[bowman] fundamental PROCESSES IN HISTORICAL SCIENCE 567 



the individual in the throngs of a crowded city; but no such verbal 

 description can vie in interest with the living face, or even with a good 

 portrait or photograph; and, to a person who has never seen a loco- 

 motive, the best account of it that he has ever heard will seem but a 

 tame affair compared with the real thing, when for the first time it 

 comes thundering down the track at his side. So it is as between 

 contemporary and later records. The change in interests, in modes 

 of thought, and in outlook upon life generally, that creeps impercep- 

 tibly over a community even within a generation, may be accurately 

 narrated in review; but the very person who has lived through it all, 

 and can appreciate from his own experience the accuracy of the nar- 

 rative, will realize more sharply the distance he has travelled from the 

 starting-point, by turning to a book, a magazine, a newspaper, a let- 

 ter, or any other literary product of forty, thirty, or twenty years ago. 

 This sharper realization, however, is not a matter of trustworthiness 

 in the contemporary product, nor of untrustworthiness in the later 

 narrative. In the pages of the book, magazine, newspaper, or letter, 

 even though untrustworthy, the interests and the whole mental and 

 social atmosphere of the earlier circle and time will be a living, pulsing, 

 breathing thing; but in the later narrative, however accurate, this 

 life cannot live again. 



8. Harmonization of Discrepancy. The practice of harmonizing 

 conflicting statements in trustworthy records originates as a natural 

 development in the present historical method, because the method as 

 a whole and this practice in particular both rest essentially on the use 

 of reasoned probability as a positive criterion. In this practice, not 

 only is value ascribed to reasonable harmonizations, but the pos- 

 sibility of suggesting one is made the test of error in the conflicting 

 statements. The value of such harmonizations, and the validity of 

 this test, are both disproved experimentally in the paper on the 

 Origin and Treatment of Discrepancy in Trustworthy Records men- 

 tioned in the paragraph on General features of the erratic nature of 

 probability (p. 538 above). The view that, if a reasonable harmoniza- 

 tion seems possible, neither of the conflicting statements is in error, 

 but if such a harmonization seems impossible, then one of the state- 

 ments at least must be erroneous, is so deeply rooted in our modes of 

 thought that, at first sight, it seems strange that the validity of the 

 test should be questioned at all. Nevertheless even such points should 

 be examined wholly without prejudice; and this test marks in fact as 

 extreme a development as there is in the unscientific use of probability 

 in history, because the probability involved is no longer probability 

 on the evidence (which would also be unscientific), but mere probabil- 

 ity in itself. The most apparent reason for the popular vogue of the 



