hairs and scales are everywhere mixed, but the hairs predominate 

 ventrally. There is here, therefore, a distinct beginning of the 

 aggregation of hairs ventrally, and of scales becoming less in propor- 

 tion on that aspect. In /. prce/atfl/a, in the male the scales are more 

 abundant dorsally, sparser ventrally ; the hairs also exist all over, but 

 are less abundant dorsally. In the female the scales practically form 

 two rows or rings, with hairs everywhere sparsely mixed amongst 

 them. This type of antenna is that which is developed in the Tineidae, 

 a family that appears to have originated in the Adelidaa. 



In Lampronia capilella and L. rubiella, which closely follow the 

 Incurvarias in phylogenetic relationship, the antennae are completely 

 encircled with hairs, and the scales have disappeared, except on some 

 eight or ten segments at the base, which carry scales on their dorsal 

 surface. 



This is a type of antenna that Dr. Jordan does not appear to have 

 met with, and he even asserts its non-existence in frenate Hetero- 

 cera, recognising it only in the lycaenid stirps. 



Although I think it is most probable that the lycaenid antenna 

 originated from one of the hesperid type, there is, no doubt, much 

 to be said in favour of its having arisen, just as we have it here in 

 Lampronia, by near relationship to Jugatae. Such a form is clearly 

 more easily attained whilst the structures are in the plastic condition 

 of these earlier stages. Still, when we see how, even in much later 

 stages, the hairs and scales advance into and retreat from each other's 

 territory, there is no insuperable difficulty in the origin from Hes- 

 peridae. Dr. Jordan postulates such an impossibility by assuming 

 the method of origin of antennal scaling that I have already dis- 

 cussed. 



Though it is somewhat out of proper order, I may here refer to the 

 way in which the dorsal scaling of the antennas may disappear in 

 certain of the most highly specialised Heterocera. It is clearly 

 apposite here, when we are considering the possibility of a haired 

 surface existing dorsally apart from direct inheritance from the primi- 

 tive forms. 



The Noctuse, Agaristidae, and Geometrae undoubtedly have 

 antennae closely and widely scaled to the tips as their typical 

 forms ; yet in Anarta. in Alypia. and in Psodos the terminal join^of 

 the antennae have the scaling reduced to the width of one scale, and 

 in Anaria melaleuca and Psodos cordcina there are several terminal 

 joints without scales. It is, no doubt, possible to argue that in my 

 specimens the odd scale has been accidentally removed. I am 

 tolerably well satisfied, by examining several specimens, that this is 

 not so ; but if it is, the reduction of the scaling to a width of one 

 scale is nearly as strong a proof of the scaling retreating as if it had 

 been finally lost. The case of Polyploca ride/is, alluded to later, 

 is, perhaps, a still stronger case in point 



Bodine tells us that the antenna of Eriocrania is wholly scaled. 

 Dr. Jordan says there are no wholly scaled antennae ; he accompanies 



