During the period from 1753 to 1964, the balance between splitting and lumping (the median genus size) has remained unchanged in five classifications of the Leguminosae (Linnaeus, Persoon, Bentham, Taubert and Hutchinson), which span the period at approximately 50-year intervals. This is in contrast with the situation in the Gramineae where increased splitting in recent classifications has resulted in smaller median genus size. However, generic sizes in the Leguminosae do not cluster around the median but are uneven: the frequency distribution of genus size in all the classifications is very skew, with many monotypic genera and some exceptionally large genera. This parameter, equality, changes considerably over time with a trend to more inequality of generic size. This is interpreted as a psychohistorical artefact of the classification process with newly discovered species tending to accrete to existing genera which act as‘nuclei’for‘chaining’. The few species which are too anomalous to fit into the large and diverse genera may be placed in monotypic genera, thus producing a pattern of many very small genera and few very big genera. The one exception to this trend is the classification of Bentham which is considerably more even than expected.
The three subfamilies of the Leguminosae show similar overall trends to the family as a whole but exhibit consistent differences between each other in median genus size and equality of genus size. The Caesalpinoideae has even genus sizes and a low median genus size whereas the Mimosoideae shows the opposite, effects too marked to be explained entirely as psychohistorical artefact. These differences may be explained partly as artefact and partly in terms of differing evolutionary history if (1) the Caesalpinoideae is an ancient group primarily of relict species which have undergone little recent speciation, (2) the Mimosoideae are an ancient group which has undergone a massive bloom of recent speciation and (3) the Papilionoideae are a comparatively recently evolved group.