1156 
Journal of Agricultural Research 
Vol. V, No. as 
Figures i, 2, 3, and 4 show the curves for the litter frequencies in the 
three generations and indicate how close the actual numbers of litters 
come to the binomial curve (x + y) 14 . It is perhaps incorrect to call the 
theoretical frequencies recorded in Table IV “expectations,” unless it is 
clearly understood that they are the expectations founded on the nearest 
binomial. There is nothing in the inheritance to make them true expec- 
_ „ tations from an experi- 
,5* <? ^ e 9 /& ✓/ /g /3 /a mental standpoint. 
The curve of the actu¬ 
al distribution spreads 
out at the extremities 
much wider than should 
be expected on a chance 
basis. This is true in re¬ 
gard to both extremes 
of the curve and makes 
it appear as though the 
curve were compound— 
i. e., the sum of several 
curves having separate 
means from which devi¬ 
ations take place. An 
analysis on this basis 
gives two small curves 
at the extremes, which, 
while they do not give 
perfect Gaussian distri¬ 
butions, are characteris- 
- „ ,,... , . . _ ... enough to make the 
Fig. i.—C urve of litter frequencies in the P generation of swine. 
assumption of another 
mean for each perfectly valid. The modes of these three curves are as 
follows: 
Curve 1. 4 pigs per litter. 
Curve 2. s pigs per litter. 
Curve 3. 12 pigs per litter. 
It is premature to announce that these modes represent centers of 
deviation for genetic factors, although a casual observation of the individ¬ 
ual data makes it^ seem that this condition may exist. Furthermore, 
the mode of curve 1 corresponds to the degree of fertility which Simpson 
states is characteristic of the wild hog, while the mode of curve 3 is very 
close to that of the Tamworth, the most fecund of domestic breeds. 
This indicates that the two may represent basic and improved factors 
for fertility, respectively, while curve 2 represents heterozygous con¬ 
ditions. 
