[bowman] fundamental PROCESSES IN HISTORICAL SCIENCE 541 



John Doe's child is a son, the first selection is eliminated. The margin 

 of probability here, as in the balanced conclusion, is narrow, but that 

 margin is established as a definite ratio in favor of the conclusion; 

 and thus, instead of two steps with their successive probabilities both 

 of the inferior measure, the conclusion is subject only to one proba- 

 bility and that of the superior measure. Although, as a matter of 

 pure science, this conclusion as to the sex of the child is therefore 

 much the superior, no one would think of attaching importance to it. 

 The chances opposed to its correctness are too definitely in view. The 

 balanced conclusion, on the contrary, notwithstanding its scientific 

 inferiority, has a vogue, for two reasons. In the first place, the fact 

 that the chances opposed to its correctness are practically as great as 

 those in favor, is obscured and overlooked, or blinked, in a confused 

 multiplicity of reasonings pro and con ; and secondly, there is a wrong 

 impression that the reasons favorable to the conclusion are to be re- 

 garded as express ground for its acceptance, and the unfavorable as 

 express ground for its rejection, so that, if one decides that the favor- 

 able outweigh the unfavorable, the excess lends an air of necessity 

 to the acceptance on the principle that the majority should rule. The 

 fact is, however, that in these balanced probable conclusions the favor- 

 able and unfavorable reasons are both probable only, and if there 

 were no unfavorable reasons at all to set against the favorable, these 

 would still leave the conclusion only probable. Balanced probable 

 conclusions of the above description, moreover, even with a weight 

 of favourable reasons twice as great as the unfavourable, are essentially 

 on the^same footing as those in which the margin either way is so slight 

 that impartial judges will differ as to the side on which it lies. Such 

 a double weight of reasons would practically remove all question in 

 the preliminary selection of the stronger side, but the final inference 

 in favor of this side would still have to be drawn. This inference, 

 since it is made, just as in the case where the margin is slight, on the 

 basis of probability of the inferior measure, must have less value than 

 a conclusion with a corresponding degree of probability of the superior 

 measure (2/3 or 2:1); and concerning conclusions with this degree of 

 probabihty of the superior measure (2/3 or 2:1), it can be shown, 

 just as in the conclusion concerning the sex of John Doe's child, that 

 no value at all is attached to them in practice. The point can be 

 illustrated by features in connection with Case 4 of the experimental 

 test. It was found there that the total amount of insurance held in 

 Canada was divided between fire, life and all other risks in the pro- 

 portion of 20-3 : 8-5 : 4-5 respectively. If then it be placed in 

 evidence that John Doe personally held insurance to the amount of 

 $1,000, this would establish a conclusion with a degree of probability 



