258 Becker — Fractional Crystallization of Rocks. 



walls a part of its less fusible component, and at each com- 

 pleted revolution it will have a different composition. This 

 composition will always tend towards that which represents the 

 most fusible mixture of the component compounds. When 

 this composition is attained, the magma will no longer undergo 

 change by circulation and partial solidification ; and the 

 residual mass will gradually solidify as a uniform material. 

 Unless then the injected magma happened to be a mixture of 

 maximum fusibility, the dike would exhibit a gradation in 

 composition from the sides towards the center. In a very nar- 

 row dike solidification might take place before an opportunity 

 was afforded for the complete elimination of the less fusible 

 material ; while in wide dikes solidified from mobile magmas 

 one might expect the central sheet to approximate to maximum 

 fusibility. 



It is evident that the process of . solidification in a laccolite 

 closely resembles that in a dike, particularly if the section of 

 greatest area is not absolutely horizontal. Convection will 

 then be set up and solidification from the walls must tend to 

 the evolution of a residuum of extreme fusibility. 



Convection in dikes and laccolites. 



The process sketched is one of the most familiar in chem- 

 istry and is usually known as fractional crystallization. It has 

 been employed in the purification of compounds ever since 

 chemistry was pursued, and indeed before ; for the preparation 

 of salt from sea water or brine depends upon it. It can be and 

 has been employed also to strengthen solutions. A familiar 

 instance is the freezing of weak alcoholic liquids. A bottle of 

 wine or a barrel of cider exposed to a low temperature deposits 

 nearly pure ice on the walls, while a stronger liquor may be 

 tapped from the center. If a still lower temperature were 

 applied the central and more fusible portion would also solidify. 

 Such a mass would be, so far as I can see, a very perfect ana- 

 logue to a laccolite. A similar concentration is effected in the 

 Pattinson desilverization process. 



Though fractional crystallization is said to have been familiar 

 to Parcelsus and even to Aristotle, the process has been studied 

 most thoroughly by Mr. F. Guthrie.* As is well known, he 



*Phil. Mag. (5), vol. xvii, 1884, p. 462. 



