Most interesting in this respect is, perhaps, the Asiatic-African genus Elymnias, a Satyrid whose numerous (over thirty) species all seem to be in need of protection, for many of them mimic immune butterflies, while the rest are inconspicuous and are provided with protective colouring on the under surface. On Plates II and III some of the former are depicted beside their models. The single African species (Elymnias phegea) ([Pl. II], Fig. 20) mimics, as has been already mentioned, the prevalent Acræa gea ([Pl. II], Fig. 21). Many of the Asiatic Elymniidæ are mimics of the immune Euplœæ, especially the dark-brown species with steel-blue shimmer, such as E. patna in India, E. beza in Borneo, and E. penanga in Borneo. In Amboina there flies an E. vitellia, the female of which mimics accurately the plain, light-brown, inconspicuous Euplœa climena which occurs there. The male of Elymnias leucocyma ([Pl. III], Fig. 26) resembles the brown and blue shimmering Euplœa binotata ([Pl. III], Fig. 25), while the female mimics the dusky, radially-striped female of Euplœa midamus ([Pl. III], Figs. 27 and 28): the male of Elymnias cassiphone resembles the blackish-brown and deep-blue iridescent Euplœa claudia, while the female is like the female of Euplœa midamus. A number of species of Elymnias copy Danaids: thus both sexes of E. lais are like Danais vulgaris ([Pl. III], Figs. 29 and 30), and E. ceryx and E. timandra are like another similar Danaid, D. tytia. The female only of E. undularis of Ceylon mimics the brown-yellow D. genutia ([Pl. II], Fig. 22) in general appearance, though not minutely, while the male ([Pl. II], Fig. 24) seems to attempt an imitation of the blue Euplœæ. A rare form, not often represented in collections, Elymnias künstleri, bears a striking resemblance to the Danaid, Ideopsis daos Boisd., with its white wings spotted with black, while three species mimic the probably immune Pierid genus Delias, especially on the under surface, which is decorated with yellow and red. Perhaps the one which has diverged farthest from the original type is Elymnias agondas Boisd. ([Pl. II], Fig. 32) of the Papua region and the island of Waigeu, for it bears two large blue eye-spots on the posterior wings, and thus, especially in the case of the almost white female, closely resembles Tenaris bioculatus ([Pl. III], Fig. 31). There are thus seven or eight types of marking and colouring differing from one another, and belonging to six different genera and a much greater number of species, which are mimicked by this one genus Elymnias.

It is most interesting to note how these mimetic species give up, more or less, the original sympathetic colouring of the under surface, and use in establishing their mimicry the marking elements which were originally directed towards concealment. According to the beautiful observations of Erich Haase on this genus Elymnias, the ground-colouring on the under surface must have been 'a grey, darkly mottled protective one,' as still occurs, for instance, in several mimetic species, such as Elymnias lais ([Pl. III], Fig. 30). This leaf-colouring disappears more and more the more perfect the mimicry of the model becomes, until, finally, the model is repeated on the under surface also. Compare, for instance, Figs. 30 and 32. From this we may conclude that a dress which makes Lepidoptera appear unpalatable morsels is a more effective protection than resemblance to a leaf. That might indeed be deduced even from the theory, for resemblance to a leaf never protects absolutely, and does so, in any case, only during rest, while apparent unpalatability repels assailants at all times.

