96 PALEONTOLOGY 



indicate that they must have been a special offshoot from the 

 Pahrodictyoptera. In the genus Brodia Send. Tillyard (1928b) 

 claims that we have a type ancestral to the Protodonata and 

 Odonata and that these two orders are collateral developments 

 from the Megasecoptera (Fig. 46). Great stress has been laid 

 upon the disco\ cry of the primitive true dragon-fly Kemiedya 

 Till., and its bearing on the evolution of the Odonata. This 

 theory of Tillyard involves the derivation of the broad-winged 

 Anisopterous forms from narrow-winged Protozygoptera which 

 has been disputed by Carpenter (1931). As this writer points 

 out, Tillyard's grounds for his conclusions rest on a slender basis, 

 and, in view of more recent discoveries, his theory will have to be 

 abandoned. The discovery of the ancient sub-order Protanisoptera, 



a 



Fig. 47. \Ving of Ditaxineura anomalostigma Till. 

 (From Carpenter.) 



represented by Ditaxineura (Fig. 47) from the Lower Permian, 

 shows that this type extends equally far back in geological time 

 as the Protozygoptera, represented by Kennedya. Without 

 going into a long and complex argument it may be affirmed that 

 no group ancestral to these two ancient sub-orders of dragon-flies 

 has yet been discovered. In Kennedya Cu.^ + lA combine to 

 form a long stalk traversing the petiole of the wing, whereas in 

 Ditaxineura there is no petiole and these two veins remain separate. 

 It would seem clear, therefore, that on this feature alone the 

 last-named genus is the more primitive. Tillyard's belief that the 

 Protozygoptera were derived from the Megasecoptera has very 

 slight evidence for its support. Carpenter sees no close affinity 

 between the last-named order and either the Protanisoptera or 

 the Protodonata. In our present state of knowledge we liave to 



