Chapter III: Part I (1)

ON THE SEASONAL DIMORPHISM OF BUTTERFLIES.

I.

THE ORIGIN AND SIGNIFICANCE OF SEASONAL DIMORPHISM.

The phenomena here about to be subjected to a closer investigation have been known for a long period of time. About the year 1830 it was shown that the two forms of a butterfly (_Araschnia_) which had till that time been regarded as distinct, in spite of their different colouring and marking really belonged to the same species, the two forms of this dimorphic species not appearing simultaneously but at different seasons of the year, the one in early spring, the other in summer. To this phenomenon the term “seasonal dimorphism” was subsequently applied by Mr. A. R. Wallace, an expression of which the heterogeneous composition may arouse the horror of the philologist, but, as it is as concise and intelligible as possible, I propose to retain it in the present work.

The species of _Araschnia_ through which the discovery of seasonal dimorphism was made, formerly bore the two specific names _A. Levana_ and _A. Prorsa_. The latter is the summer and the former the winter form, the difference between the two being, to the uninitiated, so great that it is difficult to believe in their relationship. _A. Levana_ (Figs. 1 and 2, Plate I.) is of a golden brown colour with black spots and dashes, while _A. Prorsa_ (Figs. 5 and 6, Plate I.) is deep black with a broad white interrupted band across both wings. Notwithstanding this difference, it is an undoubted fact that both forms are merely the winter and summer generations of the same species. I have myself frequently bred the variety _Prorsa_ from the eggs of _Levana_, and _vice versâ_.

Since the discovery of this last fact a considerable number of similar cases have been established. Thus P. C. Zeller[3] showed, by experiments made under confinement, that two butterflies belonging to the family of the ‘Blues,’ differing greatly in colour and marking, and especially in size, which had formerly been distinguished as _Plebeius_ (_Lycæna_) _Polysperchon_ and _P. Amyntas_, were merely winter and summer generations of the same species; and that excellent Lepidopterist, Dr. Staudinger, proved the same[4] with species belonging to the family of the ‘Whites,’ _Euchloe Belia_ Esp. and _E. Ausonia_ Hüb., which are found in the Mediterranean countries.

The instances are not numerous, however, in which the difference between the winter and summer forms of a species is so great as to cause them to be treated of in systematic work as distinct species. I know of only five of these cases. Lesser differences, having the systematic value of varieties, occur much more frequently. Thus, for instance, seasonal dimorphism has been proved to exist among many of our commonest butterflies belonging to the family of the ‘Whites,’ but the difference in their colour and marking can only be detected after some attention; while with other species, as for instance with the commonest of our small ‘Blues,’ _Plebeius Alexis_ (= _Icarus_, Rott.), the difference is so slight that even the initiated must examine closely in order to recognize it. Indeed whole series of species might easily be grouped so as to show the transition from complete similarity of both generations, through scarcely perceptible differences, to divergence to the extent of varieties, and finally to that of species.

Nor are the instances of lesser differences between the two generations very numerous. Among the European diurnal Lepidoptera I know of about twelve cases, although closer observation in this direction may possibly lead to further discoveries.[5] Seasonal dimorphism occurs also in moths, although I am not in a position to make a more precise statement on this subject,[6] as my own observations refer only to butterflies.

That other orders of insects do not present the same phenomenon depends essentially upon the fact that most of them produce only one generation in the year; but amongst the remaining orders there occur indeed changes of form which, although not capable of being regarded as pure seasonal dimorphism, may well have been produced in part by the same causes, as the subsequent investigation on the relation of seasonal dimorphism to alternation of generation and heterogenesis will more fully prove.

Now what are these causes?

Some years ago, when I imparted to a lepidopterist my intention of investigating the origin of this enigmatical dimorphism, in the hope of profiting for my inquiry from his large experience, I received the half-provoking reply: “But there is nothing to investigate: it is simply the specific character of this insect to appear in two forms; these two forms alternate with each other in regular succession according to a fixed law of Nature, and with this we must be satisfied.” From his point of view the position was right; according to the old doctrine of species no question ought to be asked as to the causes of such phenomena in particular. I would not, however, allow myself to be thus discouraged, but undertook a series of investigations, the results of which I here submit to the reader.

The first conjecture was, that the differences in the imago might perhaps be of a secondary nature, and have their origin in the differences of the caterpillar, especially with those species which grow up during the spring or autumn and feed on different plants, thus assimilating different chemical substances, which might induce different deposits of colour in the wings of the perfect insect. This latter hypothesis was readily confuted by the fact, that the most strongly marked of the dimorphic species, _A. Levana_, fed exclusively on _Urtica major_. The caterpillar of this species certainly exhibits a well-defined dimorphism, but it is not seasonal dimorphism: the two forms do not alternate with each other, but appear mixed in every brood.

I have repeatedly reared the rarer golden-brown variety of the caterpillar separately, but precisely the same forms of butterfly were developed as from black caterpillars bred at the same time under similar external conditions. The same experiment was performed, with a similar result, in the last century by Rösel, the celebrated miniature painter and observer of nature, and author of the well-known “Insect Diversions”--a work in use up to the present day.

