The courtship of another species, Dendryphantes capitatus, in which the sexes are entirely different, is described as follows:

In another species, Dendryphantes elegans, both sexes are brilliantly colored.

If we lay no emphasis on the implied emotional elements in the behavior of the spiders in this description—terms of emotion borrowed direct from human psychology—there still remain the several types of apparently significant reactions associated with courtship. The statements leave no room for doubt that vision plays an important rôle in the complex reflexes that lead gradually to successful mating. The Peckhams insist that the display of the male is always of a kind to bring before the female the special adornments of the male in whatever part of the body they may lie. The chance of subjective interpretation here is so great that unless the results are carefully checked up by studies of the attitudes assumed by males in species in which the males are without ornament, their interpretation must be taken with the greatest reserve. Assigning, as our authors do, so much by gratuitous implication to the emotional side of the picture prejudices one, perhaps too greatly, against accepting a special (even an implied intentional) exhibition of the specially ornamented parts. On the other hand, if it be conceded that the conspicuousness of the male is an element in the reaction, the very special adornments visible from the front might be supposed to enhance the effect produced in the female. Similar displays of special ornamentation in the male have been described both for birds and insects, but here, too, the question has been raised as to whether such exhibitions are more than an accidental accompaniment of the posturing of the male, for the same kind of behavior is known to occur in other cases where the male is unornamented and resembles the female. Had such a male special ornamentation it would no doubt appear to us that his behavior was “calculated” to display his ornaments.

Dr. and Mrs. Peckham point out that their observations are entirely inconsistent with Wallace’s interpretation of the origin of secondary sexual characters. They find no evidence in favor of his view that the male possesses greater “vital activity.” On the contrary, the female is the more active and pugnacious of the two. They also object to Wallace’s statement of a total absence of any evidence that the female notices the display of the male. In spiders the females “observe” the males with close attention during their courtship. They point out also that, in spiders at least, as the female gradually becomes adult, a male if preferred will have a chance of mating with several females, “and as the mating season lasts for two or three weeks the more brilliant males may easily be selected again and again.” In regard to Wallace’s argument as to the distribution of accessory plumes in humming birds, the Peckhams point out that—

From this brief survey of the family we see that, contrary to what we should expect from Mr. Wallace’s theory, although the breast muscles are the seat of the highest activity, breast plumes are the least frequent of all the forms of ornamental plumage.

What is true for birds is even more obvious for spiders where the special ornaments are not confined to parts of the body with high muscular development, etc. The writers make the very pertinent criticism that while Wallace objects to assuming the emotional states in females, he is less careful in regard to the males’ emotions when he speaks of the display “under the influence of jealousy or sexual excitement.... The males, in their rivalry with each other, would see what plumes were most effective; and each would endeavor to excel his enemy as far as voluntary exertion would enable him.”[13]

In regard to another interpretation of the courtship, the Peckhams point out:

Finally, in regard to the specific character of the display of the males, the Peckhams make the following significant statement:

In the tarantula, Petrunkewitsch finds that sight plays no rôle in mating—that it is due entirely to accidental contact between the male and female. Here the sexes are closely alike, except for a pair of hooks on the front legs of the male, by means of which he grasps the mandibles of the female, holding them during the elaborate process of transference to her genital opening the sperm that he has already collected in the genital spoon on his palpi. The hooks serve to guard the male against injury or death, while at the same time they aid him in the act of coitus.

In a common spider, Mœvia villata, two kinds of males exist. Both have been seen to mate with the same female. No preference is given to either type. The difference between them, according to Painter, is connected with or caused by an additional pair of chromosomes in the gray male. The two types may therefore have no connection with sexual selection, but be directly due to a difference in the chromosome group.

