Diopside — Forsterite — Silica. 237 



tion of this eutectic and the crystalline phases separating are 

 tridymite and pure diopside. 



It will be recalled that we started with any of the mixtures 

 M, P or D. Any of these mixtures would, then, if crystallized 

 in this manner, consist of forsterite, tridymite, and pyroxene 

 varying in composition from L to pure diopside. The actual 

 amount of pyroxene approaching L in composition would be 

 relatively large ; the amount approaching diopside, relatively 

 very small ; the amount of pure diopside infinitesimal. 



Importance of Distinguishing between the Two Types of 

 Crystallization. 



It is of great practical importance in working out such a 

 system to bear in mind the distinction between crystallization 

 of these two types. Crystallization of the first type, viz.: that 

 in which there is perfect equilibrium between the liquid and 

 the solid phases, is favored by very slow cooling during which 

 plenty of time is allowed for interaction between the liquid 

 and the solid phases which have already separated from it. 

 Crystallization of the second type is favored by quick cooling 

 during which the liquid is in equilibrium at any instant only 

 with the solid phases separating at that instant. It is very 

 doubtful if, in practice, the period of cooling could ever be 

 sufficiently prolonged to realize crystallization of the first type. 

 Crystallization of the second type is often, however, very 

 closely approached, if not actually realized, by heating a charge 

 of 20 to 50 grams to the liquid state and then removing it from 

 the furnace and allowing it to cool to room temperature. This 

 is, of course, true only of those mixtures which crystallize 

 readily, that is, those rich in pyroxenes or forsterite; the 

 silica rich mixtures may not crystallize at all with such treat- 

 ment. 



If, then, a mixture of composition P (fig. 17) is heated till 

 completely liquid and then allowed to cool, crystallization tak- 

 ing place freely, the result will be forsterite, tridymite and 

 pyroxene varying from L to diopside.* The mixture P should, 

 however, consist, when completely crystalline, simply of homo- 

 geneous pyroxene of composition P. The heterogeneous mate- 

 rial prepared by rapid crystallization is not at equilibrium— 

 though it will persist indefinitely — and should not be used to in- 

 vestigate equilibrium whether by the method of quenching or by 

 the method of noting the heat effect of any change recorded by 

 the thermoelement in running a heating curve. The initial mate- 



* In the actual case, owing to undercooling, the composition of the pyroxene 

 varies between somewhat narrower limits. The essential point is that the 

 pyroxene is of variable composition. 



Am. Jour. Sci.— Fourth Series, Vol. XXXVIII, So. 225.— Sept., 1914. 



17 



