296 Shepherd, Rankin, Wright — Binary Systems of, 



fluidity, as in the case of Al 2 Si0 5 . The best the observer can 

 do at present in the region above 1600° C. is to determine 

 the maxima and eutectics. Determination of the liqnidus is 

 bej^ond onr present facilities. 



There is also a tendency to expect reactions to proceed with 

 the same dispatch which we are accustomed to meet with in 

 the case of aqueous solutions. This has not only never occurred 

 in our experience, but unstable forms often require much urging 

 to bring them into the stable condition. 



The very iirst difficulty which we encounter is in obtain- 

 ing pure homogeneous preparations with which to begin. The 

 natural minerals are rarely pure enough to give constant data. 

 A natural mineral melts at a given temperature, depending 

 (as indicated by the Van't Hoff-Raoult law) upon the amount 

 of impurity which it chances to contain; but the next speci- 

 men, having different impurities and in different amount, will 

 melt at a different temperature. That is, the data obtained 

 apply only to the specimen examined and furnish no basis for 

 determining general relations. 



The Chemical Purity of the Ingredients. — One must begin 

 with the purest possible components. The oxides of lime, 

 alumina and silica on the market are not usually pure enough. 

 The influence of small amounts of imparities, 1 to 3 per cent, 

 cannot, be neglected. A glance at the slope of the freezing 

 point curves shows at once that the presence of one per cent of 

 impurity may cause a variation of five or ten degrees in the 

 melting temperature. 



Calcium carbonate can be obtained very pure, but high purity 

 cannot be taken for granted without testing it. 



The C. P. alumina of commerce is. likely to contain *5 to 

 2*0 per cent of alkalies, and one sample of especially pure (!) 

 alumina contained over 3 per cent of S0 3 . Baker and Adam- 

 son succeeded in making for us a hydrated alumina which 

 contained only *2 to *4 per cent alkali, and this alkali was then 

 reduced to less than one-tenth of one per cent by further 

 purification in platinum. 



Pure silica is obtained from quartz, carefully selected with 

 the microscope, and treated with aqua regia. 



Magnesium carbonate usually contains from 1 to 3 per cent 

 of CaC0 3 , and this impurity is usually not mentioned in the 

 manufacturer's analysis. In this case also, Baker and Adamson 

 produced a special preparation of high purity. 



Homogeneity of the Mixtures. — After pure materials have 

 been obtained, we have next to obtain chemically homogeneous 

 products. It has been found necessary to melt the charge at 

 least three times with fine grinding and mixing between the 

 fusions, in order to obtain a product which is chemically 