Montgomery, who made observations on the courting habits of several species of spiders, states that his “general theoretical conclusions were quite different from those of the Peckhams.” It turns out, however, that his objection to their view is based entirely on their assumption that the male is conscious of his display and that the female is guided by an esthetic sense in selecting the more beautiful male. It should be pointed out that even after the removal of these gratuitous assumptions as to the cause of the evolution of the male and female, enough still remains in Montgomery’s own observations to include his results on courtship under Darwin’s theory of sexual selection. For example, Montgomery says:

Now, it is obvious that if a more brightly colored male has a better chance of “advertising himself” to the female all the essential requirements of Darwin’s theory are fulfilled, regardless of whether the male is conscious of his ornamentation or the female makes use of an “esthetic sense.” In another passage (p. 173) Montgomery concedes all that any modern critical advocate of Darwin’s theory is likely to ask:

It is evident that Montgomery has only shifted the situation, although to advantage, I think, but is essentially in accord with Darwin’s theory of sexual selection, despite his protest to the contrary. The difference lies in Darwin’s and especially in the Peckhams’ use of the term “choice,” “aesthetic sense,” etc., to stand for the fact that the female more promptly mates (as Montgomery prefers to put it) with a male peculiarly ornamental.

The most critical observations on sexual selection that have been made in the group of insects are those by Sturtevant on the pomace fly. The courtship is described as follows:

To test whether the wings have any significance in courtship, the wings of a male were clipped off and he was put into competition with a normal male of the same stock, age, and size. A virgin female sexually mature was given to these two males. The normal male mated 72 times before the other, the clipped male 53 times. It might appear that the female selected the normal male in preference to the clipped one, or possibly that the male with normal wings drove the other male away. That the operation on the wings may have an influence on the male himself is shown in McEwen’s results. He found that clipped males lost their heliotropism. It was also possible that the courtship of the normal male might make the female ready to copulate and then she would mate with either male. Sturtevant tested the last supposition by placing single pairs in vials, testing each day an equal number of normal and clipped males. The length of time before copulation was noted. The clipped male began to court as soon as the normal, but a larger number of normal males mated in the first 12 minutes than clipped males (50 to 25). Had the females discriminated against the clipped males to an equal extent we would have expected a much greater excess than 72 to 53 when they were in competition. It appears, then, that the wings are useful in shortening the time between the meeting of the individuals and copulation. The display acts, however, almost as favorably for the other male as for the exhibitor himself. The results show, therefore, that here an esthetic sense of the female need not be postulated, for she actually shows little preference when she has been brought to the point of mating between the male that aroused her and the other male that did not. This critical test puts the problem in a different relation from that which Darwin’s theory of female choice was meant to throw light upon.

The reverse experiment—a clipped and a normal female of the same age, size, etc.—showed that the mate did not discriminate between them, for in 52 first trials the normal female was paired with 25 times, the clipped 27 times.

PART III.

THE GENETIC AND THE OPERATIVE EVIDENCE.

The genetic evidence shows, in the case of cock-feathering versus hen-feathering in birds, that only one or two Mendelian factor differences are involved. The result may seem to mean that the secondary sexual characters themselves have been acquired historically by a single evolutionary step, and that in consequence the opportunity for selection to have accomplished such a result has been enormously facilitated. Such an argument rests, however, as we know to-day, on a false interpretation of Mendelian heredity. What the evidence really shows is that one or two genes if present cause the testes to produce some substance that prevents the cock-feathering from developing. The genetic complex may require a hundred or a thousand or more special factors that are directly and indirectly concerned with the development of the cock-feathering, but one or two other factors may suffice to block this machinery; or, to change the metaphor, these dominant factors may be no more than so much sand poured into the clock. The clock may have been slowly built up historically by many contributory “factors,” but a little sand may spoil its activity. Similarly in the hen something produced by the ovary prevents the fullest possible genetic action from taking place. Here at present we do not know whether a single factor or a hundred “special” factors are necessary to produce such an inhibition, but if, as one would like to suppose, it is the same or partly the same genes involved in the ovary, and in the testes of hen-feathered males, then a relatively few, one or two, factors will suffice to bar cock-feathering from the female.

