274 H. S. Summer. 



Total 



I. Average of analyses of dacite from Cherokees, Willimigongong Creek 

 and Braemar House. 

 II. Average of analyses of granite from Dog Eocks, Geeloug, and Cape 

 Woolamai. 



The lines drawn tlirough the points representing the percentage 

 of the oxides are produced both ways. It will be seen that in the 

 direction of increasing silica percentage the magnesia line vanished 

 at 78 per cent, silica, and is closely followed by tlie lime. It is 

 considered that a rock containing 78 per cent, silica is the most 

 acid product that this magma could produce under this type of 

 differentiation. A more acid rock than this could, however, be pro- 

 duced by a complementary subdivision of one of the possible pro- 

 ducts of the linear type of differentiation. 



In the direction of decreasing silica percentage the potash line 

 vanished at approximately 50 per cent, silica, and this point is a 

 possible limit of differentiation in the basic direction. In working 

 out other variation diagrams it has been found that whereas the 

 total alkalies may show a linear type of variation the soda and 

 potash lines may become curved. In this case, then, there is the 

 possibility that differentiation may not reach a limit until the total 

 alkalies become nil at about 40 per cent, silica. In reading of? any 

 composition below 50 per cent, tlie soda lino must lie disi-egarded 

 and the soda i-ead off from the total alkalies line. Compositions 

 have been read off the diagram for every 5 per cent, increase of 

 silica, and are given in Table VII., together with the most acid 

 differentiation product containing 78 per cent, of silica. 



