Ill] THE WINGS 57 



to M 4 not far from its point of departure from M 1 _ 3 , the latter 

 being closer up to R. Hence, separation of the sectors of the 

 arculus (i.e. M t _ s and M 4 ) is an archaic character. Union of these 

 sectors occurs in many Libellulidae, and is evidence of high speciali- 

 zation in this region. Further, the sectors primitively arose from 

 the upper portion of the arculus. Movement of their points of 

 origin downwards towards Cu, as in Hemicordulia (Plate II, fig. B), 

 is a sign of specialization. 



The essential conditions (fig. 20 a) for the formation of the 

 peculiar discoidal cell in the Odonata appear to have been two, 

 (a) the presence of the arculus, (6) a bifurcation of Cu at some point 

 distad from the level of the arculus. The second condition is 

 absent in the Protodonata ; though Cu is generally very beautifully 

 waved, it is always unbranched. The connection between the un- 

 branched Cu of these insects and the once-branched Cu of the true 

 Odonata seems to be entirely missing. If we grant, however, that 

 conditions (a) and (6) must be fulfilled before a true discoidal cell 

 can be formed, it is quite easy to see how it was formed under those 

 conditions. The development of the cross-vein above and proxi- 

 mally nearest to the bifurcation of Cu into a strong support would 

 at once complete the closing-in of a large quadrilateral area with 

 thickened sides, viz. M above, Cu below, the lower part of the 

 arculus proximally, and this new support distally (fig. 20 6). This 

 is essentially the original discoidal cell of the Odonata. Further, 

 since the main veins M and Cu both arose fairly close together 

 from the narrow wing-base, and diverged as they traversed the 

 wing distad, we may fairly assume that this quadrilateral was not 

 a rectangular one, but of irregular form, with the distal side some- 

 what greater than the basal side. The evidence of palaeontology 

 supports this view, for neither the regular quadrilateral of modern 

 Calopterygidae, nor the extremely sharply distally pointed quadri- 

 lateral of the Lestinae and Agrioninae, is found in the Jurassic or 

 Miocene Zygoptera (see fig. 164). These two forms, indeed, repre- 

 sent two divergent lines of specialization from the older type 

 (fig. 20 c' to e', and c' to/"). 



As the cross-veins of all ancient forms of Odonata were abun- 

 dant, it is clear that originally the discoidal cell enclosed a series 

 of parallel cross- veins between its two ends. This was clearly a 



