OF TESTIMONIES OR JUDGMENTS. 653 



Archbishop Whately's rule, but the conditions of its valid application are evi- 

 dently not fulfilled in the example which he has given. To satisfy these condi- 

 tions, this problem ought to be changed into the following : " There exists a certain 

 quality of style, the possession of which would prove the work to be by the au- 

 thor supposed. The probability that the work possesses that quality is §. There 

 is a person so well informed that his attributing the work to the supposed 

 author would be conclusive. The probability that he does attribute it to the 

 author in question is f. Required the probability, on these grounds, that the 

 supposed author is the real one." But this is evidently not the sense in which 

 the problem was meant to be understood. Thus to take one point, it is not the 

 quality of the style that is a matter of probability, but the mode in which a 

 known and observed quality affects the question of authorship. 



Taking the problem in its intended meaning, each of the fractions f , f, mea- 

 suring, not the probability of the truth of certain premises, but the probability 

 drawn from these premises, as conditions, in favour of a certain supposition 

 (I use this word in preference to conclusion), we are no longer permitted to 

 apply the formula above determined. And we are not permitted to do so, because 

 the probabilities with which we are concerned are conditional, and their possession 

 of this character greatly increases the difficulty of the problem. Its rigorous formal 

 solution is given in Art. 34, and shows that the probability sought is, generally 

 speaking, indefinite, — a result which agrees with the conclusions of Bishop Ter- 

 rot, by whom the error to which attention has been directed was first pointed 

 out, Transactions of the Royal Society of Edinburgh, vol. xxi., p. 369. 



I trust that I have not in any way misrepresented Archbishop Whately's 

 reasoning ; and I am the more encouraged to believe that I have not, as a defence 

 of it which appeared in the United Church Journal expressly proceeds upon the as- 

 sumption that the probabilities with which we are concerned are probabilities that 

 the authorship is proved. To this view Bishop Terrot justly demurred. Nor 

 was its inconsistency materially diminished by assigning to proof a meaning less 

 absolute than belongs to demonstration. For whatever degree of cogency, — of 

 power to produce conviction, — we suppose to characterize proof, the thing itself 

 belongs to consciousness, and the question whether given evidence is sufficient 

 to convey proof to our minds or not, is a matter of knowledge, not of probability. 



VOL. XXI. PART IV. 8 N 



