SPHINGID^E. 497 



of such pupae and larvae respectively as they have been able to 

 accumulate, and the former furnishes the following important 

 contribution to the phylogeny of the group. He writes (in litt.): 

 " In dealing with the phylogeny of the Sphingidae, it is necessary 

 to review the whole group, and it is impossible to escape the division 

 into Amorphidae (Smerinthids) and Sphingidae, which is pointed out 

 by so many differences, and has been recognised, apparently easily 

 by so many observers. But here at once we have to reconcile this 

 position with the fact to which Bacot calls our attention, and which is a 

 fundamental one. He shows us that the Eumorphids in the earliest 

 larval stage possess discrete thoracic tubercles, whilst all the other 

 divisions have them conjoined ; unless, therefore, we postulate a 

 diphyletic origin for the group, for which there is really no other good 

 reason, we must believe that the primitive Sphingid had such a larva, 

 i.e., with the thoracic tubercles discrete and not conjoined. This 

 position is not affected by any question as to the homologies of these 

 tubercles and what they are to be called. This was the character of 

 the larva of the primitive Sphingid, but only in its first stages. The 

 adult Sphingid larva may have been much the same as that we now 

 find in any of the groups, I am not now concerned to guess which. 

 This character taken alone places the Enmorphinae alone as primitive, 

 and throws together the Sphingi/iae, Amorphidae, Sesiinae (Macroglos- 

 sinae) and Hemarinae, into one group (anted, p. 366). The pupal facts 

 flatly contradict this; they show that the Sphingids, as distinguished not 

 only from the Amorphids but from all other lepidoptera, have a 

 specialised structure, which one cannot believe to have been 

 diphyletic without the very clearest and most cogent evidence. Can 

 anything of this kind be said about the larval tubercles ? The larval 

 tubercles are often of very persistent type throughout large groups, 

 and their evidence is not to be regarded as otherwise than very 

 strong ; still this change from a separate to a conjoined condition is 

 one that is very frequent, and occurs over and over again 

 throughout the lepidoptera ; that it should occur twice over instead 

 of only once within one particular family is perhaps a little unusual, 

 but still more or less in the ordinary course of things. The 

 primitive Sphingids then had discrete thoracic tubercles in the 

 first larval stage, but they had not the pupa of the Sphi?igidae. 

 It was a pupa more like that of Smerinthus but not necessarily the 

 same. Did the proboscis of this pupa reach to the end of the 

 wings ? If we assume that it did not, but followed the type of 

 Amorpha, then we may believe that the difficulties of again separating 

 the wings and getting the proboscis to their tips, led to the attempt 

 to accommodate the proboscis by throwing the head backwards. 

 If we assume that it did, then the throwing back of the head was 

 a variation, the precise reason of which is not clear, any more 

 than we know why precisely similar requirements led to a pouch 

 being formed at the end of the wings in Plusia and Cticullia. 

 Something of this sort had to be done, Plusia did it in one way, 

 Sphinx did it in another, why, we do not know, but if this be so, 

 one ought to have some hope of meeting with a pupa, essentially 

 Amorphid, but with the proboscis to the end of the wings, not 

 as a special modification of either the Amorphid or Sphingid pupa, 

 but as a reminiscence of the ancestor of both, and without 



2 H 



