When with the thoughts suggested in the last chapter we contemplate the problem of Evolution at large the hope at the present time of constructing even a mental picture of that process grows weak almost to the point of vanishing. We are left wondering that so lately men in general, whether scientific or lay, were so easily satisfied. Our satisfaction, as we now see, was chiefly founded on ignorance.

Every specific evolutionary change must represent a definite event in the construction of the living complex. That event may be a disturbance in the meristic system, showing itself in a change in the frequency of the repetitions or in the distribution of differentiation among them, or again it may be a chemical change, adding or removing some factor from the sum total.

If an attempt be made to apply these conceptions to an actual series of allied species the complexity of the problem is such that the mind is appalled. Ideas which in the abstract are apprehended and accepted with facility fade away before the concrete case. It is easy to imagine how Man was evolved from an Amoeba, but we cannot form a plausible guess as to how Veronica agrestis and Veronica polita were evolved, either one from the other, or both from a common form. We have not even an inkling of the steps by which a Silver Wyandotte fowl descended from Gallus Bankiva, and we can scarcely even believe that it did. The Wyandotte has its enormous size, its rose comb, its silver lacing, its tame spirit, and its high egg production. The tameness and the high egg production are probably enough both recessives, and though we cannot guess how the corresponding dominant factors have got lost, it is not very difficult to imagine that they were lost somehow. But the rose comb and the silver colour are dominants. The heavy weight also appears in the crosses with Leghorns, but we need not at once conclude that it depends on a simple dominant factor, because the big size of the crosses may be a consequence of the cross and may depend on other elements.

Now no wild fowl known to us has these qualities. May we suppose that some extinct wild species had them? If so, may we again make the same supposition in all similar cases? To do so is little gain, for we are left with the further problem, whence did those lost wild species acquire those dominants? Suppositions of this kind help no more than did the once famous conjecture as to the origin of living things—that perhaps they came to earth on a meteorite. The unpacking of an original complex, the loss of various elements, and the recombination of pre-existing materials may all be invoked as sources of specific diversity. Undoubtedly the range of possibilities thus opened up is large. It will even cover an immense number of actual examples which in practice pass as illustrations of specific distinction. The Indian Rock pigeon which has a blue rump may quite reasonably be regarded as a geographically separated recessive form of our own Columba livia, for as Staples-Browne has shown the white rump of livia is due to a dominant factor. The various degrees to which the leaves of Indian Cottons are incised have, as Leake says, been freely used as a means of classification. The diversities thus caused are very remarkable, and when taken together with diversities in habit, whether sympodial or monopodial, the various combinations of points of difference are sufficiently distinctive to justify any botanist in making a considerable number of species by reference to them alone. Nevertheless Leake's work goes far to prove that all of these forms represent the re-combinations of a very small number of factors. The classical example of Primula Sinensis and its multiform races is in fact for a long way a true guide as to the actual interrelations of the species which systematists have made. That they did make them was due to no mistake in judgment or in principle, but simply to the want of that extended knowledge of the physiological nature of the specific cases which we now know to be a prime necessity.

But will such analysis cover all or even most of the ordinary cases of specific diversity between near allies? Postponing the problem of the interrelations of the larger divisions as altogether beyond present comprehension, can we suppose, that in general, closely allied species and varieties represent the various consequences of the presence or absence of allelomorphic factors in their several combinations? The difficulty in making a positive answer lies in the fact that in most of the examples in which it has been possible to institute breeding experiments with a view to testing the question, a greater or less sterility is encountered. Where, however, no such sterility is met with, as for instance in the crosses made by E. Baur among the species of Antirrhinum there is every reason to think that the whole mass of differences can and will eventually be expressed in terms of ordinary Mendelian factors. Baur has for example crossed species so unlike as Antirrhinum majus and molle, forms differing from each other in almost every feature of organisation.[1] The F₂ generation from this cross presents an amazingly motley array of types which might easily if met with in nature be described as many distinct species. Yet all are fertile and there is not the slightest difficulty in believing that they can all be reduced to terms of factorial analysis.

