CRYSTALLIZATION-DIFFERENTIATION IN MAGMAS 401 



melting temperatures of olivine, magnetite, and other components 

 of monomineralic rocks." This is in reality a collateral considera- 

 tion, the reason being rather that natural rocks fail to show the 

 kind of evidence that they would of necessity show were immisci- 

 biHty a fact; indeed, it is incredible that anyone who has followed 

 the process to its inevitable consequences could favor immiscibility 

 in magmas. 



A feature of igneous rocks that has led Grout,^ Daly, and a 

 number of others to favor immiscibility is the fact that two adja- 

 cent rocks that are evidently closely related frequently show a very 

 abrupt transition from the one to the other. Yet a brief considera- 

 tion of hquid immiscibility should show that it is not as likely to 

 give discontinuous variation as is crystallization. It is true that 

 if two liquids that are only partially miscible are shaken together in 

 a flask two different liquids are formed, and if the flask be set 

 aside they will become two separate layers with a definite bounding 

 surface. If the temperature is kept constant these two distinct 

 and sharply bounded layers will persist. However, if the immisci- 

 bility is the result of cooling a homogeneous solution the behavior 

 is not so simple. In this case a certain amount of immiscible 

 globules should form in the liquid when a certain temperature is 

 reached, and, even if time were allowed then for the collection of 

 the globules as a separate layer, more immiscible globules would 

 form in each layer as soon as cooling was resumed. And when cool- 

 ing had proceeded to the point where crystallization ensued a 

 marked increase in the separation of immiscible globules would 

 occur in association with, and as a necessary consequence of, the 

 separation of crystals. We thus see that immiscibility is not a 

 process taking place at an early stage of cooling as a result of 

 which a sudden separation of a liquid into two liquid layers occurs. 

 The separation is rather a formation of small globules that grow 

 slowly by diffusion and can collect as a separate layer only by 

 comparatively slow movement in response to gravity. Neither is 

 immiscibility a process that is completed at a very early stage in 

 the cooling history, and of which all evidence is destroyed. It is 

 a process that may begin very early but must continue until the 



' "A Type of Igneous Differentiation, "/ai<r. GeoL, XXVI (1918), 656. 



