EFFECTS OF IISTBREEDI^STG AND CROSSBREEDING. 



47 



■ Table 4. — Correlation between the family averages for various pairs of characters. 

 Calculated separately for 1906 to 1910 and 1911 to 1915. 



[Significant correlations indicate physiological interrelation. Effect of size of litter on mortality and 

 growth eliminated from consideration by use of indexes.] 



1906-10. 



VitaUty 



Growth 



Fertility 



Vitality with growth. 



Per cent born alive with per cent raised (BA) ' 



Birth weight with gain 



Size of litter with Litters per year 



Per cent born alive with birth weight 



Per cent born aUve with gain 



Per cent raised (B A) ^ with birth weight 



Per cent raised (BA) i with gain 



Per cent raised with 33 days' weight . 



VitaUty with fertility. 



Fertility with growth. 



Per cent born ahve with size of litter 



Per cent born alive with Utters per year 



Per cent raised (BA)' uith size of Utter 



Per cent raised (BA) i with Utters per year. 



Per cent raised with young per year 



Size of litter with birth weight 



Size of Utter with gain 



Litters per year with birth weight . 

 Litters per year with, gain 



+0.03 

 + .75 

 + .04 



- .08 

 + .03 

 + .07 

 + .02 



- .04 



- .10 



+ .04 



+ .28 



.00 



+ .03 



+ .26 



+ 



Young per year with 33 days' weight. 



+ .09 

 4- .21 



+0.30 

 + .59 



- .03 

 + .01 



- .28 



- .21 



- .23 



+ .12 



+ .01 

 + .17 

 + .2:3 



+ .29 



+ .62 

 + .62 



- .34 



- .22 



+ .22 



1 Per cent raised of those born aUve. 

 I 



The demonstration of significant differences among the families 

 in rate of growth, fertihty, and mortality among the yomig, raises 

 the question as to whether the variations in these characters are in- 

 herited independently of each other, or are merely so many mani- 

 festations of a differentiation in general vigor. Such a differentiation, 

 if present, might be due to the transmission of disease in families 

 as well as to genetic causes. To settle this point, the various char- 

 acters were correlated with each other in each period, again using 

 the product-moment method. The results are given in Table 4. 



Before examining these results in detail it will be well to point out 

 that if the differentiation among the families were merely in general 

 vigor we should expect to find even higher correlations than where 

 the families were compared at different periods of time with respect 

 to a given character. In the present case we are comparing char- 

 acters of the very same animals. 



The results, however, contain few correlations which are at all 

 significant, remembering that only values above 0.40 can be looked 

 on as definitely significant, while values below 0.30 can be given 

 little weight. Among the 36 correlations, only 4 are above +0.40, 

 and only 2 more are between +0.30 and +0.40. The latter are 

 balanced by two correlations between —0.30 and —0.40. This, 

 of course, is directly antagonistic to the hypothesis of a differentiation 

 in general vigor, genetic or otherwise. 



Two of the four significant correlations are between birth weight 

 and gain— namely, +0.75 ±0.06, and +0.59 ±0.09 in the two 