The question next arises, as to whether the causes originating the phenomena are not the same as those to which we ascribe the change of winter and summer covering in so many mammalia and birds--whether the change of colour and marking does not depend, in this as in the other cases, upon the _indirect action_ of external conditions of life, i.e., on adaptation through natural selection. We are certainly correct in ascribing white coloration to adaptation[7]--as with the ptarmigan, which is white in winter and of a grey-brown in summer, both colours of the species being evidently of important use.

It might be imagined that analogous phenomena occur in butterflies, with the difference that the change of colour, instead of taking place in the same brood, alternates in different broods.[8] The nature of the difference which occurs in seasonal dimorphism, however, decidedly excludes this view; and moreover, the environment of butterflies presents such similar features, whether they emerge in spring or in summer, that all notions that we may be dealing with adaptational colours must be entirely abandoned.

I have elsewhere[9] endeavoured to show that butterflies in general are not coloured protectively during flight, for the double reason that the colour of the background to which they are exposed continually changes, and because, even with the best adaptation to the background, the fluttering motion of the wings would betray them to the eyes of their enemies.[10] I attempted also to prove at the same time that the diurnal Lepidoptera of our temperate zone have few enemies which pursue them when on the wing, but that they are subject to many attacks during their period of repose.

In support of this last statement I may here adduce an instance. In the summer of 1869 I placed about seventy specimens of _Araschnia Prorsa_ in a spacious case, plentifully supplied with flowers. Although the insects found themselves quite at home, and settled about the flowers in very fine weather (one pair copulated, and the female laid eggs), yet I found some dead and mangled every morning. This decimation continued--many disappearing entirely without my being able to find their remains--until after the ninth day, when they had all, with one exception, been slain by their nocturnal foes--probably spiders and _Opilionidæ_.

Diurnal Lepidoptera in a position of rest are especially exposed to hostile attacks. In this position, as is well known, their wings are closed upright, and it is evident that the adaptational colours on the under side are displayed, as is most clearly shown by many of our native species.[11]

Now, the differences in the most pronounced cases of seasonal dimorphism--for example, in _Araschnia Levana_--are much less manifest on the _under_ than on the _upper_ side of the wing. The explanation by adaptation is therefore untenable; but I will not here pause to confute this view more completely, as I believe I shall be able to show the true cause of the phenomenon.

If seasonal dimorphism does not arise from the _indirect_ influence of varying seasons of the year, it may result from the _direct_ influence of the varying external conditions of life, which are, without doubt, different in the winter from those of the summer brood.

There are two prominent factors from which such an influence may be expected--temperature and duration of development, i.e., duration of the chrysalis period. The duration of the larval period need not engage our attention, as it is only very little shorter in the winter brood--at least, it was so with the species employed in the experiments.

Starting from these two points of view, I carried on experiments for a number of years, in order to find out whether the dual form of the species in question could be traced back to the direct action of the influences mentioned.

The first experiments were made with _Araschnia Levana_. From the eggs of the winter generation, which had emerged as butterflies in April, I bred caterpillars, and immediately after pupation placed them in a refrigerator, the temperature of the air of which was 8°-10° R. It appeared, however, that the development could not thus be retarded to any desired period by such a small diminution of temperature, for, when the box was taken out of the refrigerator after thirty-four days, all the butterflies, about forty in number, had emerged, many being dead, and others still living. The experiment was so far successful that, instead of the _Prorsa_ form which might have been expected under ordinary circumstances, most of the butterflies emerged as the so-called _Porima_ (Figs. 3, 4, 7, 8, and 9, Plate I.); that is to say, in a form intermediate between _Prorsa_ and _Levana_ sometimes found in nature, and possessing more or less the marking of the former, but mixed with much of the yellow of _Levana_.

It should be here mentioned, that similar experiments were made in 1864 by George Dorfmeister, but unfortunately I did not get this information[12] until my own were nearly completed. In these well-conceived, but rather too complicated experiments, the author arrives at the conclusion “that temperature certainly affects the colouring, and through it the marking, of the future butterfly, and chiefly so during pupation.” By lowering the temperature of the air during a portion of the pupal period, the author was enabled to produce single specimens of _Porima_, but most of the butterflies retained the _Prorsa_ form. Dorfmeister employed a temperature a little higher than I did in my first experiments, viz. 10°-11° R., and did not leave the pupæ long exposed, but after 5½-8 days removed them to a higher temperature. It was therefore evident that he produced transition forms in a few instances only, and that he never succeeded in bringing about a complete transformation of the summer into the winter form.

In my subsequent experiments I always exposed the pupæ to a temperature of 0°-1° R.; they were placed directly in the refrigerator, and taken out at the end of four weeks. I started with the idea that it was perhaps not so much the reduced temperature as the retardation of development which led to the transformation. But the first experiment had shown that the butterflies emerged between 8° and 10° R., and consequently that the development could not be retarded at this temperature.

