Miscellanea 
121 
TABLE II. 
Correlation of Sex-Ratios of First Four Members of Mother and Laughter Mares' 
Produce. Thoroiirjhhred Horses front The General Stud Book. 
Sex-Eatio of First Four Members of Daughter's Produce. 
O X S 
o 
Oh 
■^.^ 
0) 
C/3 
4 
1 
i 
t 
.3 
5 
4 
+ 
Totals 
0 
+ 
3 
12 
35 
20 
10 
80 
1 
i 
20 
65 
119 
72 
15 
291 
t 
19 
104 
137 
97 
23 
380 
3 
I 
8 
45 
84 
63 
13 
213 
4 
-I 
1 
8 
17 
10 
36 
Totals 
51 
234 
392 
262 
61 
1000 
of sex-ratios derived from completed families of eight and ujjwards and from the first four 
members of these families respectively, while the means and standard deviations are given 
in Table III. 
TABLE III. 
Group 
Mean 
S. D. 
Completed Families 
First Four Members Only 
(Mother's Produce 
(Daughter's Produce 
(Mother's Produce 
(Daughter's Produce 
•4620+ -0029 
•5020 ± -0033 
•4585 + ^0052 
•5012+ ^0052 
•1379 ± ^0021 
•1546+^0023 
•2425 + ^0037 
•2426 + ^0037 
In the completed family of the mother, at least one female must appear and this of course 
reduces the mean sex-ratio. When the oflspring were not selected in this way, the mean sex- 
ratio was •502. If we now apply the condition that at least one member is to be a female, the 
n—1 
sex-ratio ought to become — — ('502) where ?i is the number in the family. The average family 
was found to be ir433 and substituting this for n we get the sex-ratio to be '458. Actually it 
was found to be '462, an excellent agreement. 
When the question of variability is considered we find as before that o-^. is zero. Taking the 
first four members of the family, the standard deviation is "2426 ± ^0037, while that which would 
/•502 X ^498 
arise from a chance distribution in families of four is -y/ = •2499. The difference 
is within the probable error and it is to be noted that the actual s.d. is less than that which 
would arise from a chance distribution. According to Mr Cobb's suggestion, it ought to be 
greater. 
If we consider completed families (only the 1000 cases in the second generation were taken) 
the theoretical variability is given by 
Biometrika vi 16 
