628 THE PRINCIPLES OF SCIENCE. [CHAP. 



are similar. There is no identity of nature between a word 

 and the thing it signifies ; the substance iron is a heavy 

 solid, the word iron is either a momentary disturbance of 

 the air, or a film of black pigment on white paper ; but 

 there is analogy between words and their significates. 

 The snbstance iron is to the substance iron-carbonate, as 

 the name iron is to the name iron-carbonate, when these 

 names are used according to their scientific definitions. 

 The whole structure of language and the whole utility of 

 signs, marks, symbols, pictures, and representations of 

 various kinds, rest upon analogy. I may hope perhaps 

 to enter more fully upon this important subject at some 

 future time, and to attempt to show how the invention of 

 signs enables us to express, guide, and register our thoughts. 

 It will be sufficient to observe here that the use of words 

 constantly involves analogies of a subtle kind ; we should 

 often be at a loss how to describe a notion, were we not 

 at liberty to employ in a metaphorical sense the name of 

 anything sufficiently resembling it. There would be no 

 expression for the sweetness of a melody, or the brilliancy 

 of an harangue, unless it were furnished by the taste of 

 honey and the brightness of a torch. 



A cursory examination of the way in which we popu- 

 larly use the word analogy, shows that it includes all 

 degrees of resemblance or similarity. The analogy may 

 consist only in similarity of number or ratio, or in like re- 

 lations of time and space. It may also consist in simple 

 resemblance between physical properties. We should not 

 be using the word inconsistently with custom, if we said 

 that there was an analogy between iron, nickel, and 

 cobalt, manifested in the strength of their magnetic 

 powers. There is a still more perfect analogy between 

 iodine and chlorine ; not that every property of iodine is 

 identical with the corresponding property of chlorine; 

 for then they would be one and the same kind of sub- 

 stance, and not two substances; but every property of 

 iodine resembles in all but degree some property of chlo- 

 rine. For almost every substance in which iodine forms 

 a component, a corresponding substance may be dis- 

 covered containing chlorine, so that we may confidently 

 infer from the compounds of the one to the compounds 

 of the other substance. Potassium iodide crystallises in 



