ONE SCIENCE DEPENDS ON ANOTIIEK. 223 



confirmation which wore repeated in respect to the electric 

 and magnetic forces. Thus again the discovery of the 

 polarization of light led to experiments which ended in the 

 discovery of the polarization of heat a discovery that 

 could never have been made without the antecedent 

 one. Thus, too, the known refrangibility of light and 

 heat lately produced the inquiry whether sound also is not 

 refrangible ; which on trial it turns out to be. 



In some cases, indeed, it is only by the aid of concep 

 tions derived from one class of phenomena that hypoth 

 eses respecting other classes can be formed. The theory, 

 at one time favoured, that evaporation is a solution of 

 water in air, was an assumption that the relation between 

 water and air is like the relation between salt and water ; 

 and could never have been conceived if the relation be 

 tween salt and water had not been previously known. 

 Similarly .the received theory of evaporation that it is a 

 diffusion of the particles of the evaporating fluid in virtue 

 of their atomic repulsion could not have been entertained 

 without a foregoing experience of magnetic and electric 

 repulsions. So complete in recent days has become this 

 consensus among the sciences, caused either by the natural 

 entanglement of their phenomena, or by analogies in the 

 relations of their phenomena, that scarcely any consider 

 able discovery concerning one order of facts now takes 

 place, without very shortly leading to discoveries concern 

 ing other orders. 



To produce a tolerably complete conception of this pro 

 cess of scientific evolution, it would be needful to go back 

 to the beginning, and trace in detail the growth of classifi 

 cations and nomenclatures ; and to show how, as subsidiary 

 to science, they have acted upon it, and it has reacted upon 

 them. We can only now remark that, on the one hand, 

 classifications and nomenclatures have aided science by con 

 tinually subdividing the subject-matter of research, and giv- 



