229 



The author then examined the probabilites arising from combined 

 testimony, as expressed by the formula, 



1 



-(T)"^ 



where p is the antecedent probability, v the average veracity of the 

 witness, and n the number of witnesses. The cause of an antecedent 

 probabihty was stated to be a natural tendency to believe in the con- 

 tinuity of series, and to be, in fact, a modification of the inference 

 from experience. 



It was then shewn, that this expression is applicable to the pro- 

 bable correctness of the verdicts of juries, assuming v to represent 

 the probability of correctness in each juryman. Taking the number 

 of the jury at three, it appeared that the probability of getting a 

 correct unanimous verdict, was to the probability of getting a correct 

 verdict by a majority as 3 to 4. But, on the other hand, it appeared 

 that assuming p to be ^^, the probability of error in the unanimous 

 verdict is ^3^5, while the probability of error when two divide against 

 one is -^ ; that is to say, about eighty times as great as in the for- 

 mer case. Hence it was inferred that verdicts, by a bare majority, 

 are admissible in civil cases, where it is sufficient that the verdict be 

 probably correct ; but not in criminal cases, where it is desirable 

 that the probability of correctness should be carried as near to cer- 

 tainty as possible. 



Lastly, Whereas there are, in every criminal charge, a certain 

 number of points, all of which the jury desire to know, and which, 

 if all proved, whether for or against the accused, would give a defi- 

 nite probability of his guilt or innocence, but of which one or more 

 may remain untouched by the evidence, it was shewn that the jury 

 would not deal fairly with the case if they threw the non-established 

 point out of view, or if they gave the benefit of the doubt, as it is 

 called, to the prisoner. Supposing ^^ to be the probability of guilt, 

 independently of the non-established point, which must be considered 

 as equally likely to be really for as agahist the accused, we must see 

 that the consideration of this adds i to a, and the same to b. So that 

 ° becomes °"*"^ . ; and these, reduced to a common denominator 



a + 6 ft + 6 + 1 



D, are -"'' + '^" ''+^" and ?^!+^y:?±?. So that the considers- 



