444 NEW AND RARE AUSTRALIAN AGRrONID^, 



of the legion. When these fail, recourse must be had to non- 

 venational characters, but it is advisable to use unisexual charac- 

 ters as little as possible. It is a great pity that de Selys divided 

 his "grand-genre" Ayrion into two main divisions depending 

 purely on a single female character (the presence or al)sence of a 

 spine at the end of segment 8, underneath); for there are, and 

 probably always will be, a large number of rare species whose 

 females are not known, that must remain incertce seclis in such a 

 classification. I, therefore, propose that, for generic subdivision 

 in this legion, the following order of preference should be ob- 

 served : — (1 )Constant venational difference. (•2)Constant bisexual 

 characters, not venational. (3)Constant unisexual characters, 

 only to be used when (1) and (2) fail, and precedence to be given 

 to unisexual vial>' characters, owing to the comparative rarity of 

 the females. 



I now propose to study these in detail, in oi-der to obtain a 

 satisfactory classification at least of the Austi'alian species of the 

 legion. 



[.Venational differences. — The following are of value for generic 

 subdivision :■ — a. The amount of petiolation of the wing. h. The 

 form and length of the inferior sector of the triangle. 



Taking a and h together, they form a most valuable character 

 for subdivision. Those forms are the least asthenogenetic in 

 which the basal postcostal nervule is closest to the base of the 

 wing, the petiolation consequently least marked, and the inferior 

 sector of the triangle longest and best developed. Such forms, 

 while marking the " high-water mark " of successful cajnogenetic 

 specialisation in the legion, must nevertheless be considered to be 

 more in line with the ancestral stem of the group than are their 

 asthenogenetic offshoots. We may indicate four main stages of 

 development of the combined characters a and h, as follows : — 



{\.)Argia and allies (not Australian). Wings very slightly 

 petioled, the petiolation ceasing rvell before the basal postcostal 

 nervule; inferior sector of triangle reaching back basally just 

 beyond this nervule to the point at which petiolation ceases. 



(ii.) Agrion, Ischnura and their allies. Wings a little more 

 petioled (to about level of first antenodal or beyond it); inferior 



