ATTACIDES. 269 



link between the true Citheroniids and Attacids, and points out 

 that, as Duponchel and Poulton have shown, the larva wholly 

 discards the congenital characters exhibited in its spinous armature 

 (Citheroniid) at its last moult, and assumes an entirely different 

 shape, being destitute of spines, and taking on the general appearance 

 of an Attacid larva. Side by side with this change in appearance, 

 Packard notes a difference in habit, the larva now feeding passively 

 on its foodplant, and depending upon its colour to escape its enemies, 

 and, if discovered, utilising the Sp/it/ix-like attitude and its large 

 terrifying eye-spot to frighten away any aggressor. 



Smith followed Packard in separating the Hemileucidae as 

 a distinct family from Citheroniidae {Ceratocamfiidae) and Atta- 

 cidae {Saturniidae), but Dyar {Trans. N. Y. Acad. Sci., xiv., 

 p. 55) criticised this conclusion, stating that the characters on 

 which the Hemileucids and Attacids (Saturniids) were separated 

 by Smith, viz., that the antennae of the latter were doubly 

 bipectinate in the $ , and those of the former singly so, was not 

 sound, " many genera presenting a most interesting gradation in this 

 respect, the females being generally behind the males in degree of special- 

 isation," and he further suggests that it is clearly an arbitrary division 

 to draw the line between these families on this character alone, without 

 further evidence that this separation really corresponds to a dichoto- 

 mous division in the line of descent. He points out that if the £ sex 

 had been chosen instead of the $ , the division would have corres- 

 ponded with one based on larval characters and the families could 

 then be described : 



1. Hemileucids: Antennae of ? moth singly bipectinated, of s either 

 singly or doubly so. Larvae with primitive first stage, a dorsal tubercle on 9th 

 abdominal segment and none on anal plate ; tubercle shafts densely covered with 

 sharp defensive spines. 



2. Saturniidae*: Antennae of both sexes doubly bipectinated. Larvae lacking 

 the primitive first stage ; no dorsal tubercle on 9th abdominal segment, but a pair 

 on the anal plate ; tubercle shafts, short or smooth, with few weak spines or hairs. 



Bodine, on antennal characters, shows {Antennae, etc., p. 42) that 

 the Bombycidae are not a constituent part of the superfamily Saturniina 

 (as the superfamily Attacides has been called by some American 

 authors). He places the Attacidae ( Saturniidae) near the Cithero- 

 niidae ( Ceratocampidae) and finds both rather distinct from the 

 Hemileucidae. His grouping is as follows : 



* In the European Aglia tan, the larva possesses all the characters of the Satur- 

 niidae {Attacidae), but differs in the great inequality in development of the tubercles. 

 Their final disappearance in the last stage is less distinctive (compare the American 

 Samia ceanothi). In the imago the male antennae are doubly bipectinated. the 

 ? are serrate (singly). This combination of characters probably entitles the genus 

 to family rank. Packard places (Proc. Am. Phil. Soc, xxxi., p. 140) it as a subfamily 

 of the Ceratoca?nj)idae (= Citheroniidae) ; but this is negatived by the arrangement 

 of the larval tubercles (though favoured by their unequal development) and by the 

 structure of the antennae of the moth (Dyar). Packard says (toe. cit., p. 

 139) : " Aglia appears to be a Ceratocampid in its earlier larval stages, 

 the caterpillar in its final stage, however, and the moth being very near to 

 the Saturnians." Grote considers (Can. Ent., 1895, p. 283) that this quite 

 excludes Aglia from the Citheroniidae, and states that it may be entitled to family 

 rank. Dyar further notes that, in Anisota, i and ii are both present, i unconsolidated ; 

 although earlier, he had erroneously stated i to be absent (Ann. N.Y. Acad. Sci., viii., 

 p. 232). 