In a case like the clover butterfly, where the genetic relations work out on the theory of one pair of factors that produce two types of females and one type of male, it seems more reasonable to infer that such a difference has not been slowly acquired by many smaller mutational changes, because the two types are not adapted to live under two different environments for which their differences fit them respectively, but to live in the same environment. It has never been claimed, so far as I know, that these two types of females have arisen through some males preferring one, some another kind of female, so that even although it may seem probable that the genetic situation is simple, the simplicity can not be turned to the advantage of the theory of sexual selection. It is unnecessary to discuss further the origin of the factor or factors suppressing the development of one type in the male or the probability of the multiplicity of such factors. In the case of such species as Papilio memnon and P. polytes, with three types of females, the situation is the same as above, with the addition of the theory of mimicry, that “explains” some advantage accruing to each type of female. Since the latter is only a form of natural selection, we are not further concerned with the change here. Punnett’s excellent treatment of the problems involved in his recent book on mimicry brings the subject down to date.

Meager as is the genetic and surgical evidence at present, it is enough to show that only by further work along these lines can we hope to lay a firm foundation for a scientific study of the subject. It is equally important that critical evidence be obtained in regard to the effect on the female of males of different types in competition. The instinctive reactions of animals in these respects, their first reaction, the associations that may or may not result, are practically an open field for investigation. The entire equipment of human psychology of the introspective school, that has been appealed to for help in a situation itself little understood, reads often more like fiction than like science.

So far as one branch of the subject goes—the possible interpretation of ornamentation in the male—there seem to be two ways at least in which the subject calls for immediate investigation: First, if it can be shown that, other things being equal, a more adorned male rouses the female to prompter mating, it may be inferred with some probability that in the long run such conduct would lead to the establishment of the more effective individual, but this would not be true unless the males mate, as a rule, more than once, for any advantage that might accrue to a more ornamented male would not affect the course of evolution of the species if every other male found a mate too. Second, if it could be shown that the special ornamentation of the male is only one of several effects of a gene whose main effect is in some other direction, then the advantage gained through natural selection in this other direction would carry in its wake the advance in ornamentation, and if the change affects one sex more than the other, owing to the difference in the genetic complex of the two sexes, it would be called a secondary sexual character.

A. Evidence from Mammals.

Owing to the differences in the secondary sexual characters of different breeds of sheep, we have more genetic information about such characters in this group than in other groups of mammals. Fortunately, also, in some of the breeds both castration and ovariotomy have been performed, and consequently we are in position to utilize both sources of information for interpreting the situation. In certain breeds both males and females have horns (Dorsets), in which case the horns of the male are larger than those of the female. In other breeds neither males nor females have horns (Suffolks). In still other breeds the males have horns and the females are hornless (Merinos and Herdwicks). The clearest evidence that we have, both genetic and operative, is that obtained by Woods, as reported by Bateson, in which horned (Dorsets) and hornless (Suffolks) breeds were crossed. In the Dorsets, where both sexes have horns, those of the male are larger than those in the female. When the young male is castrated the horns develop, but only as far as in the female. It appears, therefore, that the presence of the testis, probably through some secretion from it, contributes to the development of the horns. The other race, the Suffolks, have no horns in either sex. Castration produces no change in their hornless condition.

When a Dorset ram is crossed to a Suffolk ewe the sons have horns, the daughters lack them. The reciprocal cross gives the same results. The factor or factors involved are therefore not sex-linked. When the F₁’s from the cross or from its reciprocal are inbred, four classes of offspring are produced, namely: Horned male, 3; hornless male, 1; horned female, 1; hornless female, 3. The ratios, as above, are approximately 3:1:1:3.

