114 PART II. SOME IMPORTANT METHODOLOGICAL TERMS. 



fication. Just as it is unscientific to generalise in an unknown 

 field from one or a few instances, so verification, save in a 

 crucial experiment, is unscientific if it only embraces one or a 

 few occurrences, or if it rests "upon anything but a very ex- 

 tensive comparison with a great mass of observed facts". 

 (Herschel, Discourse, [219.].) Verification is peculiarly a process 

 employed in the sciences, a process which is mostly omitted in 

 ordinary life, or else is performed in a perfunctory manner. 



Verification may signify in a certain connection re-examina- 

 tion, as when we desire to examine the correctness of a previous 

 observation ; or it may mean examination, as when a generali- 

 sation is to be tested. Verification may also proceed by cal- 

 culation, as in the discovery of the planet Uranus; by reason- 

 ing, as when the matter to be verified is a geometrical pro- 

 position; or by reasoning and feeling, as when beauty and 

 goodness are involved. Rules as to repeated examination, 

 exhaustion of conditions, a critical attitude, especially apply here. 

 (See Conclusions 22 to 24.) 



Those who favour the unceremonious framing of hypotheses 

 frequently maintain that no .harm will ensue provided we are 

 careful to verify them. In this they assume that verification 

 is not encircled by obstacles, whereas verifying a justifiable 

 hypothesis may be the work of a generation, whilst the attempt 

 to verify unwarranted hypotheses usually connotes an endless 

 task leading often to deeper and deeper misconceptions. The 

 decline of Rome, for instance, has found many hypothetical 

 explanations, any one of which may conceivably be true. It 

 is contended that nations decay like individuals; that the 

 superior types had been eliminated by the wars; that the 

 Barbarians rushed and crushed Rome; that growing immorality 

 and luxurious living robbed Rome of its stamina; and that the 

 introduction of malaria sapped the health of its inhabitants. 

 Here are five hypotheses, and it would be easy to augment 

 their number. The difficulty, almost an insuperable one, mani- 

 festly lies in proving one or more of them to be correct. 

 Likewise, the apparent differences in the size of the full moon 

 on the horizon and at the zenith, have led to the formation of 

 sundry hypotheses, none of which have yet been substantiated. 

 Premature indulgence in speculation tends, in fact, to make 

 confusion worse confounded, and supplies the unwary with a 

 trap instead of with a bridge. 



There are two grades of verification simple and deductive, 

 and the references above have only been to the first. In the 

 former case we only examine a number of impartially and 

 judiciously selected specimens to test the truth of a hypothetical 

 proposition. In the latter, we reason that if the proposition A 

 holds, the propositions Al, A2, A3 will also hold, and that if 

 we therefore perceive that Al, A 2, A3, do hold, A probably 

 represents the facts correctly. The supreme anxiety should be 