A very different result was obtained from the experiment made at a lower temperature.[13] Of twenty butterflies, fifteen had become transformed into _Porima_, and of these three appeared very similar to the winter form (_Levana_), differing only in the absence of the narrow blue marginal line, which is seldom absent in the true _Levana_. Five butterflies were uninfluenced by the cold, and remained unchanged, emerging as the ordinary summer form (_Prorsa_). It thus appeared from this experiment, that a large proportion of the butterflies inclined to the _Levana_ form by exposure to a temperature of 0°-1° R. for four weeks, while in a few specimens the transformation into this form was nearly perfect.

Should it not be possible to perfect the transformation, so that each individual should take the _Levana_ form? If the assumption of the _Prorsa_ or _Levana_ form depends only on the direct influence of temperature, or on the duration of the period of development, it should be possible to compel the pupæ to take one or the other form at pleasure, by the application of the necessary external conditions. This has never been accomplished with _Araschnia Prorsa_. As in the experiment already described, and in all subsequent ones, single specimens appeared as the unchanged summer form, others showed an appearance of transition, and but very few had changed so completely as to be possibly taken for the pure _Levana_. In some species of the sub-family _Pierinæ_, however, at least in the case of the summer brood, there was, on the contrary, a complete transformation.

Most of the species of our ‘Whites’ (_Pierinæ_) exhibit the phenomenon of seasonal dimorphism, the winter and summer forms being remarkably distinct. In _Pieris Napi_ (with which species I chiefly experimented) the winter form (Figs. 10 and 11, Plate I.) has a sprinkling of deep black scales at the base of the wings on the upper side, while the tips are more grey, and have in all cases much less black than in the summer form; on the underside the difference lies mainly in the frequent breadth, and dark greenish-black dusting, of the veins of the hind wings in the winter form, while in the summer form these greenish-black veins are but faintly present.

I placed numerous specimens of the summer brood, immediately after their transformation into chrysalides, in the refrigerator (0°-1° R.), where I left them for three months, transferring them to a hothouse on September 11th, and there (from September 26th to October 3rd) sixty butterflies emerged, the whole of which, without exception--and most of them in an unusually strong degree--bore the characters of the winter form. I, at least, have never observed in the natural state such a strong yellow on the underside of the hind wings, and such a deep blackish-green veining, as prevailed in these specimens (see, for instance, Figs. 10 and 11). The temperature of the hothouse (12°-24° R.) did not, however, cause the emergence of the whole of the pupæ; a portion hibernated, and produced in the following spring butterflies of the winter form only. I thus succeeded, with this species of _Pieris_, in completely changing every individual of the summer generation into the winter form.

It might be expected that the same result could be more readily obtained with _A. Levana_, and fresh experiments were undertaken, in order that the pupæ might remain in the refrigerator fully two months from the period of their transformation (9-10th July). But the result obtained was the same as before--fifty-seven butterflies emerged in the hothouse[14] from September 19th to October 4th, nearly all of these approaching very near to the winter form, without a single specimen presenting the appearance of a perfect _Levana_, while three were of the pure summer form (_Prorsa_).

Thus with _Levana_ it was not possible, by refrigeration and retardation of development, to change the summer completely into the winter form in all specimens. It may, of course, be objected that the period of refrigeration had been too short, and that, instead of leaving the pupæ in the refrigerator for two months, they should have remained there six months, that is, about as long as the winter brood remains under natural conditions in the chrysalis state. The force of this last objection must be recognized, notwithstanding the improbability that the desired effect would be produced by a longer period of cold, since the doubling of this period from four to eight weeks did not produce[15] any decided increase in the strength of the transformation. I should not have omitted to repeat the experiment in this modified form, but unfortunately, in spite of all trouble, I was unable to collect during the summer of 1873 a sufficient number of caterpillars. But the omission thus caused is of quite minor importance from a theoretical point of view.

For let us assume that the omitted experiment had been performed--that pupæ of the summer brood were retarded in their development by cold until the following spring, and that every specimen then emerged in the perfect winter form, Levana. Such a result, taken in connexion with the corresponding experiment upon _Pieris Napi_, would warrant the conclusion that the direct action of a certain amount of cold (or of retardation of development) is able to compel all pupæ, from whichever generation derived, to assume the winter form of the species. From this the converse would necessarily follow, viz. that a certain amount of warmth would lead to the production of the summer form, _Prorsa_, it being immaterial from which brood the pupæ thus exposed to warmth might be derived. But the latter conclusion was proved experimentally to be incorrect, and thus the former falls with it, whether the imagined experiment with _Prorsa_ had succeeded or not.

I have repeatedly attempted by the application of warmth to change the winter into the summer form, but always with the same negative result. _It is not possible to compel the winter brood to assume the form of the summer generation._

_A. Levana_ may produce not only two but three broods in the year, and may, therefore, be said to be _polygoneutic_.[16] One winter brood alternates with two summer broods, the first of which appears in July, and the second in August. The latter furnishes a fourth generation of pupæ, which, after hibernation, emerge in April, as the first brood of butterflies in the form _Levana_.

