ABSTRACT
In our day it is becoming almost an established principle that any human function which cannot be performed by a machine is 'subjective' and has no place in science; Sneath and Sokal probably think of themselves as engaged in a gallant attempt to 'save' systematics by giving it a properly scientific basis. The advantages claimed for numerical taxonomy are objectivity and mechanisability. A very interesting test case for numerical taxonomy has been the recent study by Rohlf of larval and adult classificatory characters in mosquitoes of the genus Aedes. Rohlfs own conclusion from the comparison of the two dendrograms was that 'highly significant correlations were found between larval and adult relationships. Despite the correlations, there was also significant heterogeneity between larval and adult relationships'. In addition to criticisms based on fundamental principle, the Sneath-Sokal conception of a 'unit character' will encounter numerous practical difficulties.
