IN THE LARV^ OF ANISOPTERID DRAGONFLIES. 167 



than those of any other Cordulme genus known to me. There are 22-24 

 lamella3 in a hemibranch. Delicate purplish-mauve pigment is present as 

 in Hemicordulia. There can be no doubt that the form of the gill-basket in 

 this genus is further definite evidence in favour of its being a highl}^ 

 specialized offshoot of the Eucorduliini. 



(5) Orthetrum (Plate 21. figs. 13, 14). — In this genus we see the latnellse 

 taking on a somewhat more symmetrical shape than is to be noticed in the 

 CorduUince. This tendency towards symmetry in the form of the upper part 

 of the lamella seems, indeed, to be typical of the Libellulince. The lamellar 

 vortex is absent, the tips are wide, well rounded, and rather flat^ so that the 

 bent or " nodding ^' effect is almost wholly lost. The lamellae lie very close 

 together, so that, when a portion of a hemibranch is spread out on a slide, 

 each lamella overlaps not only the one next to it posteriorly, but also a portion 

 of a second one as well. The general effect is one of greatly increased 

 regularity as compared with the type of lamella seen in the CorduUince. 



In Orthetrum caledonicum there are 24—26 lamellse in a hemibranch. The 

 whole gill-basket is of a pure shining white colour, without any trace of 

 pigmentation. 



(5) Diplacodes (text-fig. 12 ; Plate 21. fig. 21). — This genus shows the 

 most advanced development of the Libelluline type of lamella known to me. 

 The general shape of the lamellse, which number from 26 to 30 in a hemi- 

 branch, is the same as in Orthetrum ; but, if anything, the symmetry of the 

 tip is even more marked. A further advance on Orthetrum is that the whole 

 gill-basket is so closely pigmented as to appear almost jet-black when cut 

 open. When examined under a low power this pigment is seen to be of a 

 deep purple colour, and is much denser at the bases of the lamellae than 

 towards their tips. In both Diplacodes and Orthetrum the capillaries a;re 

 excessively fine and numerous. 



Reviewing the above variations of structure in the Lamellate Type of 

 gill-basket, we see that there is a progressive development from the lower to 

 the higher forms, from which Austrocordidia stands out as a separate side- 

 branch, in which an attempt is made to reproduce the papillse so successfully 

 developed in Anax. In some lamellae taken from a larva of Hemicordulia 

 tau, I have noticed the beginnings of a similar attempt at papilla-formation. 

 It seems, therefore, unnecessary to name any sub-types within the Lamellate 

 Type — especially as, when the number of known forms is increased, we 

 may expect to find further intermediate links in the evolutionary chain. 



It remains true, nevertheless, that the gill-basket of the Synthemini is at 

 once distinguished from all the rest by the size of its basal pads and the 

 small number of lamellse in a hemibranch, while, amongst the rest, the oill- 

 baskets of the Libellulince exhibit a general difterence from those of the 

 CorduUince by the greater symmetry of form and the closeness of arrange- 

 ment of their lamellae. Hence, if at any time it is desired to subdivide the 



12* 



