1909-10.] The Significance of the Correlation Coefficient, etc. 507 
that given in Case I. (par. 33). The second is nearer that given in Cases 
II. or III. 
That is, the height of the first can be explained on the ground that 
three races mix with the production of distinct hybrids, that of the second 
on the ground that dominance of some sort manifests itself with growth, an 
explanation as possible as that of the authors who attribute the fall in the 
correlation coefficient to the selection of puppies. The fraternal correla- 
tions are also high, being r = ’6 76 for brethren of same litter and r — ' 559 for 
the selected record, both much higher than the Mendelian formulae given 
in par. 38 warrants. A cause of such high coefficients has been shown in 
par. 40, where it is noted that if three races mix fraternal correlation is 
raised. 
(Issued separately August 1 , 1910 .) 
