( 410 ) 



brauclied off, or after the separation of this branch ; the Hesperids may be a very old 

 branch of the phylum, or they may be a relatively youug one, that stands in a some- 

 what similar relation to the Lycaenidae as the Itismorjjhiinae do to the other 

 Pierulae: the antennae do not tell us which assumption is correct. 



However, it is not necessary to assume that the Hesj/eriklae branched oil' after 

 the Butterflies had developed into two phylnms; it is quite intelligible that tliey 

 separated from the ancestral stock before a separation into the two main pliylums 

 had taken ])lace. The presence of scaling on the dorsal side of all the joints in all 

 the species — a character which is equally constant only in families of Moths; the 

 very constantly great development of ventral scaling; the long tapering club of most 

 species — a specialisation which is peculiar to the Hesperiidae among Butterflies, but 

 is found also in some Moths, ('oei/fia for instance ; and especially the absence of all 

 those specialisations by which the L;/caeno-Krycino-Pieridae are distinguished from 

 the Pajjilioni-yi/mphalitlac, suggest that the Ilcxperiidae originated before the two 

 main phylums of Butterflies had separated; but the evidence for this third possi- 

 bility, the separate origin of the skippers, is also entirely inconclusive. 



The connection between the various families thus deduced may be illustrated by 

 the following diagram: — 

 NvMPHAi.inAE. Papii.ionidae. Hesperiidae. Lycaexidae. Krvcinidae. Hikridae. 



The uncertainty as to the jiosition of the Hesperiidae is very suggestive, if we 

 remember that the Hesperids and Lycaenids are so often very similar to each other 

 in their antennae. This similarity consists in both families possessing ancestral 

 characters, which, as said before, find their explanation in the origin of the twd 

 families from the same ancestral stock, and do not imply that the Hesperiidae and 

 Lycaenidae separated relatively lately. As we have seen that, notwithstanding the 

 agreement of the two families in several generalised characters, the Hesperiidae can 

 very well have branched off before the Pajjilioni-yymplinlidae parted from the rest 

 of the Butterflies, we have here an instructive illustration of the fact — so very 

 often entirely disregarded in classiticatory work — that the presence of the same 

 character in two diflerent families (or higher or lower categories, down to in- 

 dividuals), though demonstrating origin of both from a common ancestral form, can 

 be, or is, evidence of closer relationship only, if the character is a specialisation and 

 not of the ancestral type. 



