IN THE LARY^: OF ANISOPTEEID DRAGONFLIES. 147 



■sets of cross-folds, whose total area probably does not exceed one-half of the 

 total area of the six main folds, we thus obtain an approximate total of 

 10,800 complete capillary loops in the whole gill-basket. 



In Petalura gigantea, a larva which is many times larger than that of 

 Austrogomplius ochraceus, the gill-tracheoe are not only much shorter, but 

 more numerous. As the only larvse available for examination had been pre- 

 served for four years in alcohol, none of the capillaries were visible, and it 

 was therefore, of course, quite impossible to attempt an estimate of the 

 number of finer capillaries. We should, I think, be well under the estimate 

 in assigning a total of ^0,000 complete loops to this large larva, which even 

 then would be poorly supplied in comparison with the much smaller Austro- 

 gomplius larva, after making due allowance for its much greater size. 



In Cordulegaster, the gill-folds resemble those of Petalura very closely in 

 the thickness and arrangement of the efferent trachea and the firmness of the 

 cuticle. 



The occurrence of the Undulate Ti/pe in Austrogomplius is very remarkable, 

 since all other Gompliince whose larvae have so far been examined have been 

 found to possess the ptapillate. type of gill. The discovery of this important 

 difference, then, suggests that the genus Austrogomplius is not closely related 

 to any of the Palsearctic genera, and is very probably an isolated remnant of 

 an even older Gromphine fauna. 



It should, however, be borne in mind that none of the larvae of the laroer 

 Gompliime of the tribe Ictini have yet been studied in this connection. In 

 these larger, less reduced, and presumably more archaic genera, we may well 

 hazard the supposition that some at least, if not all of them, will be found 

 to possess the more primitive undulate type of gill. 



2. Papillate Type. (Text-fig. 3 ; Plate 18. fig. 2.) 



This type of gill has been found by me only in the larva of Hemiqomphus 

 lieteroditus (text-fig. 3). Dr. Ris has shown it to exist also in the larvse of 

 the Palsearctic Gomphine genera Gomplms and Onycliogomphus. It seems to 

 be characteristic of the main mass of the Gompliince — or, at any rate, of the 

 tribe Gompliini. 



As in the Undulate Type of gill-basket, each of the six main folds is a 

 complete longitudinal eversion of the rectal wall, with supporting cross-folds 

 arranged in the same manner. But, in the present case, these complete 

 eversions of the wall only project a very short distance into the cavity, and 

 carry along their distal border a very conspicuous layer of dark purplish- 

 brown pigment. Below this layer of pigment each gill-trachea is found to 

 split up into a set of five or six branches. Each of these branches pro- 

 jects through the pigment-layer, carrying with it the wall of the rectum, so 

 as to form a long slender papilla of cylindrical form, projecting far into the 