Those unversed in butterfly lore usually ask, when these mimetic relations are expounded to them, how we know that copies which are so like their models really belong to a different genus, or even family. There are certainly cases in which model and copy resemble each other so closely that even a zoologist cannot tell one from the other without close examination, as, for instance, in the case of certain transparent-winged Heliconiidæ of Brazil (Ithomiides) and their mimics belonging to the family of Whites. But even in such cases the likeness only extends as far as is theoretically requisite, that is, only to those characters that make the butterfly appear to the eye of its pursuer like another species, known to it to be unpalatable. The likeness does not extend to details, which can only be seen with a magnifying-glass or a microscope, and above all, it does not extend to the caterpillar, pupa, or egg. Thus, in the case cited, we may be certain that the caterpillar of Ithomia is quite different from that of the mimicking White, since the former will be, in structure, of the type of Ithomia caterpillar, and the other of the usual type of Whites. As yet, indeed, these two species are not known in their caterpillar stages, but other cases are known. A species belonging to the same genus as our indigenous 'kingfishers' (Limenitis populi), a diurnal butterfly of North America, Limenitis archippus ([Pl. I], Fig. 9), strongly resembles the brown-yellow, immune Danais erippus ([Pl. I], Fig. 8), while the caterpillars of both species are quite different, that of Danais erippus possessing the remarkable, soft and flexible horn-like processes of the Danaid caterpillars ([Pl. I], Fig. 10a), while the caterpillar of Limenitis archippus ([Pl. I], Fig. 11a) is at once recognizable by its blunt, club-shaped and spinose papillæ as a Limenitis caterpillar. The adaptation of the butterfly to its protected model has thus exercised no influence upon the caterpillar. Nor has it affected the pupa, which in both cases exhibits the very different and quite characteristic form of the Danais pupa and the Limenitis pupa respectively ([Pl. I], Fig. 10b, and 11b).

But even in the butterfly itself nothing is altered, except what increases the resemblance to the model. All else has remained unchanged, above all, the venation of the wings. Since the painstaking and valuable work of Herrich-Schäfer the venation has been made the basis of the whole systematic arrangement of butterflies, and it enables us, in point of fact, to distinguish with precision, not the families alone, but often even the genera, for the course of the veins in the different species of a single genus is the same, and that is true for the mimetic species as well as for others. Thus the Danaid-like Limenitis has the usual Limenitis venation, of the kind seen in our own indigenous species of Limenitis, and the already described Elymnias species of the African and Indian forests and grassy plains have always the venation characteristic of this genus, whether they be protected only by sympathetic colouring or imitate an immune Euplœa, a Danais, an Acræa, or a Tenaris. However much the contour of the wing may vary, the venation is unaffected, and we can distinguish model from copy by this means alone, so that, even when there is the closest resemblance, no doubt is possible. In its theoretical aspect this constancy of venation is obviously important, for as nothing about the organism is incapable of variation, the veining of the wings might have varied, as indeed it has varied from genus to genus in the course of the phylogenetic history; but as changes in venation could not be detected by the butterflies' enemies, however sharp-sighted, there has been no reason in these cases for variation in this respect.

In this connexion Poulton has brought forward interesting facts showing that the mimics of one model, belonging to different genera, often secure the same effect in quite different ways. Thus the glass-like transparency of the wings in the Heliconiidæ of the genus Methona depends on a considerable reduction of the size of the scales, which ordinarily cover both sides of the wing as thickly as the tiles on a roof, and produce the colour. In another quite similar species, also transparent-winged, the Danaid Ituna ilione, the transparency is due to the absence of most of the scales, and in a third mimic, Castnia linus, var. heliconoides, the scales are not altered either in size or number, but have become absolutely unpigmented and transparent. In a fourth mimic, a Pierid, Dismorphia crise, the scales have not decreased in number, but have become quite minute, while in a fifth case, the nocturnal Hyelosia heliconoides Swains., the same thing has happened as in Castnia, but the scales are also fewer in number. Thus in each of the mimics the changes which have taken place in the scales are quite different, but they bring about the same effect, the glass-like transparency of the wings, on which the resemblance to the model depends: what we have before us is, therefore, not a similarity of variation, but only an appearance of similarity in external features.

