174 THE SCIENCE OF LOGIC 



agencies actually observed, or by quite other agencies, and under 

 quite other conditions, than those under actual observation. 



To take a simple illustration : &quot; Suppose you mix three dif 

 ferent kinds of poison [B, C, D\ with water [A] and give it to 

 three people : they all die, but you cannot argue that the water 

 is the cause of death, though it is the only invariable antecedent &quot;.* 

 You cannot even argue with certitude that the water is a cause of 

 death. And why ? Because you are not certain that it contains 

 the &quot; necessitating and indispensable &quot; cause of death : there may 

 be another &quot;invariable antecedent&quot; present in the three cases, 

 namely, something (say X} contained in, and common to, the 

 poisons, B, C, D , each of which therefore may be a cause ; while 

 the indispensable element, X, though not in A, is not eliminated 

 by the absence of C D, or of B D, or of B C, but is present and 

 operative in all three instances, A B, A C, and A D, first in B, 

 then in C, then in D. From all of which we see that the causal 

 hypothesis which this method can merely suggest, may con 

 ceivably be even a wrong hypothesis. Simple observation reveals 

 to us only what are causes in the wider sense ; and in the absence 

 of further analysis the method of agreement may possibly elimi 

 nate many such causes successively, leaving the indispensable factor 

 present every time, not in the observed &quot; invariable&quot; element, but 

 in one or other of the &quot; varying&quot; elements : &quot; If heat, for instance, 

 is produced by friction, combustion, electricity, all these real causes 

 would be eliminated by this method, for they are points in which 

 the different instances of heat differ &quot;. 2 At best, then, this 

 method can merely suggest, as a more or less probable hypothesis 

 to be otherwise tested, the supposition that the observed invariable 

 antecedent is, or contains, the cause. 



The rule was entitled by Mill the &quot;method of agreement,&quot; because, 

 though the instances differ in details that are presumably irrelevant to the 

 phenomenon, they agree in the one presumably essential, and therefore im 

 portant, point. He formulated the rule in the following way : 



&quot; If two or more instances of the phenomenon under investigation have 

 only one circumstance in common, the circumstance in which alone all the 

 instances agree, is the cause or effect of the given phenomenon.&quot; 



He means, of course, &quot;one circumstance in common&quot; besides the pheno 

 menon itself, which is common to all the instances. &quot; Two &quot; instances would 

 give very little probability, for this depends on the number and variety of the 

 instances. Of course, z/we could be certain that the observed instances of 

 any phenomenon had really &quot;only one circumstance in common&quot; we could 



1 Palaestra Logica, p. no, 338. * MELLONE, of. cit., p. 297. 



