82 Mr. M. M. Pattisoii Muir on Chemical Classification. 



deal concerning the new object. The difficulties of such a 

 classification are also apparent. We may find certain proper- 

 ties correlated together in two individuals ; these individuals 

 we form into a class. Yet we may discover another individual 

 having properties which almost oblige us to admit it into this 

 class, and nevertheless in one or more of its properties it shall 

 differ from the 100 individuals originally examined. For 

 instances, the metals formed a tolerably well-defined class of 

 substances before the time of Davy. One of the apparently 

 best established characteristics of this class was high specific 

 gravity ; this mark was correlated in all the known metals 

 with malleability, lustre, a certain position in the electric 

 series, &c. But Davy discovered two substances which were 

 possessed of specific gravities less than that of water, but 

 which the chemist was nevertheless obliged to admit into 

 the class of metals. 



The mark, the characteristic of a class, is not therefore 

 always that property which strikes the casual obsener as in a 

 high degree diagnostic of the class. 



There is another allied source of difficulty in classification : 

 although we have examined with the greatest care the indi- 

 vidual objects forming a class, and have established the univeisal 

 correlation of certain properties, it is nevertheless possible that 

 new substances may be discovered v>'hich shall exhibit the whole 

 of these properties save one. We shall probably be obliged to 

 admit the new substances to the class which we have formed ; but 

 Ave shall at the same time be obliged to define anew the proper- 

 ties of the members of the class. Of course it would be possible 

 to make a new class for the newly discovered substances ; but 

 if the general analogies of these substances point to a close con- 

 nexion between them and the members of the class already estab- 

 lished, the formation of a new group, for the sake of maintain- 

 ing unchanged the old definition, would be equivalent to doing 

 away with one of the chief advantages of a rational system of 

 classification. 



2. It appears, therefore, as if every system of classfi cation 

 of natural objects and phenomena must of necessity be subject 

 to frequent, and sometimes to radical alterations. Each scheme 

 of classification is a more or less perfect representation of the 

 state of knowledge in that branch of science to which it ap- 

 plies at the time of its production. 



With the advance of knowledge the scheme of classification 

 must change. Moreover there must be minor and subsidiary 

 systems of classification underlying the wider and more gene- 

 ral scheme. To take an instance from chemistry: the modern 

 system of classification is mainly founded on the valency of 



