The Development of Population Genetics 313sires and dams.” (p.34). In other words, any extremes caused by selection would
be returned to the centre by regression; hence, “sports”, which he considered quite
stable, must be the only effective source of evolutionary variation. This of course
was in direct conflict with Darwinism, which saw “sports” as having no role in
evolution due to their obliteration by blending inheritance.
Galton’s arguments against blending inheritance, which were backed up by pow-
erful statistical analyses of the data turned out to be incorrect. And, what initially
looked like a clever and potentially devastating strategy against the Darwinians
was actually turned to their advantage by Karl Pearson (1857–1936). Pearson
graduated with mathematical honours from Cambridge in 1879 but before em-
barking on his work in statistics he was a scholar of German history, folklore and
philosophy. In addition, he worked in the philosophy of science, the theory of
elasticity and was active in politics at the University of London, especially on
issues regarding socialism and the education of women. With the assistance of
his colleague the marine biologist W.F.R. Weldon 1860–1906), Pearson extended
Galton’s statistical work into biology while his Mendelian views were developed in
Britain primarily by the biologist WilliamBateson (1861–1926).^3 Although Wel-
don was not a mathematician he was particularly interested in Galton’s methods
but required the help of the more able Pearson to fully apply them in the context
of animal evolution. Bateson was interested in another aspect of Galton’s work,
the discontinuously varying traits such as white and red flower colour. This issue
was discussed by Galton inNatural Inheritancewhere he put forward his views
about evolution involving a discontinuous process. Pearson saw Galton’s work not
as “a biological hypothesis, but the mathematical expression of statistical vari-
ates... [which] can be applied... to many biological hypotheses” [Pearson, 1930,
21]. His goal was to give the work a more rigorous foundation and hence to secure
a scientific foundation for natural selection, a task which involved an evaluation of
the heredity law, specifically the regression constants in the geometric series (1/2,
1/4... ).
In a paper in 1895 Pearson pointed out that the multiple regression coefficients
1/4, 1/16, 1/64 etc. were incompatible with the assumption that correlations of
the offspring with the individual parent, grandparent etc. form the series r, r^2 ,
r^3 , etc. He showed that any such series caused all the coefficients except the first
or parental coefficient to vanish and reduced the ancestral multiple regression to
a simple biparental inheritance. Hence, the parental characters would determine
completely those of the offspring and contrary to what Galton had assumed, after
the relaxation of selection and commencement of inbreeding, there would be no
further regression after the first reproduction.^4 In 1895 Pearson formulated a
version of the ancestral law showing its connection to the law of regression. This
new law took the form of a multiple regression equation of offspring on mid-
midparental ancestry:
(^3) Although it is sometimes suggested that Bateson’s account of Mendelism deviated from
Mendel’s own formulation but that is not a topic I want to explore here.
(^4) See also [Pearson, 1930, 39].