( 400 ) 



method will yield results satisfactory to the inqniring mind of tlip naturalist. 

 The difficulty in the speculative method is, that one has to account for all the 

 similarities and differences in the various organs of the groups classified, and the 

 danger in it is, that only sach characters are brought forward which suit the purpose 

 best, while the others are left alone. In the case of u classification of the Butterflies 

 based upon the characters of the antennae that difficulty does not seem to me to be 

 insurmountable, and I hoj)e to avoid that danger altogether. 



The scaling is a special feature of Lepidoptera. Scales are modified hairs. 

 Only in Lejndoptera do we find the antennae clothed with scales; in all other 

 insects they have a covering of hairs and bristles which are mostly sensory in 

 function. In the nearest allies of Lepidoptera, in Trichoptera, the antennae are 

 furnished all over with a covering of sensory hairs. The scaled Le])idopterous 

 antennae are, therefore, doubtless derived from a not-scaled hairy ancestral type. 



The development of the sense-hairs of the antennae into scales may have taken 

 place together with the appearance of scaling on other parts of the body, or the 

 antennal scaling may have been acquired after the scaling of other organs had become 

 a relatively constant character of Lepidoptera. If the first alternative is correct, we 

 should exjject that intermediate stages between the hairs and scales as regards form 

 would be met with in such Lepidoptera where mutation is obviously in progress, the 

 scaling either varying in extent individually, or being different in extent in closely 

 allied species. However, the excess in the extent of scaling of one individual over 

 the other, or of one species over its close ally, consists of scales like those of the 

 rest of the scaled area; this new, or more recently acquired, scaling does not show 

 any greater similarity to hairs thau does the phyletically older scaling. On the 

 other hand, this sharp demarcation between scaled and hairy areas of the antennae, 

 and the abrupt appearance of fully developed scales in certain individuals, or species, 

 in excess over the scaling of other individuals, or species, are decidedly in iiivour of 

 the second alternative. For, if the sensory hairs of the Lepidojiterous antenna began 

 partly to lose their sensory character in the course of evolution after the non-sensory 

 hairs of other organs had become modified into scales, the physiological forces 

 which produce in each Lepidopteron, instead of hairs, the scales, would have free 

 play also with these non-sensory Lairs of the antennae; the non-development of the 

 nerve-cell of the antennal hair in pupal life would directly lead to the development 

 of a full-sized scale. This suggestion, that the ancestral scale-winged insects had 

 hairy antennae, is fully borne out by what wc know about the extent of the covering 

 of tine sense-hairs in certain Lepidoptera. 



We have seen above that the area covered with fine sense-hairs is in some 

 species much smaller than in closely allied forms, part of the hairy area being in the 

 latter species occupied by scaling. As the presence of scales brings always with it 

 absence of sense-hairs from that place, it is obvious that the hairs have become 

 modified into scales. The fine sense-hairs, which we have so often mentioned in the 

 descriptions of the Butterfly antennae, represent, therefore, a type of hair from which 

 the antennal scaling of Lepidoptera may have been derived. The most generalised 

 tyj)e of antenna covered with hairs would be one in which the hairs were equally 

 distributed over the whole surface, and such a type we find preserved in certain 

 Ilejjialidae (in Ilepialn.s kumuU, for instance). If we consider further that this 

 type cannot have developed from a scaled antenna, because such a derivation would 

 necessitate the assumption that the sensory function which was lost when the hairs 



