Andersen — System Anortlnte-Forsterite- Silica. 409 



I. Experimental Work. 



The experiments were undertaken on artificial mixtures of 

 pure components. No fluxes were used. All thermal work 

 was done in air under ordinary pressure. 



Preparation of Mixtures. 



The preparations were made up from : Ground quartz puri- 

 fied with HC1 ; alumina treated with a boiling solution of 

 NH 4 C1 ; pure analyzed calcium carbonate, magnesia obtained 

 by igniting the purest precipitated hydrated carbonate. All 

 the substances were dried to constant weight and kept in 

 desiccators. 



Most of the mixtures were prepared in the following way : 

 The compounds were weighed and then carefully mixed by 

 grinding them together in an agate mortar. Charges of 5 gr. 

 were generally used. The finely ground mixture was brought 

 without loss into a 10 cc platinum crucible and heated in the 

 platinum resistance furnace for about one hour at a temper- 

 ature high enough to ensure a complete melting. The melt 

 was then suddenly chilled to form glass, which in turn was 

 crushed to a fine powder and sifted. Then the whole oper- 

 ation of melting, chilling, and powdering was repeated and 

 the powdered glass of the second treatment was crystallized by 

 proper heating. With mixtures rich in silica this required 

 several hours or even several days. With less viscous mixtures, 

 on the other hand, it could be done in a few minutes. During 

 the crystallization the powder sintered together to a solid cake, 

 which was crushed and ground. In this way perfectly homo- 

 geneous mixtures were always obtained. 



In cases where larger quantities of preparations were made 

 up, the melting had to be undertaken in a Fletcher furnace. 

 Otherwise the procedure was the same as that described. 



Thermal Study. 



The platinum-resistance furnace was used in all the thermal 

 work. The temperatures were measured with the Pt : Pt-Rh 

 thermoelement in connection with the usual potentiometer 

 installation of the Geophysical Laboratory.* 



Heating curves. — In the preliminary work heating curves 

 were run on mixtures of the series CaAl 2 Si 2 8 -MgSi0 3 . 

 Charges of about 2 gr. were used, the bare thermoelement 

 being immersed in the charge. The temperature was read 

 every half minute. The results are stated in Table 1. It is 



* W. P. White, Phys. Eev. xxv, 834, 1907. 



