444 SCIENCE PROGRESS. 



But the curious feature which particularly deserves 

 notice is this : In proportion as our mineral classification 

 becomes more purely a chemical one as interpreted by 

 isomorphism, so does it depart more and more widely from 

 what was originally meant by a chemical classification ; as 

 our views on isomorphous replacement become extended, 

 and radicles which are apparently diverse are recognised as 

 capable of mutual replacement, so is the genetic kinship of 

 minerals as referable to certain acids or bases more and 

 more lost from sight, until ultimately a classification of 

 crystallisable substances on mineralogical principles bears 

 little resemblance to the classification of chemical text- 

 books. 



It must be borne in mind that the purpose of systematic 

 mineralogy is not to co-ordinate those characters which are 

 useful for determinative purposes ; yet the latter have 

 always exercised an influence upon the classification. At 

 the beginning of the century the external characters alone 

 were used for determination, they were also the basis of 

 classification ; somewhat later the necessity of chemical 

 analysis was fully recognised, and the system became purely 

 chemical. It may be that now, since improvements in 

 physical methods render it possible to identify a mineral 

 with accuracy by optical characters, specific gravity, and 

 form, as is constantly done in petrographical research, 

 determinative mineralogy does not lend such strong support 

 as formerly to a rigidly chemical definition and classification 

 of mineral species. 



To be bound now to any system which necessarily 

 classes together all silicates or all oxides is only to be 

 hampered by preconceived interpretations of the analyses. 

 The symptoms of dissatisfaction with such a system are 

 widespread, and in extreme cases almost express themselves 

 by an apparent return to the old Natural methods in which 

 the composition was of no account ; well-defined natural 

 groups are accepted on grounds of general similarity in 

 characters, and attempts are then made to explain the 

 chemical constitution as also indicative of the same 

 similarity. 