I frequently placed pupæ of this fourth brood in the hothouse immediately after their transformation, and in some cases even during the caterpillar stage, the temperature never falling, even at night, below 12° R., and often rising during the day to 24° R. The result was always the same: all, or nearly all, the pupæ hibernated, and emerged the following year in the winter form as perfectly pure _Levana_, without any trace of transition to the _Prorsa_ form. On one occasion only was there a _Porima_ among them, a case for which an explanation will, I believe, be found later on. It often happened, on the other hand, that some few of the butterflies emerged in the autumn, about fourteen days after pupation; and these were always _Prorsa_ (the summer form), excepting once a _Porima_.

From these experiments it appeared that similar causes (heat) affect different generations of _A. Levana_ in different manners. With both summer broods a high temperature always caused the appearance of _Prorsa_, this form arising but seldom from the third brood (and then only in a few individuals), while the greater number retained the _Levana_ form unchanged. We may assign as the reason for this behaviour, that the third brood has no further tendency to be accelerated in its development by the action of heat, but that by a longer duration of the pupal stage the _Levana_ form must result. On one occasion the chrysalis stage was considerably shortened in this brood by the continued action of a high temperature, many specimens thus having their period of development reduced from six to three months. The supposed explanation above given is, however, in reality no explanation at all, but simply a restatement of the facts. The question still remains, why the third brood in particular has no tendency to be accelerated in its development by the action of heat, as is the case with both the previous broods?

The first answer that can be given to this question is, that the cause of the different action produced by a similar agency can only lie in the _constitution_, i.e., in the _physical nature_ of the broods in question, and not in the external influences by which they are acted upon. Now, what is the difference in the physical nature of these respective broods? It is quite evident, as shown by the experiments already described, that cold and warmth cannot be the _immediate_ causes of a pupa emerging in the _Prorsa_ or _Levana_ form, since the last brood always gives rise to the _Levana_ form, whether acted on by cold or warmth. The first and second broods only can be made to partly assume, more or less completely, the _Levana_ form by the application of cold. In these broods then, a low temperature is the _mediate_ cause of the transformation into the _Levana_ form.

The following is my explanation of the facts. The form _Levana_ is the original type of the species, and _Prorsa_ the secondary form arising from the gradual operation of summer climate. When we are able to change many specimens of the summer brood into the winter form by means of cold, this can only depend upon reversion to the original, or ancestral, form, which reversion appears to be most readily produced by cold, that is, by the same external influences as those to which the original form was exposed during a long period of time, and the continuance of which has preserved, in the winter generations, the colour and marking of the original form down to the present time.

I consider the origination of the _Prorsa_ from the _Levana_ form to have been somewhat as follows:--It is certain that during the diluvial period in Europe there was a so-called ‘glacial epoch,’ which may have spread a truly polar climate over our temperate zone; or perhaps a lesser degree of cold may have prevailed with increased atmospheric precipitation. At all events, the summer was then short and comparatively cold, and the existing butterflies could have only produced one generation in the year; in other words, they were _monogoneutic_. At that time _A. Levana_ existed only in the _Levana_ form.[17] As the climate gradually became warmer, a period must have arrived when the summer lasted long enough for the interpolation of a second brood. The pupæ of _Levana_, which had hitherto hibernated through the long winter to appear as butterflies in the following summer, were now able to appear on the wing as butterflies during the same summer as that in which they left their eggs as larvæ, and eggs deposited by the last brood produced larvæ which fed up and hibernated as pupæ. A state of things was thus established in which the first brood was developed under very different climatic conditions from the second. So considerable a difference in colour and marking between the two forms as we now witness could not have arisen suddenly, but must have done so gradually. It is evident from the foregoing experiments that the _Prorsa_ form did not originate suddenly. Had this been the case it would simply signify that every individual of this species possessed the faculty of assuming two different forms according as it was acted on by warmth or cold, just in the same manner as litmus-paper becomes red in acids and blue in alkalies. The experiments have shown, however, that this is not the case, but rather that the last generation bears an ineradicable tendency to take the _Levana_ form, and is not susceptible to the influence of warmth, however long continued; while both summer generations, on the contrary, show a decided tendency to assume the _Prorsa_ form, although they certainly can be made to assume the _Levana_ form in different degrees by the prolonged action of cold.

The conclusion seems to me inevitable, that the origination of the _Prorsa_ form was gradual--that those changes which originated in the chemistry of the pupal stage, and led finally to the _Prorsa_ type, occurred very gradually, at first perhaps remaining completely latent throughout a series of generations, then very slight changes of marking appearing, and finally, after a long period of time, the complete _Prorsa_ type was produced. It appears to me that the quoted results of the experiments are not only easily explained on the view of the _gradual_ action of climate, but that this view is the only one admissible. The action of climate is best comparable with the so-called cumulative effect of certain drugs on the human body; the first small dose produces scarcely any perceptible change, but if often repeated the effect becomes cumulative, and poisoning occurs.

This view of the action of climate is not at all new, most zoologists having thus represented it; only the formal proof of this action is new, and the facts investigated appear to me of special importance as furnishing this proof. I shall again return to this view in considering climatic varieties, and it will then appear that also the nature of the transformation itself confirms the slow operation of climate.

