Carpenter — Relationships between Classes of Arlhropoda. 323 



of Insects ; and this speculation is tempting to those who seek to derive 

 Insects, independently of other Arthropods, from A nnolidan ancestors. 

 But the view will not bear examination. Brauer ('69), Lubbock ('74), 

 ;ind, more recently, MiaU ('95), and Packard ('98), have shown con- 

 clusively that the active, campodeiform, armoured larva characteristic 

 of the lower orders of Insects must have preceded, in the evolu- 

 tion of insect-metamorphosis, the worm-like cruciform larva charac- 

 teristic of the more specialised orders. Not only is this evident 

 from a study of the various orders, but from comparison between the 

 families of any one order. Among the Lepidoptera, for example, we 

 find that the caterpillar of a low- type moth, like Hepialus, has, in 

 addition to a chitinous tergite on the first thoracic segment, paired 

 tergal plates on the second and third segments, and in some species on 

 the abdominal segments also, while the legs are strong and relatively 

 long, recalling those of a beetle-larva. But the caterpillar of a high- 

 type moth — a Sphinx, for example — has no distinct tergal plates on 

 any body-segment, excepting the first thoracic, those of the other 

 segments being reduced to tubercles, while its legs are relatively 

 shorter and weaker than those of the Hepialus caterpillar. Thus we 

 see that the worm-like characters of the larva are most markedly 

 shown by the higher moths. Among the beetles a complete transition 

 from the campodeifonn to the cruciform type of larva can be traced ; 

 while the fact that, in the life-history of certain genera, the former 

 type precedes the latter in the development of the individual, shows 

 conclusively that the active armoured grub preceded the worm-like 

 caterpillar or maggot, which is undoubtedly a specialized secondary 

 larval form. We may, therefore, safely accept the conclusion that 

 the primitive insects were thysanuriform. 



But in connexion with the object of this essay it is of the greatest 

 importance to arrive at a correct view as to the segmentation and 

 appendages of the primitive insects. To this we are guided partly by 

 morphological and partly by embryological evidence. Taking, first of 

 all, the head, at least six limb-bearing segments, all primitively postoral 

 in position, can now with certainty be recognised. Foremost of these 

 is the antennal segment bearing the feelers, innervated by the deuto- 

 cerebral ganglia. Then comes the tritocerebral segment with evanes- 

 cent appendages clearly detected by Wheeler ('93) in the embryo of 

 Anurida, and by Uzel ('97) in that of Campodea. In the latter insect, 

 indeed, these appendages persist as paired tubercles in the adult. The 

 next postoral segment of the head is that which bears the mandibles. 



The next segment has only recently been clearly demonstrated ; 



