17 



the scales here seems to be parallel with the cases of the Satur- 

 nias and Danaids, but, as in Hepialus there were hairs and scales 

 mixed all over the antenna, this disappearance of the scales leaves 

 a hair-covered and not a naked surface. The Hepialids are certainly 

 on their special stirps a very high family indeed. 



The few scales that remain on the antennae of our European 

 Hepialids suggest to me that they are the (degenerated ?) descend- 

 ants of forms with pectinate antennae. 



I have above noted the disappearance of scales from the dorsal 

 surface of the pectinations in the highest Psychids. 



When we come to the butterflies we find quite different types of 

 antennae to those we have been used to in the moths, just as we 

 find also different types of pupae (and of other organs, structures, etc.). 



There is first, perhaps, the circumstance of the scales being much 

 more numerous than two rows — a condition very rare amongst the 

 moths except in association with pectination. 



Secondly, there is the very strong apical direction of the hairs. 

 In moths the hairs have a ventral, the scales a dorsal, position, 

 usually with tolerable uniformity from base to apex of the antenna, 

 a tendency for the haired surface to be wider apically is not com- 

 mon, and a marked instance of such a tendency is decidedly rare. 

 In butterflies the dorsal and ventral orientation of scales and hairs 

 obtains throughout, but it is equalled and often exceeded l)y the 

 tendency of the hairs to occupy the terminal joints dorsally as well 

 as ventrally, yielding up the ventral surface basally to the scales. 



Dr. Jordan has worked out the morphology of butterfly antennae so 

 fully that I have no further facts to add to his. 



I think his facts rather confirm than throw further doubt on the 

 hypothesis that the Castnias are an early foreshadowing of the butter- 

 flies, and that the Hesperids are not only the lowest family but 

 possibly ancestral to the others. 



The chief difficulty of making the Hesperids ancestral, at least of 

 the Papilionid stirps, appears to me to be not in the dorsal so much as 

 in the ventral scaling in the Hesperids. I see no difficulty in the 

 necessary changes taking place, that is the small apical ventral sen- 

 sory surface might spread in any direction, if natural selection 

 approved, but the apical tendency of this surface is so strong in all 

 the sections of the Rhopalocera, that I very much doubt its return- 

 ing entirely to the base in any instance after securing so apical a 

 position as it has in Hesperids. 



Another remarkable feature of butterflies is the acquisition of 

 depressions or grooves on the ventral aspect of the antennae. This 

 does not occur in Hesperids or Lycasnids ; but in Erycinids a longi- 

 tudinal groove appears in the higher forms on the ventral aspect of 

 the club, and extending up the stalk. 



In the Papilionids a slight tendency of this sort occurs, which is 

 very evident in its derivatives. 



The Parnassininii have a row of rather irregular pits. The Pierina 



3 



