526 N. L. BOW EN 



when it would attain the composition Ab^Ani, and then add the for- 

 eign crystals AbiAru, that we filtered off. It is plain that we would 

 have available by the making over of these crystals at 1450° all the 

 heat that would have been evolved between 1500° and 1450° by the 

 continuous process of making over of these crystals had they been 

 left in contact with the liquid. Though we have used the plagio- 

 clase series as an illustration it is clear that the exothermic reaction 

 taking place at 1450° is merely a deferred result of the principle of 

 Le Chatelier which states that an equilibrium reaction proceeds, 

 with falling temperature, in the direction resulting in evolution of 

 heat. It is a perfectly general property of any sohd solution series 

 that if, at any temperature, crystals which are at equilibrium with 

 liquid at a higher temperature are added to saturated liquid, the 

 reaction which ensues between liquid and crystals is exothermic. 

 The case of the addition of an inclusion of composition nearer 

 the low temperature (more sodic) end of the solid solution series may 

 now be examined. A liquid of composition AbiAuz is just saturated 

 at 1490°. We cannot add to it a more sodic solid inclusion at the 

 same temperature because any more sodic inclusion will be liquid 

 at this temperature. But if we add inclusions of Ab^Ani at such a 

 temperature that they are solid, say 1200°, it is plain that the actual 

 temperature of the liquid is adequate to melt these inclusions; the 

 only question is the source of the quantity of heat. The liquid must, 

 of course, be cooled off in supplying the heat required to heat up 

 the inclusions, but, since the liquid is saturated, it cannot be cooled 

 without some crystallization taking place. However, it will take a 

 very small amount of crystallization to supply the heat necessary to 

 heat up a considerable amount of inclusions. The enormous dis- 

 crepancy between the specific heats of sihcates and their solution 

 heats is plainly of double significance in connection with these prob- 

 lems. Not only can the heat necessary to heat up the inclusions be 

 supplied by crystallization of some of the liquid, but so also can the 

 heat required to melt the inclusions. To accompHsh this it will 

 require, however, the formation of crystals approximately equal in 

 amount to the amount of inclusions melted. If we added 20 per 

 cent of inclusions of AbaAn^ at 1200°, to a Kquid of composition 

 AbiAn2 at 1490° we would obtain (assuming that these thermal 



