THE SOLDIERLESS TERMITES OF AFRICA 



29 



different aspects of behaviour and environment their variation is unlikely to be 

 closely linked, except by a general size factor. However, the correlations between 

 corresponding characters of imagos and workers were sometimes lower than might 

 be expected on the assumption that they are adapted to the same functions. The 

 inter-caste correlations, expressed as percentages, are tabulated for the common 

 characters of imago and worker as follows: 



Head width ........ 94- 1 5 



Pronotum width ....... 87-38 



Hind tibia length . . . . . . .9184 



Postclypeus width ....... 90-46 



Postclypeus length . . . . . . . 7417 



Left mandible, apical to first marginal tooth . . 92-23 



Left mandible, first to third marginal .... 88-47 



Left mandible, third marginal to molar . . . 76-09 



Right mandible, apical to fust marginal . . . 91 -14 



Right mandible, first to second marginal . . . 87-15 



Right mandible, second marginal to molar . . . 80-18 



This table would seem to show that while the lengths of apical teeth of the 



PRINCIPAL COMPONENT WEIGHTINGS 

 eigenvector 



15 



CANONICAL VARIATE WEIGHTINGS 

 eigenvector 



coefficient levels 



CHARACTERS MEASURED 



Width of head 



Pronolum, width 



Fore tibia. greatest width 



Fore tibia, length 



Hind tibia, length 



Postclypeus. width 



Postclypeus, Length 



Left mandible- apical to l»t marginal 



Left mandible- 1st to 3rd marginal 



Left mandible- 3rd marginal to molar 

 \ TTll)Ki^ht mandible- apical to 1st marginal 

 ^^g^[l2) Right mandible- 1st to 2nd marginal 



J(T3)Right mandible- 2nd marginal to molar fj3) 



2-5-4 ! 4- 0-5-5; 



Fig. 15. Worker caste-weightings of measured characters obtained from two multivariate 

 analyses as in 14. In the principal component analysis the weighting coefficient on 

 character 3 in eigenvector 1 was about half the size of the rest. 



