RELATION OF RACE CROSSING TO THE SEX RATIO. 1 99 



Is the excess of male births in cross niatings numerically great 

 enough to be considered significant in comparison with the prob- 

 able errors involved? The evidence on this point is presented in 

 Table III., which compares the sex-ratio for each cross mating 

 with that for each of the two pure matings related to it. The last 

 column of the table gives the ratio of the difference in each case 

 to the probable error of the difference. In interpreting this last 

 column it will be remembered that a difference which is three or 

 more times as large as its probable error is to be regarded as 

 significant ; a difference which is between two and three times 

 its probable error is probably significant ; while a difference less 

 than twice its probable error when taken by itself is probably 

 not significant. In general, the technical biometrical use of the 

 term "significant" intends to convey the idea that the odds are 

 so great as to amount to practical certainty that a so-called 

 "significant" result did not arise simply as a purely chance 

 effect of random sampling, but represents a direct causal nexus 

 between phenomena. 



Table III. 



Comparison of the Sex-Ratios of the Offspring of Pure and Cross 



Matings. 



From this table it appears that the excess of male births in the 

 cross matings as compared with the pure is in general large in 

 proportion to its probable error. In only one out of the four 