If allowance be made for the complicating effects of sterility, is there anything which prevents us from supposing that such good species as those of Veronica or of any other genus comprising well-defined forms may not be similarly related? I do not know any reason which can be pointed to as finally excluding such a possibility. Nevertheless it has been urged with some plausibility that good species are distinguished by groups of differentiating characters, whereas if they were really related as the terms of a Mendelian F₂ family are, we should expect to find not groups of characters in association, but rather series of forms corresponding to the presence and absence of the integral factors composing the groups of characters. I am not well enough versed in systematic work to be able to decide with confidence how much weight should be attached to this consideration. Some weight it certainly has, but I cannot yet regard it as forming a fatal objection to the application of factorial conceptions on the grand scale. It may be recalled that we are no longer under any difficulty in supposing that differences of all classes may be caused by the presence or absence of factors. It seemed at first for example that such characters as those of leaf shape might be too subtle and complex to be reducible to a limited number of factors. But first the work of Gregory on Primula Sinensis showed that several very distinct types of leaves were related to each other in the simplest way. In that particular example, intermediates are so rare as to be negligible, but subsequently Shull dealing with such a complicated example as Capsella, and Leake in regard to Cottons, both forms in which intergrades occur in abundance, have shown that a simple factorial scheme is applicable. We need not therefore, to take an extreme case, doubt that if it were possible to examine the various forms of fruit seen in the Squashes by really comprehensive breeding tests, even this excessive polymorphism in respect of structural features would be similarly reducible to factorial order.

It must always be remembered also that in a vast number of cases, nearly allied forms which are distinct, occupy distinct ground. Moreover, by whatever of the many available mechanisms that end be attained, it is clear that nature very often does succeed in preventing intercrossing between distinct forms so far that the occurrence of that phenomenon is a rarity under natural conditions. The facts may, I think, fairly be summarized in the statement that species are on the whole distinct and not intergrading, and that the distinctions between them are usually such as might be caused by the presence, absence, or inter-combination of groups of Mendelian factors; but that they are so caused the evidence is not yet sufficient to prove in more than a very few instances.

The alternative, be it explicitly stated, is not to return to the view formerly so widely held, that the distinctions between species have arisen by the accumulation of minute or insensible differences. The further we proceed with our analyses the more inadequate and untenable does that conception of evolutionary change become. If the differences between species have not come about by the addition or loss of factors one at a time, then we must suppose that the changes have been effected by even larger steps, and variations including groups of characters, must be invoked.

That changes of this latter order are really those by which species arise, is the view with which de Vries has now made us familiar by his writings on the Mutation Theory. In so far as mutations may consist in meristic changes of many kinds and in the loss of factors it is unnecessary to repeat that we have abundant evidence of their frequent occurrence. That they may also more rarely occur by the addition of a factor we are, I think, compelled to believe, though as yet the evidence is almost entirely circumstantial rather than direct. The evidence for the occurrence of those mutations of higher order, by which new species characterized by several distinct features are created, is far less strong, and after the best study of the records which I have been able to make, I find myself unconvinced. The facts alleged appear capable of other interpretations.

The most famous and best studied examples are of course the forms of Oenothera raised by de Vries from Oenothera Lamarckiana in circumstances well known to all readers of genetic literature. Whatever be the true significance of these extraordinary "mutations" there can be no question about the great interest which attaches to them, and the historical importance which they will long preserve. Apart also from these considerations it is becoming more and more evident that in their peculiarities they provide illustrations of physiological phenomena of the highest consequence in the study of genetics at large.

