The subject-matter of the biological and medical sciences is remarkable for its richness and complexity. Moreover, the wide range of variation observed in both organisms and their environments is frequently analysable into simpler components only with great difficulty. Suppose we want to compare the behaviour of two different animal populations. Not only does each population consist of individuals differing amongst themselves with regard to factors like sex, age, physical measurements, coloration, susceptibility to disease, aggressiveness, etc., but the patterns of behaviour in which we are interested may themselves be fairly complicated. For these reasons, much biological work tends to be comparatively quantitative in nature. In the more exact sciences of physics and chemistry, on the other hand, we find that irreducible variation is usually fairly small, and often consists of little more than experimental errors. The latter can, as a rule, be virtually eliminated by averaging over several repeated determinations.

In biological sciences, therefore, inherent variation must be accepted as basic, and must be handled as such. This certainly makes numerical arguments more difficult. We may talk about the average number of eggs laid by a certain species of bird under particular environmental conditions or the proportion of subjects protected by an immunising vaccine. But these average figures conceal the fact that specific instances may easily show very different results. Some females will produce very large clutches, others will not lay at all.