CRYSTALLIZATION-DIFFERENTIATION IN MAGMAS 403 



globules in this manner would result in their more rapid accumula- 

 tion as a separate layer. It should be noted, however, that this 

 rapidity of accumulation could never result in the complete accumu- 

 lation of all the globules as a separate layer. If, for example, 

 coohng were interrupted at some temperature between FK and 

 ED, and time allowed for the accumulation of all the globules as a 

 separate layer, as soon as cooling was resumed new globules would 

 form in each layer, and their accumulation by the slow process of 

 gravitative adjustment would begin again. It is plain then that 

 whatever complications are assumed the magma must arrive at 

 the temperature ED in a blotchy condition, many of the blotches 

 being of rather large dimensions as a result of the coalescence of 

 globules. By large dimensions is meant a diameter several times, 

 perhaps very many times, the diameter of the crystals in the average 

 plutonic rock. At the temperature ED, when the liquids in equilib- 

 rium have the composition E and D, crystallization begins, crystals 

 of A separating. It is important to note the nature of the first 

 crystals separating, for it will be recalled that the liquid separating 

 was rich in B. Those who advocate the separation of olivine, 

 pyroxene, plagioclase, etc., as immiscible liquid tacitly assume a 

 correspondence between the kind of material that would separate 

 early as a liquid and the kind of material that we know from experi- 

 mental and petrographic experience separates early as crystals. 

 As a matter of fact there is no necessary relation, and the fact that 

 correspondence must be assumed in each individual case is sufficient 

 in itself to throw doubt on a process requiring such an assumption. 

 Continuing the consideration of the cooling of the mixture, 

 which had been carried to the stage of the beginning of crystalli- 

 zation, at the temperature ED, we find that crystals of A would 

 separate, and that as a necessary consequence more liquid of com- 

 position D would be formed and some liquid of composition E 

 would be used up. These reactions would continue at constant 

 temperature with the amount of crystals A and the amount of liquid 

 D increasing at the expense of liquid E until finally all of liquid E 

 would disappear, when the whole mass would be made up of about 

 80 per cent of crystals A and 20 per cent of liquid D. We thus see 

 that up to a time when the mass is largely crystalline two liquids are 



