EFFECTS OF INBREEDING AND CROSSBREEDING. 49 



The lack of correlation between the two elements in fertility, size 

 and frequency of litters ( + 0.04 and —0.03 in the two periods) 

 is also surprising. Here again the hereditary factors must be largely 

 independent of each other. A small amount of common heredity 

 may be obscured by a tendency toward negative correlation due to 

 independent factors. Within inbred stock large litters tend to be 

 followed by a long delay, and fertilization immediately after the 

 birth of a litter tends to result in a small litter. These tendencies 

 are not very important, however, as was shown in the first part 

 of this bulletin. 



The correlations between average weight and ability to bear 

 and raise the young successfully have more tendency to be negative 

 than positive. The lack of correlation between percentage born 

 alive and birth weight or gain is noteworthy because genetic differ- 

 entiation with respect to these characters is beyond question. The 

 absence of significant correlation between the percentage born alive 

 and size of litter is important for the same reason. The other 

 correlations which connect vitality and fertility agree in giving no 

 significant indications of heredity of general vigor. There is finally 

 a tendency toward negative correlation between body size and 

 frequency of litters. 



It will be seen that in general the differences among the families 

 are of a different kind from those which distinguish individuals in 

 a given stock or similar stocks raised under different conditions. In 

 the latter cases we are undoubtedly dealing largely with differences 

 n general condition. In summer and fall litters are both larger and 

 more frequent, larger percentages of the young are born alive and 

 also raised among those born alive, and growth is more rapid, than 

 in winter and spring. Inspection of the yearly fluctuations in the 

 averages of the various characters (given in Part I) shows that if 

 the year were made the unit very high positive correlations would 

 be obtained in all cases instead of the insignificant ones of Table 4, 

 got by making the family the unit. This would be true even though 

 the downward trend of all characters were eliminated. The conclusion 

 which is forced on us by these considerations is that there is little 

 or no differentiation among the f amilies in general vigor, but instead 

 a fixing in each family of particular traits in some particular com- 

 bination. 



A detailed study of the combinations of characters actually found 

 in the 22 families is instructive. For this purpose it is convenient to 

 arrange the families in the order of excellence in each character in 

 each period of years. A division into Hre groups makes the relations 

 more easy to grasp. In Table 5 the best three families in any 

 respect are given rank A; the next five, rank B; the middle six, 



