BLASTOGENIC VARIATIONS. 135 



the impress of their individuality upon the sex-cells, in 

 diminishing proportion according to their remoteness. 

 Such a fact can only be accounted for by assuming the 

 existence, in the germ-plasm, of definite units carrying 

 definite characters, and the regular halving in the aver- 

 age strength or amount of such characters during the 

 reducing division of the nuclear matter of the sex-cells 

 which precedes each act of sexual reproduction. 



It was stated above that Professor Pearson calcu- 

 lated the correlation constant between brothers to be 

 .4. In a remarkable memoir recently published Pro- 

 fessor Pearson * and his collaborators have collected 

 together all the statistics at present available, as to fra- 

 ternal correlation in the horse, the dog, and in daphnia, 

 as well as in man, and have found the mean of the con- 

 stants deduced from 19 series of observations to be 

 .4479. Individual constants range from .6934 down to 

 .1973, but doubtless some of the extreme values in 

 either direction are, for various reasons, invalid. It 

 seems probable, therefore, that fraternal correlation, 

 whether it concerns stature, cephalic index, eye-colour, 

 or longevity in man, or coat colour in the dog and horse, 

 may be taken to fluctuate about a mean value of .4 

 to .5. 



The greater part of this memoir, however, concerns 

 correlation in the vegetable kingdom. Professor Pear- 

 son points out that the individual puts forth a number 

 of like organs, such as blood corpuscles, spermatozoa, 

 petals of the flower, leaves of the trees, which are un- 

 differentiated, but that nevertheless there is a consider- 

 able amount of variation among these " undifferenti- 

 * Phil. Trans. 1901, A. p. 285. 



