A DEFENCE OF MENDEL’S PRINCIPLES OF HEREDITY.

“The most fertile men of science have made blunders, and their consciousness of such slips has been retribution enough; it is only their more sterile critics who delight to dwell too often and too long on such mistakes.” Biometrika, 1901.

Introductory.

On the rediscovery and confirmation of Mendel’s Law by de Vries, Correns, and Tschermak two years ago, it became clear to many naturalists, as it certainly is to me, that we had found a principle which is destined to play a part in the Study of Evolution comparable only with the achievement of Darwin—that after the weary halt of forty years we have at last begun to march.

If we look back on the post-Darwinian period we recognize one notable effort to advance. This effort—fruitful as it proved, memorable as it must ever be—was that made by Galton when he enuntiated his Law of Ancestral Heredity, subsequently modified and restated by Karl Pearson. Formulated after long and laborious inquiry, this principle beyond question gives us an expression including and denoting many phenomena in which previously no regularity had been detected. But to practical naturalists it was evident from the first that there are great groups of facts which could not on any interpretation be brought within the scope of Galton’s Law, and that by no emendation could that Law be extended to reach them. The existence of these phenomena pointed to a different physiological conception of heredity. Now it is precisely this conception that Mendel’s Law enables us to form. Whether the Mendelian principle can be extended so as to include some apparently Galtonian cases is another question, respecting which we have as yet no facts to guide us, but we have certainly no warrant for declaring such an extension to be impossible.

Whatever answer the future may give to that question, it is clear from this moment that every case which obeys the Mendelian principle is removed finally and irretrievably from the operations of the Law of Ancestral Heredity.

At this juncture Professor Weldon intervenes as a professed exponent of Mendel’s work. It is not perhaps to a devoted partisan of the Law of Ancestral Heredity that we should look for the most appreciative exposition of Mendel, but some bare measure of care and accuracy in representation is demanded no less in justice to fine work, than by the gravity of the issue.

Professor Weldon’s article appears in the current number of Biometrika, Vol. I. Pt. II. which reached me on Saturday, Feb. 8. The paper opens with what purports to be a restatement of Mendel’s experiments and results. In this “restatement” a large part of Mendel’s experiments—perhaps the most significant—are not referred to at all. The perfect simplicity and precision of Mendel’s own account are destroyed; with the result that the reader of Professor Weldon’s paper, unfamiliar with Mendel’s own memoir, can scarcely be blamed if he fail to learn the essence of the discovery. Of Mendel’s conception of the hybrid as a distinct entity with characters proper to itself, apart from inheritance—the most novel thing in the whole paper—Professor Weldon gives no word. Upon this is poured an undigested mass of miscellaneous “facts” and statements from which the reader is asked to conclude, first, that a proposition attributed to Mendel regarding dominance of one character is not of “general”52 application, and finally that “all work based on Mendel’s method” is “vitiated” by a “fundamental mistake,” namely “the neglect of ancestry53.”

To find a parallel for such treatment of a great theme in biology we must go back to those writings of the orthodox which followed the appearance of the “Origin of Species.”

On 17th December 1900 I delivered a Report to the Evolution Committee of the Royal Society on the experiments in Heredity undertaken by Miss E. R. Saunders and myself. This report has been offered to the Society for publication and will I understand shortly appear. In it we have attempted to show the extraordinary significance of Mendel’s principle, to point out what in his results is essential and what subordinate, the ways in which the principle can be extended to apply to a diversity of more complex phenomena—of which some are incautiously cited by Professor Weldon as conflicting facts—and lastly to suggest a few simple terms without which (or some equivalents) the discussion of such phenomena is difficult. Though it is impossible here to give an outline of facts and reasoning there set out at length, I feel that his article needs an immediate reply. Professor Weldon is credited with exceptional familiarity with these topics, and his paper is likely to be accepted as a sufficient statement of the case. Its value will only be known to those who have either worked in these fields themselves or have been at the trouble of thoughtfully studying the original materials.

The nature of Professor Weldon’s article may be most readily indicated if I quote the summary of it issued in a paper of abstracts sent out with Review copies of the Part. This paper was most courteously sent to me by an editor of Biometrika in order to call my attention to the article on Mendel, a subject in which he knew me to be interested. The abstract is as follows.

