[bowman] fundamental PROCESSES IN HISTORICAL SCIENCE 537 



ratio of 61:1 ought to be much reduced, and a conservative estimate 

 of the 3rd conclusion would be 6/7 or 6:1. Ordinarily where the evi- 

 dence is contradictory and a verdict is barely reached, a lengthy 

 discussion is necessary to bring about this decision, therefore the prob- 

 ability of the 4th conclusion, that the jury was slow in arriving at the 

 verdict, may be conservatively estimated at 10:1. The 6th conclu- 

 sion, that the jury which decided for the merchant did so probably 

 because they were more favorably impressed on the whole with his 

 trustworthiness than the banker's, may be conservatively estimated 

 at 25:1. Juries almost always decide according to their impression, 

 on the whole, of the respective merits of the two sides. The personal 

 trustworthiness of the plaintiff and defendant respectively is not 

 necessarily an essential feature in the merits of a case, but in this 

 instance their personal word was all the evidence offered by either 

 of them on the principal point. 



Summary of the Experimental Test. 



a. Verification of the general principle that probability even of the 

 superior measure does not produce individually correct results. The 

 12 cases in the experimental test, which have been drawn from actual 

 experience, have been selected intentionally to give as wide a variety 

 as possible within a reasonable space. For this reason the cases 

 range from the simplest incidents to the most important questions 

 engaging human attention, and they deal also with various features 

 in the practical affairs of life. It should be noted that these cases 

 are not only actual, but also typical, occurrences. It would be an 

 exaggeration to say that the number of similar cases is without limit, 

 but it is correct to say that it is impossible to set the limit. Their 

 number is indefinitely great. 



The 38 probable conclusions in Cases 1 to 11, and 3 out of the 

 10 probable conclusions in Case 12, are probable conclusions of the 

 superior form. In 34 of these 41 conclusions, degrees of probability 

 o-curred ranging from about 2/3 or 2:1 to 1,000/1,001 or 1,000:1; 

 and in the remaining 7 of these 41 conclusions, the degree of probability 

 rated as "high" was greater than the highest of the 34. The features 

 found in Case 1 continued throughout the test. The successive 

 probable conclusions continually displaced one another, and all 

 were finally eliminated, as a result of subsequent additions of evidence. 

 The test thus verifies experimentally the general principle stated in 

 the Century definition, that probability even in its superior measure 

 has no significance in producing individually correct results, and the 

 inferior form of probability is shown in the illustrative Case 12 to be 

 still less able to produce such results. The test, therefore, verifies 



