PHRYXIDI. 143 



of the theories of the phylogenesis of different groups which are 

 from time to time afforded us. It is quite natural to find T. 

 porcellus to be imaginally the lowest, and larvally the highest, of the 

 group. Being imaginally the lowest, it cannot be descended from, 

 say, H. celei'io or /. alecto, though the larval evidence says it 

 might be ; nor can I. alecto be descended from T. porcellus though 

 the imaginal evidence says it might be. The common ancestor 

 must be one as low larvally as /. alecto, as low imaginally as T. 

 porcellus. Since leaving that common ancestor, which may have 

 been more, but could not have been less, primitive than the one 

 we have just outlined, T. porcellus has used the interval in specia- 

 lising its larva, and in specialising its pupa for locomotion, but, in 

 other respects, has not much advanced its pupa and has left the 

 imago, altered it may be, but if so, very little, in the directions 

 which we regard as advance and specialisation in the Sphingides. 

 Such species again as Pergesa acteus or I. alecto have advanced to a 

 high point as pupae and imagines, but have retained the early larval 

 arrangement of a full series of spots. There is sufficient indication 

 in both Sphingidae and Amorphidae of this row of spots to suggest 

 that it arose amongst the primitive Sphingids before the first 

 definite division into the subfamilies we now recognise. The 

 relationship between these Eumorphid larvae would appear to be 

 something near to the following rough tabulation : 



A. A primitive larva as in Darapsa myron, plain subdorsal line, no retractile 

 front segments beyond what nearly all larvae, and especially Sphingid larvae, have 

 more or less (Primitive Eumorphids). 



B. A larva precisely like A, but with primitive eye-spots along all the seg- 

 ments, not quite so advanced as Weismann's " open ring-spot," fig. 50A. 



C. A larva like B, but with ring-spots rather more advanced than 50A, more 

 like those of Thaumas vespertilio (Weismann, fig. 14) (Primitive Eumorphine). 



From A would also descend the primitive Elegantes (Hiibner, antea, 

 vol. iii., pp. 350-35 1) including Philampelus, Dupo, Pachylia, Dilophonotus, 

 and probably Nephele; from a larval point of view 7 , even the more defin- 

 itely Sesiine forms would descend from A. There is, indeed, no very 

 strong line by which to divide Sesiids from Eumorphids. The difficulty 

 here is very great to define Philampelus from Daphnis, Daphnis from 

 Argeus and Argeus from a typical Eumorphid (porcellus, &c). It 

 seems necessary to assume that somewhere between A and C, these 

 (and allied genera) separated before the stirps became definitely Eumor- 

 phine. It follows from this that Darapsa myron, being really Eumor-' 

 phid, must have progressed with the Eumorphines as far as C, or 

 even on the Eumorphid side, and then reverted (as regards its larva) 

 to the more primitive form so far as eye-spots are concerned, but 

 not quite reaching the really primitive Sesiine larva, some of which, 

 however, are also specialised in various Elegantine (Philampelid) direc- 

 tions. From C we find two branches, the Phryxids and Eumorphids : 



1. More or less preserving all the spots, using them as parts of a warning 

 coloration, and not developing retractile segments. 



2. Developing retractile segments, and with them specialising the anterior 

 spots as eye-spots. 



On this special Eumorphid branch the eye-spots became 

 confined to the first 7 abdominal segments, and probably reached 

 first a form now best seen in Isoples alecto, viz., all the spots nearly 

 equal in size and importance. This larva seems also to be less 



