302 Shepherd, JZanTcin, Wright — binary Systems of 



cristobalite and sillimanite are the only phases. The eutectic 

 for Al 2 Si0 5 -Al 2 3 is only very slightly below the melting tem- 

 perature of the compound. There is, however, a rapid rise of 

 the liquidus beyond 65 per cent A1 2 3 , so that in default of a 

 pyrometric method we must depend on the modified Joly 

 method which indicates the location of the eutectic at about 

 64 per cent A1 2 3 . V 



Corundum occurs in all compositions between Al 2 Si0 5 and 

 A1 2 3 , and no phases other than sillimanite and corundum do 

 appear. The optical properties of the sillimanite remain prac- 

 tically unchanged in the presence of Si0 2 or A1 2 3 . We 

 accordingly infer that little or no solid solution of the com- 

 ponent minerals in sillimanite occurs here. 



The corundum shows slightly altered properties, and proba- 

 bly takes up a limited quantity of sillimanite in solid solution. 

 The melting point of corundum lies above the safe working 

 range of the iridium furnace, and it is hardly worth 



while to attempt determinations in the carbon furnace, 

 nature of the equilibrium diagram is shown in fig. 4. 



The 



Fig. 4. 



£ 



/ 



X 



/ 



/ 



Si0 2 



Al 2 SiO s 



Al 2 Oj 



Unstable forms. — The two natural minerals of the same 

 composition as sillimanite are cyanite or disthene, and anda- 

 lusite. Neither of these minerals occurs in igneous rocks and 

 Yernadsky has shown* that both change to something like 

 sillimanite at temperatures above 1300°. He also states that 

 the change is accompanied by an evolution of heat. We have 

 verified the change of form, but have not been able to detect 

 any heat change. Our experience led to the conclusion that 

 *Bull. Soc. Min. ? xii, 446; xiii, 257. 



