336 



SCIENCE 



[N. S. Vol. XXXIV. No. 872 



of a theory formulated too hastily, and how 

 in later times attempts have been made to 

 remedy this evil by holding the theory, pro- 

 visionally only, as a working hypothesis, 

 Professor Chamberlin points out that even 

 the working hypothesis has serious disad- 

 vantages : 



Instinctively there is a special searching-out of 

 phenomena that support it, for the mind is led by 

 its desires. . . . From an unduly favored child it 

 readily grows to be a master and leads its author 

 whithersoever it will. . . . Unless the theory hap- 

 pens perchance to be the true one, all hope of the 

 best results is gone. To be sure; truth may be 

 brought forth by an investigator dominated by a 

 false ruling idea. His very errors may indeed 

 stimulate investigation on the part of others. But 

 the condition is scarcely the less unfortunate. 



To avoid this grave danger the method of mul- 

 tiple working hypotheses is urged. It differs from 

 the simple working hypothesis in that it distributes 

 the effort and divides the affections. ... In devel- 

 oping the multiple hypotheses, the effort is to 

 bring up into view every rational explanation of 

 the phenomenon in hand and to develop every ten- 

 able hypothesis as to its nature, cause or origin, 

 and to give all of these as impartially as possible 

 a working form and a due place in the investiga- 

 tion. The investigator thus becomes the parent of 

 a family of hypotheses: and by his parental rela- 

 tions to all is morally forbidden to fasten his 

 affections unduly upon any one. In the very na- 

 ture of the case, the chief danger that springs 

 from affection is counteracted. 



For the further elucidation of Professor 

 Chamberlin 's proposals I must refer my 

 audience to his original paper, which is 

 well worthy of careful attention. He does 

 not shirk consideration of the drawbacks 

 • — "No good thing is without its draw- 

 backs," he writes. And it may be added 

 that no good thing is entirely new, or en- 

 tirely old. Perhaps it is better to say that 

 it is generally both new and old. The 

 method of multiple hypotheses is new be- 

 cause it is still necessary to remind scien- 

 tific workers of all kinds that so long as 

 they restrict themselves to the examination 



of one hypothesis only they can never reach 

 complete logical proof: they can only at- 

 tain a high measure of probability. What 

 is often called verification^ is not complete 

 proof, but only increase in probability: 

 for complete proof it is necessary to show 

 that no other hypothesis will suit the facts 

 equally well, and thus we are bound to con- 

 sider other possible hypotheses even in the 

 direct establishment of one. 



But the method is also old in that it has 

 long been adopted in practise, however 

 partially and unconsciously by scientific 

 workers of all kinds. When as a boy at 

 school I began to make physical measure- 

 ments under Mr. J. G-. McGregor (now 

 professor of physics at Edinburgh) I 

 learned from him one golden rule: "Re- 

 verse everj'-thing that can be reversed." 

 The crisp form of the rule may be new to 

 many who have long used it in their work : 

 and its use is simply that of "multiple 

 hypotheses." For when the current in a 

 wire is reversed, the hypothesis is tacitly 



' To show that the facts agree with the conse- 

 quences of our hypothesis is not to prove it true. 

 To show that is often called verification: and to 

 mistake verification for proof is to commit the 

 fallacy of the consequent, the fallacy of thinking 

 that because, if the hypothesis were true, certain 

 facts would follow, therefore, since those facts are 

 found, the hypothesis is true. ... A theory whose 

 consequences conflict with the facts can not be 

 true; but so long as there may be more than one 

 giving the same consequences, the agreement of 

 the facts with one of them furnishes no ground 

 for choosing between it and the others. Neverthe- 

 less, in practise we often have to be content with 

 verification; or to take our inability to find any 

 other equally satisfactory theory as equivalent to 

 there being none other. In such matters we must 

 consider what is called the weight of the evidence 

 for a theory which is not rigorously proved. But 

 no one has shown how weight of evidence can be 

 mechanically estimated; the wisest men, and best 

 acquainted with the matter in hand, are oftenest 

 right. — ' ' An Introduction to Logic, " by H. W. B. 

 Joseph, fellow and tutor of New College, Oxford, 

 Clarendon Press, 1906, p. 486. 



