CHEMICOMINERA L RELA TIONSHIPS IN RO CKS 233 



all the alkali were soda, by the partial replacement of soda by 

 potash would be to render it lower in the scale toward the anor- 

 thite end, so that the amount of orthoclase would be increased 

 and the accompanying lime-soda-feldspar would be more calcic. 

 This is indicated in Diagram 3. As potash replaces soda, the 

 extreme limit of pure alkali-alumina silicate would shift along 

 the line NL toward L. When they are in equal amounts the 

 limit would be at P. At the same time, the range of rocks 

 consisting wholly of anorthite, with albite and orthoclase in- 

 equal proportions, would be indicated by the line xAn. The 

 points b' , c' , d' , etc., on it are where the lime-soda feldspar 

 would have the value of Ab 6 An 1( Ab 3 An a , AbjAnj, etc. The 

 red lines connecting these points with P on one side and with Q 

 on the other represent the range of rocks composed wholly of 

 these particular lime-soda feldspars and leucite-nephelite on the 

 one hand, and of these feldspars and quartz on the other. The 

 replacement of anorthite molecules by ferromagnesian ones will- 

 have the same effect as that just stated for each extreme of 

 alkali-lime-feldspar magmas. 



IV. There remains to be discussed the modifications that 

 would be effected in the feldspathic magmas already postulated 

 by the introduction of ferroaluminous molecules. These may 

 be of two kinds : ferromagnesian-ferroaluminous molecules,, 

 appearing in pyroxenes and amphiboles, with the general formula 

 (Mg,Fe) (Al,Fe) 2 SiO G ; and alkali-ferroaluminous molecules,, 

 occurring in mica, amphiboles, and pyroxene. Among these are 

 molecules free from ferric iron, and others free from aluminium.. 

 They are KHgAlgSigO, 3 , (K,H)(Al,Fe) Si0 4 , NaAlSi 3 6 , 

 NaFeSi 3 6 . 



Let us first consider those free from alkalis. Upon the gen- 

 eral assumption that the alkalis control an equal amount of 

 alumina, the introduction of ferromagnesian-ferroaluminous 

 molecules would demand an increase in the alumina over that 

 necessary to satisfy the alkalis. This must take place at the 

 expense of the ferromagnesian and calcium molecules in cases 

 I and II, or at the expense of anorthite molecules, in case III.. 



