240 JV. L. B 010671 — The Ternary System : 



represented only by means of isotherms. Such a diagram is 

 shown in fig. 20. Thus the behavior of composition Di 19 En gl 

 is shown in the figure to be as follows: at 1552° forsterite 

 crystallizes out and continues to crystallize down to 1523°. 

 At this temperature pyroxene begins to crystallize out, and 

 until the temperature 1420° is reached, the three phases 

 forsterite, pyroxene and liquid co-exist. At 1420° both 

 forsterite and liquid have disappeared and the whole consists 

 of pyroxene. The diagram, then, expresses the nature of the 

 phases present at any temperature in any mixture of given 

 total composition. It does not, however, express the compo- 

 sition of the individual phases present in a heterogeneous 

 mixture, inasmuch as this can be done only in terms of three 

 components. It records the temperature at which any phase 

 appears or disappears in a mixture of given total composition 

 and therefore indicates the temperature at which a heat-effect 

 takes place in that mixture, or, in other words, the temperature 

 at which a heating-curve break is to be expected. It has been 

 demonstrated that the true ternary equilibrium diagram, when 

 boundary curves, isotherms and three-phase-boundaries are 

 shown, also expresses these facts and has the advantage that it 

 expresses the composition of the phases present at any temper- 

 ature as well, but the ternary diagram is more difficult of 

 rapid interpretation. 



Comparison of the Present Results with Earlier Work. 



The system CaSiO s -MgSiO, had been formerly studied at 

 this laboratory, the method of attack being principally the 

 method of heating curves. The heat-effects indicated by the 

 heating-curve breaks are usually quite adequate for the correct 

 interpretation of equilibrium in the system, but it happened 

 that in the present case they were not. The difference between 

 the conclusions arrived at in the early work and those arrived 

 at after making quenching experiments is the result of two 

 factors. In the first place the heating-curve method is likely 

 to overlook a change involving a small heat-effect when this 

 change occurs at a temperature only a little higher than another 

 involving a considerable heat-effect. The large effect at the 

 lower point may completely mask the smaller effect which 

 should be obtained at the higher point. It is true that by 

 special treatment the higher point can often be fixed, in such 

 cases, by the method of heating curves, but the fact remains 

 that the existence of the higher point may not even be 

 suspected and the special treatment not applied. Heating- 

 curve results are, then, likely to err in the direction of incom- 

 pleteness. 



. 



