M. Beeton and K. Pearson 
51 
(4) In an earlier memoir* by one of tlie present axithors some attempt has 
been made to resolve the mortality curve into components, of which the most 
important are those of extreme childhood and of old age. The mortality of youth 
and that of middle age are far less important. In keeping with this resolution we 
have mentioned in our first study f the existence of an inheritance of brachybioty 
as well as an inheritance of longevity. Not otjly for the purposes of manipulating 
the statistics, but also because we believe there is a real physical distinction, did 
we consider it desirable to separate brachybioty and longevity. But in order to 
avoid the immense labour of the cross-correlatious if we divided the span of life 
into four or five sections, and considered how far death in one of those periods 
influenced duration of life of brethren dying in any one of tliem, we adopted only 
a twofold division, distinguishing only between minors and adults. Even this 
involved the preparation of no less than twelve fraternal correlation tables for 
duration of life, and in the case of cross-coi'relations, such as brother or sister 
minor with brother or sister adult, we were rarely able to collect more than four to 
five hundred pairs of cases. 
Now a noteworthy result of our investigations is this : That while the corre- 
lation of adult and adult brethren and of minor and minor brethren are both quite 
sensible, in no one case is the correlation between the durations of life of an adult 
and a minor sensible within the limits of our probable error. The following table 
summarises our results : 
Eelationsbip 
Correlation 
Cases 
Adult brother and minor brother 
- -0262 + -0246 
753 
Adult brother and minor sister 
- -0062 + -0349 
374 
Adult sister and minor sister 
- -0260 + -0291 
537 
Adult sister and minor brother 
- -0274 + -0328 
421 
Now although none of these correlations standing alone are sensible, it is note- 
worthy that they are all negative; an examination of the Tables XIV. XVIII. 
XXI. and XXII. will at once reveal the source of this. It is the effect of inherit- 
ance of the mortality of youth ; minors dying from 16 to 20 are associated with 
adults dying from 21 to 2.5, i.e. minors dying late correspond to adults dying early. 
Diagrams of the regression lines| show this negative slope in the later years of 
minority. Here is the source of the negative sign. It might be thought, that if 
this effect of the mortality of youth were removed, we should get sensible positive 
correlations between the age at death of an adult brother or sister and of an infant 
brother or sister. But an examination of the regression lines as far as 10 or 12 
years of age shows that there is really but little correlation between the dura- 
* Phil. Trans., Vol. 186 a, p. 407. 
t R. S. Proc, Vol. 65, p. 299, 1899. 
X Not reproduced iu this paper. 
