Alumina with Silica, Lime and Magnesia. 301 



when cooled down should consist entirely of the compound 

 AB. By taking the fused charge of composition AB, grind- 

 ing to a very fine powder and heating for a long time between 

 temperatures t 1 and £,, we were successful in bringing about 

 this reaction in the case of 3CaO.Al 2 3 . 



Alumina-Silica. — In this series there is but one compound 

 which is stable in contact with the melt. This is the mineral 

 sillimanite, Al 2 SiO., composed of A1 2 3 62'85 per cent, Si0 2 

 37*15 per cent. This occurs in nature widely distributed. 

 The same compound has been found by Mellor in the crystal- 

 lized glaze of porcelain. The pure artificial compound is 

 colorless and occurs in well crystallized prisms of density of 

 3'031, slightly lower than the natural mineral, which averages 

 3*32. Hardness is 6 to 7. The compound is unaffected by 

 water, hot or cold, and is but very slowly attacked by acids 

 or alkalies. Sillimanite is practically unaffected by HC1, 

 HN0 3 or H 2 S0 4 , hot or cold, or by cold HF. It is decom- 

 posed slowly in mixed HC1 and HF, and by fused JSTa 2 C0 3 . 

 The melting temperature of the pure compound is 1811° C. 







Table I 







a. Melting point fused 



b. 



Melting point 



of mixed and 



sillimanite made from 





heated oxides of the same 



pure 



oxides. 





composition as a. 



Millivolts 



T 





Millivolts 



T 



•475 



1812° 





•472 



1802 u 



•475 



1812 





•485 



1850 



•474 



1810 





•483 



1850 



•474 



1810 





•483 



1850 



Mean 1811° 





•479 



1833 









•480 



1835 



Table I, h, is given to show how great an error is introduced 

 by observing incompletely combined oxides. These charges 

 had been heated several times to about 1600° in the gas fur- 

 nace, but still showed free silica and alumina. After fusion, 

 they gave concordant results. The flash is also much sharper 

 when the oxides are properly combined. It will be noticed 

 that the observed melting temperature varies irregularly, and 

 is not altogether dependent upon the rate of heating. 



The eutectic Al 2 Si0 5 -Si0 2 is very hard to place, owing to 

 the extreme viscosity of the silica. Heated for one hour at 

 1550° in a platinum furnace, nearby mixtures show no evidence 

 of fusion. The 15 per cent and 20 per cent A1 2 3 charges show 

 traces of fusion before pure Si0 2 . The eutectic must there- 

 fore fall, at about 10 per cent and melt slightly below 1600° C. 

 Throughout the range of concentrations from Si0 2 to Al 2 3 .Si0 2 