During the transition from the glacial period to the present climate _A. Levana_ thus gradually changed from a monogoneutic to a digoneutic species, and at the same time became gradually more distinctly dimorphic, this character originating only through the alteration of the summer brood, the primary colouring and marking of the species being retained unchanged by the winter brood. As the summer became longer a third generation could be interpolated--the species became polygoneutic; and in this manner two summer generations alternated with one winter generation.

We have now to inquire whether facts are in complete accordance with this theory--whether they are never at variance with it--and whether they can all be explained by it. I will at once state in anticipation, that this is the case to the fullest extent.

In the first place, the theory readily explains why the summer but not the winter generations are capable of being transformed; the latter cannot possibly revert to the _Prorsa_ form, because this is much the younger. When, however, it happens that out of a hundred cases there occurs one in which a chrysalis of the winter generation, having been forced by warmth, undergoes transformation before the commencement of winter, and emerges in the summer form,[18] this is not in the least inexplicable. It cannot be atavism which determines the direction of the development; but we see from such a case that the changes in the first two generations have already produced a certain alteration in the third, which manifests itself in single cases under favourable conditions (the influence of warmth) by the assumption of the _Prorsa_ form; or, as it might be otherwise expressed, the _alternating_ heredity (of which we shall speak further), which implies the power of assuming the _Prorsa_ form, remains latent as a rule in the winter generation, but becomes _continuous_ in single individuals.

It is true that we have as yet no kind of insight into the nature of heredity, and this at once shows the defectiveness of the foregoing explanation; but we nevertheless know many of its external phenomena. We know for certain that one of these consists in the fact that peculiarities of the father do not appear in the son, but in the grandson, or still further on, and that they may be thus transmitted in a latent form. Let us imagine a character so transmitted that it appears in the first, third, and fifth generations, remaining latent in the intermediate ones; it would not be improbable, according to previous experiences, that the peculiarity should exceptionally, i.e., from a cause unknown to us, appear in single individuals of the second or fourth generation. But this completely agrees with those cases in which “exceptional” individuals of the winter brood took the _Prorsa_ form, with the difference only that a cause (warmth) was here apparent which occasioned the development of the latent characters, although we are not in a position to say in what manner heat produces this action. These exceptions to the rule are therefore no objection to the theory. On the contrary, they give us a hint that after one _Prorsa_ generation had been produced, the gradual interpolation of a second _Prorsa_ generation may have been facilitated by the existence of the first. I do not doubt that even in the natural state single individuals of _Prorsa_ sometimes emerge in September or October; and if our summer were lengthened by only one or two months this might give rise to a third summer brood (just as a second is now an accomplished fact), under which circumstances they would not only emerge, but would also have time for copulation and for depositing eggs, the larvæ from which would have time to grow up.

A sharp distinction must be made between the first establishment of a new climatic form and the transference of the latter to newly interpolated generations. The former always takes place very slowly; the latter may occur in a shorter time.

With regard to the duration of time which is necessary to produce a new form by the influence of climate, or to transmit to a succeeding generation a new form already established, great differences occur, according to the physical nature of the species and of the individual. The experiments with _Prorsa_ already described show how diverse are individual proclivities in this respect. In Experiment No. 12 it was not possible out of seventy individuals to substitute _Prorsa_ for the _Levana_ form, even in one solitary case, or, in other words, to change alternating into continuous inheritance; whilst in the corresponding experiments of former years (Experiment 10, for example), out of an equal number of pupæ three emerged as _Prorsa_, and one as _Porima_. We might be inclined to seek for the cause of this different behaviour in external influences, but we should not thus arrive at an explanation of the facts. We might suppose, for instance, that a great deal depended upon the particular period of the pupal stage at which the action of the elevated temperature began--whether on the first, the thirtieth, or the hundredth day after pupation--and this conjecture is correct in so far that in the two last cases warmth can have no further influence than that of somewhat accelerating the emergence of the butterflies, but cannot change the _Levana_ into the _Prorsa_ form. I have repeatedly exposed a large number of _Levana_ pupæ of the third generation to the temperature of an apartment, or even still higher (26° R.), during winter, but no _Prorsa_ were obtained.[19]

But it would be erroneous to assume a difference in the action of heat according as it began on the first or third day after transformation; whether during or before pupation. This is best proved by Experiment No. 12, in which caterpillars of the fourth generation were placed in the hothouse several days before they underwent pupation; still, not a single butterfly assumed the _Prorsa_ form. I have also frequently made the reverse experiment, and exposed caterpillars of the first summer brood to cold during the act of pupation. A regular consequence was the dying off of the caterpillars, which is little to be wondered at, as the sensitiveness of insects during ecdysis is well known, and transformation into the pupal state is attended by much deeper changes.

Dorfmeister thought that he might conclude from his experiments that temperature exerts the greatest influence in the first place during the act of pupation, and in the next place immediately after that period. His experiments were made, however, with such a small number of specimens that scarcely any safe conclusion can be founded on them; still, this conclusion may be correct, in so far as everything depends on whether, from the beginning, the formative processes in the pupa tended to this or that direction, the final result of which is the _Prorsa_ or _Levana_ form. If once there is a tendency to one or the other direction, then temperature might exert an accelerating or a retarding influence, but the tendency cannot be further changed.

