374 Annals of the South African 



II D. FAMILY LIBELLULIDAE. 

 II DA. SUBFAMILY CORDULINAE. 



As shown in the table (p. 388), it is not possible to give a short 

 definition of the two subfamilies of Libellnlid<.n> without giving at 

 least one unisexual character (the angulate hind wings of male in 

 Cordulinae), and even this character is not universal, as it is wanting 

 in the rather important genus HenticurdnHn (like Anax in the 

 Aeschninae, and also like Anax a highly specialised type). The only 

 bisexual character of Cordulinae, the temporal projection of the eyes, 

 is rather inconspicuous and also suffers exceptions, since a few 

 LibelluUnae (of the group Urothemis) show similar projections. Other 

 characters of Cordulinae are also unisexual : semitranspareut narrow 

 lamellae at interior margin of tibiae in males (strictly proper to 

 Cordulinae if confronted with LibeUulinae alone, but present also 

 in ChlorogompJiinae) ; strongly developed and differentiated terminal 

 appendages in males (a character of only relative value, more for the 

 Cordulia- than for the Macromia-line of the subfamily) ; unbraiiched 

 hamuli in genitalia of male (regular for all the Cordulinae, but 

 existing also in not a few LibeUulinae) ; a pair of anterior hamuli 

 visibly developed, against the anterior lamina of same organs, seems 

 more characteristic for the Macromia-line, though existing in rudiments 

 also in the Cordulia-\me. The bisexual character of predominant 

 metallic colours is a very conspicuous one, though not fully dis- 

 tinctive : there are some Cordnlinae with no metallic colours, and not 

 a few LibeUulinae with highly metallic ground-colour or pattern. 

 With all these exceptions and restrictions the two subfamilies are 

 well justified, and an expert will scarcely ever have a moment's doubt 

 about the correct placing of any given specimen. 



Cordulinae are much less homogeneous than LibeUulinae ; evidently 

 some very ancient developmental lines have been retained and exist 

 in a few, now isolated, representatives. Various attempts to a rational 

 classification have been recently made ; the best of them (according to 

 the writer's opinion) is by Mr. E. B. Williamson. Without regard 

 to minor and partly intermediate groups two main branches may be 

 distinguished : the Macromia-line with Aeschnid (more especially 

 Chlorogomphine) affinities in general build and also to some extent in 

 veuational characters, and the CorduHa-line which approaches 

 LibeUulinae in both these characters. 



In Africa only the Macromia-lme is fairly represented, and only a 

 few members of this line are met with in our present faunal limits. 