De Vries found, as is well known, that Oenothera Lamarckiana gives off plants unlike itself. These mutational forms are of several distinct and recognizable types which recur, and several of them breed true from their first appearance. The obvious difficulty, which in my judgment should make us unwilling at present to accept these occurrences as proof of the genesis of new species by mutation, is that we have as yet no certainty that the appearance of the new forms is not an effect of the recombination of factors, such as is to be seen in so many generations of plants derived from a cross involving many genetic elements. The first question is what is Oenothera Lamarckiana? Is it itself a plant of hybrid origin? To this fundamental question no satisfactory answer has yet been given. All attempts to find it as a wild plant in America have failed. It existed in Europe in the latter half of the eighteenth century. Whence it came is still uncertain, but the view that it came into existence in Europe and perhaps in Paris, seems on the whole the most probable. The question has been debated by Macdougal, Gates, and Davis. From historical sources there is little expectation of further light. Those who favour the notion of a hybrid origin look on Oenothera biennis as one of the putative parents. It has been conjectured that a species called grandiflora lately re-discovered on the Alabama river was the other parent. Experiments have been instituted by Davis to discover whether Lamarckiana can be made artificially by crossing these two species. The results so far have shown that while plants approximating in various respects to Lamarckiana have thus been produced, none agree exactly with that form. Davis, to whom reference should be made for a full account of the present state of the enquiry, points out that there are many strains of biennis in existence and that it is by no means impossible that by using others of these strains a still closer approximation can be made. None of Davis's artificial productions as yet breed at all true, as Lamarckiana on the whole does. In such a case, however, where several characters are involved, this is perhaps hardly to be expected.

One feature of the Oenotheras is very curious. Not only Lamarckiana, but all the allied species so far as I am aware, have a considerable proportion of bad and shrivelled pollen grains. This is undoubtedly true of species living in the wild state as well as of those in cultivation. I have had opportunities of verifying this for myself in the United States. No one looking at the pollen of an Oenothera would doubt that it was taken from some hybrid plant exhibiting partial sterility. On the other hand, it is difficult to suppose that numbers, perhaps all, of the "species" of the genus are really hybrids, and many of them breed substantially true. I regard this constant presence of bad pollen grains as an indication that the genetic physiology of Oenothera is in some way abnormal, and as we shall presently see, there are several other signs which point in the same direction.

Discussion of the whole series of phenomena is rendered exceedingly difficult first, by reason of the actual nature of the material. The characteristics of many of the types which de Vries has named are evasive. A few of these types, for instance, gigas, nanella, albida, brevistylis, and perhaps a few more are evidently clear enough, but we have as yet no figures and descriptions precise enough to enable a reader to appreciate exactly the peculiarities of the vast number of forms which have now to be considered in any attempt to gain a comprehensive view of the whole mass of facts. It is also not in dispute that the forms are susceptible of great variations due simply to soil and cultural influences.

The fact that no Mendelian analysis has yet been found applicable to this group of Oenotheras as a whole is perhaps largely due to the fact that until recently such analysis has not been seriously attempted. Following the system which he had adopted before the rediscovery of Mendelism, or at all events, before the development of that method of analysis, de Vries has freely applied names to special combinations of characters and has scarcely ever instituted a factorial analysis. Before we can get much further this must be attempted. It may fail, but we must know exactly where and how this failure comes about. There are several indications that such a recognition of factorial characters, could be carried some way. For example, the height, the size of the flowers, the crinkling of the leaves, the brittleness of the stems, perhaps even the red stripes on stems and fruits, and many more, are all characters which may or may not depend on distinct factors, but if such characters are really transmitted in unresolved groups, the limitations of those groups should be carefully determined. The free use of names for the several forms, rather than for the characters, has greatly contributed to deepen the obscurity which veils the whole subject.

I do not mean to suggest that these Oenotheras follow a simple Mendelian system. All that we know of them goes to show that there are curious complications involved. One of these, probably the most important of all, has lately been recognized by de Vries himself, namely, that in certain types the characters borne by the female and the male germ-cells of the same plant are demonstrably different. There can be little doubt that further research will reveal cognate phenomena in many unsuspected places. The first example in which such a state of things was proved to exist is that of the Stocks investigated by Miss Saunders.[2] By a long course of analysis she succeeded in establishing in 1908 the fact that if a plant of Matthiola is of that eversporting kind which gives a large proportion of double-flowered plants among its offspring (produced by self-fertilisation), then the egg-cells of such a plant are mixed in type, but the pollen of the same plant is homogeneous. Some of the egg-cells have in them the two factors for singleness, but some of them are short of one or both of these factors. The pollen-grains, however, are all recessives, containing neither of these factors. The egg-cells, in other words, are mixed, "singles" and "doubles," while the pollen-grains are all "doubles." The same is true of the factor differentiating "white," or colourless plastids from cream-coloured plastids in Matthiola, the egg-cells being mixed "whites" and "creams," while the pollen-grains are all "creams," viz: recessives. Later in the same year (1908) de Vries[3] announced a remarkable case which will be discussed in detail subsequently. It relates to certain Oenotheras heterozygous for dwarfness, in which (p. 113) the ovules were mixed, tails and dwarfs, while the pollen is all dwarf.

