( 399 ) 



Bntterflies according to the varions organs and strnctnres of the antennae, we see 

 again and again the Hesperiidae brought together with the Li/cncniflai', and find the 

 IsijrnphaUdae in several instances separated from the rest of the Butterflies, the 

 Papilionuhie linked witii the yi/mph<iliclae, and the Picridae with the Eri/rinidar. 

 Such agreements leave no doubt that there is some truth in ever\- artificial classifi- 

 cation, even if its basis of division is " habitus " ; and it seems very well possible to 

 arrive at a fairly correct grouping by comparinu- a number of artificial classifications, 

 or, as it is generally styled, by liasing a classification on the difiPerences and simi- 

 larities exliibited by the insects in several organs, taking those forms as more 

 closely related which agree in the greater number of characters. 



Although the grouping thus arrived at might ultimately prove to be correct, it 

 is nevertheless artificial, since that evidence is taken as the more weighty, i.e. as 

 liualitatively the better, which is merely numerically, i.e. quantitatively, the higher, 

 and also unsatisfactory, as it does not account for all those many cases in which 

 members of groups standing widely apart in the system have characters in common 

 which other members of these groups lack. A satisfactory insight into the true 

 connection between the members of any group of animals will not be gained, unless 

 the classifier takes as his aim to ascertain, so far as that is possible from the 

 necessarily incomplete knowledge of the organs, the probable phyletic development 

 of each single distinguishing character, so that we get a jiicture of the gradual 

 modification of the various organs from the ancestral stage of develojnnent into 

 those stages of mxitation which we now observe in the dift'erent members of the 

 group to be classified. 



Though at first thought this speculative method of building up a classification 

 appears to be of a very hypothetical nature, it will be seen on closer examination 

 that the method works with less assumptions than that described above, and that, 

 moreover, it does not admit any such assumj)tions to be made without a close 

 inquiry into their admissibOity. The separation of the Pieridae from the other 

 Bntterflies on account of their divided claws is arbitrary, unless it can be shown tliat 

 the divided claw is not a specialisation of any other Butterfly claw comparatively 

 lateh' acquired, and that it was also not a character of the common ancestor of the 

 Butterflies independently lost at different times by the varions branches into which 

 the Butterflies developed, being kept only by a portion (the Pieridae^ of one of the 

 branches. The presence of the Papilionid vein on the forewing of the Papilinnidae 

 (the vein that runs into the hinder margin) will justify attributing to the Papi- 

 lionidae a separate origin from the other families of Butterflies, if there is no 

 homologon of that vein in these other Butterflies (which there is), and if it is also 

 out of the question that the vein has obliterated independently in them. The 

 preservation of six fully developed legs, clearly an ancestral character, in Jlesperiidae, 

 Papilionidue, and Pieridae, is no more an argument for a closer phyletic connection 

 between these families on the one side, and between the other Butterflies with 

 reduced forelegs on the other, than is in Moths the reduction of the rostrum or 

 of the wings, or the absence of a frenulum or of tibial spines, or in Butterflies and 

 Moths the loss of scales, etc., an argument for a close relationship of tlie i'orms in 

 which such a reduction or loss is observed. 



It will be apparent from these illustrations, that there is a wide dilfereuce 

 between the method of building up a classification on assumptions, and the 

 speculative method in which the phylogenetic value of the differential characters 

 has to be investigated before thev can be made use of, and thar onlv tliis latter 



