302 
PROFESSOR KARL PEARSON AND MR. LESLIE BRAMLEY-MOORE, 
Now this is precisely what we might expect; the mares belonged to a class, of 
6 
w hich we are not certain whether their daughters have or have not recorded fecundity. 
The mean fecundity is therefore decreased and the variability increased. Add to tiiis 
group 2400 mares, all of which had had their daughters’ fertility recorded, and w^e 
find for 3909 mares, = *6345 and cr^ = *1910, i.e., the mean fecundity ascends 
and the variability falls. Illustration of this law will be found in the following two 
groups :— 
Mifi, 
1200 mares. 
•6337 
•1888 
1200 dams. 
■6525 
•1643 
Thus we send up the mean fertility and lower the variability by separating into two 
groups the pedigree of one which has a longer record. This is precisely in accordance 
with the theory already developed. Our mean fecundity and variability for brood¬ 
mares may be considered as constant characters, and variations in their values beyond 
their probable errors due to conscious or unconscious selection in the record itself, 
or in our extracting from it. 
The reader will notice at once, if he turns to the diagram of the above frequency, 
(i.) that there is a small hump at {a) of no practical importance, and a larger one 
at {q), perfect fertility being fairly frequent with only four coverings, and there being 
from the arithmetical processes involved a bias towards ( 2 ) as compared with {p). 
(ii.) The distribution of frequency, although somewhat ragged, is quite clearly not 
normal, but of the character which in other pajjers I have called shew. Were there 
any occasion, it wmuld be easy to fit it with one of my skew curves. To mark 
how (i.) will disappear and (ii.) become still more apparent, I have placed on the 
diagram the frequency distribution for 2000 mares reduced to the same scale. 