Again in Petunia Miss Saunders's[4] work has shown that a somewhat similar state of things exists, but with this remarkable difference, that though the egg-cells are mixed, singles and doubles, the pollen-grains are all singles, viz: dominants. All the Petunias yet examined have been in this condition, including some which in botanic gardens pass for original species. Whether actual wild plants from their native habitats are in the same state, is not yet known, but it is by no means improbable. The case may be compared with that of the moth Abraxas grossulariata studied by Doncaster and Raynor, in which the females are all heterozygous, or we may almost say "hybrids" of grossulariata and the variety lacticolor. Similarly we may say that at least garden Petunias are heterozygous in respect of singleness. The proof of this is of course that when fertilised with the pollen of doubles they throw a mixture of doubles and singles. The statements which de Vries has published regarding the behaviour of several of the Oenotheras go far to show that they must have a somewhat similar organisation. On the present evidence it is still quite impossible to construct a coherent scheme which will represent all the phenomena in their interrelations, and among the facts are several which, as will appear, seem mutually incompatible. The first indication that the Oenotheras may have either mixed ovules or mixed pollen appears in the fact that Lamarckiana and several of its "mutants" used as males, with several other forms as females, give a mixed offspring. For example, de Vries (1907) found that

biennis ? × Lamarckiana ?
biennis cruciata ? × Lamarckiana ?
muricata ? × Lamarckiana ?
biennis ? × rubrinervis ?
biennis cruciata ? × rubrinervis ?

all give a mixture of two distinct types which he names laeta and velutina, consisting of about equal numbers of each. On account of the fact that the two forms are produced in association de Vries has called these forms "twin hybrids," a designation which is not fortunate, seeing that it is impossible to imagine that any kind of twinning is concerned in their production. The distinction between these two seems to be considerable, laeta having leaves broader, bright green in colour, and flat, with pollen scanty, while velutina has leaves narrower, grayish green, more hairy, and furrow-shaped, with pollen abundant.

We next meet the remarkable fact that these two forms, laeta and velutina breed true to their respective types, and do not reproduce the parent-types among their offspring resulting from self-fertilisation. This statement must be qualified in two respects. When muricata ? is fertilised by brevistylis the forms laeta and velutina are produced, but each of them subsequently throws the short-styled form as a recessive (de Vries, 1907, p. 406). It may be remembered that de Vries's previous publications had already shown that the short style of brevistylis, one of the Lamarckiana "mutants," behaves as a recessive habitually (Mutationstheorie, II, p. 178, etc.).

Also when nanella, the dwarf "mutant" of Lamarckiana is used as male on muricata as female, laeta and velutina are produced, but one only of these, namely, velutina, subsequently throws dwarfs on self-fertilisation. The dwarfs thus thrown are said to form about 50 per cent. of the families in which they occur (de Vries, 1908, p. 668). The fact that the two forms, laeta and velutina, are produced by many matings in which Lamarckiana and its mutant rubrinervis are used as males is confirmed abundantly by Honing, who has carried out extensive researches on the subject. After carefully reading his paper, I have failed to understand the main purport of the argument respecting the "double nature" of Lamarckiana which he founds on these results, but I gather that in some way laeta is shown to partake especially of the nature of Lamarckiana, while velutina is a form of rubrinervis. The paper contains many records which will be of value in subsequent analysis of these forms.

Before considering the possible meaning of these facts we must have in our minds the next and most novel of the recent extensions of knowledge as to the genetic properties of the Oenotheras. In the previous statement we have been concerned with the results of using either Lamarckiana itself or one of its "mutants" rubrinervis, brevistylis, or nanella as male, on one of the species biennis or muricata. The new experiments relate to crosses between the two species biennis and muricata themselves.

