160 MR. R. J. TILLYARD ON THK RECTAL BREATHING-APPARATUS 



both Anax? and JEschia have gills of exactly similar structure ; and, in 

 Anax, the papillse do not begin to develop until the sixth instar. Yet the 

 young larva of Anax grows very much more rapidly than the young larva 

 of y^sclina. 



It seems, therefore, that other factors must also play a part in the differ- 

 ence between the two larvse. I think that one of these two factors is the 

 earlier appearance of the imagines of Anax, and the consequent hatching of 

 the larva at a time when the organic life of the ponds is at a maximum. 

 The imagines of the various species of ^"Eschia almost all appear late in the 

 summer. Hence the larvse must hatch out in the autumn, or possibly not 

 until the following spring. Another factor which must play some part is the 

 more sluooish character of the ^-Esclina larvse. I have alwavs found the 

 larvse of Anax pajjuensis, in my aquaria, to be far more greedy and rapacious 

 than any other Odonate larvse known to me. The lawn of yEscJ ma hrevistyla, 

 on the other hand, though rapacious enough compared with other Odonate 

 larvse, cannot be compared with that of Anax, either in its alertness in 

 watching for prey or in its capacity for taking food. 



It would seem then that all four causes mentioiied above may play a part 

 in dififerentiating these larvse one from the other. 



3. Lamellate Type. (Text-figs. 11-11 ; Plate 20. figs. 9-12 ; 

 Plate 21. figs. 13, 14, 20, 21 ; Plate 22. figs. 21, 27.) 



This very distinct and beautiful type of gill-structure is found throughout 

 the whole family of the Libellulidse. I have studied it in the genera 

 Syntlieniis, Metathemis, Austrocordulia, Hemicordulia, Cordulephya, Ortlietrum, 

 and Diplacodes. Sadoues's investigations were carried out entirely on the 

 single species Lihellula depressa, while His has studied the gills of European 

 species of the genera Cordtdia, Ortlietrum, Lihellula, and Spnpetrum. 

 Though, of course, many minor variations in structure are bound to occur 

 throughout so large a mass of forms, yet there are no outstanding differences 

 which seem to me to warrant the erection of sub-types. I propose, there- 

 fore, first of all to deal with the general characteristics common to all 

 Lamellate Types of gill, and then to describe the more noticeable differences 

 that appear under the various generic headings. 



When a lamellate gill-basket is everted by a longitudinal cut, it becomes 

 thrown into a very strongly convex form (the internal portion forming the 

 convexity) so as to resemble a Chinese paper lantern partlv opened. On 

 examining this with a low power, six double rows of very regularly arranged 

 flat leaf-like gills or lamellse are seen projecting from its surface. This 

 very neat and orderly arrangement is in very strong contrast to that exhi- 

 bited by any other type of gill-basket, and points to the lamellate type as the 

 most perfect form of gill-basket yet evolved. 



