THE REACTION PRINCIPLE IN PETROGENESIS i8i 



the liquids that are in equilibrium with two solid phases. Moreover, 

 it lies along a valley in the fusion surface and when the object is to 

 emphasize these features no serious objection can be raised to the term 

 used by Vogt. On the other hand, such a usage would tend toward 

 a non-recognition of the importance of the reaction relation be- 

 tween plagioclase crystals and the liquids lying along the boundary 

 curve and, in the writer's opinion, petrology will better be served by 

 emphasizing this relation^ — calling the curve, say, a reaction curve. 

 But quite apart from the question of nomenclature, the distinction 

 is really one of fundamental importance in petrogenesis. In the 

 case of a true eutectic line in a ternary system, the two kinds of 

 crystals separating bear what may be called a mere subtraction 

 relation to the liquid and once subtracted they are no longer 

 concerned in the equiHbrium. Moreover, the true eutectic line 

 ends in a ternary eutectic point which represents the composition 

 that all liquids of the system must finally attain and beyond which 

 they never pass. No such rigidity obtains in a system where 

 the reaction relation enters. The composition of the final liquid 

 depends upon the freedom with which reaction between crystals 

 and liquid may take place. If the reaction is complete in the 

 example taken, the crystals are continually made over and the last 

 of the liquid is used up by the reaction at 1 200°, when the composi- 

 tion of the liquid is M and all the crystals have the composition Z. 

 But if for any reason early crystals do not participate to the full in 

 the reaction — and this may occur when zoning of crystals or sinking 

 of crystals supervenes — the composition of the liquid may then 

 pass beyond M and the final liquid have a composition represented 

 by some point between M and D, with corresponding effect on the 

 composition of the final crystals.^ 



This flexibility in the behavior of the Hquid is entirely the 

 result of the reaction relation between Hquid and crystals. The 

 course of crystallization is rendered responsive to the conditions 

 under which crystallization takes place, for removal of crystals by 

 zoning, sinking, or otherwise is entirely dependent on these condi- 

 tions. No such response is possible in a eutectic system, for, 



' A similar effect may of course be brought about in the binary system of the 

 plagioclases, but it was considered unnecessary to discuss it for each case separately. 