De Vries found:

1. That the reciprocal hybrids from these two species differed, biennis × muricata producing one type of F₁ and muricata × biennis producing another. Each F₁ resembled the father more than the mother.

2. That each of the hybrids so produced breeds true on self-fertilisation.

3. That if we speak of the hybrid from biennis × muricata as BM and of the reciprocal as MB, then

BM × MB

gives exclusively offspring of biennis type but that

MB × BM

gives exclusively offspring of muricata type. Evidently, apart from all controversy as to the significance of the "mutants" of Lamarckiana, we have here a series of observations of the first importance.

The fact that reciprocal crossings give constantly distinct results must be taken to indicate that the male and female sides of one, if not of both, of the parents are different in respect of characters which they bear. This is de Vries's view, and he concludes rightly, I think, that the evidence from all the experiments shows that both biennis and muricata are in this condition, having one set of characters represented in their pollen-grains and another in their ovules. The plants breed true, but their somatic structures are compounded of the two sets of elements which pass into them from their maternal and paternal sides respectively. This possibility that species may exist of which the males really belong to one form and the females to another, is one which it was evident from the first announcement of the discovery of Mendelian segregation might be found realised in nature.[5]

Oe. biennis and muricata were crossed reciprocally with each other and with a number of other species, and the behaviour of each, when used as mother, was consistently different from its behaviour when used as father. De Vries is evidently justified by the results of this series of experiments in stating that the "Bild," as he terms it, or composition of the male and female sides of these two species, biennis and muricata, are distinct. On the evidence before us it is not, however, possible to form a perfectly clear idea of each, and until details are published, a reader without personal knowledge of the material cannot do more than follow the general course of the argument. For fuller comprehension a proper analysis of the characters with a clear statement of how they are distributed among the several types and crosses is absolutely necessary. According to de Vries the female of biennis possesses a group of characters which he defines as "conica" in allusion to the shape of the flower-buds. Besides the conical buds, this group of features includes imperfect development of wood, rendering the plant very liable to attacks of Botrytis, and comparatively narrow leaves.

The female of muricata carries a group of features which he calls "frigida," and, though this is not quite explicitly stated in a definition of that type, it is to be inferred[6] that its characteristics are regarded as greater height, strong development of wood with comparative resistance to Botrytis, and broad leaves.

The characters borne by the male parts of the two species are in general those by which they are outwardly distinguished. For example, the leaves of Oe. biennis are comparatively broad and are bright green, while those of muricata are much narrower and of a glaucous green, and I understand that de Vries regards these properties as contributed by the male side in each case and to be carried by the male cells of each species. The suggestion as regards biennis and muricata comes near the conception often expressed by naturalists in former times (e. g., Linnaeus) and not rarely entertained by breeders at the present day, that the internal structure is contributed by the mother and the external by the father.

On the other hand, the offspring of each species when used as mother is regarded as possessing in the main the features of the maternal "Bild," but the matter is naturally complicated by the introduction of features from the father's side, and it is here especially that the account provided is at present unsatisfactory and inconclusive. There seems, however, to be no serious doubt that biennis and muricata each in their outward appearance exhibit on the whole the features which their pollens respectively carry, and that the features borne by their ovules are in many respects distinct.

The types are thus "hybrids" which breed true. The results of intercrossing them each way are again "hybrids" which breed true. It will be remembered that on former occasions de Vries has formulated a general rule that species-hybrids breed true, but that the cross-breds raised by interbreeding varieties do not. One of these very cases was quoted[7] as an illustration of thisprinciple, viz: muricata × biennis. The grounds for this general statement have always appeared to me insufficient, and with the further knowledge which the new evidence provides we are encouraged to hope that when a proper factorial analysis of the types is instituted we shall find that the phenomenon of a constant hybrid will be readily brought into line with the systems of descent already worked out for such cases as that of the Stocks, and others already mentioned.

