320 Margaret Morrison
Mendelian principles using biometric techniques, Pearson claimed that a general
pure gamete theory was too inelastic to cover the facts. In other words, there was
no room for manipulation of existing theoretical assumptions that could result in
a convergence of the regression coefficients to the same value. From the point of
view of a Mendelian analysis there were simply too many details that needed to be
considered. Although there was nothing inconsistent between biometry and such
a general Mendelian theory, the laws required for a formulation of the latter (e.g.
laws governing reproduction as a result of random mating, to name just one type)
also needed to be tested against the inheritance statistics for the relevant popu-
lations. However, deducing all of these effects in the field of inheritance involved
extremely complicated mathematical analysis — too complicated for the consid-
eration of single cases. Pearson considered his pure gamete theory a “generalized
Mendelian theory” for the case he had considered and claimed it would lead to the
same system of inheritance developed by statistical description of observed facts.
In other words, although this version of the Mendelian theory could not stand
alone as providing general principles of inheritance it created no inconsistencies
with the biometric account. The statistical discrepancies could be resolved by
adding some additional assumptions to the theory or by introducing some real life
complexities like homogamy or differential fertility into the theoretical scheme.
The advantage of biometric analysis was that it could attain a level of certainty
that was impossible using Mendelian schemes. If one was interested in how char-
acteristics were inherited and how populations changed over time and then one
needed statistical techniques that would permit the measurement of changes in
frequency of phenotypic characters, as well as large representative samples from
which to measure population parameters. Once the observations were analysed
into statistically defined populations they formed a “model” of the real popula-
tion. Biometrical statistics was designed to apply to large samples drawn from
human, plant and animal populations. And, if the samples were large enough
then one could supposedly substitute the sample statistics for the population pa-
rameters (e.g. mean, standard deviation, correlation coefficient etc.). This kind
of statistical practice together with the notion that species needed to be defined
in population terms separated Weldon and Pearson from most biologists (not just
Mendelians). Indeed, it was this concept of a “model population” that marked the
differences between Pearson and Fisher, and it was Fisher’s reconstruction of the
notion of a population that ultimately facilitated the synthesis of Mendelism and
Biometry.
Fisher studied physics and mathematics at Cambridge and had a keen inter-
est in Eugenics. As early as 1911, during his second year at Cambridge he had
familiarized himself with Pearson’s work and the issues surrounding multifacto-
rial Mendelian models. Although he had no obvious objections to Mendelism as
a theory of heredity he was concerned with its lack of predictive power beyond
the offspring of a particular set of parents. Although Mendelism was capable of
predicting with certainty the possible types of children of given parents, biometry
was more vague but capable of wider application. The probable measurement of