METHOD OF DISCOVERING CAUSAL LAWS 169 



6 , into separate elements or factors, a, &, c . . . /, #, 0, /, ?, r . . . , and 

 the supposition that some one factor or group of factors (say m) is the cause of 

 p. Here is where the difficulty increases ; and where some writers have left 

 themselves open to the charges of conveying false impressions as to the sim 

 plicity of the process by which the real cause is to be discovered, and of setting 

 up wrong ideals of this process. The difficulty lies in the fact that the breaking 

 up of any field of phenomena, coexisting in space or successive in time, into 

 separate entities, capable of being expressed and dealt with symbolically, as 

 a, b, c, etc., must be to a large extent a mental analysis, which cannot claim 

 to give us an adequate view of the complex reality as it actually is. It is only 

 abstract and incomplete aspects of the reality that can be so represented. No 

 doubt, it is only by such analysis that we can hope to detect causal connexions 

 between phenomena. But nature does not present its materials to us thus 

 analysed or broken up into separate factors. Its agencies act and interact ; 

 they often counteract and neutralize one another. Its influences cross and 

 recross and combine with one another in hidden and intricate ways. It is on 

 our making suggestive and fruitful analyses and syntheses of the materials 

 which constitute our sense-experience, that the progress of science depends. 

 When this work is done on the proper lines, the materials for induction are 

 prepared, and the rules laid down by Mill are easy and obvious ; but this 

 work of preparation is the most difficult stage in the whole inductive process. 

 And the rules in question rather suppose it to be done than help us to do it. 

 Mill made the mistake of practically overlooking this part of the process ; his 

 treatment of induction conveys the impression, the erroneous impression, that 

 nature furnishes us with prepared materials : with simple, isolated causes and 

 effects, which have only to be observed, enumerated, and expressed by separate 

 symbols (cf. 245). 



The same is true of Bacon and Jevons (209, 210). Their treatment of 

 the subject ignores the difficulty of making a successful analysis of the field 

 of investigation into separate elements, about each of which we may next go on 

 to inquire whether or not it is the cause of the phenomenon, p. Into how 

 many alternatives are we to break up the field of investigation when we attempt 

 to state the problem in the form of a disjunctive proposition : &quot; The cause of 

 p is either a, or , or &amp;lt;r, . . . or /, or m, or , or . . . &quot; ? In other words, how 

 many hypotheses are we to make and test ? This is certainly not determined 

 for us by the way in which the data are given to us ; for they are not given 

 catalogued into elements. Nor is it possible to determine the number of 

 hypotheses, as Jevons suggests, by any merely formal counting of instances 

 and &quot; calculation of mathematically possible combinations &quot; J of elements 

 in those instances ; for even if the elements were in each instance really 

 distinct and independent of one another, as they certainly are not, 

 the number of instances would be an &quot;unattainable infinite series,&quot; 2 

 and, hence, no certain conclusion could be obtained by such a method. 



How, then, are we to proceed ? Are we, perhaps, to go on selecting 

 empirically and by haphazard, one after another, all the factors, a, b, c, d, . . . 

 which we can detect in the field of investigation, S, and test each separately, 



1 WELTON, op. /., p. 59; cf. ibid., pp. 53-55; VENN, Empirical Logic, pp. 

 o. 



2 WELTON, ibid., p. 59. 