In further discussion of these facts de Vries makes a suggestion which seems to me improbable. Since the egg-cells of muricata, for instance, bear a certain group of features which are missing on the male side, and conversely the pollen bears features absent from the female side, he is inclined to regard the bad pollen grains as the bearers of the missing elements of the male side and to infer that there must similarly be defective ovules representing the missing elements of the female side. No consideration is adduced in support of this view beyond the simple fact that the characters borne by male and female are dissimilar, whereas it would be more in accord with preconception if the same sets of combinations were represented in each—as in a normal Mendelian case. There is as yet no instance in which the absence of any particular class of gametes has been shown with any plausibility to be due to defective viability, though there are, of course, cases in which certain classes of zygotes do not survive owing to defective constitution (e. g., the albinos of Antirrhinum studied by Baur, and the homozygous yellow mice). I am rather inclined to suppose that in these examples of hybrids breeding true we shall find a state of things comparable with that to which we formerly applied the terms "coupling" and "repulsion." In these cases certain of the possible combinations of factors occur in the gametic series with special frequency, being in excess, while the gametes representing other combinations are comparatively few. In a recent paper on these cases Professor Punnett and I have shown that these curious results vary according to the manner in which the factors are grouped in the parents. If A and B are two factors which exhibit these phenomena we find that the gametic series of the double heterozygote differs according as the combination is made by crossing AB × ab, or by crossing AB × aB. In a normal Mendelian case the F₁ form, AaBb, produces gametes AB, Ab, aB, ab, in equal numbers; but in these peculiar cases those gametes which contain

the parental combinations are in excess. This excess almost certainly follows the system indicated by the accompanying table. In the general expressions n is half the number of gametes required to express the whole system. Now if we imagine that sex-factors are involved with the others concerned in such a relationship as this we have a system of distribution approximating to that found in biennis and muricata. The difference in reciprocals is represented in a not improbable way. It cannot yet be said that the rarer terms in the series are formed at all, and perhaps they are not. As we pointed out in our discussion of these phenomena, the peculiar distribution of factors in these cases must be taken to mean that the planes of division at some critical stage in the segregation are determined with reference to the parental groups of factors, or in other words, that the whole system has a polarity, and that the distribution of factors with reference to this polarity differs according to the grouping of factors in the gametes which united in fertilization to produce the plant. Subsequent proliferation of cells representing certain combinations would then lead to excess of the gametes bearing them. It is on similar lines that I anticipate we shall hereafter find the interpretation of the curious facts discovered by de Vries, though it is evident that a long course of experiment and analysis must be carried through before any certainty is reached. The work must be begun by a careful study of the descent of some single factor, for example, that causing the broader leaf of biennis, and we may hope that the study of Oenothera by proper analytical methods will no longer be deferred.

We have now to return to the relations of laeta and velutina. These two forms, it will be remembered are frequently produced when Lamarckiana or one of its derivatives is used as male, and the most unexpected feature in their behaviour is that both breed true as regards their essential characteristics, on self-fertilisation. If one only bred true the case might, in view of the approximate numerical equality of the two types, be difficult to interpret on ordinary lines, but as both breed true it must be clear that some quite special system of segregation is at work. What this may be cannot be detected on the evidence, but with the results from the biennis-muricata experiments before us, it is natural to suspect that we may here again have to recognise a process of allocation of different factors to the male and female sides in laeta and velutina. That some such system is in operation becomes the more probable from the new fact which de Vries states in describing the group of characters which he calls conica, namely that this type is the same as that of velutina.

There are many collateral observations recorded both by de Vries and others which have a bearing on the problems, but they do not yet fall into a coherent scheme. For example, we cannot yet represent the formation of laeta and velutina from the various species fertilised by Lamarckiana ?. That this is not due to any special property associated with the pollen of Lamarckiana is shown by the fact that a species called Hookeri gives laeta and velutina in both its reciprocal crosses with Lamarckiana (de Vries, 1909, p. 3), and also by the similar fact that Lamarckiana ? fertilised by the pollen of a peculiar race of biennis named biennis Chicago throws the same types. Before these very complicated phenomena can be usefully discussed particulars must be provided as to the individuality of the various plants used. This criticism applies to much of the work which de Vries has lately published, for, as we now know familiarly, plants to which the same name applies can be quite different in genetic composition.

