102 J. P. ID DINGS 



from silicate minerals, such as certain iron ores 1 whose analyses 

 occur at the extreme lower right hand corner of Diagram I. 

 From these it is evident that molecules of iron oxides may 

 separate by differentiation from silicate molecules. The maxi- 

 mum limit for soda will be when the soda-silica ratio is 0.5, 

 that is when Na0 2 =21.75 P er cen t- and Si0 2 =42.10 per cent. 

 The maximum limit for potash will be when the potash-silica 

 ratio is 0.25, that is when K 2 = 2i.5i percent, and Si0 2 =54.92 

 per cent. Both alkalis sink to a minimum of zero in certain 

 kinds of peridotites, pyroxenites, and in certain eruptive iron 

 ores. They may disappear in anorthosite composed of pure 

 anorthite. 



The limit of distribution of analyses toward low silica appears 

 to be a line which would correspond to a variable mixture of 

 the lowest alkali-alumina sil.icate — nephelite for soda magmas, 

 and leucite for potash-magmas — and the least siliceous ferro- 

 magnesian silicate, fayalite. This limit will probably be modified 

 when series of rocks are analyzed grading from gabbros to the 

 iron ores. The range of variations in the proportions of iron 

 oxides, magnesia, lime and alumina, does not appear in these 

 diagrams. 



In this connection it may be pointed out that the variations 

 in all of the chemical constituents, other than silica, must 

 increase in proportion as silica decreases. When 75 per cent, 

 of the magmas is silica, only 25 per cent, remains for other con- 

 stituents, but when there is only 40 per cent, of silica, 60 per 

 cent, must consist of other compounds. It is known that rocks 

 high in silica contain much more alumina and alkalis than 

 other constituents, hence the variation in these other constitu- 

 ents cannot be great. The greatest variation occurs when part 

 of the alumina is replaced by ferric oxide. But in rocks low 

 in silica there is no general law controlling the other constitu- 

 ents, which may, therefore, vary widely in different rocks. 

 Consequently the number of different kinds of rocks possible 



1 Vogt, J. H. L., Bildung von Erzlagerstatten durch Differentiationsprocesse in 

 basischen Eruptivmagmata, Zeitschr. fur prakt. Geol., Berlin, 1893, Jan., Apr. 



