GEOLOGICAL HISTORY OF INSECTS. 357 



A carious and. somewhat unexpected fact is found in the present universal prevalence of 

 membranous front wings in all the orders of Metabola, similar to what is found in the direct 

 paleozoic ancestors of Heterometabola; while most existing Heterometabola, though lower 

 in general organization than the Metabola, have passed beyond this feature of uniformity 

 to one of greater differentiation, the front wings being more or less coriaceous, while the 

 hind wings are still membranous. This, together with the direct relation of some paleozoic 

 insects to later types, would lead us to believe that we are to look at the neuropteroid 

 Palaeodictyoptera as the ancestors not only of later Neuroptera but also of all Metabola, 

 and would account in a measure for the somewhat close relationship of the Phryganidae 

 and lower Lepidoptera. 1 



Allusion has been made to Brongniart's discovery of an insect's wing in the middle 

 Silurian — a long way removed from the upper Devonian, which had hitherto been their 

 lowest known horizon. But though he quickly published a rude figure of his fossil, it is 

 insufficient for critical purposes, and it would probably be hard to obtain from a single 

 discovery the clew we need as to the ancestry of the Palaeodictyoptera. We may safely 

 conclude, however, that the winged Palaeodictyoptera came in as early as the middle 

 Silurian and that up to the close of the paleozoic epoch their divergent stems were still 

 admissible into one general order. 



Now when we look at the insects of later formations, we find types of every one of the 

 existing orders of insects — speaking of these orders in their broadest sense, as we have 

 everywhere done in this essay — we find every one fully developed in the Jurassic period. 



In the Orthoptera we find as good a proof as anywhere, since cockroaches are the oidy 

 insects found in any numbers in the very lowest mesozoic rocks. Their presence in the 

 Trias and its significance will be alluded to later. In the Jurassic rocks nearly forty 

 species are known, of which about one-third are in the lower Jurassic, and nearly all are 

 true Blattariae. So too in the Liassic rocks we recognize all the families of saltatorial 

 Orthoptera and the Forficulariae, so that the Orthoptera may be considered as well 

 established early in mesozoic times. Unfortunately no Phasmida have yet been recovered. 



Only one or two Neuroptera have been recognized in the Trias, but in the Lias we have 

 a considerable number, including Megaloptera, Sialina, Panorpidae, Phryganidae, Ephe- 

 meridae, Termitina and Odonata, showing that the differentiation into the non-existing 

 families was apparently complete early in mesozoic times, and that forms of nearly all 

 recognized families were abundant in the middle and later Oolite. 



The two orders just mentioned are almost the only ones that have yet been recognized 

 in the scanty fauna of the Trias, but the moment we reach the lower Jurassic rocks we 

 find traces of nearly all the others; thus several families both of Hoinoptera, and of Heter- 

 optera are found in Liassic rocks, including such diverse types as the Coreidae, Belosto- 

 midae, (Jieadina and Cicadellina, while Fulgorina and Aphidina are added in the Oolite. 



The (Joleoptera, of which we found only indefinite traces in paleozoic rocks, have been 

 found in the Trias (Chrysomelites), and the adjacent Rhaetic has disclosed forms as differ- 

 ent as Hydrophilit.es, Buprestites and Curculionites, while the Lias already claims some 

 one hundred and twenty-five species referred to as many as seventeen distinct families. 



1 In this connection it would be well to call attention to "synthetic type." See Bost, Journ. Nat. Hist., vn, 590. 

 one of Dr. A. S. Packard's early papers on Neuroptera as a 