Attention should also be called to one curiously paradoxical series of results. When the dwarf "mutant" of Lamarckiana which de Vries names "nanella" is used as father on muricata, F₁ consists of laeta and velutina in approximately equal numbers. Both forms breed true to their special characteristics, but velutina throws dwarfs of its own type, while laeta does not throw dwarfs. Subsequent investigation of the properties of these types has led to some remarkable conclusions, and it was in a study of these plants that de Vries first came upon the phenomena of dissimilarity between the factors borne by the male and female cells of the same plant, a condition which had been recently detected in the Stocks as a result of Miss Saunders's investigations. The details are very remarkable. We have first the fact that muricata ? × dwarf nanella ? gives about 50 per cent. laeta and about 50 per cent. of velutina.

As regards Velutina it was shown that:

The three experiments taken together prove, as de Vries says, that the ovules of velutina are mixed, talls and dwarfs, and that the pollen is all dwarf. The condition is almost the same as that of the Stocks. It may be noted also that in the Stocks the egg-cells of the "double" type are in excess, being approximately 9 to 7 of the "single" type, but de Vries regards the two types in velutina as probably equal in number. The figures (169:231) rather suggest some excess of the recessives, perhaps 9:7, and the point would be worth a further investigation.

As regards laeta, by self-fertilisation no dwarfs were produced, but in all other respects it behaved almost exactly like velutina. The ovules are evidently mixed talls and dwarfs, and whether fertilised by dwarfs or by the pollen of velutina, which is already proved to be all dwarf, the result was a steady 50 per cent. of talls and 50 per cent. of dwarfs. The pollen of laeta used on dwarfs gives nothing but dwarfs, and in three series of such experiments 226 dwarfs were produced.

We are thus faced with this difficulty. Since the egg-cells of laeta are evidently mixed, talls and dwarfs, and the pollen used on dwarfs gives all dwarfs, why does not self-fertilisation give a mixed result, talls and dwarfs, instead of all talls? De Vries regards the result of self-fertilisation as showing the real nature of the pollen, and declares it to be all talls, while he represents the behaviour of the same pollen used on dwarfs by stating that in these combinations the dwarf character dominates. This does not seem to me a natural interpretation. I should regard the pollen of laeta as identical with that of velutina, namely dwarf, and I suspect the difficulty is really created by the behaviour of laeta on self-fertilisation. Until a proper analysis is made in which the identity of the different individuals used is recorded, no further discussion is possible.[8]

Other results of a complicated kind involving production of laeta and velutina together with a third form have been published by de Vries in his paper on "Triple Hybrids." To these also the same criticism applies. Some of the observations seem capable of simple factorial representation and others are conflicting.

Taking the work on Oenothera as a whole we see in it continually glimpses of order which further on are still blocked by difficulties and apparent inconsistencies. Through such a stage all the successful researches in complicated factorial analysis have passed and I see no reason for supposing that with the application of more stringent methods this more difficult set of problems will be found incapable of similar solutions. To return to the original question whether in Oenothera we can claim to see a special contemporaneous output of new species in actual process of creation, it will be obvious that while the interrelation of the several types is still so little understood, such a claim has no adequate support. It is true that many of the "mutants" of Lamarckiana can well pass for species, but this is equally true of many new combinations of pre-existing factors as we have seen in Primula Sinensis and other cases. Still less can it be admitted that these facts of uncertain import supply a justification for the conception which has played a prominent part in the scheme of the Mutationstheorie, namely that there are special periods of Mutation, when the parent-species has peculiar genetic properties. To conclude: The impression which the evidence leaves most definitely on the mind is that further discussion of the bearing which the Oenotheras may have on the problem of evolution should be postponed until we have before us the results of a searching analysis applied to a limited part of the field. In such an analysis it is to be especially remembered that we have now a new clue in the well-ascertained fact that the genetic composition of the male and female germ-cells of the same individual may be quite different. When with this possibility in view the behaviour of the types is re-examined I anticipate that many of the difficulties will be removed.

Outside the evidence from Oenothera, which, as we have seen, is still ambiguous, I know no considerable body of facts favourable to that special view of Mutation which de Vries has promulgated. Of variation, or if we will, Mutation, in respect of some one character, or resulting from recombination, there is proof in abundance; but of that simultaneous variation in several independent respects to which de Vries especially attributes the origin of new specific types I know only casual records which have yet to undergo the process of criticism.

