Karl Pearson and David Heron 
247 
But let us look at the matter from other standpoints. We have the stature 
data for Father and Son arranged in the eye-colour groups. Let us arrange it 
in a fourfold table, precisely as we did the eye-colour in order to demonstrate 
Mendel'ism, i.e. 
Stature of Father. 
Son. 
1+2 + 3 
4+5+6+7+8 
Totals 
o 
1+2+3 
464 
155 
619 
are 
4+5+6+7+8 
158 
223 
381 
Totals 
622 
378 
1000 
CO 
We may give these a class-name, say, Short and Tall — corresponding to Short 
Muzzle and Long Muzzle, or to "Over 30" and "Under 30" egg hens, etc., etc. 
We have at once a Mendelian table : 
Short Fathers 
Tall Fathers 
Totals 
Short Sons 
464 
155 
619 
Tall Sons 
158 
223 
381 
Totals 
622 
378 
1000 
Now assume a discrete unit between Short and Tall and find the Boas-Yulean <£. 
It is 
$ = -335, 
and as its Mendelian value should be "333, the agreement is extraordinary ! Tall- 
ness and shortness in man are clearly Mendelian presence and absence of a unit 
character ! The eye-colour tables at the same divisions show exactly the same 
result. Surely Mr Yule is satisfied that the correlation of stature in man is of 
the order | ? Yet the tetrachoric r, in these cases at the same division gives for 
eye-colour "55 and for stature "51, i.e. is some 60 to 70 °/ Q greater. Now let us 
treat these two cases by an identical process ; we use a normal horizontal and a 
normal vertical scale, and we determine all ranges in terms of the range of the 
frequency of groups 3 + 4 treated as of length h. We find the mean values of the 
arrays of sons for each group of fathers by assuming that the means will be 
approximately given if the array be treated for this purpose as Gaussian (see 
p. 236). The results are given in the table on p. 249. 
Can any reader who examines these data deny the remarkable parallelism of the 
two cases ? Let him also look at the Diagrams VIII and IX (p. 248) drawn for 
stature and eye-colour and ask himself whether it is possible to make any distinc- 
tion between the cases of inheritance of eye-colour and inheritance of stature, 
beyond the irregularity depending on the group of 36 cases (3"6 °/ 0 of the whole) 
of dark blue-eyed fathers. The two regression lines run with almost complete 
agreement, the stature data being a little more regular than the eye-colour data. 
