Dlopside — Forsterite — Silica. 259 



of the melts, the behavior of a mixture of composition M, fig. 

 17. was described. If we imagine that the olivine crystals 

 settled towards the lower layers, then, as the temperature of 

 the point K is approached, the composition of these layers 

 would lie between M and A and of some of the upper layers 

 between P and K. On further cooling the various layers 

 would behave quite differently, those having an abundance of 

 olivine would crystallize to a mixture of pyroxene and olivine, 

 those with only a small amount of olivine would have the 

 olivine completely resorbed and would finally crystallize to a 

 mixture of pyroxene and silica (cristobalite or tridymite). 

 Thus the settling of crystals may bring about not merely a 

 difference in the proportions of the minerals in various layers, 

 but may produce a partially different assemblage of minerals. 



The portions which crystallized to pyroxene and silica 

 would, moreover, become completely crystallized at a lower 

 temperature than would the other portions and might, perhaps, 

 show an intrusive relation to the olivine-bearing portions. 

 Such a result would be impossible if the minerals showed the 

 simple eutectic relation, even if there were settling of crystals, 

 for in this case all parts would have some share of liquid until 

 the time of simultaneous final solidification of all the layers, 

 assuming, of course, all the layers to have the same tempera- 

 ture. In the case of a difference of temperature in different 

 layers, the cooler layers would necessarily become completely 

 solid before the warmer layers when eutectic relations pre- 

 vailed, however great a quantity of crystals had settled. In 

 the melts described, however, the settling of olivine could 

 bring it about that a lower, olivine-rich layer might become 

 completely solid when at a higher temperature than an upper 

 layer, poor in olivine, and still largely liquid. 



But this result may be accomplished, not by the settling of 

 olivine alone, but also by the settling of the early-formed, 

 more magnesian pyroxenes and the continued concentration of 

 calcic pyroxenes in the liquid. The crystallization would then 

 tend towards crystallization of the second- type and the result 

 would be the same as when zoned crystals are formed. Here 

 again one portion might become completely solid even when at 

 a higher temperature than another portion still partly liquid. 

 A similar behavior is possible in any melt that deposits mix- 

 crystals of which one constituent has a considerably higher 

 melting point than the other, as is notably the case with that 

 very important rock-forming series, the plagioclase feldspars. 

 Thus it seems possible that a homogeneous magma, contain- 

 ing substances capable of forming mix-crystals, as do all mag- 

 mas, could become differentiated through crystallization into 

 parts which gave evidence of having solidified at different 



