The Ternary System MgO-Al,0,-SiO t . 307 



liar circumstance, however, that the excess glass from 

 compositions containing between 65% A1 2 3 and 85% 

 A1 2 3 did not crystallize readily even when slowly 

 cooled from the high temperature (2100°) obtained in 

 the iridium furnace, it was possible to obtain A1 2 3 

 (corundum) and spinel 6 each alone imbedded in glass in 

 certain compositions. This enabled us to determine 

 more accurately than would otherwise have been possi- 

 ble the limits of certain portions of the fields of A1 2 3 and 

 spinel. Aside from these data the limits of the portions 

 of the fields mentioned which are at temperatures too 

 high for direct study were determined indirectly from 

 certain theoretical reasonings based on experimental 

 evidence. This evidence and theory will be taken up 

 later when we come to consider the boundary curves and 

 quintuple points. 



Fields of Unstable Compounds. 

 Clinoenstatite, MgO.Si0 2 , and cordierite have no 

 definite melting points, each being a crystalline substance 

 of such a nature that when pure it dissociates when melt- 

 ing begins, therefore their compositions lie outside of 

 their respective fields. The determinations of the limits 

 of these two fields involved no especial difficulty since 

 each lies wholly within the range of temperatures possi- 

 ble in the platinum quenching furnace. The determina- 

 tion of the composition of such unstable crystalline 

 substances may be, however, a matter of extreme diffi- 

 culty. Of the two such substances which we have to 

 consider in this ternary system one, MgO.Si0 2 , is a 

 definite compound of constant composition; the other, 

 cordierite, is affected by solid solution, and is so fine- 

 grained when crystallized from a glass at temperature 

 below that at which it begins to melt, that its homegeneity 

 cannot be accurately tested. (See p. 316.) 



Melting Tempebatuees Within" the Fields. 



Of the eight fields of the diagram two have melting 



temperatures all of which may be determined in the 



platinum quenching furnace. These are the fields of 



MgO.Si0 2 and the ternary compound (cordierite). 



6 Mieroscopial examination of various iridium furnace melts containing 

 only spinel and a small percentage of glass would seem to indicate that 

 spinel in ternary mixtures takes up a small amount of Si0 2 in solid solu- 

 tion. This together with the extent of the solid solution between ALO :! and 

 MgO.ALO : , in the binary system is indicated in fig. 1. 



Am. Jour. Sci.— Fourth Series, Vol. XLV, No. 268.— April, 1918. 

 22 



