47 8 SCIENCE PROGRESS. 



to important exceptions, presumably result from fundamental 

 principles upon which we can at present only speculate ; but 

 it seems safe to assert that they will be embodied in any 

 natural system of classification which may ultimately be 

 arrived at. 



Since we cannot have a chemical analysis of every rock 

 it is clear that the practical application of a chemical classi- 

 fication must be in general an indirect process. The object 

 kept in view must be to construct such a system, based on 

 mineralogical characters, as shall most closely correspond 

 with the chemical grouping taken as a guide. The rela- 

 tively small number of the common rock-forming- minerals 

 — a consequence of the chemical laws already alluded to — 

 facilitates such a scheme, but there are obvious considera- 

 tions which limit its application. In the first place the 

 mineralogical analysis of a rock is necessarily a qualitative 

 rather than a quantitative one. This difficulty is only very 

 partially evaded by ranking some mineral constituents as 

 " essential " and others as " accessory," or by trying roughly 

 to take account of the relative proportions of the essential 

 minerals in a rock. The implied discrepancy, however, 

 between a rough mineralogical and an accurate chemical 

 analysis is perhaps smaller than it appears. It may even 

 be argued that, where the two estimates differ, a mineral- 

 ogical is not less likely than a chemical characterisation to 

 lead to a true interpretation of relationships, for in cases 

 where the relative proportions of the minerals vary widely 

 {e.g., in gabbros) we often find the extreme types forming 

 parts of one and the same body of rock. 



A more difficult question arises in the case of the vol- 

 canic rocks, most of which contain more or less unindivid- 

 ualised glassy matter. A mineralogical description, how- 

 ever complete, cannot express the chemical composition of 

 a rock which consists only partly of minerals. It may be 

 said that the minerals which are actually found give an in- 

 dication of the composition of the magma from which they 

 have crystallised out ; but, granting this, can we safely as- 

 sume that that magma had the same composition as the rock 

 finally consolidated? The case has been well put by Judd 



