BY n. J. TILLYARD. 423 



under certain circumstances, be made to function as orifices of 

 inspiration at any time that the larva chooses to leave the water. 

 But we are concerned here only with the problem of aquatic 

 respiration, as carried on, day in and day out, throughout the 

 whole life of the growing larva. The admitted facts that ce7'tain 

 Dragonfly larvse do occasionally crawl out of the water and 

 breathe air directly through their spiracles, and that probably 

 all Dragonfly larvse also use their spiracles in this way for a 

 short time just prior to metamorphosis, do not affect the general 

 question, and cannot be used as an argument against the con- 

 clusions to be arrived at. 



I have chosen to restrict the problem in two directions. 

 Firstly, I shall deal only with the Suborder Anisoptera, i.e., those 

 dragonflies whose larvse breathe by means of recto! gills, and 

 shall not consider the somewhat different problem presented by 

 the Zygoptera, whose larvse breathe by means of caudal gills. 

 This is done because our knowledge of the morphology of these 

 latter organs is still in a very backward state, and does not as 

 yet offer a sufficiently sound basis on which to build any general 

 physiological conclusions. The morphology of the rectal gills of 

 Anisoptera, on the other hand, is by now fairly well known, so 

 that we are able to approach the physiological problem with 

 some hope of success. Secondly, I do not intend to offer an ex- 

 planation of the working of the whole respiratory system, includ- 

 ing a detailed discussion of the elimination of nitrogen and 

 carbonic acid gas from the body. The problem before us is to 

 find a satisfactory explanation of the method of obtaining the 

 oxygen, necessary for the metabolism of the larva, from the 

 water, and its distribution to all parts of the body. 



It will not be necessary to explain in detail the differences 

 between the various types of rectal gills. I have already dealt 

 very fully with these in a previous paper(8), to which I would 

 refer anyone desirous of studying the comparative morphology 

 of these beautiful and interesting structures. It will be sufficient 

 here to state that the different forms can be classified primarily 

 into two main divisions, the Simplex System and the Duplex 

 System. In the former, the rectal wall is thrown into six main 



