HYPOTHESIS 155 



the phenomena. In other words, the material reason why we can 

 not regard the suggested law as certain, or verified, is because we 

 have not sufficiently analysed the facts. In each case, the points 

 of resemblance between the facts under observation and other 

 known facts, suggest that the causal law which accounts for those 

 points in the known facts may also account for them in the facts 

 now being examined. We are trying to extend a known law 

 to a new case. But this extension is an hypothesis which awaits 

 verification. The connexion of the two cases by a common law 

 is not verified by mere analogy as such. Analogy, as such, &quot; sticks 

 in the particular instances,&quot; and gives only a more or less probable 

 conclusion. 



There are good and bad arguments from analogy. The 

 probability of the conclusion may range indefinitely from zero 

 and practical certitude. On the one hand, the conclusion may 

 be far less probable than its contradictory, when, as in ($), the 

 points of resemblance seem to have little causal relation to 

 the conclusion inferred from them, and not to include such 

 essential conditions for this conclusion as the presence of oxygen 

 or air, and absence of extremes of temperature. 1 Or, again, the 

 conclusion may be no more probable than its contradictory, when 

 conflicting analogies produce absolute doubt ; as, for example, in 

 the case of some lower form of living thing which may present 

 certain resemblances to animal life, and certain other equally 

 marked resemblances to vegetable life. Or, finally, it may 

 become evident that the resemblances are causally connected 

 with the inferred property, in which case the argument passes 

 beyond the stage of analogy, its premisses take the form of the 

 first figure of syllogism, and its conclusion becomes a certainly 

 established law. If some symptom common to cholera and the 

 other disease in example (a), above, could be shown to be due to 

 no other cause than the action of a bacillus, the conclusion would 

 become certain that the disease in question was due to such 

 action. In order, therefore, that an inference from resemblances 

 may lie within the limits of analogy, it must be probable, but 

 only probable, not certain, that the common characteristics are 

 causally connected with the conclusion based upon them. 



235. WORTH OF ANALOGY : ITS FUNCTION IN VERIFICATION. 

 How, then, are we to estimate the value or force of an argument 

 from analogy? On what will the probability of its conclusion 

 1 C f . MELLONE, op. cit., p. 263. 



