Ethel M. Elderton and Karl Pearson 
505 
These values were so startling and so contradictory, that we proceeded to 
eighth differences with the results : 
5,m,'l\mi.Ssm./= + '5481 + -0067, 
which emphasised as well as confirmed the previous results. 
Now it seems absurd to suppose that the deaths of female infants in one year 
can organically influence the deaths of males of the same group in the next year, 
or male infants the deaths of females in the successive year. But the extraordinary 
feature of these results is that wliile a high deathrate of female infants lessens the 
deathrate of males in the second year of life of the same group, a high deathrate 
of male infants increases the deathrate of females in the second year of life of the 
same group. 
In order to throw further light on the matter we investigated male and female 
deathrate correlations in the third and fourth years of life. We found 
s„"./5...'.5 , = --2640 + -0887, 
5..»3"\"'.-5o< = - ■"082 ±-0954. 
The second is practically zero, the first of no importance having regard to the 
high values of the correlation of deathrates of groups of the same sex in the third 
and fourth years of life (^-.--lOS ± -085; ? : - -731 + -078). Had we come to 
these values at first we should have been content, but the cross relation between 
the infant deaths of one sex and the deaths in the second year of life of the 
opposite sex was undoubtedly puzzling. 
We then proceeded to still further limit our conditions by determining the 
partial correlation between female infants in one year and males in the second 
year of life of the same birth-year when the deathrates of the males "in the first 
year of life and of the females in the second were both constant. We obtained 
5„»<,.5„»,,"W.««m, = + -1632 ±-0928, 
So»,'.5«,»,'5..,.s«W = + -2997 + -0868. 
Having regard to their probable errors these are of a quite different and 
negligible significance when compared with the values of 
given above. 
It is worth while noting that 
5«m2"\<.s„„, = - -2188 + -0908, 
S,n,,%,n,.S,m./ = + -1088 ± -QMS 
also give values of no practical importance. Or, to annul the spurious influence 
of infantile deaths of one sex, A, on deaths in the second year of sex, B, of the 
same group, it is more effective to render constant the deaths of A in the second 
year of life than of B in the first year of life. 
