402 Prof. Whewell on Isomorphism^ 



as an illustration, " that amphibole consists essentially of a 

 single atom of trisilicate of lime, and that all else which may 

 be discovered by analysis is accidental mixture." 



The difficulties of this theory appear to me to be absolutely 

 unconquerable; and if such were the best conjecture we could 

 make concerning the constitution of our mineral specimens, 

 we should be compelled, I think, to despair of ever attaining 

 any distinct or consistent knowledge of them. For if we take 

 any table of the analyses of amphibole, for instance that in 

 Leonhai'd's Handhuch^ we find, besides the lime and silica, 

 which Mr. Brooke supposes to be essential, a proportion of 

 magnesia varying from 2 to 25 per cent., protoxide of iron 

 from to 20 per cent., alumina from to 26 per cent, and 

 various other ingredients. The smallness of the amount of 

 these ingredients in some cases, shows that, on any hypothesis 

 which looks to essential elements, they are not essential; 

 while the largeness of their amount in other cases shows that 

 we can acquire no useful or applicable knowledge of the com- 

 position of minerals by taking up a view in which they go for 

 nothing. 



If we are to neglect 25 per cent, of magnesia or of alumina 

 in some specimens of amphibole as accidental mixture, by 

 what right do we take account of the lime as essential, which 

 in no case amounts to more than 14 j per cent.? The che- 

 mical constitution of a mineral according to such a view must 

 be a matter of mere assertion ; for it not only does not ap- 

 proximately represent all the good analyses (which is what it 

 ought to do), but it does not represent any one, within any rea- 

 sonable limits. It does not represent the analyses any better 

 than a dozen other suppositions would do taken at random. 

 It appears to be, on this theory, a matter of ai'bitrary assump- 

 tion which part of the analysis is to be supposed essential 

 and which accidental : and if we could not find any better 

 mode of considering our analyses, I cannot conceive what 

 reason there could be for ever analysing a mineral at all. In 

 interpreting the result, we should always have the difficulty 

 which Dr. Johnson complained of with regard to an inaccurate 

 narrator; — it is of no use to be content with believing half of 

 what he says, for we do not know "which half. 



In some cases we may go further, and show that there 

 cannot be any essential composition, common to all the speci- 

 mens of the same species. What are the essential ingredients 

 of garnet? If we again turn to Leonhard's table, we find 

 silica indeed common to all the analyses, but not one single 

 ingredient besides. Alumina, lime, protoxide of iron, per- 

 oxide of iron, protoxide of manganese, are the most common 



of 



