IN THE LAKV^ OF ANISOPTERID DRAGONFLIES. -x 159 



view o£ parts o£ two hemibranchs from a somewhat younger larva (7th 

 instar). The drawing in text-fig. 9 is taken from a photograph of the 

 everted gill-basket of a full-grown larva, part of which is reproduced in 

 Plate 21. fig. 19. 



Number of Capillaoy Loops in the Gill-basket. 



For the larva studied by him (probably Anax Junius) Oustalet calculated 

 the number of capillaries at 120,000, allowing only 5 loops to each papilla, 

 100 papill?e to each hump^ and 20 humps in a hemibranch. In Anax 

 papuensis there are only 12 humps in a hemibranch, but the average number 

 of loops in each papilla appears to be six rather than five. Hence the 

 approximate total for A. papuensis is 6 x 100 x 12 x 12, or 86,400 complete 

 loops. Oustalet''s larva was larger than the one studied by me — hence his 

 total may be well within the mark for that species. Anax papuensis is one 

 of the smallest species of the genus, so that we cannot be wrong in stating 

 the average number of loops in larvse of this genus at over 100,000. 

 Probably an investigation of the larva of one of the largest species, such as 

 A. gibbosulus, would yield a very much higher total. 



Origin of the Papillo-foliate Sub-type. 



It is difficult to indicate precisely the reason ior the caenogenetic develop- 

 ment of the papillae on a form of gill already so highly specialized as the 

 Foliate Type. In studying the larvse of Anax papuensis and JEsclma brevi- 

 styla, it seemed to me that the reason might be sought for primarily in the 

 difference of habitat. The ySscJma larva lives in slowly-running water, 

 while that of Anax inhabits stagnant pools. Hence it seemed probable that 

 the poorer aeration of the water breathed by the A7iax larva might have 

 called forth, in the process of evolution, a correspondingly more efficacious 

 type of gill. An objection to this view is that the larvse of those Palsearctic 

 species of /Esclina which inhabit stagnant water have not developed papillae. 

 But it seems probable that the larvse oi the genus ^schna all originally 

 inhabited running water. Hence, if the habit of breeding in still water, in 

 the case of these Palsearctic species, is of very recent requirement compared 

 with the same habit in Anax larvfc, my supposition may well be correct. 



Dr. Pis has suggested to me (in litteris) that the papillae are correlated 

 with quickness of development. The larvse of all species of Anax reach 

 maturity in one year, while those of yEsclina nearly always take two or 

 three years. Dr. Pis thinks that the rapid growth of the Anax larvse may 

 be principally due to the increased oxygenating power of the respiratory 

 system, due to the presence of the papillae. This explanation seems to me to 

 be a good one. But it cannot account entirely for the difference in speed of 

 growth. For, as will be shown later (p. 179), the newly-hatched larvae of 



