BY R. J. TILLYARD, 415 



alteration of habits. Being unable to compete with the 

 invaders along the LibcUnlid line of development, Cor- 

 dulepliya began a cmiogenetic dtyarture towards zygopterous 

 lines. This was confined to the imago, the larva being appar- 

 ently well able to hold its own. This departure, aiming at 

 reduction in size together with an alteration of habits of 

 flight, necessitated a new line of development in wing-vena- 

 tion. The large triangles and anal area, so necessary to the 

 soaring and skimming flight of Lib'-llulidc', were no longer of 

 value, and, to whatever degree these may have been developed 

 in Cord III e])/ii/a, they had now to be undone and altered to 

 the new requirements. There set in, therefore, the opposite 

 tendencies to those associated with the Lihelhdido' in general. 

 Reduction in size meant particularly reduction in basal 

 areas ; as the hindwing became narrower, the anal loop 

 (whether it was well-formed or not matters little here) became 

 more and more reduced, until we see it represented now in 

 C. pygnio'd by two sti'ong and well-formed cells ; a procession 

 instead of a recession of the hindwing-triangle began, accom- 

 panied by a reduction in actual size and an ascent of the 

 upper cross-vein, re-forming the original "quadrilateral" 

 triangle of the older LihcIUdidce. In all this, C. pyymcea far 

 outran C. montana, as would have been expected. In C. 

 niDiitaiia, we see an intermediate stage, which is very strong 

 evidence in favour of our view of the case. The hindwing- 

 triangle is much irtdrr than in ('. pyijinaa, and much less 

 recessed. It appears that the triangle has been stretched or 

 widened along the wing-length by the gradual narrowing of 

 the basal areas, but that the basal side has not yet reached 

 a position of stable equilibrium. ('. pyymxea has solved the 

 problem by shortening the triangle, and supporting it by a 

 second cubital cross-vein placed well after the arculus. The 

 forewing-triangle of C. niontana is also much larger and 

 wider than that of (' . pyymcea^ and the ascent of the upper 

 cross-vein is less, so that the triangle of the former is more 

 normally shaped. As regards the anal loop, the reduction 



