568 



SCIENCE 



[N. S. Vol. XXXIII. No. 850 



phenomena, as seen to-day, represent a 

 phase in the evolution of thermal springs. 

 Arnold Hague 

 U. S. Geological Suevet 



HISTOSIOMETBY AS AN EXACT SCIENCE 

 . In the issue of Science for November 19, 

 1909, under the title " A New Name for a 

 New Science " I proposed the term histori- 

 ometry for that ' class of researches in which 

 the facts of history have been subjected to 

 statistical treatment according to some 

 method of measurement more or less objective 

 or impersonal in its nature. These researches 

 have chiefly had in view the listing and 

 grading of historical characters, either for the 

 purpose of studying mental heredity or for the 

 better appreciation of problems associated 

 with the psychology of genius. Researches of 

 this type are capable of a far greater espan- 

 sion and application than is generally sup- 

 posed. They can be applied to events as well 

 as to individuals. They can, by treating the 

 vast store of human records which exist in 

 books as material for the construction of an 

 exact science, work towards the solution of a 

 wide range of historical problems, such as the 

 causes underlying the rise and fall of nations 

 or other fundamental questions in history. 



Before anything can be done which shall 

 give general satisfaction and agreement it will 

 be necessary for this subdivision of science to 

 justify itself, to measure its own shortcom- 

 ings, to appreciate its own limitations, as well 

 as to prove its own right to recognition of in- 

 dependent estate. 



If we are to fathom historical causation by 

 objective methods it is obligatory first to prove 

 that history itself, as we commonly find it in 

 the printed records, is a sufficiently valid ac- 

 count of what actually happened. Second, it 

 is equally necessary to find proof that the ob- 

 jective methods correctly deal with these facts. 

 It might be supposed that the second proof 

 awaits the first; but this is not necessarily so. 

 If the records themselves were very much at 

 fault, so that the statements of historians 

 were very far from ideal truth, or if the ob- 

 jective methods of collecting and analyzing 



these statements were subject to a large error 

 (or if both these forces were in play) then it 

 would be difficult to find wherein the trouble 

 lay. But if, on the contrary, it fortunately 

 be that history as we find, it is in its important 

 statements a fair representation of the truth, 

 and if the methods of historiometry which 

 deal with these records are also sound, then it 

 is not difiicult to prove both propositions at 

 the same time. 



I will give some instances to illustrate this, 

 which show that such is the case for several 

 types of historical records and for several 

 methods of history measurement. This could 

 not be done did we not possess some third cri- 

 terion, some third standard of comparison of 

 a non-historical nature. One such non-his- 

 torical criterion is furnished by the known 

 correlation ratios for resemblances between 

 close blood relatives as determined in the 

 anthropometric laboratory. These have been 

 worked out and accurately measured for men- 

 tal and moral traits, stature, head index and 

 length of forearm. I have shown in " Hered- 

 ity in Eoyalty "^ that if the members of royal 

 families are graded by the adjectives applied 

 to them by historians and encyclopaedists and 

 then the coefficients of resemblance are meas- 

 ured between the near of kin, who have been 

 so graded, these coefficients (historiometrie) 

 substantially agree with the anthropometric. 

 Such would not be the case if historians per- 

 verted the truth greatly, or if for any other 

 reason the truth were largely unattainable. 

 To make this clear it is only necessary to 

 think what the result would be if history were 

 merely " a pack of lies agreed upon " as the 

 extreme view puts it. We should then fail to 

 properly pick out our true intellectual giants 

 and runts. The result would be nothing but 

 confusion. A whole series of errors would be 

 distributed at random. This would act like 

 rain on waves and fiatten down to a common 

 level the real difierences between the individ- 

 uals. The correlation measurements would 

 fall and we should get no results comparable 

 to those obtained from the delicate and ac- 

 ' New York, Henry Holt, 1906. 



