RECENT ADVANCES IN SCIENCE 39 



in that it usually contains some ferrous oxide and is also 

 hydrated. Natural cordierite, heated to 1440 C. for thirty 

 minutes, breaks down to glass and sillimanite in a manner 

 similar to the behaviour of the form of the ternary compound 

 mentioned above. 



The application of these results to petrology calls attention 

 to several important problems which have not yet been solved ; 

 these include the nature of the alumina-bearing molecule in 

 the non-alkaline pyroxenes *and amphiboles, the part played 

 by water in many pyrogenetic minerals, and the influence 

 of ferrous oxide in the ferromagnesian metasilicates. The 

 ternary compound mentioned above can hardly constitute 

 the aluminous molecule in the pyroxenes, as, though it can 

 apparently take up silica, and to a small extent spinel, in solid 

 solution, it shows no signs of being isomorphous with the 

 metasilicate. As no other ternary compound is stable in con- 

 tact with the melt, the alumina-bearing molecule isomorphous 

 with the metasilicate must, so far as the dominantly magnesian 

 pyroxenes and amphiboles are concerned, be one which dissoci- 

 ates below its melting-point. Hence the fact that Tschermak's 

 hypothetical silicate, which has been suggested as the aluminous 

 molecule in these minerals, dissociates on fusion (Sitz. Akad. 

 Wiss. Wien. 121, 897, 191 2) is not a valid argument against its 

 existence. 



The instability of MgAl 2 Si0 6 at high temperatures has been 

 verified by V. Schumoff-Deleano (Centr. Min. 191 7, 290), who 

 has attempted to determine the constitution of aluminous 

 augite by synthetic means. This writer finds that diopside 

 (CaMgSi 2 6 ) can take up 1 5 per cent, of alumina in solid solution, 

 a similar result being obtained in the case of enstatite (MgSi0 3 ) 

 and pseudowollastonite (CaSi0 3 ). So far as enstatite is 

 concerned, this result is at variance with the work of G. A. 

 Rankin and H. E. Merwin, who state (loc. cit. p. 307) that in 

 the ternary system MgO — A1 2 3 — Si0 2 , magnesium meta- 

 silicate is a compound of constant composition, while the result 

 for the lime compound is contradicted by the work of G. A. 

 Rankin and F. E. Wright (Amer. Jour. Set. 39, 1, 191 5), who 

 find that the amount of alumina that CaSi0 3 can hold in solid 

 solution is less than 1 per cent. So far, the experimental 

 evidence is insufficient to allow discrimination amongst the 

 numerous theories that have been advanced. 



