BREWSTER. — RANGE OP VARIABILITY. 279 



Conclusion. 



These four cases are the only oues to which I have applied this method 

 of measuring variability. They are, however, taken entirely at random, 

 and are in no wise selected cases. The conclusion to which they all 

 point is that which, on general grounds, seems most likely to be true. 

 This conclusion is, that there is so intimate a causal connection between 

 the characters of individuals and those of the allied groups into which 

 they are combined, that, in proportion as any character is variable in the 

 individuals of one group, it is different in the allied groups. 



Finally, I have to express my great indebtedness for assistance, criti- 

 cism, and suggestion to Dr, Charles B. Davenport and Mr. Frederick H. 

 Satford, both "of Harvard University. 



WOLFBOROCGH, N. H., 



March 5, 1896. 



