546 



N. L. BOWEN 



When we have spoken of adding calcite and dolomite to the mix- 

 tures mentioned we have imagined that the lime and magnesia 

 have been taken into solution as they might be in a laboratory fur- 

 nace where the furnace suppHed the requisite quantity of heat. 

 This quantity would be very large, for the conversion of carbonate 

 into silicate is an endothermic reaction and its conversion into sili- 

 cate in solution is undoubtedly still more strongly endothermic. 

 Now if the liquid were originally a saturated liquid and calcite or 



•So/id SO/ufWm 



Fig. io. — Equilibrium diagram of the system CaO-MgO-SiOj. (After Ferguson 

 and Merwin.) 



dolomite were added without any provision for additional heat 

 other than that already contained in the liquid, the lime and mag- 

 nesia would not be simply dissolved. Instead, precipitation of 

 olivine (forsterite) would occur, no matter whether calcite or dolo- 

 mite was added, and as the forsterite separated the liquid would 

 attack the calcite, converting it into lime silicates. Since this is an 

 endothermic reaction the heat for it can be supplied only by crystal- 

 lization of something from the liquid which will be, of course, the 

 phase with which it is saturated, viz., forsterite. There is another 



1 



I 



