POSTNATAL GROWTH IN THE ALBINO RAT 45 



SO there appears to be a tendency to closer correlation, as well as 

 to greater variability, in the males. It will be observed, on com- 

 paring this table with the preceding, that (excepting the head) 

 the organs most closely correlated with the body weight are not 

 those of least variability in weight, but a group which is some- 

 what more variable than the body as a whole (kidneys, liver, 

 lungs). In these organs, the average coefficient of correlation 

 varies from 0.80 to 0.86. 



Next come the brain and heart (0.78). The gonads are more 

 closely correlated (0.75) with the body weight than might be 

 anticipated from their variability in weight; while the eyeballs 

 (0.45) and suprarenals (0.40) are lowest in the scale. 



The average coefficient of correlation for the organs studied 

 is 0.70. It will be noted that there seems to be a direct relation 

 between the variability of the viscera and their correlation with 

 the body weight. In both it is evident that the average coeffi- 

 cient is lowest at birth and seven days, and highest at three 

 weeks. 



As a matter of fact, however, a higher coefficient of correlation 

 is to be expected during growth at times when the body weight 

 and organ weight are most variable. This is due to a 'spurious 

 correlation,' which has been referred to repeatedly in connection 

 with the discussion of correlation of the individual organs with 

 the body weight at the definite age periods. When all the rats 

 at a given age are grouped together there is naturally, on account 

 of their unequal growth, a considerable scattering of the body 

 weights and corresponding organ weights. This causes an aug- 

 mentation of the real coefficient of correlation, in accordance with 

 the general principle of ''correlation due to heterogeneity of 

 material" (Yule '11, p. 214). To determine the true correlation 

 between the body weight and organ weight at any given age, it 

 would be necessary to have a sufficient number of animals so 

 that they could be separated into groups of approximately the 

 same body weight or organ weight. The present data are unfor- 

 tunately inadequate for this purpose. It is evident, however, 

 that all the coefficients of correlation above given are somewhat 

 too high, and that no conclusions can be drawn from them as 



