448 ACTUAL WORKING IN ORIGINAL RESEARCH. 



of comparing things which require to be practised before 

 we can extensively reaspn respecting them. 



In order to explain the entire collection of the results 

 of a research, we usually require to devise a theory or an 

 idea sufficiently great to include and agree with the whole 

 of the circumstances and results. 



The explanation of results is essentially a logical pro- 

 cess, and especially requires a capacity for accurate in- 

 ference. The power of inference is based upon the 

 universal principle that we may substitute like for like in 

 material phenomena without altering the effects, and like 

 for like in our thoughts without weakening the argu- 

 ment ; two things also which are similar or equivalent to 

 a third one are like or equivalent to each other, and may be 

 substituted in a similar manner. In proportion as two things 

 are alike in the essential points which influence the effect or 

 conclusion, so far may we substitute the one for the other 

 in our experiments, and the conception of one for that of the 

 other in our classifications and reasoning. But frequently 

 the two things are not exactly alike in close essential points, 

 and it usually requires extensive knowledge and experience 

 to be able to judge how far and to what extent they are 

 really similar, and therefore how far the effect or conclu- 

 sion derived from the one may be inferred from the other. 



In order to explain phenomena correctly, we must 

 draw correct conclusions. That great mistakes are fre- 

 quently made in inferring explanations, is proved by the 

 very different and frequently incompatible causes assigned 

 for the same phenomenon by different scientific men. We 

 frequently explain (or rather seem to explain) one mystery 

 by stating another. One of the commonest errors is that 

 of generalising too widely, or drawing conclusions from an 

 insufficient variety or too small a number of instances. 

 The inferences we draw from each observation or example 



