376 JOHN C. PHILLIPS 
there is a difference of only 200 grams. The coefficient of varia- 
bility (c. v) is 5.82 for the males and 3.43 for the females. 
In 1911 an attempt was made to raise another lot of Fi’s but 
only two fertile eggs were obtained and these were discarded. 
The F, generation (table 4) was obtained by mating the three 
F, females, Nos. 83, 101 and 106, to male No. 105. -This male 
is slightly above the mean size of the F; males, which fact perhaps 
accounts, in part at least, for the larger size of the F, mean, as 
compared with that of the F;. The eggs from this mating were 
set in three lots, May 29, June 8 and June 15. The entire class 
was killed and weighed on October 29. The ages therefore varied 
from one hundred and thirty-four to one hundred and fifty-one 
days. The mean weight of the seven youngest males, hatched 
June 15, is 1738 grams, as against 1863 grams for nine males 
hatched May 29, a difference of about 7 per cent. This shows 
that adult weight is very nearly reached at four and one-half 
months of age. 
A glance at table 4 shows the very remarkable variation which 
these F,’s exhibit. There is an actual range of variation in weight 
of 887 grams among the males and of 650 grams among the fe- 
males. The smallest male, No. 158, is of substantially the same 
weight as the largest pure mallard. The largest F, male, No. 175, 
is only slightly smaller than the mean of the pure Rouen males for 
1911, the difference being only 2 per cent. 
The F, coefficient of variation is 12.07 for the males as against 
5.82 for the F, males, while the F, females have a coefficient of 
variation of 11.07 as against 3.43 for the F, females. 
It is-interesting to note that the coefficient of variation is larger 
in the males than in the females, in all groups except the pure 
large race, where the reverse is the case. 
GENERAL CONSIDERATIONS 
A fuller statistical analysis of the foregoing data may help 
to make their significance plain. For this purpose the observed 
weights may be classified in a variation table. Since males are 
on the average heavier than females, it is evident that separate 
tables must be made for males and females. Let us take the mean 