In the face of such facts there can be no further question of the often repeated objection, that the resemblance of model and copy depend on the similarity of external influences upon species living in the same latitude, even if that were not already sufficiently refuted by the frequent restriction of the mimicry to the female. And that mimicry should be a mere matter of chance is negatived even by the single fact that model and copy always live in the same area, and that the local varieties of the model are faithfully followed by the mimic. An interesting example of this is furnished by Elymnias undularis, already mentioned, for in this case the female ([Pl. II], Fig. 23) mimics the brown-yellow Danais plexippus ([Pl. II], Fig. 22), not wherever E. undularis occurs, but only in Ceylon and British India. In Burmah, where another Danaid, D. hegesippus, is common, it mimics that; and in Malacca it does not copy a Danaid at all, but resembles the male of its own species, which in India is very different from it, since there the female mimics one of the blue iridescent Euplœæ ([Pl. III], Fig. 24). It cannot therefore be a matter of 'chance,' and we should have to give up all attempt at a scientific interpretation if we were not prepared to accept that of natural selection. Even the interference of a purposeful Power can hardly be seriously considered in this case, even by those who are inclined to such a view, for the gradual approximation to the model, which is a matter of course in a process of evolution, could only appear, if referred to the benevolent intelligence of a Creator, as an unworthy trick, designed to lead humanity astray in its strivings after knowledge. On the other hand, this gradual increase of resemblance, which becomes apparent when we compare several mimetic species—this carrying over, step by step, from the female to the male—and many other facts point to the working of natural forces according to law, and, if there is to be found anywhere in living nature a complicated process of self-regulation, it certainly lies before us here, clearer and less open to objections than almost anywhere else. I do not mean to say, however, that we can verify it statistically in detail, as has been demanded by the fanatical opponents of natural selection. A direct testing of natural selection is, as has been already shown, nowhere possible: we can never exactly estimate how great the advantage is which a species requiring protection derives from a slight increase in the resemblance to an immune model; and I for one do not know how we could even definitely prove that a certain species needed a greater degree of protection than it had previously enjoyed in order to ensure its persistence in the struggle. It would be necessary to know the total number of individuals living on a certain area for many generations. If it appeared that there was a progressive diminution in the number of individuals, we should be justified in concluding that the species had not an adequate power of persistence, and that it therefore required a more effective protection. But it is impossible for us to collect such exact data for any species living under natural conditions, although we can often say approximately that a species is progressively decreasing in numbers. Even this, however, we can usually do only in cases which are influenced directly or indirectly by the interference of Man in nature, and in which the falling off in the species occurs so rapidly that there is no time for the slow counteractive influence of natural selection. We shall see later that in this way many species have been eliminated even within historic times.

I have just spoken of the 'need of protection,' and I have a few remarks to add on that subject. It is a mistake to believe that every 'rare' species, that is, one represented by few individuals, is already in process of disappearing. It is not the absolute number of individuals that determines the survival of a species, but the fact of the number remaining the same. It is equally mistaken to suppose that an amelioration of the conditions of existence for any species by natural selection is possible only when its persistence is already threatened; that is, when the number of individuals (the 'normal number') is steadily decreasing. On the contrary, it is of the essence of natural selection that every favourable variation which crops up is, ceteris paribus, preserved, and becomes the common possession of the species, quite independently of whether this improvement is absolutely necessary to its preservation or not. In the latter case it will simply become a commoner species instead of a rare one; and every species is, so to speak, striving to become common and widely distributed, since every advantageous variation that can possibly be produced is accumulated and made the common property of the species. But this has its limits, not only in the constitution and the structure of each species, but also in the external conditions of its life. If a species of butterfly be restricted, in the caterpillar stage, to a single, rare species of plant, its normal number will be, and must remain, a small one. But if there arise within it a variation in the food-instinct whereby a second and it may be a commoner plant becomes available, then the normal number of the species will rise, and perhaps the original number of individuals may be more than doubled. It is, however, by no means necessary to assume that the species was previously in process of decadence; on the contrary its normal number may have remained quite constant.