It is also possible--indeed, probable--that a period may be fixed in which warmth or cold might be able to divert the original direction of development most easily; and this is the next problem to be attacked, the answer to which, now that the main points have been determined, should not be very difficult. I have often contemplated taking the experiments in hand myself, but have abandoned them, because my materials did not appear to me sufficiently extensive, and in all such experiments nothing is to be more avoided than a frittering away of experimental materials by a too complicated form of problem.

There may indeed be a period most favourable for the action of temperature during the first days of the pupal stage; it appears from Experiment No. 12 that individuals tend in different degrees to respond to such influences, and that the disposition to abandon the ordinary course of development is different in different individuals. In no other way can it be explained that, in all the experiments made with the first and second generations of _Prorsa_, only _a portion_ of the pupæ were compelled by cold to take the direction of development of _Levana_, and that even from the former only a few individuals completely reverted, the majority remaining intermediate.

If it be asked why in the corresponding experiments with _Pieris Napi_ complete reversion always occurred without exception, it may be supposed that in this species the summer form has not been so long in existence, and that it would thus be more easily abandoned; or, that the difference between the two generations has not become so distinct, which further signifies that here again the summer form is of later origin. It might also be finally answered, that the tendency to reversion in different species may vary just as much as in different individuals of the same species. But, in any case, the fact is established that all individuals are impelled by cold to complete reversion, and that in these experiments it does not depend so particularly upon the moment of development when cold is applied, but that differences of individual constitution are much more the cause why cold brings some pupæ to complete, and others to partial, reversion, while yet others are quite uninfluenced. In reference to this, the American _Papilio Ajax_ is particularly interesting.

This butterfly, which is somewhat similar to the European _P. Podalirius_, appears, wherever it occurs, in three varieties, designated as var. _Telamonides_, var. _Walshii_, and var. _Marcellus_. The distinguished American entomologist, W. H. Edwards, has proved by breeding experiments, that all three forms belong to the same cycle of development, and in such a manner that the first two appear only in spring, and always come only from hibernating pupæ, while the last form, var. _Marcellus_, appears only in summer, and then in three successive generations. A seasonal dimorphism thus appears which is combined with ordinary dimorphism, winter and summer forms alternating with each other; but the first appears itself in two forms or varieties, vars. _Telamonides_ and _Walshii_. If for the present we disregard this complication, and consider these two winter forms as one, we should thus have four generations, of which the first possesses the winter form, and the three succeeding ones have, on the other hand, the summer form, var. _Marcellus_.

The peculiarity of this species consists in the fact that in all three summer generations only a portion of the pupæ emerge after a short period (fourteen days), whilst another and much smaller portion remains in the pupal state during the whole summer and succeeding winter, first emerging in the following spring, and then always in the winter form. Thus, Edwards states that out of fifty chrysalides of the second generation, which had pupated at the end of June, forty-five _Marcellus_ butterflies appeared after fourteen days, whilst five pupæ emerged in April of the following year, and then as _Telamonides_.

The explanation of these facts is easily afforded by the foregoing theory. According to this, both the winter forms must be regarded as primary, and the _Marcellus_ form as secondary. But this last is not yet so firmly established as _Prorsa_, in which reversion of the summer generations to the _Levana_ form only occurs through special external influences; whilst in the case of _Ajax_ some individuals are to be found in every generation, the tendency of which to revert is still so strong that even the greatest summer heat is unable to cause them to diverge from their original inherited direction of development, or to accelerate their emergence and compel them to assume the _Marcellus_ form. It is here beyond a doubt that it is not different external influences, but internal causes only, which maintain the old hereditary tendency, for all the larvæ and pupæ of many different broods were simultaneously exposed to the same external influences. But, at the same time, it is evident that these facts are not opposed to the present theory; on the contrary, they confirm it, inasmuch as they are readily explained on the basis of the theory, but can scarcely otherwise be understood.

If it be asked what significance attaches to the duplication of the winter form, it may be answered that the species was already dimorphic at the time when it appeared in only one annual generation. Still, this explanation may be objected to, since a dimorphism of this kind is not at present known, though indeed some species exhibit a sexual dimorphism,[20] in which one sex (as, for instance, the case of the female _Papilio Turnus_) appears in two forms of colouring, but not a dimorphism, as is here the case, displayed by both sexes.[21] Another suggestion, therefore, may perhaps be offered.

In _A. Levana_ we saw that reversion occurred in very different degrees with different individuals, seldom attaining to the true _Levana_ form, and generally only reaching the intermediate form known as _Porima_. Now it would, at all events, be astonishing if with _P. Ajax_ the reversion were always complete, as it is precisely in this case that the tendency to individual reversion is so variable. I might, for this reason, suppose that one of the two winter forms, viz. the var. _Walshii_, is nothing else than an incomplete reversion-form, corresponding to _Porima_ in the case of _A. Levana_. Then _Telamonides_ only would be the original form of the butterfly, and this would agree with the fact that this variety appears later in the spring than _Walshii_. Experiments ought to be able to decide this.[22] The pupæ of the first three generations placed upon ice should give, for the greater part, the form _Telamonides_, for the lesser portion _Walshii_, and for only a few, or perhaps no individuals, the form _Marcellus_. This prediction is based on the view that the tendency to revert is on the whole great; that even with the first summer generation, which was the longest exposed to the summer climate, a portion of the pupæ, without artificial means, always emerged as _Telamonides_, and another portion as _Marcellus_. The latter will perhaps now become _Walshii_ by the application of cold.

