Correlation and Variation ni the Toad 6ll 



tion classed as "healthy," /. e., the relation between variability 

 and correlation is inverse. 



General Hospital "Healthy 



Population Postmortems'' 



Coefficient of variability in (males) (malesj 



Weight of heart 3^-39 i7-7i 



Weight of liver 21.12 14.80 



Weight of spleen 5°-S^ 38.21 



Weight of kidney 24.62 16.80 



CoefEcient of correlation between 



Heart and liver -1931 .2780 



Heart and spleen . 1827 . 2654 



Heart and kidney -2577 .4004 



And again, with increase in age the variability of the heart and 

 spleen was found to remain constant but the correlation coefficient 

 increases steadily from .08 (25 to 35 years) to .25 (45 to 55 years). 

 How this relation between variation and correlation can be 

 explained mathematically is not clear but doubtless a mathe- 

 matical "description" will be forthcoming once the biological fact 

 is established. 



The last point to be mentioned is the relation of the groups of 

 larger and smaller individuals, /. ^., those above and below the 

 average, to the regression line. We have seen that among those 

 above the average in any respect the "scatter" about the regression 

 line is markedly less than among those below the average. The 

 full significance of this relation needs farther investigation but I 

 take it that this is an indication that the larger, and therefore in 

 general the older, individuals gradually approach more closely the 

 type as measured by the coefficients of correlation. From the charac- 

 ter of the distributions found in this material this would seem to be 

 the result of growth rather than of selection upon the basis of 

 variability. That the coefficients of correlation as well as of 

 variability change with age is well known. That is to say in 

 growth, organs behave as more or less independent units. It 

 seems likely that normal processes of growth may consist to an 

 important extent in the gradual approach toward a condition of 

 more perfect general correlation. Data bearing upon this impor- 

 tant point are still meager; it is unfortunate that the age of these 



