OPINION AND PROBABILITY 265 



do not warrant a certain, but only a probable, conclusion. This may 

 happen in two ways : (a) either because the premisses contain a 

 judgment which is only probably true ; or (fr) because, though 

 the premisses are all true, they do not necessarily involve, but 

 only suggest, or point to, the truth of the conclusion we may 

 seek to derive from them. 



(a) The former class of argument was described by Aristotle 

 as a syllogism from probabilities : &amp;lt;7tA.Xo7ioyi09 e elxormv the 

 eiVo? being one of those rough generalizations or &quot; moral uni- 

 versals &quot; commonly accepted as &quot; practical &quot; certainties. Where 

 we have a number of such probable syllogisms depending on one 

 another, in the form of a sorites (188), and leading up to a conclu 

 sion, we have what is called &quot; chain evidence &quot;. And as a chain 

 cannot be any stronger than its weakest link, the probability of 

 the conclusion in such a case cannot be greater than that of the 

 weakest premiss. It is usually considerably less ; for, since each 

 probable premiss inherits, besides its own innate weakness, that 

 of all its antecedents, the conclusion must inherit the combined 

 weakness of them all. 1 



Care should be taken to distinguish chain evidence from cir 

 cumstantial evidence; in the latter the various probable signs 

 are independent of one another, and their combination, therefore, 

 forms a cumulation or addition of probabilities which may pos 

 sibly be so strong as to issue in moral certitude. 



(b) The second kind of probable argument is that in which 

 the premisses, though true, do not necessarily involve the conclu 

 sion sought to be derived from them. This is what Aristotle 

 described as the syllogism from signs, or symptoms, or indications : 

 a-v\\oyt,a-fio&amp;lt;f e a-rjfieiojv. And to both arguments the syllogism 

 from probabilities, and the syllogism from signs he gave the title 

 of Enthymeme. The Enthymeme, therefore, in Aristotle s meaning 

 of the latter term the &quot; Aristotelean &quot; enthymeme, as it is now 



1 &quot; For instance, a man is accused of murder. There is very strong evidence that 

 a man just like the prisoner was in the company of the murdered man on the night 

 when the murder was committed. It is almost certain that the man who was known 

 to be in his company did the deed. There is, moreover, a strong presumption against 

 the theory urged by the counsel for the defence, that the deceased made an unpro 

 voked attack on his companion on the night in question and met his death from him 

 in self-defence. But it does not follow that the accused should be convicted of mur 

 der. For if the probability of the three circumstances pointing to guilt is three to 

 one, the balance of probability is nevertheless rather against than in favour of their 

 being all of them true, and this means that it is more likely that the accused was 

 innocent than that he was guilty.&quot; CLARKE, Logic, p. 430. 



