SECTION 14.-VERIFICA TION AND PROOF. 1 1 5 



to frame legitimate inferences, and for both types of verification 

 it is of crucial moment that the basal generalisations shall be 

 lucidly expressed. 



47. (B) Proof may be regarded as the process whereby 

 the truth of a statement is established beyond a reasonable 

 doubt. Bacon's synthetic procedure, which we follow in a 

 modernised form in Part V, is one eminently suitable for proving 

 propositions of a certain general order. By registering all rele- 

 vant affirmative class instances, then class instances similar 

 but negativing the former, then determining the degree of the 

 presence of a quality, then excluding automatically what is 

 not common to all cases, and then framing a general statement 

 in the form of a guarded definition, the greatest practical 

 assurance is obtained that we have a fully proved statement 

 before us. The comprehensive nature of the procedure, and 

 the reliance on fact rather than on sweeping hypotheses, further 

 accentuate its evidential value, especially when supplementary 

 dialectical methods, such as Bacon suggests, or as are suggested 

 in Conclusions 27 and 28, are employed to ensure that nothing 

 escapes from the closely woven meshes. 



John Stuart Mill, following Bacon and Herschel, proposed, 

 as we have seen, five decisive methods of testing statements 

 dealing with causes: the Canons of Agreement, Difference, 

 Joint Method of Difference and Agreement, Residue, and Con- 

 comitant Variation. (For details, see 17.) These should be 

 all applied when practicable. However, two factors need to 

 be taken into account in respect of them. First, that the pro- 

 gressive nature of truth most rarely admits of the rigorous 

 application of these Canons, and, secondly, that, as a matter 

 of fact, most present-day scientific enquiries are in too un- 

 developed a condition to permit the Canons being frequently 

 utilised, except tentatively. And what holds of causal, holds 

 equally of static, facts. 



Canons of probability are, therefore, needdd, both in regard 

 to static and dynamic phenomena, to supplement Mill's Canons. 

 The principal ones applied to a hypothesis are general agreement 

 with existing knowledge, withstanding deductive tests, leading 

 to new truths, and not being invalidated by fresh information. 

 However, many hypotheses are working hypotheses, and admit- 

 tedly do not harmonise with all the recognised facts. Further- 

 more, the special sciences apply, in addition, auxiliary tests. 

 In chemistry, reagents; in geology and in the historical sciences, 

 analogy; in botany, the test of fitting into the botanical scheme 

 of classification ; in physiology, staining; and in other sciences, 

 other acknowledged tests, are resorted to. Simple verification, 

 instrumental or otherwise, is a further criterion. To be sure, 

 every science acts as such as a check upon alleged additions 

 to knowledge, and almost invariably it possesses special methods 

 for testing its particular "class of truths. In art and in conduct 



