178 N. L. BOWEN 



mixtures. In investigation of this kind, both on its theoretical 

 and experimental sides, Vogt took a leading part and the importance 

 of his pioneer work cannot be overestimated. 



As might have been anticipated, it was soon found that the 

 eutectic relation did not always obtain, for, after all, it is but one 

 of many possible relations between two or more components in 

 any system. Very early it was realized that the existence of soHd 

 solution between two components might eliminate the eatectic 

 relation between them. Later, as more accurate methods of 

 experimentation were introduced, examples of incongruent melting 

 were found, which again lead to the elimination of the eutectic. 

 Indeed, it seems that relations between the components such that 

 the eutectic disappears are so common in mineral systems that it 

 may be doubted whether considerations based upon the doctrine 

 of eutexia can longer be regarded as of any great service to petro- 

 genic theory, in spite of their great suggestive utihty in the past. 



It should be realized that these non-eutectic relations should 

 not be termed anomalies nor even apparent anomalies. They are 

 just as definitely the result of the laws of solution as are eutectics. 



This paper has been written with the object of readjusting the 

 emphasis in our conception of mineral relations. It will be pointed 

 out that another relation between the phases, here called the 

 reaction relation, is very common in silicates, and is at the same 

 time a very much broader concept, with which to approach the 

 study of petrogenesis, than the simple doctrine of eutexia, so 

 that it might well be set up as a principle to serve as a guide and 

 stimulus to the search for facts. 



OF CONTINUOUS REACTION SERIES 



The reaction relation in a common form is illustrated by the 

 case of the plagioclase feldspars. The equiHbrium diagram of the 

 plagioclases (Fig. i) is of a simple and famiHar type.^ Those 

 features of it that require emphasis in the present connection may 

 be brought out very briefly. The crystallization of any random 

 Hquid, say AbjAui, takes place in the following manner. At 1450° 

 crystals of the composition AbiAn4 begin to separate. As the 



^ N. L. Bowen, American Journal of Science, Vol. XXXV (1913), p. 583. 



