BY R. J. TILLYARD. 45 



median gill respectively. The median ^ill is bilaterally symme- 

 trical about the mid-sagittal plane of the larva. The lateral gills 

 are asymmetrical in tliemselves, but the two gills are symmetri- 

 cally placed, to right and left, with respect to the mid-sagittal 

 plane. 



The median yUl is formed from the appendix dov^alis of the 

 larva, Tliis is a median outgrowth from the eleventh abdominal 

 tergite. Tn position, it corresponds with the telson of Crustacea, 

 and may be considered analogous with that organ. It is not, 

 however, a true homologue of the telson, since there is strong 

 evidence that the appendix dorsaUs of the Odonate larva is a 

 more recent development than the ai'chaic cerci. 



The latPi-al (jilh are formed from the two cerci of the larva. 

 These are the original abdominal appendages of the last or 

 eleventh segment, i.e., outgrowths from the l)ipartite sternite of 

 that segment. They are. therefore, the true homologues of the 

 uropods of Crustacea, 



Although the cerci, and the appendix dorsalis (when present), 

 in other insect-larvae, ai-e frequently many-jointed, yet in the 

 Dragonfly larva they are always either unjointed or only two- 

 jointed. Throughout the Sub-Order Anisoptera, where caudal 

 gills are not developed, these three processes are always compara- 

 tively short, hard, and unjointed, forming together a strong anal 

 pi/ramid, which guards the anal opening, and can be opened or 

 closed at will. In the Zygoptera, as we shall see later, the two- 

 jointed forms appear to be the older, and we are able to establish 

 the descent of certain unjointed forms fi'om older two-jointed 

 forms. There are, however, other unjointed forms which show- 

 no evidence of an ancestral two-jointed form. Further, reduc- 

 tions from a two-jointed to an unjointed form appear to have 

 taken place along several separate lines of descent. Hence we 

 cannot divide the gills into two main types according to the 

 number of joints, but must search for a more natural method of 

 classifying them. 



Each gill is attached to a short basal piece, viz., the reduced 

 remnant of the eleventh tergite in the case of the median gill. 



