2o8 HOLLAND: SIVAMALA1 SERIES. 



form of corundum if at the same time there is no considerable 

 quantity of magnesia and iron present, whilst spinel (or spinel and 

 corundum) is formed if there is more than 0*5 per cent, of MgO 

 and FeO contained in the magma. Reference to the analysis on 

 page 34 will show that these conditions with regard to magnesia and 

 ferrous oxide are exceeded in the case of the prevalent form of 

 elaeolite-syenite at Sivamalai, and that corundum would not have 

 separated from it unless the percentage of alumina had been 

 sufficiently increased to satisfy all the bases. The presence of 

 considerable quantities of ferromagnesian minerals in the elaeolite- 

 syenite and the remarkable absence of these compounds from the 

 associated corundiferous rock would thus alone account for the 

 separation of free alumina in the latter case, and its absence from 

 the elaeolite-bearing rock. 1 



The corundum is very unevenly distributed through the coarse- 

 grained felspar-rock and the material collected is insufficient to 

 obtain an analysis representative of the whole rock. We are not 

 able, therefore, to state the amount of alumina in the magma from 

 which this rock crystallized, but it was probably not much greater 

 than that contained by the elaeolite-syenite, and the mineralogical 

 difference between the two rocks would thus be due to the higher 

 proportion of other bases contained in the elaeolite-syenite. So far as 



1 It is not unlikely that the law of Morozewicz referred to above does not 

 represent the whole truth ; for magnetite sometimes occurs locally in the corundi- 

 ferous rock in quantities sufficiently large to represent more than 0*5 per cent, of 

 FeO. It is possible that when Fe 2 3 is also present, magnetite may form instead 

 of hercynite (FeO. A1 2 3 ). I have frequently found these two minerals together, 

 with the magnetite included by the hercynite, indicating that FeO. Fe 2 3 forms 

 before FeO. A1 2 3 , and it is possible, therefore, that when there is enough Fe 2 O s 

 to unite with the FeO, the alumina must (a) unite with some other available pro- 

 toxide to form a spinelloid, (b) form other alumino-silicates, or, if still in excess, 

 (c) separate as corundum. Nevertheless, it is only locally that magnetite is 

 abundant in the corundiferous felspar-rock. An average analysis would show 

 less than 0*5 per cent. FeO. 



( 40 ) 



