422 CLAUDE FULLER. 



of both the imago and worker castes ; and, as joint IV of the 

 XVIII-jointed form, it is also small. In the XVII-joiuted 

 form the fifth joint is practically the smallest of the series ; 

 and, as this repi-esents the relative large joint VII of the 

 XlX-jointed form, it gives evidence that the shape and size 

 of the joint is influenced by the distance it is removed from 

 the head. 



A somewhat similar state of affairs prevails in the antennal 

 pattern of T. natal en sis, where in the XlX-jointed imago 

 antenna V is narrower and smaller than IV and VI. It is 

 reproduced as joint IV in the XVIII-jointed ; but, in the 

 XVII-jointed expression, it is again V that is smaller than IV 

 or VI. In the XVIII-jointed antennae. III and IV are small 

 and about equal, whilst Vis always larger than VI ; froni this it 

 is inferred that the XVIII-jointed worker antenna is due to the 

 failure of V and A^I of the full expression of XIX to separate, 

 and that the XVII-jointed form of both major and minor 

 soldier is due to the fusion of III with IV of the XVIII-jointed 

 form ; by this means the small joint VII of the imago becomes 

 the small joint V of the soldier. A pseudo XVII-jointed 

 worker-major antenna is due to the fusion of the joints V 

 and VI of the XVIII-jointed form. Upon the other hand, 

 the XVII-jointed antenna of the worker-minor is not due to 

 the failure of joints to separate after being produced, but 

 rather due to the fact that joints have not been produced. 



As it was thought that a series of antennal formulte, such 

 as have been used by workers with coccids, might be useful 

 in fixing species, a number of exact measurements of many 

 antennae of T. natalensis and T. badius were made. 

 Although of a constant general pattern, the variability in the 

 actual length of the joints was such that it Avas rare that 

 more than two formulae in ten coincided. My colleague, 

 Mr. C. K. Brain, of the Union Division of Entomology, having 

 met with similar difficulties with the formulee for coccid 

 antennas, had some time ago devised a means of reflecting the 

 antennal patterns by charts. This he has improved to show 

 the range in the lengths of the individual segments. By 



