534 N- L. BOWEN 



constant. The liquid is not at all desilicated, even though it has 

 caused the conversion of the inclusions into a more siliceous phase. 

 By precipitation of the phase with which it is saturated, it has 

 adjusted its composition in such a way as to remain on the same 

 saturation curve. This again is a principle that can be applied in 

 general to the reaction between a liquid and inclusions belonging 

 at an earlier stage of the reaction series than the phase with which 

 the liquid is saturated. 



Further considerations relative to this reaction pair will be 

 developed in connection with their behavior in the more complex 

 liquids containing anorthite as worked out by Andersen"^ and shown 

 in Figure 8. From this figure we may very readily predict what 

 would happen to "inclusions" of the various solid phases immersed 

 in liquid. Let us take first a liquid just saturated with forsterite 

 (say M at 1450°). Ordinarily crystals of forsterite would separate 

 first; they would then react with liquid to form pyroxene; pyroxene 

 would continue to separate for a time and would then be joined by 

 silica; finally, at the ternary eutectic 1222°, anorthite would join 

 these, and the product would consist of pyroxene, silica, and anorth- 

 ite. If to the original liquid some "foreign inclusions" of forsterite 

 were added the liquid would, on cooling, attempt to make the 

 forsterite over into pyroxene, but now there would not be enough 

 liquid to accomplish this entirely, and it would be used up at 1260° 

 while there was still some forsterite left; so that the solidified mass 

 would now consist of pyroxene, forsterite, and anorthite. Thus we 

 see that even though it is precisely the substance with which the 

 liquid is in equilibrium, the addition of forsterite has a considerable 

 effect upon the crystalline product formed. The exact effect is a 

 tendency to limit the scope of the products to the early members of 

 the crystallization sequence. In this case, silica, a later member of 

 the sequence, does not appear. It is plain, too, that the effect does 

 not depend on the particular properties of this reaction pair and 

 that we may conclude that it would be true of any discontinuous 

 reaction series. We have already seen that the same effect is found 

 in the continuous reaction series (solid solution series). We may 



'The system anorthite-forsterite-silica, Amer. Jour. Sci., Vol. XXXIX (1915), 

 p. 440. 



