536 N. L. BOWEN 



Suppose first that the amount of inclusions was such as to give a 

 total composition represented by the point K. The liquid would 

 then be completely used up by the reaction while some inclusions 

 yet remained, so that the finally soHdified mass would consist of 

 forsterite, clino-enstatite, and anorthite. If, on the other hand, the 

 amount of inclusions was such as to give a total composition repre- 

 sented by the point L, all the inclusions would be changed to clino- 

 enstatite while some liquid yet remained, and this liquid would then 

 pass onward to the deposition of silica and anorthite in the ordinary 

 way. Thus the addition of inclusions of this kind, if in sufficient 

 quantity, restricts the career of the liquid and confines the crystalline 

 products to adjacent members of the reaction series. 



It is perhaps not necessary to add that if the liquid did not react 

 with the inclusions no effect on the course of the liquid would ensue. 



Just as was the case in the binary system, the hquid is not desili- 

 cated by this addition of silica to the forsterite inclusion, but 

 by precipitating an appropriate amount of the phase with which it 

 is saturated (clino-enstatite), it maintains its position on the same 

 (clino-enstatite) saturation surface. 



Nothing has been said of the heat effect of this reaction and it has 

 been tacitly assumed that the temperature remains constant. The 

 reaction is, in fact, exothermic, as can be readily shown by applying 

 the same reasoning as was applied to the similar case in a continuous 

 reaction series. If no heat were abstracted from the system it would 

 heat itself up and equilibrium would be established at a slightly 

 higher temperature with a somewhat more magnesian liquid than 

 the initial liquid. But the formation of this somewhat more mag- 

 nesian liquid is not properly to be taken as an indication that the net 

 result of the process is a direct solution of some of the inclusions. A 

 direct solution of inclusions would mean a decrease in total solids 

 and an increase in liquid, whereas the reaction referred to results in a 

 diminution in the amount of liquid and a corresponding increase in 

 the amount of solids even when this heating effect takes place. If 

 heat is being taken from the system this process would act as a 

 deterrent upon the rate of cooling. 



All of these effects we have found to be true of analogous inclu- 

 sions in the case of the continuous reaction series. 



