[bowman] fundamental PROCESSES IN HISTORICAL SCIENCE 557 



in the table (except where he has to contrary grounds) without fur- 

 ther test, simply because such a test is unreasonably inconvenient. 

 The historian, in the strength of the same essential correctness, is 

 justified and required as a reasonable man to use any individual state- 

 ment in the record or part of the record (except where he has contrary 

 grounds) without further test; and this, (where the record is 

 the only source) for the much better reason that such further test is 

 impossible. 



2. Pure probability, regardless of its height, should not be used 

 where formal probability, i.e., a correct process, is available. The 

 surgeon was not justified in substituting a pure probability even of 

 1,000,000:1 for the correct process available. In so far as formal 

 probability cannot be applied, pure probability may be used in scienti- 

 fic, including historical, investigation in any series of decisions, if the 

 degree of pure probability applied be of a height which will develop 

 in the series an average of correct results equal to, or greater than, 

 the average of essential correctness required in the branch of science 

 involved. 



3. Reasoned probability ought not to be used as a positive cri- 

 terion of conclusions in scientific, including historical, investigation; 

 but reasoned probability has a most important function in guiding 

 the investigator in the search for available evidence, and also as a re- 

 sisting negative force to assist in the exclusion of reasonable doubt 

 from the final conclusions, and to detect, if possible, latent defects in 

 the application of correct processes in the inexact sciences, such as the 

 exemplification of the requisites of trustworthiness in history, thereby 

 reducing the percentage of incidental error in the final results reached 

 by the application of correct processes or formal probability. 



4. Obligatory decisions in practical applications of the sciences 

 and in practical affairs generally. In this class of decisions, if a cor- 

 rect process is not available, reasoned probability or pure probability 

 should be applied in so far as either is available, and is better 

 adapted to the conditions; and in any height available. The same is 

 true of semi-obligatory decisions under the above circumstances, 

 i.e., when a decision is not insistent, and yet preferable, because a 

 failure to decide seems to involve on the whole a greater risk than that 

 of a wrong decision. 



5. Required average of essential correctness in science, especially 

 in history. The average of essential correctness that can and ought 

 to be attained cannot be exactly determined even in an exact science 

 such as mathematics, and still less in an inexact science like history; 

 but the average required is manifestly much less in inexact sciences 

 than in the exact. An erroneous quantity, e.g., in a mathematical 



