541 THE PANORPOID COMPLEX, iii.. 



The condition found in the higher Lepidoptera is well illus- 

 trated by Chapman, in Comstock's latest work (15, fig. 21) . The 

 genus figured is Antheraea (family Satumiidae), one of the most 

 highly specialised of all Lepidoptera. In this type, as might 

 have been expected, the alar trunk is complete, and the median 

 trachea has migrated completely across to the cubito-anal group. 



It seems obvious that the archetypic condition of the bases of 

 the main trachea? for the Lepidoptera ought not to be deter- 

 mined by the selection of such a highly specialised type as 

 Antheraea. The llepialidae and Cossidae, at any rate, might 

 reasonably be expected to show a more primitive condition. 

 Moreover, it is not yet really known by what means the con- 

 necting piece between the two main groups of trachea? has been 

 developed, nor by what exact method the media has migrated 

 across it. I therefore paid careful attention to these points, 

 when dissecting pupa? of the more archaic Lepidoptera. 



In the Cossidae, I found, within the single genus Xyleutes, 

 much variation in the position of M. Sometimes it was close 

 to the costo-radial group ; sometimes 'close to the cubito-anal 

 group ; and less frequently it lay about half-way between them, 

 much as in the Planipennia. In all cases, the alar trunk was 

 complete . 



In the Hepialidae (Text-fig. 36 b), I met with a very great 

 surprise. In many pupal wings the alar trunk is not complete; 

 but the two groups of trachea? are brought together by the ap- 

 proximation of two large trachea?, viz., the most posterior of the 

 costo-radial group (this trachea should be M) and the most 

 anterior of the cubito-anal group. These two trachea? pass into 

 the wing alongside one another, and together supply the branches 

 of M. 



Several pupa? of Charagia showed this condition, and I began 

 to think that it might be the normal one for the family. How- 

 ever, further dissections showed that the condition was very 

 variable. In some pupa?, the cubito-anal branch to M was very 

 weakly formed, while the principal supply to M consisted of a 

 huge trachea connected with the costo-radial group. This, I 

 think, is probably the most archaic condition. The tendency 

 towards the splitting-back of trachea? towards their bases is very 

 strong in all Lepidoptera. If then, we grant that the alar trunk 

 was not completed in the ancestor of the Lepidoptera, it is easy 

 to see how, when the tracheal connection from the cubito-anal 