So, in the case of the mimetic butterflies, we do not need to assume that they all previously required protection in the sense that they would have become extinct had they not assumed a likeness to an immune species. We may indeed conclude, on other grounds, that it was the rarer species which increased their number of individuals by the mimetic protection, and in doing so they certainly enhanced at the same time their chance of survival as a species. In the more abundant species mimetic resemblance to species whose unpalatability rendered them immune could not have been evolved, as it would have been disadvantageous, not only for the model, but for the mimicking species itself, while in species less rich in individuals, such resemblance would necessarily have a protective value, no matter whether the species was in danger of extinction or not. The process of selection must have started simply because the mimetic individuals survived more frequently than the others, and the mimetic resemblance must have gone on increasing as long as the increase brought with it a more effective protection. It is, therefore, a fallacious objection to say that a species, whose existence was threatened, would, considering the slowness of the process of selection, have died out altogether before it could have acquired effective protection by mimicking an immune species. The assumption is false—the widespread, hazy idea that the process of natural selection can only begin when the existence of the species is threatened. On the contrary, every species utilizes every possibility of improvement; and every improvement for which variation supplies the necessary material is possible. The augmentation of the profitable variations follows as a necessity from the more frequent survival of the best-adapted individuals, and this 'more frequent survival' will be not only a relative one, due to the fact that the better adapted individuals will be less decimated, it will also be absolute, because more individuals of the species will survive than before. Of this Papilio merope may serve as an example; in Madagascar it now flies about only slightly varied from the original form, var. meriones. Here, therefore, the species is maintained, without the aid of mimetic protection. We do not know if the reason for this lies in the absence of an immune model, or in the non-appearance of suitable mimetic variants, or in other conditions; but we know that without mimicry the species holds its own against its enemies. But if, in Abyssinia, a female of this butterfly exhibited variations which would make her resemble, in any degree, the unpalatable Danias chrysippus, these mimetic variants would be less decimated than the original form of female, and would, therefore, gain stability, and gradually increase both in mimetic resemblance and in the number of individuals. But is this any reason why the original form of the female should diminish in numbers? In itself, certainly not; the red mimetic females could increase in number without causing any decrease of the yellow ones, for the red are in no way in conflict with the yellow, and we must not think of the number of individuals as so fixed for each species that it cannot increase. On the contrary, it must increase, as soon as the conditions of existence are permanently improved, and this happens, in this case, through the mimetic protection of the red female. We can thus easily understand how mimetic and non-mimetic females can live side by side in Abyssinia.

In all the rest of Africa, however, there are only mimetic females of Papilio merope, and none of the colour of the male; these last, therefore, have been crowded out by the mimetic form, not actively, but through the more frequent survival of the mimetic form, so that those like the male became gradually rarer, and finally died out—that is, ceased to occur. The matter is not so simple as it seems, and we shall best understand it by thinking of the dimorphism of the caterpillars of our hawk-moths, which we discussed before, in which the green form in the full-grown caterpillar is less well protected than the brown. In many species the brown form has crowded out the green, in others brown and green occur side by side, but the green is less abundant, and in some species very rare. This must be regarded as the simple result of the circumstance that a higher percentage of the green than of the brown caterpillars fall victims to enemies, and thus, in the course of generations, the green form becomes slowly but steadily rarer. This will be the case even if the newer and better adaptation raises the number of individuals (the 'normal number') in the species, for this increase must always be a limited one, even if it be very great, which is hardly likely in this case. For the normal number is not determined by the mortality at one stage, but by that at all the stages of life taken together. Thus a normal number always persists, notwithstanding the improved conditions for the species, and, on this assumption, the form under less favourable conditions cannot permanently hold its own with that under better conditions, but must gradually disappear. We can understand, then, that the primitive form of the Papilio merope female may persist even for a long time side by side with the mimetic form in certain habitats. It is, probably, not a mere chance, that this should have happened just in Abyssinia, for, in that region, the mimetic female is still tailed—that is, she has not yet reached the highest degree of resemblance to her immune model. In the whole of the rest of Africa the process of the transformation of the female has already reached its highest point, and on the east and west coasts, as well as in South Africa, the primitive form of the species is now represented only by the male.

The gradual dying out of the less favourably conditioned forms of a species is a law which follows as a logical necessity from the essence of the process of selection, but its reality may be inferred from the phenomena themselves. On it depends, as far at least as adaptations are concerned, the transformation of species.