340 Annals of the South African Museum. 



subfamily is rich in various adaptations to live on the ground, or even 

 buried in sand or mud ; some curious and even grotesque forms exist 

 in tropical Africa. 



The South African representation of the subfamily is comparatively 

 poor ; but it is not improbable that more species remain to be 

 discovered. 



1. Legs short, third femora not reaching beyond the joint between thorax and 



abdomen. Discoidal field in front wings more distally and less widened 

 than in (2), divergence of M 4 and Cuq beginning at the level of nodus 

 or more distally. Between M rx and M^ in hind wing a single cross-vein 

 (Plate VIII, figs. 3,4,5, 6). Lateral margin of abdominal segments 9, or 

 8 and 9, foliate. 



Legs long, third femora reaching to the joint between abdominal seg- 

 ments '2 and 3. Discoidal field in front wings much dila.ted to the 

 border, divergence of M, and Cuq beginning proximally to the level 

 of nodus. -Between MI- S and JV/ 4 in hind wing two cross-veins. No 

 indication of anal loop in hind wing, all branches of A running parallel 

 in about right angles to the anal margin (Plate VIII, fig. 2). Abdomen 

 cylindrical. Lateral margin of terminal abdominal segments not foliate. 

 Superim- appendages of male scarcely longer than tenth segment, simple ; 

 inferior appendage of about equal length, divided in two widely divergent 

 branches . . . .... Podogomphtis. 



2. A distinct anal loop in hind wing, by convergence of A 3 and A 4 at their 



beginning and coincidence of cross-veins (Text-fig. 5, Plate VIII, figs. 



5, 0) 4. 



No distinct anal loop in hind wing, at most some convergence of A s and 

 J 4 at their beginning, but no coincidence of cross-veins (Plate Vlll.figs. 

 3,4). Segments 8 and 9 foliate ....... 3. 



3. 1'terostigma large. Third to seventh abdominal segments comparatively 



robust. Superior appendages of male but little longer than tenth seg- 

 ment, not approximate nor touching each other in the mid-dorsal line ; 

 inferior appendage of about equal length. Structure of female genital 

 segments much as in preceding genus . . . Crenigomplnij. 



Pterostigma moderate. Third to seventh abdominal segments very 

 slender. Superior appendages of male considerably longer than tenth 

 segment, approximate in the mid-dorsal line, so much as to be con- 

 tiguous at least in the basal half or two-thirds ; inferior appendage 

 considerably shorter, deeply divided in two contiguous branches. 

 Female, with vulvar scale broad and comparatively short; on ventral 

 plate of ninth segment a corresponding shallow groove, bordered by a 

 low ridge . . . . . . . - . . . Mesogmnphns. 



4. Anal loop two-celled (Plate VIII, fig. 5). Segments !S and 9 foliate. Male 



with appendages much longer than tenth segment. 



Female with vulvar scale and ninth ventral plate much as in Mesogomphus 



Onychogomphus. 



Anal loop of more than two cells (Plate VIII, fig. G). Only the eighth seg- 

 ment foliate at lateral margin. Male with appendages much shorter 