A simple Mendelian explanation covers the results. If we assume that the Dorsets, both male and female, are homozygous in a factor for horns, H, that is not in the sex chromosome, and that the Suffolks “lack this factor,” i. e., that they have an allelemorphic factor for hornlessness, the germ-cells are H-H and h-h, respectively. Only one kind of individual, Hh, results in F₁. Since the male with this formula develops horns, we must conclude that the presence of the testis (through its secretions) causes horns to develop, while in the female of this same composition horns are not produced because of the absence of the testes. The sex-cells in these F₁ individuals are H-h and H-h. Chance meeting of these gametes will give 3 classes of individuals, irrespective of sex, namely, (1) HH, (2) Hh, (1) hh. The expectation for the males is that those of the composition (1) HH and (2) Hh will develop horns, while those of the composition hh will not develop horns. There should be 3 horned to 1 hornless male. In the females we expect those with the composition (1) HH to develop horns, since they have the same formula as the pure Dorset; those with the formula Hh are not expected to develop horns, because the F₁ females of this composition do not have horns; those with the formula hh are not expected to develop horns, because they have the same composition as have the pure Suffolk. There should be 3 hornless to 1 horned female. Combining both sexes, the expectation for F₂ is 4 horned to 4 hornless. Arranged according to sex, these give the classes realized: Horned male, 3; hornless male, 1; horned female, 1; hornless female, 3. That this is the correct explanation is borne out by back-crossing the hornless F₁ female to a hornless Suffolk ram. The former has two kinds of gametes, H and h, the latter only gametes that bear the h factor. Half the sons should be horned, half hornless, because half of them are Hh and half hh. But none of the daughters should have horns, because neither the Hh nor the hh females produce horns. This is the result realized, viz, 3 hornless offspring to 1 horned.

The preceding account of the inheritance of the factor for horns is based on the combination of Dorsets and Suffolks used by Wood. That other conditions may exist in other breeds and even in races of the same breed is claimed by Arkell as a result of a large number of crosses that he has carried out. He states, for instance, that in the great Merino class, with its various sub-breeds, there are flocks in which the males only are horned, but even here there may be individual males that are hornless “and at times the females may also show some signs of horn growth.” In America, Arkell states, there are three types of Merinos—the American, the Delaine, and the Rambouillet. He quotes Plumb (Types and Breeds of Farm Animals, Boston, 1906) as stating that “the American Merino ram carries heavy, spirally twisted horns, but the ewes are hornless; ... that the rams of the National Standard or Victor-Beald Delaines may or may not have horns; that the Dickinson Delaines may have small horns, but a polled head is preferred,” etc. These conditions suggest that there may be more than a single factor for horns in sheep or that there may be modifying factors in certain breeds. In fact, Arkell and Davenport attempt to cover the results of Arkell’s experiments by assuming that there is an inhibiting factor for horns that is carried by the sex chromosome. Such an inhibitor (I) would be double in the XX female and single in the X male. It is assumed to be incapable of preventing the development of horns in the heterozygous Hh male, the inhibitor being there simplex (i.e., one I), while the double inhibitor is capable of preventing the horns in the heterozygous (Hh) condition, but not of preventing the development of horns when the homozygous (HH) condition occurs. There are several objections to this scheme: first, that there is no evidence that a sex-linked inhibitor is present that affects the hornless breeds, for the evidence indicates rather that there is no factor for horns present in them, at least in the Suffolks; second, the peculiar balance between the factors for horns and the inhibitor seems an extremely artificial statement. Arkell and Davenport intimate that races with horned males and hornless females do not exist in a pure state. That breeds impure in these respects may exist need not be denied, but that pure races for such a dimorphic condition do exist seems probable. Castle states, for instance, that he knows at first hand of such races of Merinos. Castle also states that castrated Merino rams in this race do not develop horns, and this result is in accordance with statements made by Marshall for Herdwicks (a race with horned males and hornless females). Under the circumstances it is certain that the presence of the testes is one of the factors in determining whether horns develop at all (as in Merinos), or in determining the extent to which they develop (as in the Dorsets), rather than that the difference between the sexes is due only to an inhibiting genetic factor. Nevertheless, it may be well to keep open the possibility that there may be different factors for horns in different races (allelomorphs or others), or conversely, that the genetic composition of the races is different, the factor for horns remaining the same, but producing a different effect.

