56 JOURNAL OF GEOLOGY—SUPPLEMENT 



but rather one characterized by diopsidic pyroxene. But such 

 granites are rare — many times less common than hornblende 

 granites, and these in turn many times less common than biotite 

 granites. The formation of the molecules which enter into the 

 make-up of biotite results from various equihbrium reactions in 

 which water plays an essential role. If any of the products of 

 these reactions are removed from participation, through continuance 

 of the processes of separation of liquid from crystals, further differ- 

 entiation beyond the biotite granite stage is thereby permitted. 



It has been shown that at the biotite granite stage, and to a 

 lesser extent in preceding stages, reactions take place in the Hquid 

 whose principal feature is the breakdown of polysilicate molecules, 

 probably under the influence of water, to the simpler orthosihcate 

 molecules, among them KAlSi04 and NaAlSi04. The precipitation 

 of KAlSijOs, NaAlSiaOs, KAlSi04 in mica, and SiOa as quartz means 

 the concentration in the liquid of all the other molecules indicated in 

 the reactions given on pp. 44 and45. These are principally NaAlSi04 

 and the volatile constituents, water, chlorine, etc., with their com- 

 pounds. If the crystals of the biotite granite stage, including 

 quartz, sink out of this Hquid,' then the concentration of NaAlSi04 

 will finally reach a stage where nephelite will begin to precipitate. 

 There may also result a concentration of CO2, S, SO3, CI, etc., suffi- 

 cient to cause their precipitation in compounds such .as cancrinite, 

 lazurite, hauynite, and sodalite, minerals which are peculiar to 

 nephelite syenites and related rocks. 



The field association of many nephelite syenites and their miner- 

 alogy point clearly to their crystallization at this stage of the 

 magma. Both the mineralogy of the rocks, and the high fluidity of 

 the magma indicated by quick changes of composition in alkaline 

 rocks point to the high concentration of volatile constituents at this 

 stage. 



Differentiation during crystallization from these very fluid 

 magmas will take place very freely and the formation of both 



' Or, as seems more likely, if squeezing out of residual liquid is accomplished at the 

 appropriate stage. The appropriate time may, perhaps, be that immediately preceding 

 the period of marked resorption of quartz which is well attested in most rhyolites. A 

 very important effect on the future course of the liquid would ensue, analogous to that 

 which results if crystals and liquid are separated when oUvine is about to be resorbed. 



