258 E. H. Sellards — Types of Permian Insects. 



the earlier members to be grouped with the parent or associated 

 phyla, while the later stand in our classification under distinc- 

 tive names. The line selected as the point at which to break 

 the various insect phyla was the imaginary line dividing Pale- 

 ozoic and Mesozoic — a line subject to change with the advance 

 of stratigraphy and paleontology and in any case confessedly a 

 line of convenience. A more natural classification is attained 

 by recognizing under a single head an entire phylum. Diffi- 

 culties will be met in applying the principle, owing to the near 

 approach of the first ancestor of a phylum to the parent phylum 

 (unless the origin of phyla should prove to be much more sud- 

 den than has been heretofore generally supposed). In practice 

 it is clearly safe to recognize a phylum under a common head 

 as far toward its point of origin as the evidence available per- 

 mits the line to be with certainty determined. Thus, with the 

 dragon-flies, there is apparently no doubt of the origin of the 

 phylum as a distinct line in Carboniferous or pre-Carbonifer- 

 ous time and continuing to the present time. They are there- 

 fore entitled to a group name distinctive of the phylum as a 

 whole. 



Regarding the relative rank to be assigned the Protodonata, 

 whether ordinal as proposed by Handlirsch* or subordinal, I 

 would urge again what has been shown above, namely, that 

 the main veins of the wings of the Coal Measures and Permian 

 dragon-flies present an arrangement in agreement in their 

 major characters with that of both Mesozoic and modern 

 dragon-flies. The differences found are due to the specializa- 

 tion of cross veins, with which are associated minor changes in 

 the direction of some of the main veins. These secondary differ- 

 ences are to my mind insufficient characters on which to base 

 ordinal rank. The characters which are now known to exist 

 entitle the Protodonata in my view to not more than sub- 

 ordinal rank. According to this view the order Odonata con- 

 sists of three suborders, as follows : Protodonata, Zygoptera, 

 and Anisoptera. The term Odonata is thus still available as 

 an ordinal term, covering the Odonate phylum as a whole. 



* Eevision of American Paleozoic Insects, Proc. U. S. Nat. Museum, vol. 

 xxix, p. 689, 1906. 