It may be pointed out in passing that if, as Arkell assumes, the hornless races are due to the presence in them of an inhibitor for horns, the results can be worked out without postulating that the inhibitor is sex-linked. For example, if the hornless male and female be HHII and the horned male and female HHii, the F₁ horned males and hornless females will be HHIi. The germ-cells will be HI and Hi in each sex, which, by chance meeting, as shown below, gives the results obtained by Wood. Thus:

These formulæ give 3 horned males, 1 hornless male, 1 horned female, 3 hornless females. This formulation, while appealing apparently to a different set of factors from those used by Arkell, is in reality the same in principle, since the heterozygous condition is here represented by Ii (instead of Hh) and sex determines that the heterozygous male is horned and the female hornless.

The genetic relations of the Merino with horned males and hornless females to the Dorsets, in which both sexes are horned (but in the male the horns are larger), must be different from the genetic relation in the other cross. There are two theoretical possibilities, viz., that a different factor for horns is present that is either an allelomorph or another different factor; or second, that a modifier is present in the Merino that keeps down the development of the horns in the female. An answer could be obtained by breeding Merinos to horned and to hornless and getting F₂ from both crosses. Arkell’s data is not sufficient to settle the question, because his numbers are often too small, but chiefly because it appears that there were two genetic types present in his flock of Merinos, one of which is characterized by scurs (very short horns) in the females, the other by hornlessness in the female. He found in a cross between a hornless father and Merino mother (that had knobs or scab-like growths) that the daughters had horns or scurs and carried a determiner for horns (as subsequent generations showed). On the other hand, in other cases where the Merino mother was without horns, her F₁ daughters had no horns. In both cases the F₁ sons had horns. Arkell cites this cross as “proving” that the knobs of Merino ewes depend for their development upon two horn determiners (H´H´). It is not at all evident that the results lead to such a conclusion, as other explanations will cover the case as well.

Arkell’s mating between Dorsets and Merinos (tables IX and XVI) corroborates his view “that the knob of the Merino female is represented in the germ-plasm by the double determiner.” The 5 F₁ sons had long horns, 3 F₁ daughters had horns present, and 2 had them absent (table XVI). If some of the Merino mothers used were homozygous for a factor that inhibits the development of horns in the female we can account for the hornless daughters, and if other mothers did not have this factor (or were heterozygous for it) we can account for the horned daughters. Evidently more evidence is needed. Arkell himself assigns a corresponding difference to the mothers in these cases, based on the observed fact that the mother that had knobs or scurs were the ones that gave birth to the horned daughters. If the above suggestion proves true, it shows that the Merino condition dominates the Dorset condition. The result is in harmony with the view that both have a common factor for horns, but that in addition the Merinos have a non-sex-linked modifier that holds down the development of the horns in the ewe.

What bearing have these results on the theory of sexual selection? Clearly the Merino male, as constituted at present, develops horns because he is a male, but only in the sense that his testes secrete some substance that makes his horns grow. That maleness does not in itself necessarily produce horn is shown by the absence of horns in the Suffolk breed. Is it the same factor, present in the Merino, that produces horns in both sexes of Dorsets when homozygous and in the male only when heterozygous? If originally the ancestral race had no horns, the appearance of factors for horns would, even in a heterozygous condition, have sufficed in the males for the development of horns. If this gave them any advantage either over the enemies of the race or in the eyes of the female, such factors might be perpetuated, and through transferrence to the females ultimately become homozygous in both sexes. Both would then have horns, whether horns were or were not of any advantage to the female, which would have them because they have an advantage to the other sex.

