LEPIDOPTERA PHAL^N^J OF THE WHOLE WORLD. 27 



imaginal and pupal states do not stand greatly in the way. For 

 a Cossid or Psychid larva to be derived from a Zygsenid one 

 seems impossible. As positions that are not impossible, but 

 very unlikely, the collocation of Sphingids and Cymatophorids 

 seems open to question. We should remove the Cymatophorids 

 to the Geometrician, and both nearer the Pyralidae. The rest of 

 the flat-egged families on the Notodontid stirps should be brought 

 to the Limacodid stem. 



However, it is easier to criticise than to construct ; and no 

 phylogeny of Lepidoptera yet suggested has quite pleased anyone 

 but the author, whom we may guess to be often as dissatisfied 

 as anyone else. The materials are not yet nearly abundant 

 enough. The ' Key ' is a much better one than in the ' Moths 

 of India,' being founded for the most part on significant 

 characters, or rather these are taken more in order of their 

 importance, giving a more natural result. The families placed 

 in the ' Key,' under A. b. b 1 . have no greater claim to be other 

 than subfamilies of Tineidae than many that are not differentiated, 

 except perhaps the Pyralidae. They differ by superficial characters, 

 chiefly the good old one of size. The table and key still show 

 traces of fa* fault to which all systematists are prone, and, from 

 having had the field so much to themselves, neurationists per- 

 haps more than others. That is, the assumption that a similar 

 neuration (or whatever character the systematist affects) means 

 alliance, a different one divergence, not merely usually and gene- 

 rally, as is indeed truly the case, but practically always and 

 without exception. Where other structures dispute this position, 

 they are prepared to say that these other structures may differ 

 widely in allied forms, or may have reached a practically iden- 

 tical specialisation in very distinct forms ; no doubt a correct 

 proposition, but they appear to be unable to apply it to the 

 character they have adopted. This is no doubt largely in- 

 evitable, as a specialist does not know what value to attach to 

 structures that he has not studied in detail. We recognise in 

 the table that its author is still largely a neurationist, but not 

 absolutely so, or altogether unable to use his wide knowledge to 

 modify neurational conclusions. 



T. A. C. 



d2 



