IN THE LARV^ OF ANISOPTERID DRAGONFLIES. 



191 



or provide us with another new type. In passing, it may be remarked that, 

 if the larva of the Petaliini can be shown to possess gills of the Duplex 

 System, that fact will practically decide once and for all the claim of this 

 group — already a strong one — to be included in the true j^sclmince. The 

 study of the Implicate Type as compared with the Foliate Type confirms the 

 view already held that the Bracliytronini {Implicate Type) are more primitive 

 than the ^UscJinini, though they have branched away soraev\^hat from the 

 line of ascent of the latter. The study of the Foliate Types marks out 

 yUscJma and Anaa; as naturally closely allied, and also confirms the view 

 already held that Anax as now constituted was formed by rapid csenogenetic 

 specialization from the older .^5c/ma-stock. 



The most valuable phylogenetic evidence afforded by our study is un- 

 doubtedly that which concerns the Libellulid stock, whose origin is still 



PrimiCivt UnduJatc Type 



Text-fig. 21. — Phylogenetic Diagram. 



a matter of doubt. The form of the gill-basket points at once to the two 

 facts that they are not only a very highly specialized side-branch from the 

 main line of Anisopterid advance (which we may take to be represented by 

 the upgrowth of the jFiSclinince from Undulate-Typed ancestors), but that the 

 origin of this highly successful stock goes very far back, almost to the very 

 root-beginnings of Anisopterid history. By no other supposition than this 

 can we explain the complete suppression of the underlying Undulate Type 

 even as early as the third or fourth instar in the Libellulid larva. Further 

 light may be thrown on this by studying the gill-basket in the newly-hatched 

 Libellulid larva ; we may even have to examine its formation within the 

 embryo. Even within the evidence before us, we are able to single out 



