30 JOURNAL OF GEOLOGY— SUPPLEMENT 



as a result of very rapid cooling would give a mixture of forster- 

 ite, clino-enstatite, and anorthite. Sinking of crystals would have 

 the same result as in the case of the liquid x, but the total amount of 

 forsterite formed would be greater and a considerable peridotite-like 

 layer might be formed. If the composition of the liquid were 

 that of the point s, slow crystallization with perfect equilibrium and 

 no sinking of crystals and also very rapid cooling would give a 

 final product of forsterite, clino-enstatite, and anorthite rather 

 rich in forsterite but entirely free from spinel. Slow crystalliza- 

 tion with sinking of crystals would remove the early-formed spinel 

 crystals from that part of the Hquid in which they formed and 

 thereby prevent the re-solution which takes place with perfect 

 equihbrium. These spinel crystals, collecting with the earHer 

 forsterite crystals, would give a mass simulating a spinel-hearing 

 peridotite. 



It is to be noted that during the crystallization of the Uquid x 

 in which complete re-solution of olivine takes place with perfect 

 equilibrium, there is a stage at which olivine crystals constitute 

 about i^ per cent of the whole mass. In case forsterite crystals 

 sank during the crystallization of this mixture, the Kquid would be of 

 the same composition in all parts until the temperature y was 

 reached, after which the Uquid in the part containing olivine 

 crystals would change in composition in a different manner from the 

 parts of the liquid free from olivine crystals. The latter would 

 depart from the boundary curve FM, cross the clino-enstatite 

 field, and become completely crystalHne when the temperature N 

 was reached. The former would follow the boundary curve and 

 become completely crystalline at the higher temperature M. Dur- 

 ing the period when the liquid in contiguous parts was forced to 

 follow different courses by the presence of olivine crystals in the one 

 part and their absence in the other, it seems likely that a squeezing 

 out of the liquid might give rise to phenomena similar to those 

 noted in the rather rare natural case of the intrusion of a hetero- 

 geneous magma — might, in fact, give a product simulating a banded 

 gabbro. The liquids concerned would not, however, be immiscible. 



It may be pointed out also that certain layers into which con- 

 siderable olivine had settled would be peridotitic (UierzoHtic) in 



