BY R. J. TILLYARD. 693 



and showed how remarkable it was that these two genera should 

 possess exactly the same type of branching of all the veins, in 

 both fore and hind wings. At first sight, one is tempted to 

 postulate "a Megalopterous origin for the Triehoptera, and to 

 say that Rhyacophila, and therefore the Order Triehoptera, must 

 have originated from a type very close to Austrosialis. But a 

 little consideration will show us that this would be a very serious 

 misinterpretation of the actual facts of the case. The Sialidae 

 ace the most highly specialised of the aquatic Megaloptera, and 

 are undoubtedly derived from ancestors that were more like the 

 Corydalidae. Austrosialis is much more highly specialised than 

 Rhyacophila in the very long and complete fusion between Cui 

 and M]_4 ; the consequent loss of all traces of the cubito-mediui 

 Y-vein; the removal distad of the secondary radial fork, and 

 consequent reduction in the length of the branches of Ks; and 

 the distal fusion of Sc with R 1# Rhyacophila, on the other hand, 

 is specialised in quite a different direction, in having the anal 

 veins of the forewing completely looped up into a double Y- 

 vein. There are, of course, many other differences, but those 

 mentioned will suffice for our purposes. Complete proof or 

 disproof of any suggested origin of the Triehoptera from the 

 Sialidae could only be supplied by a study of many other charac- 

 ters besides the venation; and the possibility of such a descent 

 will again be discussed in the part of this research which has 

 to deal with larval characters. For the present, we must put 

 it aside, and see only in the resemblances between Austrosialis 

 and Rhyacophila a very striking instance of parallel develop- 

 ment, between two widely separated types which, as we shall 

 sec later on, have nevertheless had a common origin in a more 

 remote period. 



If we compare Austrosialis (Text-fig. 47a) with Sialis (Text- 

 fig. 104), we can see in the latter the preservation of the old 

 CorydaJid tendency to pectination of R2+3. For, by comparing 

 the radial sector of this latter genus w T ith that of Archichauliodes, 

 we see at once that suppression of the original terminal forking 

 of R 2 , with a shifting of the tips of R2+3 somewhat costalwards 

 along the wing-margin, would result at once in the form found 

 in sialis. Moreover, the extra branch of R 2 is also frequently 

 present in specimens of the genus Sialis (though not in the 

 example figured), but is usually found to have migrated across 

 to the base of R 3 ; so that the type of pectination found in this 



