192 MR. R. J. TILLYARD ON THE RECTAL BREATHING-APPARATUS 



Cordulegaster (on its totality of larval characteristics) as the nearest living 

 relative of this huge family. But the gap between Cordulegaster and the 

 Libellulidse is still a large one. This gap is considerably lessened if we 

 compare Cordulegaster with Synthemis, for the gill-basket of the latter agrees 

 with that of the former in possessing only the archaic number of twelve sets 

 of cross-folds, albeit they are wholly disguised as lamellae. The large size 

 also of the basal pads indicates to us a way of approach between the localized 

 pads of Libellulidse and the generalized pads of other Anisoptera. It seems 

 probable, perhaps, that we shall in the end have to admit that the origin of 

 the Libellulidae is quite lost to us in the mists of antiquity, unless some 

 fortunate fossil find in the still unworked fresh-water beds of the Trias-Jura 

 in Australia may yet hold the key to the puzzle. 



Finally, our study emphasizes the homogeneity and closeness of descent of 

 the two subfamilies forming the Libellulidas. The characters separating the 

 Corduliince from the Libellidinoi are everywhere of a lower order of value 

 than those separating other subfamilies. Yet there can be no doubt of their 

 distinctness. It seems, too, that much of the difficulty of separating the two 

 may be due to convergence between their highest members. (Certainly the 

 difference between Synthemis and, say, Libellula is very much more marked 

 than that between the Eucorduliini and Trameini, which are regarded at 

 present as the highest members of the two subfamilies. We may, perhaps, 

 have to alter our views later on as to the position of the Trameini, and assign 

 to the Symjjetrini — those Passerines amongst Odonata — the pride of place in 

 the Libelluline stock. 



In conclusion, if we had to select from amongst all the forms studied 

 the two most highly specialized larvae, we should undoubtedly choose Anax 

 and Diplacodes, standing at the extreme ends of development of their respec- 

 tive branches. To select from these two the most highly specialized of all 

 Anisopterid larvae would be no easy task; but I am inclined to award the 

 palm to Anax, which seems to me to combine in the highest degree all those 

 qualities essential to the carnivorous aquatic larva. 



Bibliography. 



1. Amans, p. 1881. — Recherches anatomiques et physiologiques sur la 



Larve de VJEscIma grandis. Lab. Zool. Fac. Sci. de Montpellier, 

 pp. 5-16, 1 plate. 



2. Chun, C. 1876. — Ueber den Ban, die Entwicklung und physiologische 



Bedeutung der Rectaldrlisen bei den Insekten. Abhandl. d. Senckenb. 

 Naturf. Ges., Bd. 10. 



3. CuviER, G. 1798. — Memoire sur la maniere dont se fait la nutrition des 



insectes. Mem. Soc. Hist. Nat. de Paris, vii. p. 49, plate 4. 



