THE REACTION PRINCIPLE IN PETROGENESIS 197 



The composition of this liquid lies at some point on the 200° 

 isotherm. If the cooling were continued to 100° and no farther 

 there would again be Hquid left whose composition would be rep- 

 resented by a point on the 100° isotherm, and this Hquid would never 

 crystalhze. And so it would be for any temperature except some 

 negative temperature at which the water itself would appear as ice. 



A similar condition will prevail in the aqueous system repre- 

 sented by magmas. The cooling of rocks is always Hmited by the 

 temperature of their surroundings. There is always a hquid left 

 that does not crystalhze and it is probable that to the very last 

 the rock constituents bear a reaction relation to this Hquid. 



In this connection the behavior of liquids crystalHzing along 

 the boundary Qs-HjO (Fig. 4) may be instructive. At the higher 

 temperatures crystalHzation consists of the simple separation of 

 both KHSiaOs and quartz, but at lower temperatures KHSiaOj 

 separates and quartz redissolves, or, better, the liquid reacts with 

 quartz, converting it into KHSiaOg. If the reaction is not free to 

 take place the Hquid leaves the boundary curve, passes into the 

 KHSiaOg field, and thereafter only KHSiaOg separates from it, no 

 quartz. Thus we have a reaction relation between liquid and 

 crystals at the very latest stages and, in consequence of it, alter- 

 native behavior of the Hquid. 



There are some indications that in granitic Hquid at the peg- 

 matite stage, more or less similar conditions prevail. Quartz may 

 react with Hquid to produce feldspar and sometimes the graphic 

 structure may be a result of this reaction (replacement). Why 

 replacement should produce graphic structure in some instances and 

 not in others is not apparent, but the possibiHty should be con- 

 sidered that graphic structure may often be the result of replace- 

 ment both in rock minerals and in ore minerals. The graphic 

 intergrowth of spinel and pyroxene seen in certain rocks can scarcely 

 be interpreted otherwise than as due to reaction.^ Experimental 

 work on systems involving these phases shows pretty definitely that 

 anything suggesting a eutectic relation between spinel and pyroxene 

 is not to be entertained. - 



' W. N. Benson, Jour. Proc. Roy. Soc. N. S. Wales, Vol. XLIV (1910), p. 521. 



^ Rankin and Merwin, "The Ternary System MgO-Al203-Si02, " Anier. Jour. Sci., 

 Vol. XLV (1918), pp. 301-25; also Olaf Andersen, "The System: Anorthite-Forster- 

 rite-Silica, " Amer. Jour. Sci., Vol. XXXIX (1915), p. 437. 



