GROUNDS OF DISBELIEF. 197 



The error of this argument seems to be the same 

 which we pointed out in a former chapter*, that of ap- 

 plying a theorem, only true of the degrees of probabi- 

 lity of causes, to the probability of what are neither 

 causes nor indications of causes, nor in any other way 

 specially connected with the effect. The point in ques- 

 tion is, the comparative probability of two suppositions, 

 that the witness lies, and that he speaks truth. But 

 these are not two possible causes of the given effect 

 (the witness's assertion) ; they are merely two possible 

 qualities of it. The truth of the assertion is, indeed, 

 on the supposition of veracity, the cause of its being 

 made ; but the falsity of it is not, on any supposi- 

 tion, a cause of its being made. It is not incompatible 

 with the dishonesty of the witness, that he should 

 have spoken the truth : the difference between the 

 two suppositions of honesty and dishonesty is, that 

 on the one he would certainly speak the truth, while 

 on the other he was just equally likely to speak that 

 or anything else. If the falsity of the proposition 

 were a real cause for his asserting it, and there were 

 no possible mode of accounting for a false assertion 

 but by supposing that it is made precisely because of 

 its falsity, I do not see how Laplace's argument could 

 be resisted. The case where there are 999 possible 

 false assertions, and that in which there is but one, 

 would then present a vast difference in the probability 

 that the assertion actually made proceeded from 

 falsity; because in the one case a mendacious witness 

 was sure to assert the one false fact, in the other 

 there would be an equal chance of his asserting any 

 one of the 999. But as it is, the falsity was a mere 

 accident of the assertion, not the cause of it ; and 



Supra, pp. 79, 80. 



