Diopside — Forsterite — Silica. 227 



The Temperature Method. — The three-phase-boundaries can 

 be determined by another method which gives more accurate 

 results. Any composition along the line AK (fig. 13) shows 

 the two solid phases pyroxene and forsterite at 1400°. If the 

 temperature is raised a very little, liquid begins to form. This 

 fact may be used to determine the figure corresponding to 

 fig. 13 for any temperature. Thus if a mixture of com- 

 position y is held at various temperatures and the temperature 

 of beginning of melting determined, this will be found to be 

 1400°. If a mixture of composition y 1 is taken it is found to 

 begin to melt at a somewhat lower temperature. If, then, 

 Ay 1 is produced to meet the pyroxene line CD at K x and the 

 point G 1 on the isotherm for this lower temperature is joined 

 with the point K„ the line K^ will be the three-phase-bound- 

 ary for this temperature. 



It is easy to see that the three-phase-boundaries LH and 

 LJHj can be determined in a similar manner, namely, by find- 

 ing the temperatures at which various mixtures of pyroxene 

 and silica begin to melt. This method (the temperature 

 method) is more accurate than the composition method and 

 was the one followed. The results were confirmed by the 

 composition method in some instances during the course of 

 some experiments in which pyroxenes of various compositions 

 were prepared, for optical purposes, by holding a mixture at a 

 temperature at which it gave pyroxene and glass. (Table IX.) 



The figure used in the discussion (fig. 13) is the experiment- 

 ally determined figure for 1100°, determined by the tempera- 

 ture method. It shows that the pyroxenes (K to L) are very 

 much richer in magnesia than the liquids with which they are 

 in equilibrium (G to H). As will appear later, this is uniformly 

 true in the present system except in compositions very close to 

 diopside. 



The three-phase-boundaries for several other temperatures 

 were determined by this method of finding the temperature of 

 beginning of melting for various mixtures. The procedure is 

 to hold the completely crystallized mixture at successively 

 higher temperatures until the temperature is attained at which 

 the first trace of glass is found in the quenched product. The 

 results are given in Table IV. The information given in 

 Table IT is expressed graphically in figs. 14 and 15, i. e. the 

 three-phase-boundaries are drawn. Only those drawn in full 

 lines have been experimentally determined. It was not possible 

 to fix them accurately for the lower temperatures because over 

 a considerable range of composition these lower-melting mix- 

 tures show practically the same temperature of beginning of 

 melting. Dotted lines are given to show the general direction 

 which the undetermined lines must take. 



