LATER STAGES OF EVOLUTION OF IGNEOUS ROCKS 27 



and N and that in the upper layers by a point on the curve FE. 

 A squeezing out of the hquid from such a mass might bring these 

 different hquids together and draw them out into streaks which 

 though perfectly miscible would have only a Hmited opportunity to 

 mix before crystallizing, and would thus give a result simulating 

 primary banding, a rather rare but quite definite phenomenon of 

 igneous rocks. 



It has been supposed in the foregoing that only crystals of 

 forsterite settle out, but it may readily be imagined that in a certain 

 limited portion nearly all the forsterite has settled out at the time 

 when the crystallization of pyroxene begins and that there is 

 opportunity for the sinking of pyroxene crystals. Even in layers 

 having an abundance of forsterite crystals, pyroxenes will crystal- 

 lize and join the sinking swarm. The result is an enrichment of the 

 lower layers in the more magnesian pyroxenes of early separation 

 with a consequent raising of their temperature of final consoHda- 

 tion somewhat above N, whereas the liquid in the upper portion is 

 continually enriched in more calcic pyroxene with a consequent 

 lowering of its temperature of final consolidation below G. The 

 composition of the liquid is, in fact, continually offset toward the 

 eutectic diopside-silica. In an ideal case the composition of the 

 last minute quantity of hquid would reach the eutectic diopside- 

 siHca, but in any actual case the eutectic will not be attained, the 

 last of the liquid being used up before the temperature and compo- 

 sition of the eutectic is reached. The temperature of final con- 

 solidation of the residual liquid and its composition will he at some 

 point intermediate between G and the eutectic diopside-silica, 

 approaching closer to the latter according as the opportunity for 

 the sinking of crystals increases. The opportunity for the sinking 

 of crystals is directly related to the rate of coohng. It is obvious, 

 therefore, that the rate, of cooling is of fundamental significance 

 in controUing the ultimate products of crystallization. 



In a system "with simple eutectic relations, settling of crystals 

 could make no difference in the temperature of final consohdation 

 of the last of the liquid, nor in its composition. 



Crystallization with straining of of the liquid. — A process involv- 

 ing the separation of cr^^stals and liquid by the squeezing out of 



