to Lord Brougham. 39 



beings tire led to pursue a course of conduct conformable to 

 that which in those endowed with reason proceeds from rea- 

 sonable inference, is another question. It is also another 

 question, what is the difference in the degrees of certainty re- 

 quired to satisfy the mind of the philosopher, and to regulate 

 the conduct of the man. It is sufficiently plain that a wise 

 man will not build his house on the edge of a volcano, how- 

 ever low the presumption may be that its eruptions will last. 



The idea that there is any absolute certainty of future phy- 

 sical events rests on no grounds of reason. Mathematics 

 have an abstract and absolute certainty of their own. The 

 certainty of physics is absolute only as respects facts and laws 

 that actually exist. Our views in regard to the future are 

 necessarily, in natural knowledge, qualified and conditional, 

 for the highest no less than the lowest of the presumptions 

 which it contemplates. We believe that the sun rises and the 

 tide flows, as the volcano smokes, in conformity with laws 

 whether simple or complex, known or unknown, which will 

 not lightly be changed. Among the causes of these effects, 

 no rational philosopher ever overlooked the First: and since 

 we cannot calculate the course of His secret operations, we 

 must be content to allow that however great the difference 

 may be in the value of physical presumptions as establishing 

 actual laws, there is no such thing as physical certainty for the 

 time to come. 



The certainties of the actual laws of nature are attained to 

 by inductive observation. Where there is any regularity dis- 

 cernible in events, a few observations, on the principle above 

 explained, indicate a cause. Then the business of the induc- 

 tive philosopher is to investigate that cause, not by repeating 

 the observation, from which he would gain neither light nor 

 certainty, but by varying it till the possible causes of the event 

 by the process of elimination are reduced to one. Even when 

 for accuracy observations appear to be repeated, the value of 

 the repetition consists in the presumed variation of the cir- 

 cumstances, by means of which the accidental errors of dis- 

 turbing causes destroy one another. 



In experiments, where the circumstances are diversified at 

 our will, and we proceed, in the language of Bacon, to bind 

 the Proteus, and force nature to deliver her oracles, the more 

 dexterous and accurate the experimenter, the less need he has 

 either to repeat, or vary, his experiments. The difference be- 

 tween a learned and an unlearned experimentalist, the ad- 

 vantage which a disciple of Bacon and Newton, and Black 

 and Cavendish and Lavoisier, possesses over men unin- 

 structed in the science of induction, is — that the former has 



