PHYSICAL METHOD. 571 



obsen-ed facts, to fit them for being rapidly and accurately collated, 

 sometimes even for being collated at all, with the conclusions of theory. 

 This preparatory treatment consists in finding general propositions 

 which express concisely what is common to large classes of observed 

 facts : and these are called tho ompiricd laws of the phenomena. We 

 have, therefore, to inquire, whether any similar preparatory process 

 can be performed upon tlie facts of the social science j whether there 

 are any empirical laws in history or statistics. 



In statistics, it is evident that empirical laws may sometimes be 

 traced ; and the tracing them forms an impt)rtant part of that system 

 of indirect observation on which we must often rely for tho data of the 

 Deductive Science. The process of the science consists in inferring 

 eftects from their causes ; but we have often no means of observing 

 the causes, except through the medium of their eflects. In such cases 

 the deductive science is unable to predict the eflects for want of the 

 necessary data ; it can tell us what causes are capable of producing 

 any given eft'ect, but not with what frequency and in what quantities 

 those causes exist. An instance in point is aff'orded by a newspaper 

 now lying before rae. A statement was furnished by one of the oflicial 

 assignees in bankruptcy, showing, among the various bankruptcies 

 which it had been his duty to investigate, in how many cases the losses 

 had been caused by misconduct of different kinds, and in liow many 

 by unavoidable misfortunes. The result was, that the number of fail- 

 ures caused by misconduct greatly preponderated over those arising 

 from all other causes whatever. Nothing but specific experience 

 could have given sufficient ground for a conclusion to this purport. 

 To collect, therefore, such empirical laws (which are never more than 

 approximate generalizations) from direct observation, is an important 

 part of the process of sociological inquiry. 



The experimental process is not here to be regarded as a distinct 

 road to the truth, but as a means (happening accidentally to be the 

 only, or the best available) for obtaining the data which the deductive 

 science cannot do without. When the immediate causes of social facts 

 are not open to direct observation, the empirical law of the effects 

 gives us the empirical law (which in that case is all that we can obtain) 

 of the causes likewise. But those immediate causes depend upon 

 remote causes ; and the empirical law, obtained by this indirect mode 

 of observation, can only be relied upon as applicable to unobserved 

 cases, so long as there is reason to think that no change has taken 

 place in any of the remote causes on which tho immediate causes de- 

 pend. In making use, therefore, of even tho best statistical generali- 

 zations for the purpose of inferrinir (though it bo only conjecturally) 

 that the same empirical laws will hold in any new case, it is necessary 

 that we be perfectly well acquainted with the remoter causes, in order 

 that we may scrupulously avoid applying tho empirical law to cases 

 which differ in any of the circumstances on which the truth of the law 

 ultimately depends. And thus, oven where conclusions derived from 

 specific observation are available for practical infiircnces in new cases, 

 it is necessary that the deductive scicMice should stand sentinel f)ver 

 the whole process ; that it should be constantly referred to, and its 

 sanction obtained to every inference. 



The same thing holds true of all generalizations which can be 

 grounded on history. Not only there are such generalizations, but it 



