I02 Social Environment 



probable error (the probable limit of variation 

 from the given result by the use of more cases) 

 is not more than one-sixth of the coefficient.^ 

 On the face of the results, the correlation 

 between density of population and fertility in 

 men of letters is very decisive. It will be seen 

 from the tables that for 1850 the coefficient is 

 0.60 ±0.08; for i860 it is 0.72 ±0.06; while 

 for 1870 it is 0.76 ±0.05; or, if 46 states are 

 included, 0.70^0.05. There are, however, 

 two objections that may be immediately antici- 

 pated. It may be claimed that the results are 

 in the main due to conditions in the southern 

 states, where the Negro, though incapable of 

 contributing to the noted men, has yet been 

 included in the total population. In another 

 computation, therefore, the Negro has been left 

 out of account with the result that the coeffi- 

 cient based on the census of i860 is found to 

 be 0.74 ± 0.06, two points higher than it was 

 previously. It is clear, then, that the inclusion 

 or exclusion of the Negroes is not a significant 

 factor. Still, it may be claimed that the corre- 

 lation is due to other sectional contrasts be- 

 tween the North and the South ; or, on the other 



1 King, W. I., Elements of Statistical Method, p. 215. 

 The Macmillan Co., New York, 1912. 



