SPH1NGIDES. 859 



characters, showing in what direction, in the various stages, affinities 

 may be expected to be found. These may be noted as : 



Ovum. — Of the flat type, oval in outline, plump, green in colour, shell 

 transparent, surface smooth. 



Larva. — Tubercles i, ii, iii and iv with simple setae, v atrophied, a super- 

 numerary prespiracular, base of tubercles i on 8th abdominal developed into caudal 

 horn, and carrying tubercles i in 1st stadium (after 1st moult, i generally obsolete). 



Pupa. — Of two forms: (i) Amorphid — rough, with short maxillae; wider 

 antennae. (2) Sphingid — smooth, with long maxillae; narrowed antennae. 



Imago. — Frenulum ; two anal nervures to hindwing ; otherwise of two 

 forms : (1) Amorphid — short tongue, broad wings. (2) Sphingid — with long 

 tongue, narrow, sharp-pointed wings. 



We should, therefore, expect in the primitive Sphingid, at 

 least the following characters : 



Ovum. — Flat type, shell transparent. 



Larva.— First stage (and possibly later), with i, ii, iii, iv and v generalised 

 in structure and position ; tendency to consolidation and enlargement of bases of 

 i on abdominal segment 8. 



Pupa. — With distinct, and probably fairly strong, projecting points on 

 movable segments. 



Imago. — With distinct proboscis ; thickly scaled ; frenulum ; two anal 

 nervures to hindwing-. 



Grote, at first, fell into the error that has proved a stumbling-block 

 to most of his successors in attempting to derive one group from 

 another existent group. The lowest members that we now have 

 of any stirps, are as many generations from the common ancestor 

 as the highest, and, therefore, though they may not have specialised 

 in the same way as the highest, in so many, and what we regard 

 as so important, characters as they have, yet have had as much 

 time to specialise, and very possibly have specialised in some 

 characters quite as much as they. This gives us the result that 

 often puzzles us, that, what we cannot help thinking the lower 

 form, is, from some points of view, much the higher. Specialisation 

 is understood to mean elaboration of organs and functions, and 

 higher and lower forms can only be such, in proportion to the 

 greater or less elaboration of what we consider (often rather 

 arbitrarily) the higher organs and functions. The practical result 

 of this is that every case must be judged on its own merits, and 

 the common ancestor of a superfamily, family, subfamily, tribe or 

 genus, must be built up as possessing all the generalised characters 

 found anywhere in the group, and none of the specialised ones, 

 i.e., specialised as compared with other members of the group. 

 Having obtained this as well as we can, we must work forwards 

 to the present representatives of the group. The result almost invari- 

 ably is that no present forms can be derived from any other present 

 forms, and that nothing very definite exists between any present 

 form and its most remote ancestors, e.g., one may build up a 

 common ancestor to pinastri and lignstri, but neither is that common 

 ancestor, nor is any other existent species. If atropos be added 

 to the group, the ancestor is varied, but is equally nonexistent, 

 and so on. This will make quite clear our disagreement with 

 most of the conclusions of Poulton, Packard and others on 

 this subject, for it must be quite clear that neither Aglia nor 

 Dimorpha, nor any other already suggested existent form, can 

 be a direct ancestor of the Sphingids, for neither of these 

 has a larva which, in the first stage, has i, ii, iii, iv, so 



