscibilit}^ as a factor in the differentiation of igneous rocks. 

 Lavas could not fail to show the separation of liquid phases in all 

 its stages just as clearly as they do the separation of crystalline 

 phases in all its stages. Attention must be directed toward the 

 separation of crystalline phases — i.e., crystallization — -as the funda- 

 mental factor in differentiation though the possibility that sul- 

 phides and perhaps some oxides may become immiscible under 

 certain conditions may be freely admitted. It may be pointed out 

 that, considered on general grounds, limited miscibihty between such 

 dissimilar substances as sulphides and silicates is likely, but wholly 

 unlikely among the silicates themselves, substances which have 

 such remarkable tendencies to be miscible even in the soHd state. 



Di^erentiation through crystallization. — -Differentiation may be 

 brought about as a result of crystallization in two ways: through 

 the localization of crystallization and through the local collection 

 of crystals. 



Since the outer parts of a body of magma are always cooler 

 than the interior, there is a certain period during which crystalliza- 

 tion is taking place only near the border, the interior being still 

 above the temperature of beginning of crystallization. It has 

 been supposed that during this period precipitation of the mineral 

 or minerals of early crystalHzation takes place in the cooler parts, 

 but that this does not bring about a local impoverishment of the 

 liquid in the crystallizing material, the composition of the liquid 

 being kept uniform and continued growth of the crystals being 

 maintained by free diffusion from all parts of the liquid^ or by con- 

 vection currents.^ Thus there results a general richness of the 

 border phase in minerals of early crystallization, according to this 

 behef. Such freedom of diffusion Becker has shown to be alto- 

 gether inconsistent with what is known on that subject. Apart 



^ A. Harker, The Natural History of Igneous Rocks, p. 318, 

 ^ G. F. Becker, Am. Jour. Sci. (3), 1897, pp. 21-40. 



