16 



The existence of pectinations seems always to disturb the regular 

 arrangement of the scale in two rows to a segment in all Obtectse 

 that I have examined, the dorsal surface probably partaking of 

 the tendency to increased developments. In Drepaua {lacertinarid) 

 and in Notodonts this tendency seems to be best resisted, but the 

 tvvo rows in these cases are not smooth and regular as in the 

 most typical Nodua antenna. 



The Brephidae have antennae that show the effect of a day- 

 flying habit in the strong tenJency to a greater development of 

 hair-surface apically. 



The first thirteen joints are c )nipletely scaled, the ventral scales 

 inclining centrally and overlapping. About the middle of the 

 antenna, the gradual opening of these ventrally has proceeded till 

 the nearly normal type of obtect antenna is reached, namely, the 

 dorsal half scaled, the ventral haired. The haired surface con- 

 tinues to encroach on the scales, till there are only three or four 

 scales to a segment on the four last joints, and the terminal one 

 is scaleless. 



The Cymatophoridie have for the most part ordinary obtect 

 ante mae, with very smooth scaling and very short regular hairs, 

 but P. ridens presents most remarkable antenna;. The dorsal 

 scaling rapidly narrows, till at a quarter of the length of the antenna. 

 From the base there are two segments with only one scale each, and 

 the remaining three-quarters of the antenna are without scales and 

 densely clothed throughout with fine hairs. 



This instance seems conclusive of the fact that the haired surface 

 may invade the scaled one, and produce an antenna even more 

 extreme than that of any Lvccena, and yet have no immediate con- 

 nection with the primitive Lepidopteron. Ridt7is is certainly a 

 peculiar form, but I have never heard any doubt thrown on its 

 being a typical Cymatophorid, and its antenna must be derived 

 from the ordinary obtect form which is characteristic of the 

 other species. The cases we have considered show that scales 

 can retreat before the advance of the sense-hairs to the same 

 surface ; but the scales can also disappear apart from any advance 

 of hairs, leaving the surface naked. This takes place only in most 

 specialised forms. The two most notable instances are the Satur- 

 niadae at the summit of the bombycid stem, and the Danaidae 

 near if not at the summit of the nymphalid stem. We see in 

 the case of the butterflies some steps in the process of the dis- 

 appearance of the scales, as in some Cha'-axes {jasii/s, etc.) the scales 

 still exist, but shrivelled to minute specks stranded at remote 

 distances from each other. Dr. Jordan quotes one or two instances 

 in Arctia, sp.. The case of the Hepialidse seems parallel to these. 

 In our own humiili and sundry other species there is a row of one 

 or two scales extending along half the antenna, the rest of the 

 antenna being hair-covered In other species with pectinated 

 antennae there is a larger area of scales. The disappearance of 



