22 DONALDSON BODINE. 



From a comparative study of these corresponding forms in the 

 two sexes and of a large series of male antennae, I conclude that the 

 evolution of the pectinations was essentially as follows: A single 

 pair of pectinations originated on the ventro-distal edge of the seg- 

 ment. As development went on there was a steady selection of those, 

 forms in which the pectinations had their origin less toward the ex- 

 treme distal end. The result was that the bases of the pectinations 

 graduallv migrated proximad. In addition to this movement there 

 was also a tendency towards a migration dorsad. This last was a 

 slower and a later method of specialization, but one which more 

 directly resulted in an increase of the sensitive ventral surface. 

 Among the Saturniina we find a further development. Here, with 

 the exception of the Bombycidse proper, we find the pectinations are 

 dorsal, and where they have largely developed and have migrated to 

 the extreme proximal ends of the segments, a second pair have 

 originated as outgrowths from the distal edge of one segment towards 

 the overshadowing pectinations of the proximal edge of the adjacent 

 segment. The earlier form of this is seen in Automerb to, Fig. 8, 

 w^here the distal pair of one segment is so closely apposed to the 

 larger proximal pair of the next, that it is difficult to see the line 

 of junction except with a considerable magnification. In Tropcea 

 luna the distal pair have begun to migrate proximad and have be- 

 come independent of the older and more firmly established ones. In 

 Saniia ceeropia the two })airs of each segment are more nearly equal 

 throughout the clavola, while in Philoifamia cynthia, Fig. 3, the distal 

 pair have migrated so far proximad as to occupy almost the middle 

 of the segment. 



We have seen how natural selection has caused an increase in the 

 area and spread of the sensitive surface. Along with this process 

 there was a diflPerentiation of the hair structures themselves, and 

 doubtless a corres])onding increase in their efiiciency. It is probable 

 that in the primitive form of antenna a diflferentiation had already 

 taken place between hairs for protection and for sense-perception. 

 The simplest form of the latter sort has already been described as a 

 hair of the first type. A comparative study of types two and three 

 will show that, while they are quite distinct from the first and from 

 each other, no great change would be necessary to transform the first 

 into either the second or third. Intermediate forms do not exist in 

 the same antenna, but difierent species show difierent degrees of 

 variation from the first type. 



