17 



//. Tubercles greatly modified or absent. 

 J. Tubercle I normal (when present). 



A". Tubercles productnl into naked Heshy horns or repre- 

 sented by coloured spots. Papilionuiae, Xyniphalidae 

 (in part). 

 KK. No trace of tubercles. Xymphalidite (in part), Pieridae^ 

 Hesperidae. 

 JJ. Tubercle / consolidated with its fellow on the dorsum. 

 L. No unpaired dorsal tubercle anterior to abdo- 

 minal segment 8. 



M. Tubercles largely present. Saturnia. 

 MM. Only the dorsal tubercle on segment 8. 

 Sphingidae. 

 LL. A line of unpaired dorsal tubercles throughout 

 the length of the abdomen, anterior to segment 

 8f or largely so. Nymphalidae (in part). 



What strikes us in the first place in this list is that several 

 families belong partly to one group, partly to quite a different one. 

 This might indeed be expected in an artificial system like this. 

 In the second place we see that the determinations of many families 

 with the aid of this table will prove to be difficult e.g. the Pieridae^ 

 which are densely covered with setae, and which, though their 

 primitive pattern remains visible for a long time, are said by 

 Dyar to present: „no trace of tubercles". 



The artificiality of such a classification is evident. 



I think that Dyar in this case was under the influence of 

 Comstock's suggestive paper (1893), and that he exerted himself 

 to find a characteristic in the caterpillars, which allowed him to 

 apply Comstock's division of Lepidoptera into Jugatae and Frenatae 

 to the larvae also. 



My third and main objection is that except in group I no 

 attention has been paid to the ontogenesis. Where the setal pattern 

 undergoes rather important modifications during the larval stage 

 and some members of the family remain on a lower scale of 



2 



