( 403 ) 



a totally scaled underside : tliat never occurs ; there is always a distal area beset 

 with sensory hairs left not-scaled. Nor have I found the ventral scaling in any 

 species in a state of obliteration, a process which would lead to the scaled j)ortion of 

 the ventral surface becoming naked, as we have seen on the upperside ; but the 

 sensory hairs are very often fewer in number at the a])ices of the joints and at the 

 sides. Hence the extended ventral scaling means always higher specialisation than 

 the less extended scaling. 



The restriction of the npt-scaled ventral area to the distal joints of the antenna 

 is not a feature common to all Lejjidoptera that have extended ventral scaling ; for 

 we have seen that in the Moths with extremely long and thin antennae {(.'himahacke, 

 for instance) the apical portion is all scaled. But in Lepidoptera witli clubbed 

 antennae the sense-hair-bearing area, if restricted, is always apical, and this is not 

 only the case in Butterflies, but also in those Moths in which the clubbed antenna has 

 an extended development of ventral scaling (among (.'asfitiidae, !Sesiidfie). The 

 restriction of the sense-hair-bearing area to the apex of the antenna is, therefore, 

 dependent on the development of a club, and hence must be a character acquired 

 subsequently to the modification of the originally filiform into clubbed antennae. If 

 this is so, then we can satisfactorily explain why members of different families agree 

 in the extent of the antennal scaling. The cause of the similarity in sjiecialisatiou 

 is not direct relationship, but development in a definite direction which is the same 

 in the not nearly related Butterflies because the antennae are all clubbed. That 

 the restricted sense-hair-bearing area is always ventral, not dorsal, is also very 

 intelligible, as the current of the air, which is to be analysed by the sensory hairs, 

 strikes the ventral surface of the antenna of the flying insect. The facts that in 

 the Moths with very long antennae the not-scaled area is restricted to the basal half, 

 and in the clubbed antennae of Butterflies and certain Moths to the distal joints, do 

 not contradict each other, as it is in both cases the portion that protrudes mostly 

 frontad, and hence receives the current of the air first, which bears the sensory 

 organs, the long antennae floating backwards during flight. 



If we now api)ly these conclusions to the various families of Butterflies with a 

 view of finding the phyletic connections between these insects, wo shall arrive at 

 some interesting results. 



The nesperiuhie are, on account of the not-scaled area being restricted to the 

 club in all species, the most liighly specialised family (as to antennal scaling). As 

 not a single species has the not-scaled area extending down the stalk, all the 

 members of the family can be derived from an ancestral Hesperid in which the whole 

 antenna except the ventral surface of the club was scaled. But considering that the 

 scaling both of upper and under surface develojis in a definite, basi-ajiical, direction, 

 which must necessarily lead to the same result, as said above, it is also possible that 

 the various groups of Ilrsjicriidin' diverged from one another in other characters 

 before that large amount of scaling had been acquired. >So much, however, is 

 certain that the llesperiidae originated from a form in which the development of 

 dorsal and ventral scaling was in jnogress, else it would not be intelligible why all 

 the species are so renuirkably specialised in the same way in this resj)ect ; antl as 

 the proximal ventral scaling is a later development than the clnb, the lli'spcriidue 

 must have sprung from Lepidoptera with clubbed antennae of which the dorsal 

 aud ventral surfaces were scaled to a ])robably considerable extent, the rest of the 

 organ being covered with sense-hairs. 



The ancestors of Li)caenidae, Eryciiudae, aud Picridac mnst have had the 



28 



