554 



N. L. BOWEN 



excess of silica amounting to about lo per cent, no corundum would 

 be formed from sUlimanite, because the join of sillimanite with such 

 a composition misses the corundum field. Plainly the freeing of 

 corundimi from sillimanite depends in these liquids on the Uquid 

 being "basic," that is, haAdng not more than a moderate excess of 

 silica over the feldspar composition. Free silica associated with 



Fig. II. — Equilibrium diagram of the system CaO-Al203-Si02 

 and Wright.) 



(After Rankin 



the sillimanite does not have a comparable effect in restricting the 

 formation of corundum; indeed, the silica would require to be 

 nearly equal in amount to the sillimanite in order to neutralize the 

 tendency to form corundum. 



The system MgOrAlaOsrSiOa given in Figure 12 shows the same 

 general condition. If a mass of sillimanite were immersed in 

 molten MgSi03 and Mg2Si04 in equal parts, a layer around the 

 inclusion, which might have the bulk composition 75 per cent silli- 

 manite, 25 per cent the above mixture, would consist at higher 

 temperatures of corundum and liquid; at lower temperatures, of 



