Karl Pearson 
381 
they are assumed to prove. I fully appreciate the careful experimental work done 
by both Jennings and Hanel ; I am not pledged to any theory of life, or of inherit- 
ance ; it is the method and logic of their reasoning which concerns me in the first 
place. These are essentially inadequate and do not meet the points really at 
issue. That the degree of resemblance decreases in a geometrical pi^ogression, 
i.e. that offspring are more like their parents than their grandparents or great- 
grandparents, is not for me a theory, it is a deduction from the observed 
facts; it becomes a theory, which flows from an hypothesis, if you accept, say 
Mendelism*. Those facts would not be changed, if the theory of pure lines were 
demonstrated for either Paramecium or Hydra. But the establishment of that 
conclusion would affect biometricians in the indirect manner that any fundamental 
subversal of Darwinian principles destroys the only philosophical theory of life at 
present available, and so lessens the interest many feel in devoting time and 
energy to the measurement of life. 
(8) As I have taken Hanel to illustrate recent biological work on the in- 
heritance of variation, I will consider in this section another biological memoir, 
one on fertility. 
I have hitherto made no reference to a recent interesting paperf by Captain 
R. E. Lloyd, I.M.S. He states that he was led by some observations of mine on 
the fertility of Shirley poppies to question whether fertility was related to weiglit 
in rats. He gives two correlation tables which T reproduce below, but he makes 
no statistical reduction of these tables, but draws apparently from the mere 
examination of them the conclusion that gigantic and dwarf rats are just as 
fertile as common rats of average size. " There is clear evidence that the largest 
and smallest rats are quite as fertile as those of average size" (p. 264). 
Now the first impression I formed from Lloyd's tables before they were reduced 
was that they showed quite sensible correlation between weight of rat and number 
of young. And this on reduction proved to be the case, the dwarf rats have fewer 
offspring than the average rat, and the large rats have more offspring than the 
* A curiously ignorant account of the biometric treatment of heredity has recently been given 
by W. Weinberg: " Ueber Vererbungsgesetze beim Menschen," Zeitschrift fiir induktive Abstam- 
mungs- und VererhiuujHlehre, Bd. i. 1909, S. 377 ct seq. He does not appear while writing his paper to 
have known the difference between the correlation and regression coefdcients ; and only when penning the 
last paragraph did a vague inkling of the difference come to him — through the writings of Johannsen ! 
He overlooks entirely the allowances for assortative mating in biometric work, and accuses the biometric 
school of neglecting environment, when at any rate in the case of man it is the only group, which 
by statistical method and by direct investigation has endeavoured to allow for it. Weinberg has stated 
that the correlation between parents and offspring observed by biometricians in the case of the 
phthisical diathesis is due to a mixture of classes having different tuberculosis death-rates ; and this 
regardless of the fact that (i) the data of the biometricians were purposely drawn from very uniform 
classes, and (ii) the correlation of husband and wife for the same material was shown to be much 
smaller than that of blood relations and in some cases zero. It hardly seems needful to reply to 
criticisms of this character. 
t "The Relation between Fertility and Normality in Rats," Records of the Indian Muscutii, Vol. iii. 
Part III. pp. 261—5. 
Biometrika vii 49 
