( 409 ) 



not-grooved L_vca«iid auteuiiae, leaves no doubt, that the three-grooved antennae 

 of Dismorphiinae represents a further specialisation of the one-grooved type, not a 

 generalisation. There are no intergradations between the Pierine and Dismorphiinc 

 antennae known to me; the two groups of Butterflies stand in this resjiect more 

 widely apart than the Pierinae and Erycinidae, and one might, therefore, be 

 justified in giving the Dismorphiinne family rank. But as there is an obvious 

 gradation among the Dismorphiinae from a t3'pe with three separate grooves (f. 30) 

 to a type with the grooves joined to each other at tlie apical edge of the joint (f. 31), 

 the earlier form of tlie three-grooved antennae had probably one transverse apical 

 groove, widened in the middle, the mesial groove of Enjcinidae-Pierinae pushed 

 distad. The apical position of the grooves stands perhaps in connection with the 

 great develojiment of ventral scaling, as both IHsmorpliiiiKie and the N;/inph(ili<lae 

 which have the grooves in a subapical position on each joint (Jiithima asterope, 

 f 58) possess extended ventral scaling. The absence of an indication of transition 

 from the Pierine to the Dismorphiine antenna among Pierinae, as well as the great 

 constancy in the appearance of three grooves in T>ismorjihiinae, all the species of 

 which subfamily have this specialised character, show that the IHsmorphiinae are 

 not a further specialisation of any branch of the recent Pierinae, but must have 

 diverged at an early time. 



The relationship of the Pierid with the Eryciuid antenna is here demonstrated 

 quite independently of the above conclusion that the Papilionidae and Ntjmphalidae 

 are one phylum. It is of the greatest importance to note that we bring the above 

 two families in a second phylum of Butterflies with odd-grooved antennae, not 

 because they do not belong to the Papilioni-Nj/mphalidae, but because their 

 antennae show a sjiecialisation peculiar to them, and that we consider the Lycaenidae 

 to belong to the same jihylum on account of the close agreement of the Lycaenid 

 antenna with that Erycinid type in which the development of the grooves is 

 incipient. The independence of the arguments for a connection between Xi/mphalidae 

 and Papilionidae on the one side, of those which speak for a relationshiii between 

 Lt/caenidae, Kri/cinidae, and Pieridae on the other side, lends additional strength 

 to them. 



Of quite a different t3'pe from the fine sense-hairs and setiferous punctures are 

 the " sense-bristles " found on both the scaled and not-scaled areas of the antennae. 

 Their development is independent of that of those other sensory hairs, and hence 

 the evidence they ofter in respect to the relationship of the tamilies of Butterflies is of 

 great weight. As the number of bristles is always very limited in Lepidoptera, the 

 bristles never forming a covering of the joint, as the fine sense-hairs do, the most 

 generalised antenna, in which all the sides are similarly developed, should Iiave the 

 bristles arranged in belts running round the joints. And, indeed, wo find such an 

 arrangement among Juyatae, the belt being, however, often disturbed. Can a 

 regular belt be the character that obtained in the ancestor of Lepidoptera ? That 

 the bristles of Moths and Butterflies can be derived from a single row is certainly 

 an argument not speaking against the Hepialid arrangement representing that of 

 the ancestral type ; but much more convincing than this argument are the facts 

 that we find a nearly regular postmedian belt of bristles among Lycaenidae, that 

 many other Butterflies have on the not-scaled ventral surfoce of the club the bristles 

 also arranged in one transverse series, and that among Heterocera the ventral 

 surface has a similar row of Ijristles, for instance in Coo/tia, while in many other 



