OF PSEUDOCOCCIDAE & ERIOCOCCIDAE 



79 



and Eriococcinae. The description of the males of Kermococcus quercus (L.) by 

 Borchsenius (i960) prompted Giliomee to assume that Kermococcidae are more 

 closely related to Coccidae than to any other Coccoidea, on the basis of 6 characters 

 shared by the two families. Further investigations on Kermococcidae are required 

 before Giliomee's assumption can be supported since it was based on one species 

 only, and since the present study revealed that some of the characters he used are 

 also shared by Pseudococcidae (separate pre- and postocular ridges) and Eriococcidae 

 (separate ocular ridges ; short and transverse scutellum). 



ANALYSIS 



On the suggestion of Dr. K. Boratynski, advantage was taken of the computing 

 facilities (IBM 7090 computer) available at Imperial College, and the data obtained 

 in this study (138 characters for 29 species) were subjected to the more elaborate 

 statistical analysis by some methods of numerical taxonomy. In their book on the 

 subject, Sokal and Sneath (1963, p. VIII) define the numerical taxonomy as "the 

 evaluation by numerical methods of the affinity or similarity between taxonomic 

 units and the employment of these affinities in erecting a hierarchic order of taxa " ; 

 it " aims to develop methods which are objective and repeatable both in evaluation 

 of taxonomic affinity and in the erection of taxa ". They pointed out that assess- 

 ment of the degrees of affinity of the taxa (expressed as a coefficient of similarity) 

 should be based on a large number of characters (at least 60) to assure objectivity 

 and repeatability. All these characters should be treated as of equal taxonomic 

 value and the equivalence of all the characters is one of the basic assertions of 

 numerical taxonomy. The relationships between taxa thus calculated is " phene- 

 tic ", i.e. based on overall resemblance and is free of subjective phylogenetic 

 speculation. But the authors believe that numerical methods may open up a wide 

 field in the exact measurement of evolutionary rates and may provide a more 

 critical approach to phylogenetic problems (pp. VIII, 57 and chapter 8). 



Method 



The method employed in the main part of this thesis contains also an element of 

 objectivity in the fact that all the characters were treated as being equivalent ; 

 the assessment of the degree of similarity was based on simple counting of the 

 characters shared by each pair of species. But this method does not take into 

 consideration which of the characters are shared by various species ; moreover, a 

 certain amount of subjective selection of characters was used in the grouping of 

 taxa. For example, the separation of Planococcus and Pseudococcus groups of 

 genera was based on one exclusive character, i.e. the absence or presence of fleshy 

 setae on the main parts of the body. 



The comparison of these results with those obtained by more accurate calculation 

 of coefficient of similarity between each pair of species, was thought to be of interest. 



It should be pointed out that the application of numerical taxonomy is still in 

 an experimental stage, and the exact value of the various suggested methods for 