One would expect that the second and third generations would revert more easily, and in a larger percentage, than the first, because this latter first acquired the new _Marcellus_ form; but the present experiments furnish no safe conclusion on this point. Thus, of the first summer generation only seven out of sixty-seven pupæ hibernated, and these gave _Telamonides_; while of the second generation forty out of seventy-six, and of the third generation twenty-nine out of forty-two pupæ hibernated. But to establish safer conclusions, a still larger number of experiments is necessary. According to the experience thus far gained, one might perhaps still be inclined to imagine that, with seasonal dimorphism, external influences operating on the individual might directly compel it to assume one or the other form. I long held this view myself, but it is, nevertheless, untenable. That cold does not produce the one kind of marking, and warmth the other, follows from the before-mentioned facts, viz. that in _Papilio Ajax_ every generation produces both forms; and, further, in the case of _A. Levana_ I have frequently reared the fourth (hibernating) generation entirely in a warm room, and yet I have always obtained the winter form. Still, one might be inclined not to make the temperature _directly_ responsible, but rather the retardation or acceleration of development produced through the action of temperature. I confess that I for a long time believed that in this action I had found the true cause of seasonal dimorphism. Both with _A. Levana_ and _P. Napi_ the difference between the duration of the pupal period in the winter and summer forms is very great, lasting as a rule, in the summer generation of _A. Levana_, from seven to twelve days, and in the winter generation about two hundred days. In this last species the pupal state can certainly be shortened by keeping them at an elevated temperature; but I have, nevertheless, only in one case obtained two or three butterflies at the end of December from caterpillars that had pupated in September, these generally emerging in the course of February and March, and are to be seen on the wing in warm weather during the latter month. The greatest reduction of the pupal period still leaves for this stage more than 100 days.

From this last observation it follows that it is not the duration of development which, in individual cases, determines the form of the butterfly, and which consequently decides whether the winter or summer form shall emerge, but that, on the contrary, the duration of the pupal stage is dependent on the tendency which the forthcoming butterfly had taken in the chrysalis state. This can be well understood when we consider that the winter form must have had a long, and the summer form a short pupal period, during innumerable generations. In the former the habit of slow development must have been just as well established as that of rapid development in the latter; and we cannot be at all surprised if we do not see this habit abandoned by the winter form when the opportunity presents itself. But that it may be occasionally abandoned the more proves that the duration of the pupal development less determines the butterfly form than does the temperature directly, in individual cases.

Thus, for instance, Edwards explicitly states that, whereas the two winter forms of _P. Ajax_, viz. the vars. _Walshii_ and _Telamonides_, generally appear only after a pupal period of 150 to 270 days, yet individual cases occur in which the pupal stage is no longer than in the summer form, viz. fourteen days.[23] A similar thing occurs with _A. Levana_, for, as already explained, not only may the development of the winter form be forced to a certain degree by artificial warmth, but the summer generation frequently produces reversion-forms without protraction of development. The intermediate reversion-form _Porima_ was known long before it was thought possible that it could be produced artificially by the action of cold; it appears occasionally, although very rarely, at midsummer in the natural state.

If, then, my explanation of the phenomena is correct, the winter form is primary and the summer the secondary form, and those individuals which, naturally or artificially, assume the winter form must be considered as cases of atavism. The suggestion thus arises whether low temperature alone is competent to bring about this reversion, or whether other external influences are not also effective. Indeed, the latter appears to be the case. Besides purely internal causes, as previously pointed out in _P. Ajax_, warmth and mechanical motion appear to be able to bring about reversion.

That an unusually high temperature may cause reversion, I conclude from the following observation. In the summer of 1869 I bred the first summer brood of _A. Levana_; the caterpillars pupated during the second half of June, and from that time to their emergence, on 28th June-3rd July, great heat prevailed. Now, while the intermediate form _Porima_ had hitherto been a great rarity, both in the free state and when bred, having never obtained it myself, for example, out of many hundreds of specimens, there were among the sixty or seventy butterflies that emerged from the above brood, some eight to ten examples of _Porima_. This is certainly not an exact experiment, but there seems to me a certain amount of probability that the high summer temperature in this case brought about reversion.

