group, to illustrate these points by figures. I only bring forward 

 this criticism in order to state that our investigations, as far as they 

 have gone, disprove the old groupings on colour, that galactodactyla 

 and pentadactyla art; no more congeneric because they are white than 

 Chrysophanus phlaeas and Aglais nrticce because they are red. I also 

 bring it forward to combat the ignorant nonsense that is talked as to 

 "good" and "bad" genera by the men who still study the wings, their 

 pattern and nervures, and have neither the time nor the training to 

 study the life-histories de novo, who still act as if they believe the early 

 stages are useless for generic purposes, and who carry out the charming 

 simple-mindedness of the old author who remarked that it would never 

 do to use the legs for the purpose of classification, as most of the in- 

 sects received from abroad had had them broken off, and that the early 

 .stages were equally useless because they were not known. Of course, 

 it is true that you cannot use for classification characters that you have 

 not studied, but it is another thing to decry more complete work in 

 order that it shall be considered as no better than superficial material 

 obtained without effort. The whole' essence of classification is to 

 show the relationship of insects to each other, and the question is a 

 simple one — whether one genus for all the butterflies or plumes, or 

 many genera for each of them, will best represent that relationship. 

 In my opinion, if the larval and pupal characters separate widely 

 spilodactyla, galactodactyla, and pentadactyla, I am prepared to take 

 up a strong position against the men who know nothing about the 

 structure of the early stages, and argue that my genus for galacto- 

 dactyla is bad because they cannot see any differences between it 

 and pentadactyla, and will not take the trouble to discover them. 



So much for the things people hold opinions about, usually in 

 inverse ratio to their knowledge of the facts. Now about the facts 

 themselves. Omitting the Agdistids, there is a distinct division in 

 the true plume stirps — /. e., that, leaving out the early stages in the 

 developmental series, when the plumes had assumed forms somewhat 

 similar to those we now know, certain characters began to develop 

 that cut them markedly into two branches. One of these characters 

 is in the imaginal stage. It is this, that, though in the frenulum of 

 the c? there is practically no difference, in the ? , the frenulum splits 

 distinctly into two sections, (i) in which the $ spina is single, (2) in 

 which the 5 spina is double. By itself, it is a character of no more 

 value than any other character, but it supports the general subdivision 

 made by Hofmann on the neuration, it supports the marked characters 

 of the egg, larva and pupa, and one supposes, therefore, that it is 

 broadly of primary value in separating them into two large sections. 

 The "one spina" group corresponds with the Platyptiliids, the "two 

 spina" group corresponds with the Alucitids {sens. rest.). 



It may be well now to note the main features of the larval structure. 

 One suspects that the original Alucitid larva was an internal feeder 

 and that the external feeders are modern developments. I have 

 already stated that there is a distinct difference in the larvae of the 

 two sides of the plume phylum. On the Platyptiliid side, the larva 