Because the genetic evidence shows that a single factor difference between the breeds with and without horns accounts for the horned condition in one of them, it by no means follows that horns as they exist arose as a single mutant factor change. True, they may have arisen as a new single factor difference, but the Mendelian evidence can not be claimed as evidence for this view. The a priori argument based on the relation of horns in an adaptive sense to the rest of the body would appear rather to indicate that they could not have arisen at a single mutational step.

Concerning the still broader bearing of this evidence on the theory of sexual selection, two distinct questions are involved: first, how has the present racial difference in horns arisen in domesticated sheep, and secondly, what was the original condition of sheep. Reversing the order of these questions, we find that sheep were domesticated in Asia and Europe before the dawn of history. “Whether our well-known and useful animal is derived from any one of the existing wild species, or from the crossing of several, or from some now extinct species, is quite a matter of conjecture” (Flower and Lydekker’s “Mammals”). Most of the wild species of the genus (of which about 12 are recognized) have horns in both sexes, but larger in the male. There are 3 wild species in which the horns are lacking in the female, according to Flower and Lydekker. If these have been crossed into the domesticated breeds the condition shown by the Merino may go back to the wild state. The third condition found in domesticated races, viz, hornlessness, may have appeared under domestication. Such a change might have arisen in either of the two other types and would be comparable to well-known losses of characters shown by domesticated animals and plants. These losses of characters are usually ascribed to actual losses of genes; any lost gene in the complex of factors necessary for the production of horns might cause such a change. But there is no advantage, in fact, in ascribing the loss in the character to a loss in one of the factors producing that character, for any change of any kind in the factor complex might bring about the same result and the evidence from multiple allelomorphs should put us on our guard against the all too easy assumption that a loss in a character involves necessarily loss of a factor in the real sense in which loss is used in ordinary speech.

The operative and genetic evidence for sheep shows that if the horns in the male were developed through natural or sexual selection we should expect them to develop also in the female. The greater development in the male seems to be due to secretions from the testes which probably are due to special factors that call them forth, but whether such factors were also acquired to reinforce the effects being produced through selection or were already present (reinforcement for horns being only a by-product of their activity) can not of course be known. We can suppose that special factors that suppress the development of horns in the female may have arisen in the wild or in the domesticated races and have been perpetuated because of some imagined benefit conferred; or that in certain races factors were already present that kept down the development of horns in the female. In any case such factors do not cause their effects through secretions from the ovary, because after ovariotomy horns do not develop; nor are they sex-linked factors. Any speculation as to how natural or sexual selection has brought about the evolution of the horns in sheep must reckon with the conditions imposed on such speculation by the preceding information. So far as I can see, it leaves the situation in this respect neither better nor worse off than before.

In deer the effects of castration are well known, but there is no genetic evidence to show the kind of factors involved, since no crosses have been made between species with differences in their horns. If the young male deer is castrated before the antlers have appeared, no horns develop. If castrated at the time when the antlers have begun to develop, incomplete or imperfect development follows. The antlers remain covered with the velvet, and are said not to be thrown off periodically as in the normal male. If the adult stag with antlers is castrated, the horns are precociously dropped, and, if replaced at all, the new antlers are imperfect and are not renewed. I do not know of any cases in which females have been spayed, but no doubt the ovaries must sometimes become diseased. There are, however, a few records of horns developing in this sex in old age, or presumably after disease of the ovaries. Both male and female reindeer are horned. Castration produces no effect on the development of the horns.

In the case of deer it is evident that the presence of the testes in the male causes the horns to develop. The genetic factor, or factors, for horns may be supposed to be carried by both sexes, but the effects of the factor can be seen only when the testes are present. In the reindeer and eland, on the other hand, the genetic factor for sex can produce horns without the need of the environment produced by the testes.[15] Whether we are dealing here with the same factor or whether the rest of the hereditary complex makes the result different can not be known without breeding experiments.

There is apparently a connection between the stage of development of the horns and the age of the animal, as the following statement by Yarrell[16] (1858) indicates: