In this group also the subdorsal line is retained in a very variable degree. In some species it appears to have completely vanished (_C. Acteus_, _Celerio_); in others it is present as a light stripe extending along all the segments (_C. Alecto_); whilst in others it is retained as a broad white stripe, which extends only to the fourth segment (_C. Tersa_, Fig. 35). In species possessing eye-spots, the subdorsal line is thus a very variable character. It is, however, an interesting fact that even in the present group, which has made the greatest step forward, the subdorsal line is of general occurrence, because the eye-spots in all these species may have almost a similar development to those of _Elpenor_ and _Porcellus_. The ontogeny of the tropical species would alone give a definite reply on this point, but unfortunately we are not acquainted with any of the young forms, so that we can but presume that some of them at least would show only in the first stage the simple subdorsal line without eye-spots; that in the second stage the primary pairs of eye-spots would be formed on the fourth and fifth segments, whilst the transference of these spots to the remaining segments would take place in the last stage.
The foregoing assumption is based immediately on the ontogeny of _Elpenor_ and _Porcellus_; it is supported by the considerable size attained by the eye-spots in many species of the third group, and would receive additional confirmation by observations on the Indian _C. Celerio_, supposing that Horsfield’s statements do not arise from a confusion of species. This skilful observer, who was the first to breed systematically a large number of tropical larvæ, has given a figure of the Indian caterpillar of _C. Celerio_, according to which this species possesses eye-spots on all the segments from the fourth to the tenth. The European form of this same species has eye-spots only on segments four and five, a fact which does not appear to have been known to Horsfield, as no mention of it is made in his notice of the Indian species. If the caterpillar figured is really that of _Celerio_, which I consider to be by no means improbable, not only is it thus shown that in the species of the third group the ocelli on the hind segments have a secondary origin through a repetition of the primary ones of the front segments, but we can also establish that the same species in two different regions may arrive at two different phyletic stages.