ROCK CLASSIFICATION 103 



for any given percentage of silica is much greater the lower the 

 percentage of silica. 



Diagram 1 shows that within the limitations above men- 

 tioned there are nearly all possible transitions in relative 

 proportions of alkalis and silica, and the same is true for other 

 constituents. Diagrams 2, 3 and 4 show that the range of 

 alkali variation is greatest for rocks in which the ratio of the 

 soda to the potash is greater than 2, and least for those in which 

 it is less than 1. That is, it is greatest for distinctly sodic 

 rocks and least for those very rich in potash. 



While there is a clustering of analyses in the parts of the 

 diagrams corresponding to lower alkali-silica ratios, there are no 

 loci of specially marked aggregation, except toward the most 

 siliceous end of the more potassic groups, and no such cluster- 

 ing as to suggest natural subdivisions of the analyses on any 

 chemical basis. A chemical subdivision must be an arbitrary 

 one, making breaks in continuous series. Comparison of 

 Diagrams 2, 3 and 4 shows that the most siliceous rocks are 

 generally richer in potash than in soda, also that the rocks in 

 which alkalis decrease to nearly zero contain much more soda 

 than potash. Those parts of the diagrams in which few anal- 

 yses occur will undoubtedly be filled up as more rocks are 

 analyzed. 



It is to be noted that analyses occurring in close proximity 

 to one another in these diagrams, and nearly alike in percentages 

 of silica and alkalis, may in some cases differ more or less 

 widely in other respects, so as to belong properly to different 

 classes of rocks. They are by this method of comparison the 

 more strongly contrasted and their essential differences more 

 clearly recognized. This method of comparison brings closely 

 together rocks agreeing in chemical composition, but often 

 differing in mineral composition and physical characters, and 

 having quite different names, and in this way emphasizes the 

 necessity for improved definitions of names already in use and 

 for the creation of a systematic nomenclature. 



The occurrence of these constituents in any rock as a result of proc- 



