84 ALFRED C. LANE 



rocks or noticeable subdivisions of chemical series, that chemi- 

 cally similar rocks occur in genetically different families, that it 

 follows that the subdivisions of all igneous rocks into groups for 

 the purpose of classification must be on arbitrarily chosen lines, 

 and that it is no argument against a classification if a rock of 

 great importance belongs on the boundary of a classificatory 

 division, that any system of classification will be as natural as 

 any other system, he is rather too pessimistic. Merely as a mass 

 of raixed salts, there are certain common relationships which 

 hold, and of which any system of classification, the new one 

 included, must take account. He has pointed out a number of 

 such facts: (i) In the great majority of rocks alkalis do not 

 exist in excess of that required to make feldspar or nephelite ; 

 (2) The commonest rocks are in composition like the average of 

 all rocks which has the silica percentage of 58.7 to 59.77 and an 

 alkali-silica ratio of 0.083 to 0.088. Iddings, in fact, suggests 

 the possibility that all rocks have been derived from one com- 

 mon magma by splitting. If so, this splitting must have gone 

 on under chemical and physical laws which are to be traced and 

 followed in the classification. There is, however, another possi- 

 bility which I wish to suggest, namely, that in the process of 

 splitting there is a tendency in one of the fractions toward that 

 same common average.^ To this point we will return later. 

 There are numerous other facts which Iddings points out, such 

 as the inverse relation between silica and alkalis, and iron and 

 magnesia, and various other notes on pp. 19, 64, 65, 70-81, of 

 which any classification may take account, as, in fact, the new 

 one does, to a very large degree. Upon that depends its ser- 

 viceability. The fact that there are exceptions does not invali- 

 date the importance of this relation, provided they are not too 

 many. 



Now, if one looks over the general diagram, Plate I, which 

 Iddings has prepared to see that it is true that there is "no clus- 

 tering of analyses" and " no natural subdivisions" (p. 17), I can- 

 not agree with him in the conclusion he draws. In the first place, 



'This would fit especially well in a planetesimal theory; in fact, would be almost 

 necessary in view of the wide prevalence of basaltic magmas. 



