Apkil 14, 1911] 



SCIENCE 



569 



curate measurements of the anthropometric 

 laboratory. 



Furthermore, any weakness in the method 

 of grading, any failure to properly classify the 

 great men in the high grades and the degen- 

 erates in their proper grades would work in 

 precisely the same direction to lower the corre- 

 lation coefficients. The supposed errors of 

 history and the difficulties of grading act as 

 two united strains of tension to pull the coeffi- 

 cients down towards zero, which would be the 

 coefficient for random distribution. If the 

 coefficient can stand the strain without declin- 

 ing, then, roughly speaking, we may conclude 

 both that the historical foundation is just 

 and that the method of procedure is sound. 



There are two other illustrations of method 

 which I would like to summarize here. One 

 of these series of tests is the trying out of a 

 standard biographical dictionary (historical 

 persons) against two lists of contemporaries 

 (non-historical persons) and all three in terms 

 of still another set of facts, namely birth- 

 places of distinguished Americans. The sec- 

 ond series of tests concerns the relative fame 

 of Euripides versus Sophocles, the encyclo- 

 paedias having been used and then this com- 

 pared with expert modern criticism and both 

 with the opinions of the Athenians. 



As concerns Ajnerican history, one fact is 

 very evident at the start, whatever be the 

 method of grading as applied to Americans or 

 whatever be the mental eminence graded, some 

 states in the union, some sections of the coun- 

 try, have produced more eminence than others 

 far beyond the expectation from their respect- 

 ive white populations. In this regard Massa- 

 chusetts always leads, and Connecticut is al- 

 ways second, and certain southern states are 

 always behind, and fail to render their ex- 

 pected quota. I have already pointed out^ that 

 the ratios seem orderly for a first approxi- 

 mation. That is, the higher the grade of the 

 individuals the greater and greater becomes 

 the proportion of those born in Massachusetts. 

 This may be expressed as a ratio, p into the 



' ' ' American Men of Science and the Question 

 of Heredity," Science, N. S., Vol. XXX., No. 

 763, pp. 205-210, October 13, 1909. 



random expectation. Thus if there were no 

 forces at work beyond chance distribution the 

 ratios for all sections of the country would be 

 expressed by unity, p ^ 1. If there be found 

 twice as many persons born in a certain lo- 

 cality as one would expect from the population 

 let it be expressed as p ^ 2, three times as 

 many, p ^ 3, etc. These ratios are easily com- 

 puted and can be expressed as fractions or with 

 decimals. I have computed these ratios for the 

 thirteen original states, but will present here 

 only the statistics from Massachusetts and 

 Virginia. 



It will be seen in Table I. that Massachu- 

 setts has never failed to produce twice as 

 many eminent men as the population would 

 lead one to expect, and has for some ranks and 

 types of achievement produced about four 

 times the expectation, p ranges between 2.1 

 and 4.7. Virginia, on the contrary, has but 

 rarely produced as many as might be expected 

 from the large white population and the ratios 

 in the same table are either below the expecta- 

 tion or not significantly above it. The other 

 New England states (statistics not here given) 

 have all done more than their share, but al- 

 ways less than Massachusetts. New Tork 

 gives a trivial though constant excess above 

 the expectation. From here southward the 

 ratios drop off suddenly, so that New Jersey, 

 Delaware, Pennsylvania, Maryland, North 

 Carolina and Georgia have always furnished 

 less than their share. For South Carolina the 

 ratios again rise and exceed the expectation, 

 but only by the slightest measurable amount. 

 North Carolina, of all the thirteen original 

 states, has always had the worst record in the 

 way of producing distinguished men; the 

 ratios falling to about one quarter of what 

 might be expected from the white population. 



Kegarding the tables for the two contrasted 

 states, Massachusetts and Virginia, and fol- 

 lowing down through the columns marked 

 " ratios, or number of times the random ex- 

 pectation according to the population at the 

 approximate age of their birth," one sees first 

 that the Massachusetts ratios run from 2.1 to 

 3.9 and the Virginia from 0.2 to 1.1. The 

 higher Massachusetts ratios are associated 