Besides de Vries's "Mutationstheorie" and "Species and Varieties" the chief publications relating to the subject of the behaviour of Oenothera are the following: (Many other papers relating especially to the cytology of the forms have appeared.)

Davis, B. M.
Genetical Studies on Oenothera, I. Amer. Nat., XLIV, 1910, p. 108.
Genetical Studies on Oenothera, II. Ibid., XLV, 1911, p. 193.
Gates, R. R.
An Analytical Key to some of the Segregates of Oenothera.
Twentieth Annual Report of the Missouri Botanical Garden, 1909.
Studies on the Variability and Heritability of Pigmentation in Oenothera.
Ztsch. f. Abstammungslehre, 1911, IV, p. 337.
Honing, J. A.
Die Doppelnatur der Oenothera Lamarckiana.
Ztsch. f. Abstammungslehre, 1911, IV, p. 227.
Macdougal, D. T. (with A. M. Vail, G. H. Shull, and J. K. Small).
Mutants and Hybrids of the Oenotheras.
Carnegie Institution's Publication, No. 24, 1905.
Macdougal, D. T., Vail, A. M., Shull, J. H.
Mutations, Variations and Relationships of the Oenotheras.
Carnegie Institution's Publication, No. 81, 1907.
de Vries, H.
On Atavistic Variation in Oenothera cruciata.
Bull. Torrey Club, 1903, Vol. 30, p. 75.
On Twin Hybrids,
Bot. Gaz., Vol. 44, 1907, p. 401.
Ueber die Zwillingsbastarde von Oenothera nanella.
Ber. Deut. Bot. Ges., 1908, XXVI, a, p. 667.
Bastarde von Oenothera gigas. Ibid., p. 754.
On Triple Hybrids. Bot. Gaz., 1909, Vol. 47, p. 1.
Ueb. doppeltreziproke Bastarde von Oenothera biennis L. und Oenothera muricata L.
Biol. Cbltt., 1911, XXXI, p. 97.
Zeijlstra, H. H.
Oenothera nanella de Vries, eine krankhafte Pflanzenart.
Biol. Cbltt., 1911, XXXI, p. 129.

Note.

Since this chapter was written two contributions of special importance have been made to the study of the Oenothera problems. The first is that of Heribert-Nilsson.[9] The author begins by giving a critical account of the evidence for de Vries's interpretation of the nature of the mutants. In general this criticism pursues lines similar to those sketched in the foregoing chapter, concluding, as I have done, that the chief reason why factorial analysis has been declared to be inapplicable to the Oenothera mutants is because no one has hitherto set about this analysis in the right way. He has also himself made a valuable beginning of such an analysis and gives good evidential reasons for the belief that at least the red veining depends on a definite factor which also influences the size of certain parts of the plant. He argues further that many of the distinctions between the mutants are quantitative in nature. With great plausibility he suggests that the system of cumulative factors which Nilsson-Ehle discovered in the case of wheat (subsequently traced by East in regard to maize) may be operating also in these Oenotheras. According to this system several factors having similar powers may coexist in the same individual, and together produce a cumulative effect. Scope would thus be given for the production of the curious and seemingly irregular numbers so often recorded in the "mutating" families.

Another remarkable observation relating to the crosses of muricata and biennis has been published by Goldschmidt.[10] He finds that in the formation of this cross the female pronucleus takes no part in the development of the zygotic cell, but that when the male pronucleus enters, the female pronucleus is pushed aside and degenerates. As de Vries observed, the reciprocal hybrids are in each case very like the father ("stark patroklin"), a consequence which finds a natural explanation in the phenomenon witnessed by Goldschmidt. The results of the subsequent matings can also be readily interpreted on the same lines. Indications of maternal characters are nevertheless mentioned by de Vries, and if Goldschmidt's account of the cytology is confirmed, these must presumably be referred to the influence of the maternal cytoplasm. Clearly this new work opens up lines of exceptional interest. The interpretation I have offered above must probably be reconsidered. The distinction between the male and female cells of the types may no doubt be ultimately factorial, but it is difficult to regard such a distinction as created by a differential distribution of the ordinary factors.