AN EXAMPLE OF TAXONOMIC DISCRIMINATION BY BIOMETRIC METHODS

Taxonomy, in effect, consists in placing limits to groups recognized. These boundaries should be arranged to coincide with the points where there is the greatest discontinuity of variation. At present the boundaries between groups depend on personal judgement, although in many cases it would seem obvious that they do indeed correspond to real discontinuities. In other cases the decision of one systematist is not accepted by his colleagues who may propose alternative groupings or limits, because they have a different opinion as to where the points of discontinuity lie. It may happen that one systematist considers a certain character to be more important than others and establishes his limits accordingly, while another considers quite a different character to be the most important and groups the specimens according to this character. Unless the two characters show complete correlation the result will be two quite different groupings. Assumptions, unsupported by any large body of observations, are often inherent in existing classifications and in their use. Morphological and anatomical features are usually the most convenient for use in the separation and delimitation of groups, and it is often assumed that these groups have completely correlated physiological, genetical, ecological and other characteristics. That such correlation can be shown to occur, at least for some special character in some groups, does not justify the assumption that it occurs in all groups. Examples can be found where it has been shown that no such correlation occurs. For example, in the Oxford district Cardamine pratense shows great morphological diversity and it is found that cytologically it is also very variable: diploid, tetraploid, hexaploid and octaploid strains all occurring together. It has been found impossible, so far, to correlate the morphological and cytological characters (B.S.B.I. Conference, 1948 and personal communication, J. Burnett). Any assumption based on morphological characters regarding the cytological nature of an individual plant from these populations would not be justified. As already mentioned, the overlapping of the morphological variation of two groups is the reason why the groups can sometimes be distinguished from one another only by specialists. Much taxonomic work has been done by people with a flair for perceiving differences in morphological patterns without conscious analysis of the nature of the differences. The methods by which they distinguish their groups are not clearly defined and are difficult to communicate to others. What is required is a method of definition which is intelligible and usable by the non-specialist, and at the same time supplies a criterion for allocation permitting an estimate of the probability of correct classification. Range and distribution of variation are most easily expressed in quantitative terms, so