Abstract
Data on twenty‐six species of male Diaspididae, two of them dimorphic, are analysed in relation to the information content and other quantitative attributes of the characters used to construct analysis. It is shown that simple information theory and the use of other numerical criteria enable one to select a relatively small number of characters which effectively reproduce the salient features of referednce classification based on the entire set of available characters. The conclusion are discussed in relation to views on characters weighting and the stability of numeriocal classificatins.