VARIATION AND HEREDITY IN TWINS 177 



in the germ-plasm associated with the soma of each 

 individual. Within a single set of armadillo quad- 

 ruplets one individual having a unit character (a dou- 

 bhng of integumentary units) in the soma transmits this 

 character to its offspring, while another individual 

 (derived from the same zygote) lacking this character 

 in the soma fails to transmit it to its offspring. In 

 conclusion, I should therefore like to emphasize the 

 fact that somatic divisions may be as important agents 

 in segregating unit characters as germinal divisions 

 involved in the formation of gametes (maturation or 

 reduction divisions) are beheved to be. On this theory 

 it might well happen during cleavage that cells derived 

 from one part of a single gonad would, prior to matu- 

 ration, contain certain determiners which others in 

 another part of the same gonad lack. 



It would appear, then, that the hmits of hereditary 

 control are not to be measured by a comparison of 

 twins, or even by comparing the antimeric halves of 

 the same individual. The whole question hinges on the 

 equahty or inequality of distribution during cleavages 

 of the determinative factors; this involves what we 

 have called somatic segregation. 



STATISTICAL METHODS OF DETERMINING THE LIMITS OF 



HEREDITARY CONTROL 



Another more rehable method of testing the limits 

 of hereditary control than those applied to individual 

 cases is the statistical method which applies to large 

 groups. We can find out how strong on the average is 

 the hereditary control exercised by the predeterminative 

 mechanism of the germ-plasm with respect to certain 