Neither for the second cause to which I have ascribed the power of producing reversion can I produce any absolute evidence, since the experimental solution of all these collateral questions would demand an endless amount of time. I am in possession of an observation, however, which makes it appear probable to me that continuous mechanical movement acts on the development of the pupæ in a similar manner to cold, that is, retarding them, and at the same time producing reversion. I had, in Freiburg, a large number of pupæ of the first summer brood of _Pieris Napi_, bred from eggs. I changed residence while many caterpillars were in course of transformation and travelled with the pupæ in this state seven hours by rail. Although this brood of _P. Napi_, under ordinary circumstances, always emerges in the summer, generally in July of the same year, as the summer form (var. _Napeæ_), yet out of these numerous pupæ I did not get a single butterfly during the year 1872. In winter I kept them in a warm room, and the first butterflies emerged in January, 1873, the remainder following in February, March, and April, and two females not until June. All appeared, however, as exquisite winter forms. The whole course of development was precisely as though cold had acted on the pupæ; and in fact, I could find no other cause for this quite exceptional deportment than the seven hours’ shaking to which the pupæ were exposed by the railway journey, immediately after or during their transformation.

It is obviously a fact of fundamental importance to the theory of seasonal dimorphism, that the summer form can be readily changed into the winter form, whilst the latter cannot be changed into the summer form. I have thus far only made experiments on this subject with _A. Levana_, but the same fact appears to me to obtain for _P. Napi_. I did not, however, operate upon the ordinary winter form of _P. Napi_, but chose for this experiment the variety _Bryoniæ_, well known to all entomologists. This is, to a certain extent, the potential winter form of _P. Napi_; the male (Fig. 14, Plate I.) exactly resembles the ordinary winter form in the most minute detail, but the female is distinguished from _Napi_ by a sprinkling of greyish brown scales over the whole of the upper side of the wings (Fig. 15, Plate I.). This type, _Bryoniæ_, occurs in Polar regions as the only form of _Napi_, and is also found in the higher Alps, where it flies in secluded meadows as the only form, but in other localities, less isolated, mixed with the ordinary form of the species. In both regions _Bryoniæ_ produces but one generation in the year, and must thus, according to my theory, be regarded as the parent-form of _Pieris Napi_.

If this hypothesis is correct--if the variety _Bryoniæ_ is really the original form preserved from the glacial period in certain regions of the earth, whilst _Napi_ in its winter form is the first secondary form gradually produced through a warm climate, then it would be impossible ever to breed the ordinary form _Napi_ from pupæ of _Bryoniæ_ by the action of warmth, since the form of the species now predominant must have come into existence only by a cumulative action exerted on numerous generations, and not _per saltum_.

The experiment was made in the following manner: In the first part of June I caught a female of _Bryoniæ_ in a secluded Alpine valley, and placed her in a capacious breeding-cage, where she flew about among the flowers, and laid more than a hundred eggs on the ordinary cabbage. Although the caterpillars in the free state feed upon another plant unknown to me, they readily ate the cabbage, grew rapidly, and pupated at the end of July. I then brought the pupæ into a hothouse in which the temperature fluctuated between 12° and 24° R.; but, in spite of this high temperature, and--what is certainly of more special importance--notwithstanding the want of cooling at night, only one butterfly emerged the same summer, and that a male, which, from certain minute characteristic markings, could be safely identified as var. _Bryoniæ_. The other pupæ hibernated in the heated room, and produced, from the end of January to the beginning of June, 28 butterflies, all of which were exquisite _Bryoniæ_.

Experiment thus confirmed the view that _Bryoniæ_ is the parent-form of _Napi_, and the description hitherto given by systematists ought therefore properly to be reversed. _Pieris Bryoniæ_ should be elevated to the rank of a species, and the ordinary winter and summer forms should be designated as vars. _Napi_ and _Napeæ_. Still I should not like to take it upon myself to increase the endless confusion in the synonomy of butterflies. In a certain sense, it is also quite correct to describe the form _Bryoniæ_ as a climatic variety, for it is, in fact, established, if not produced, by climate, by which agency it is likewise preserved; only it is not a secondary, but the primary, climatic variety of _Napi_. In this sense most species might probably be described as climatic varieties, inasmuch as under the influence of another climate they would gradually acquire new characters, whilst, under the influence of the climate now prevailing in their habitats, they have, to a certain extent, acquired and preserved their present form.

The var. _Bryoniæ_ is, however, of quite special interest, since it makes clear the relation which exists between climatic variation and seasonal dimorphism, as will be proved in the next section. The correctness of the present theory must first here be submitted to further proof.

It has been shown that the secondary forms of seasonally dimorphic butterflies do not all possess the tendency to revert in the same degree, but that this tendency rather varies with each individual. As the return to the primary form is synonymous with the relinquishing of the secondary, the greater tendency to revert is thus synonymous with the greater tendency to relinquish the secondary form, but this again is equivalent to a lesser stability of the latter; it must consequently be concluded that the individuals of a species are very differently influenced by climatic change, so that with some the new form must become sooner established than with others. From this a variability of the generation concerned must necessarily ensue, i.e., the individuals of the summer generation must differ more in colour and marking than is the case with those of the winter generation. If the theory is correct, the summer generations should be more variable than the winter generations--at least, so long as the greatest possible equalization of individual variations has not occurred through the continued action of warmth, combined with the constant crossing of individuals which have become changed in different degrees. Here also the theory is fully in accord with facts.