iOi Prof. Whewell on Isomorphism, 



calc-spar any considerable per cenlage of silica? If we wish 

 to know the chemical constitution of minerais, we must take 

 those which are apparently simple, pure, and of determinate 

 character. If we take those which are plainly impure, mixed, 

 and indefinite, we are not to be surprised if we find the che- 

 mical composition also perplexed, variable, and anomalous. 

 What can we learn by looking, in the first instance, to cases 

 which we know to be exceptions to all rule, and which no 

 theory pretends, or can be expected, to include ? 



The fact which has hitherto so perplexed mineralogists is, 

 not that impure specimens are of anomalous composition ; but 

 that specimens, apparently pure, homogeneous, and well cry- 

 stallized, vary in their ingredients in an unaccountable man- 

 ner : and it is this variation which the doctrine of isomorphism 

 undertakes to reduce to laws ; so that the analyses of all well 

 selected specimens of the same species shall, within reasonable 

 narrow limits, agree with the theoretical constitution. 



If this theory do not teach us the nature of mineral species, 

 it at least promises fairer than that of accidental mixture. If 

 we learn little by comparing our analyses with a standard with 

 which we conceive them to agree (the isomorphous composition 

 of the species) we shall learn still less by comparing them with 

 an assumed standard (the essential ingredients) from which 

 it is allowed that they may differ widely and by an unknown 

 quantity. If the half of our specimen may be something dif- 

 ferent from what it seems, it does not appear what hope we 

 can have of connecting what seems with what is in minerals. 

 If extraneous substances to the amount of 50, 60, or 70 per 

 cent, may be " cemented together and cased up " in the pure 

 mineral, without our senses detecting the want of simplicity, 

 how can we ever distinguish between the case and its contents? 



I shall suppose then that when we analyse a well-crystal- 

 lized, well-characterized mineral, the analysis gives us an 

 approximation to a chemical constitution belonging to the 

 species. If this be not so, and if we do not get an approxi- 

 mation this way, it appears pretty clear that we shall not 

 obtain such information in any other manner. Any theory 

 which, like that of accidental mixture, supposes that this may 

 he no approximation at all, not only disables us from inferring 

 anything from one analysis, but makes any possible conclu- 

 sion from any number of analyses, entirely arbitrary, conjec- 

 tural, and precarious in the first place; and in the next, utterly 

 inapplicable to any new instance : while in some species, as 

 we have mentioned with regard to garnet, we can absolutely 

 prove a negative against that theory, and show that no one 

 possible chemical constitution can answer to the various re- 

 sults of analysis, even allowing admixture to any extent. 



II. On 



