BY R. J. TILLYARD. 545 



group grew out towards the base of the true median trachea 

 (still united with the eosto-radial group), it might have become 

 connected, not with the single base of this trachea, but with the 

 posterior half only, split back almost to its very base. This con- 

 nection, once established, would give rise to the condition de- 

 scribed above, in which M is supplied apparently by two strong 

 trachea?, one from each group. 



Further dissections of pupae of Hepialidae also revealed a 

 number of instances in which the alar trunk was complete, al- 

 though its middle portion is never strongly formed. In these 

 cases, M was sometimes still closely connected with the costo- 

 radial group, sometimes just as closely connected with the cubito- 

 anal group. 



The conclusion we must come to from this evidence is that, in 

 the ancestor of the Lepidoptera, the alar trunk was not fully 

 formed as a completed arch at the base of the wing. There was 

 probably present a small tracheal outgrowth from the cubito-anal 

 group, which, owing to the already strongly developed tendency 

 of the main trachea? to split back to their bases, did not always 

 make the same connection with M; and so the various formations 

 met with at the bases of Hepialid pupal wings must have been 

 arrived at. 



In so far as they have attained a moderate degree of specialisa- 

 tion, the alar trunk being complete, and the median trachea 

 placed about half-way between the costo-radial and cubito-anal 

 groups, both the Megaloptera and Planipennia must be con- 

 sidered as more specialised in this respect than the Hepialidae, 

 and. therefore, than the archetype of the Lepidoptera. But the 

 lines of evolution of these two Orders and of the Lepidoptera, for 

 this character, are not the same. The tendency towards the 

 splitting back of the trachea, which is in itself a specialisation 

 in a different direction, is only developed to any great extent 

 in the Lepidoptera; and, as we have seen, it was probably this 

 line of specialisation that prevented the complete formation of 

 the alar trunk in the earliest Lepidoptera. We might note, in 

 this connection, that, in the Jugo-frenata (27), in which this 

 splitting is not so strongly developed as in most Lepidoptera, the 

 alar trunk is completed, though its middle portion is of small 

 calibre. 



Turning next to the Orders with merotracheate wings, we may 

 take first of all the Diptera, in the older members of which 



