84 
it looks, indeed, like a separate segment, especially on the ventral 
surface, where the wing-margins reach to the true margins of the 
segment but leave the additional portion uncovered. On the dorsal 
line it is constricted, showing the arrangement characteristic of 
Nymphalids for checking all but lateral movements, though, as a 
matter of fact, no movement at all exists in this case. This incision 
seems to open slightly, in some species, on dehiscence. A similar but 
less obvious exposure of intersegmental membrane occurs along the 
hind margins of the four following segments. Kemarkable as this 
structure is in this group, and associated as it is with complete loss of 
movement, it is to be noted that the same advancement of interseg¬ 
mental membrane to a permanent and fixed surface position is quite 
plain, when looked for, even in Vanessa and still more in Pyrameis; 
in fact it is, as a constant but not always prominent structural detail, 
a feature of all butterfly pupae except those of the Lycsenids (and 
Hesperids) (figs. 1-2 Pajrilio). In the Pierids the marginal portion 
sometimes looks more like a distinct subsegment than like a portion of 
elaborated intersegmental membrane, and I am certainly not prepared 
to be dogmatic as to its true nature. It comes, however, as a strong 
support to the idea that the Lycasnids separated from the primitive 
Papilionid as soon as (or almost before) the latter had evolved from 
the Hesperids. 
Some species among the Brassolids ( Dynastor , for example) have 
the wing-cases expanded laterally like some Papilionids. 
When we come to the Lycasnids, we find that the pupas best 
known to us agree with those of the Papilionids in their mode of 
suspension, but in little else. They differ in being entirely solid, ver}^ 
rounded, and squat; in having the head curled under so as to be 
ventral rather than anterior in position ; in not possessing the interseg¬ 
mental subsegment; and in having no definite spines, horns, or 
processes, but in possessing hairs and bristles. The Lycaenidae 
are usually divided into Lycaeninae and Lemoniinae : I am inclined, 
however, to think that we should substitute the name Lycaenida 
for Lycaenidae and constitute two families Lycaenidae and Lemoniidae. 
The latter should then be divided into sub-families. The mass of 
the Lemoniidae, in a character to be presently alluded to, are higher 
than the Lycaenidae; and yet it is amongst the former sub-family 
(as at present constituted) that we must look to find the forms 
nearest to the primeval, that is, forms with both the 5th and 6th 
abdominal segments movable. Some approach to this form I 
happen to have seen in Easelasia from South America (figs. 24-25-26). 
There we have a pupa unmistakeably Ljmaenid in the ventral position 
of the head, in the squat broad outline, and in the liair-like character 
of the spines, which latter, however, exist only on the ordinary 
tubercles and not, as in the Lycaenidae, on the general surface also. 
The second legs reach the eyes as in the Lemoniidae, but both the 5tli 
and 6tli abdominal segments preserve movement. By this latter 
feature, as well as by the spinous nature of the hairs, the pupa shows a 
much nearer approach to the primary Papilionid form than does the 
solid pupa of the Lycaenidae (figs. 27-28). There is no trace of 
a marginal intersegmental subsegment. 
In the Lemoniidae, if we only had the material (apparently 
abundant enough in S. America), we might expect to find at least 
