METHOD OF DISCOVERING CAUSAL LAWS 165 



function of experiment. At the same time it is important to bear 

 in mind that all experiment is subservient to, and terminates in, 

 observation. Moreover, in many departments of scientific re 

 search as, for example, in ethics, zoology, geology, astronomy 

 experiment is practically impossible, and accurate observation 

 must be relied on as our only and ultimate channel of experience. 

 240. THE FUNCTION OF EXPERIMENT; DIFFICULTIES OF 

 ANALYSIS. When we remember that an hypothesis is verified by 

 endeavouring to show that it is sufficient and indispensable to ac 

 count for the facts (229), that the supposed cause is proved to be 

 the real cause by showing that no other cause is adequate, we 

 shall realize the necessity of analysing all the surroundings of the 

 phenomenon under investigation, of isolating the latter, of laying 

 aside all that is inoperative and unessential to its presence, in 

 order that we may be able gradually to bring to light its con 

 nexion with all the factors and conditions that really determine 

 its occurrence. This work of analysis mental analysis in the 

 first place, and real analysis, separation, isolation of factors, as 

 far as this is feasible, in the second place is really the most 

 difficult stage in scientific investigation. There may be causes 

 or influences at work, of whose existence we are totally 

 unaware : they may be part of the total proximate cause of the 

 phenomenon. And non-observation is no proof of non-existence. 

 On the other hand, certain circumstances which, in our experience, 

 have always accompanied the phenomenon, need not be in reality 

 causally connected with the latter : uniform coexistence or 

 sequence does not prove causal connexion. Hence our hypo 

 theses may at first include either too little or too much. The only 

 way to obviate these difficulties is by endeavouring to discover 

 fully and accurately ALL the circumstances accompanying the occur 

 rence of the phenomenon, so as to be able to eliminate the accidental 

 and retain the essential ones. It is for this purpose, for securing 

 a full knowledge of all the surroundings of a phenomenon, that 

 repeated observations whether simple or experimental are use 

 ful. If we could be certain, from the observation of a single 

 instance, that all the conditions we observed accompanying it, and 

 none others, were requisite for its production, we could straight 

 way formulate, and regard as established or verified, the natural 

 law connecting that phenomenon with its causes. But this is very 

 rarely, if ever, possible. Hence the need for observing a number 

 of instances. 



