312 Margaret Morrison
decreasing step-by-step in a geometrical ratio with great rapidity”.^1 This also
marked the introduction of statistical methods into biology, the defining feature
of what would become, in the hands of Karl Pearson, the science of biometry.
In 1869 with the publication ofHereditary GeniusGalton elaborated his idea
that evolution proceeded by discontinuous leaps but he needed a theory of heredity
to back it up. By 1875 he developed the view that hereditary qualities were
concentrated in the reproductive organs rather than embedded in gemmules, which
according to Darwin’s pangenesis theory were formed in all parts of an organism.
The germ plasm or “stirp”, as Galton called it, was continuously inherited by each
generation with very little alteration. Variations were caused by alterations of this
germ plasm and were distinct. The publication ofFinger Printsin 1892 marked the
forceful declaration that evolution proceeded by jerks through successive sports,
each of which is favoured by natural selection. Prior to this however, in the period
between 1877 and 1889 Galton gathered data on the inheritance of size in sweet
peas as well as the stature, eye colour, temper, disease and artistic ability in man,
all in an attempt to derive a quantative law of regression, which was given a
complete formulation inNatural Inheritance[1889]: “...the deviation of the Sons
from P [the median stature of the general population] are, on average, equal to
one-third of the deviation of the Parent from O, and in the same direction... If
P+(+- D) be the stature of the Parent , the Stature of the offspring will on the
average be P+(+-1/3 D).^2
Although the law was based on data about stature he thought it applicable
to non-blending inheritance like eye colour in humans and coat colour in Bassett
hounds where the total heritage was represented by percentages in the offspring.
In other words, since coat colour fell into one of two categories (tricolour or lemon
and white) one could accurately predict that one quarter of the litter would take
after the mother, one quarter after the father, one eighth after each grandparent,
a sixteenth after each great grandparent etc. Indeed his work on regression and
the law of ancestral heredity convinced him of the ineffectiveness of selection on
individual differences, a view first put forward inHeredity Genius[1869] where he
claimed that “because an equilibrium between deviation and regression will soon
be reached,... the best of the offspring will cease to be better than their own
(^1) Pearson pointed out that Galton should have referred to mid-parent, mid-grandparent etc.,
something that he did do in a later formulation [1885]. The mid-parent is a fictitious individual
whose height, for example, was equal to half the sum of the paternal and adjusted maternal
heights.
(^2) Although the data suggested that the average regression of mid-filial stature upon mid-
parental stature was around 3/5, Galton substituted 2/3 because it was simpler. As Provine
[1971] points out, the value 2/3 enables him to calculate the mid-filial regression on a single
parent without sufficient data. If each parent contributes equally the value will be^1 / 2 of the joint
contribution or^1 / 2 of that of the mid-parent. Hence, the average regression from the parental to
the mid-filial stature will be^1 / 2 of 2/3 which equals 1/3. His linear regression equation took the
following form: x=(2/3)y, were x=X(filial stature)-Mf,(mean filial stature), y=Y(midparental
stature)-Mp(mean midparental stature). 2/3 is called the regression coefficient and where x and
y have the same variance it is identical with the correlation coefficient, the value of which is given
by x/y.