234 JV. L. Bowen — The Ternary System : 



composition of the liquid now follows the boundary curve, 

 forsterite gradually redissolves and pyroxene changes in com- 

 position. When the liquid has the composition N the pyrox- 

 ene has the composition K. When the temperature of the 

 point K is reached, the liquid finally disappears, the vanishing 

 amount having the composition P, and the pyroxene having 

 the composition P, PP being the three-phase-boundary for the 

 temperature of the point P. The whole then consists of 

 forsterite and pyroxene P. 



If the original mixture had the composition P, forsterite 

 would separate first, as before ; at O, pyroxene L would begin 

 to separate and forsterite to redissolve. the whole behavior 

 being the same as that of composition M, until the temperature 

 of the point N is reached. At this temperature the re-solution 

 of forsterite is complete and the whole consists of pyroxene K 

 and liquid N", which is also of pyroxene composition. Thence- 

 forth the melt behaves as a binary mixture and its behavior is 

 expressed in the simple binary diagram of fig. 18, in which the 

 points N, P and K correspond with the points so lettered in 

 the ternary figure and of which no further explanation is 

 necessary. All the rearrangement between liquid and solid 

 which is necessary in order to get crystallization strictly 

 according to the binary diagram of fig. 18 is, of course, in 

 practice impossible on account of the very small temperature 

 interval involved. What really happens is the crystallization 

 of the liquid N in toto to pyroxene of composition N. 



The behavior of liquid of composition P is typical of that 

 of all liquids of pyroxene composition between N and K. 

 The behavior of liquids of composition between K and pure 

 MgSi0 3 is given in the discussion of the liquid P of fig. IT. 

 All liquids between jN" and pure diopside behave as binary 

 mixtures and the binary diagram of fig. 18 expresses their 

 behavior completely. 



The existence of both the solidus and liquidus shown in fig. 

 18 is, of course, not experimentally determinable on account 

 of the extremely small intervals involved. All that has been 

 shown is the existence of a very flat minimum, whose exact 

 position cannot be located. The occurrence of the minimum 

 was proved as follows : A charge of a given composition 

 close to diopside was compared with diopside by quenching 

 the two together from the same temperature. The charges 

 were separated only by platinum foil •02 m,n thick, so that there 

 was no possibility of a difference of temperature. In this 

 manner it was shown that the composition, diopside 82 per 

 cent, MgSi0 3 18 per cent, and also the mixture 77 per cent 

 diopside give only glass at a temperature slightly lower than 

 that at which the diopside gives only glass. Of these two 



