1 82 THE SCIENCE OF LOGIC 



DIFFERENCE AND AGREEMENT. This is really the method which 

 pushes experimental analysis to its farthest possible limits, and, 

 by doing so, enables us to eliminate by degrees the misgivings 

 that may still remain after we have applied the method of dif 

 ference. These misgivings arise, as we saw, from the fact that 

 the total cause, as determined by applications of the latter method, 

 may not after all be absolutely indispensable ; or, in other words, 

 that somewhere or somehow other agencies might conceivably be 

 substituted for it in whole or in part, and so produce the effect in 

 the total or partial absence of this cause. Now the only way of 

 allaying such suspicions is by conducting as many negative ex 

 periments as these suspicions demand, within what we consider to 

 be a reasonable sphere of investigation : experiments in each of 

 which our supposed cause is absent, and one or other of the con 

 ceivable alternatives present. Each such experiment, in which 

 the phenomenon is absent, negatives or disproves the suggested 

 alternative cause which has been introduced instead of the sup 

 posed real cause. This is, in reality, the method of analysis which 

 we have already outlined (240), and its connexion with the process 

 of verifying hypotheses by the disproof Q{ alternatives, as set forth 

 in a previous chapter (212), will be at once apparent ; for it simply 

 describes the manner in which that process of verification ought 

 to be conducted. The following is the canon in which Dr. Mel- 

 lone : embodies this method : 



&quot; When one phenomenon has been shown to be THE CAUSE OF 

 ANOTHER UNDER GIVEN CONDITIONS, by the method of single dif 

 ference ; and when we fail to find or to construct any instance where 

 the one phenomenon occurs without the other : then it is probable 

 that the first is the UNCONDITIONALLY invariable antecedent of 

 the second i.e. that the latter can be produced in no other way than 

 by the former ; and the probability increases with the number and 

 variety of the negative instances all agreeing in the absence both of 

 the effect and its suspected cause. &quot; 



The phrase &quot; unconditionally invariable antecedent &quot; means the 

 &quot; sufficient (or necessitating) and indispensable cause,&quot; or again, 

 the &quot; reciprocating cause &quot;. Dealing with the conditions for the 

 verification of an hypothesis, we saw that an hypothesis is rigor 

 ously verified when it has been shown to be the &quot; only possible &quot; 

 one that will account for the facts (229) ; and the question now oc- 

 currs : Does &quot; only possible &quot; mean &quot; conceivable absolutely in the 

 1 op. dt., p. 309. 



