BY R. J. TILLYARD. 



327 



the teeth are much reduced in size, possessing no serrate edges, 

 and varying in number from two to nearly thirty on each '-fold." 

 This similarity in the dentition of the four folds of C or dnleg aster 

 with two of the folds of Synthemis, helps to convince us fui'ther 

 of the close relationship between the two genera. The develop- 

 ment of the original Anisopterid stock had evidently proceeded 

 but a little way when the progenitors of Synthemis and Cordule- 

 gaster branched off from it; and, most certainly, in their larval 

 forms, further divergence has been very limited, though in the 

 imagines the process of evolution has effected considerable 

 separation. 



The lines of divergence of the two main grou[)S of Anisoptera, 

 viz., the Libelhdidce and jEsch^iidce, can be roughly indicated by 

 a phylogenetic diagram (text-fig. 3). In this diagram the five sub- 

 families at present dominant, representing the five lines of highly 

 successful specialisation of one kind or another, are arranged at 

 the top, while the remnants of the archaic groups (which we may 



LlBELLULiaC 



/ESCHNID/E 



/UBELLULiN.t 



CORDuuMN/E 



(s.srr.) 



. 



4 



Petal ura 



Common Anisopterid Stock 

 Fig.S — Phylogenetic Diagram. 



well believe have been, in their time, also the dominant groups) 

 are arranged lower down. Convergence of origin-lines indicates 

 convergence of the corresponding groups; e.g., I have placed the 



