364 PART V. WORKING STAGE. 



fication, however, is not equivalent to re-examining the facts, 

 for in the latter case we strive to augment, rather than to 

 revise, our information. 



In generalising we pass beyond the scrutinised facts, and, 

 therefore, unless we are, for instance, concerned with a chemi- 

 cal element or a kinetic problem, where the part fitly represents 

 the whole (Section XIII), it is necessary to test the correctness 

 of the generalisation. Wherefore, if we should surmise that 

 a low state of civilisation argues invariably a more poorly 

 endowed race, we are bound to apply every manner of test, 

 especially Conclusion 28, to ensure that we are not mistaken, 

 for possibly other explanations may more satisfactorily interpret 

 the known facts. Without verification, generalisations remain 

 hypotheses which, as a broad rule, are more likely to prove 

 erroneous than true. 



It is the same with deduction. A deduction may be made 

 for the purpose of verifying a hypothesis or in order to ex- 

 tend knowledge, as when we infer the existence of positive 

 electricity from the existence of what is called negative electri- 

 city, or posit the ether in order to avoid the problem of action 

 across empty space, or argue for the materiality of the ether 

 from the fact that light and other forces are transmitted through 

 it at a quite definite rate (as sound through air which is indubit- 

 ably material), or seek to prove the double generalisation that 

 transparent solids are good insulators and metals and good con- 

 ductors opaque by ascertaining the truth of the inference that 

 then the transparent film of a metal would have lost its con- 

 ductivity. In both connections, the process of reasoning should 

 be tested in regard to their correctness, and the facts should 

 be examined in either case, inasmuch as the hypothesis or the 

 deduction may be unwarranted. 



Lastly, where the memory, the imagination, and the processes 

 of reasoning are concerned, the like need for verification mani- 

 festly exists. 



Given, then, the universal need for verification, we can only 

 add that its methods are those of observation or meticulous 

 scrutiny, as enumerated in Conclusions 16 to 24. Casual or par- 

 tial verification is, if possible, even less admissible than casual 

 or partial observation. 



In Section XIV and Conclusions 8, 19, and 20, and in other 

 places, special expedients for verifying or testing facts have 

 been recited, and to these we would refer the student. We 

 require, then, tests, dealing with science as a whole, with special 

 classes of sciences, with particular sciences and with individual 

 portions of particular sciences, and with particular enquiries. 

 None of these can be dispensed with. 



For the sake of completeness we may indicate some of the 

 lines along which proof of an assertion may be conveniently 

 sought. Direct proof may be obtained by (a) simple and 



