310 CO. E. Tilley—Para-Gneisses vn South Australia. 
The occurrence of sillimanite in these rocks shows that it has 
arisen (a) from sericite 
(3) H,KAI,8i,0,, + Si0, = KAISi,O, + Al,Si0, -- H,O 
This equation explains the genetic association of needles of 
sillimanite in orthoclase where ferromagnesian constituents are not 
present. 
(b) from biotite. 
In these cases the sillimanite is intimately associated with garnet, 
and in certain cases the ferromagnesian constituents have been 
absorbed in spinel to the exclusion of garnet. 
The equation representing this association is of the type 
(4) {rteieisio,) * | = 3KAISi,0, + MgFe(A],0,), + Al,Si0, ++ 
MgFe(Si0,) 3H,0 
It is not improbable that some of the magnetite associated with 
sillimanite in similar cases has been derived by a degradation of 
the biotite molecule. 
The above equation is of a type expected only in the highest 
grades of metamorphism, and it is of interest to compare it with the 
products obtained by direct fusion of biotite. Backstrém, on fusion 
of biotite, found among the products olivine, leucite, spinel, and 
glass, and Doelter records the same products with the exception of 
leucite. The actual products obtained would appear to depend on 
the composition of the mica, and in highly aluminous micas olivine 
would probably not appear. According to Vernadsky ! muscovite 
yields corundum and sillimanite on fusion. These products are 
very similar to those shown in the equation above, and afford an 
interesting parallel to the processes at work in high-grade 
metamorphism. 
_ The occurrence of spinel calls for some remark. Whenever this 
mineral is present it is surrounded by a shell of garnet or sillimanite, 
and it has not been observed in direct contact with quartz, although 
this may be abundantly present in the rock. This is as might be 
expected, as its union with silica would give rise to cordierite 
according to the equation 
(5) (Mg, Fe)(AJ,0,). + 5810, = (Mg, Fe)(O),.2A1,0,.58i0, 
This reaction is doubtless a reversible one, and if so the dissociation 
temperature must exceed those temperatures normally obtaining 
in high-grade metamorphism.? 
A spinel of composition near hercynite, under the same condition, 
would yield almandine and sillimanite, assuming the pure iron 
cordierite has no stable existence. 
(6) 3FeAl,0, + 5810, = Fe,Al,(Si0,), + 2A1,810, 
Although cordierite has been carefully looked for in these rocks, 
no trace of it has been observed. 
1 Cf. Clarke, Bull. U.S.G.S., No. 616, p. 396. 
* Artificial cordierite dissociates on melting at atmospheric pressures 
(Rankin and Merwin, Amer. Journ. Sci., vol. xlv, 1918, pp. 301-25). 