“Few subjects have excited so much interest in the last year or two as the laws of inheritance in hybrids. Professor W. F. R. Weldon describes the results obtained by Mendel by crossing races of Peas which differed in one or more of seven characters. From a study of the work of other observers, and from examination of the ‘Telephone’ group of hybrids, the conclusion is drawn that Mendel’s results do not justify any general statement concerning inheritance in cross-bred Peas. A few striking cases of other cross-bred plants and animals are quoted to show that the results of crossing cannot, as Mendel and his followers suggest, be predicted from a knowledge of the characters of the two parents crossed without knowledge of the more remote ancestry.”

Such is the judgment a fellow-student passes on this mind

“Voyaging through strange seas of thought alone.”

The only conclusion which most readers could draw from this abstract and indeed from the article it epitomizes, is that Mendel’s discovery so far from being of paramount importance, rests on a basis which Professor Weldon has shown to be insecure, and that an error has come in through disregard of the law of Ancestral Heredity. On examining the paper it is perfectly true that Professor Weldon is careful nowhere directly to question Mendel’s facts or his interpretation of them, for which indeed in some places he even expresses a mild enthusiasm, but there is no mistaking the general purpose of the paper. It must inevitably produce the impression that the importance of the work has been greatly exaggerated and that supporters of current views on Ancestry may reassure themselves. That this is Professor Weldon’s own conclusion in the matter is obvious. After close study of his article it is evident to me that Professor Weldon’s criticism is baseless and for the most part irrelevant, and I am strong in the conviction that the cause which will sustain damage from this debate is not that of Mendel.

I. The Mendelian Principle of Purity of Germ-Cells and the Laws of Heredity Based on Ancestry.

Professor Weldon’s article is entitled “Mendel’s Laws of Alternative Inheritance in Peas.” This title expresses the scope of Mendel’s work and discovery none too precisely and even exposes him to distinct misconception.

To begin with, it says both too little and too much. Mendel did certainly determine Laws of Inheritance in peas—not precisely the laws Professor Weldon has been at the pains of drafting, but of that anon. Having done so, he knew what his discovery was worth. He saw, and rightly, that he had found a principle which must govern a wide area of phenomena. He entitles his paper therefore “Versuche über Pflanzen-Hybriden,” or, Experiments in Plant-Hybridisation.

Nor did Mendel start at first with any particular intention respecting Peas. He tells us himself that he wanted to find the laws of inheritance in hybrids, which he suspected were definite, and that after casting about for a suitable subject, he found one in peas, for the reasons he sets out.

In another respect the question of title is much more important. By the introduction of the word “Alternative” the suggestion is made that the Mendelian principle applies peculiarly to cases of “alternative” inheritance. Mendel himself makes no such limitation in his earlier paper, though perhaps by rather remote implication in the second, to which the reader should have been referred. On the contrary, he wisely abstains from prejudicial consideration of unexplored phenomena.

To understand the significance of the word “alternative” as introduced by Professor Weldon we must go back a little in the history of these studies. In the year 1897 Galton formally announced the Law of Ancestral Heredity referred to in the Introduction, having previously “stated it briefly and with hesitation” in Natural Inheritance, p. 134. In 1898 Professor Pearson published his modification and generalisation of Galton’s Law, introducing a correction of admitted theoretical importance, though it is not in question that the principle thus restated is fundamentally not very different from Galton’s54. It is an essential part of the Galton-Pearson Law of Ancestral Heredity that in calculating the probable structure of each descendant the structure of each several ancestor must be brought to account.

Professor Weldon now tells us that these two papers of Galton and of Professor Pearson have “given us an expression for the effects of blended inheritance which seems likely to prove generally applicable, though the constants of the equations which express the relation between divergence from the mean in one generation, and that in another, may require modification in special cases. Our knowledge of particulate or mosaic inheritance, and of alternative inheritance, is however still rudimentary, and there is so much contradiction between the results obtained by different observers, that the evidence available is difficult to appreciate.”

But Galton stated (p. 401) in 1897 that his statistical law of heredity “appears to be universally applicable to bi-sexual descent.” Pearson in re-formulating the principle in 1898 made no reservation in regard to “alternative” inheritance. On the contrary he writes (p. 393) that “if Mr Galton’s law can be firmly established, it is a complete solution, at any rate to a first approximation, of the whole problem of heredity,” and again (p. 412) that “it is highly probable that it [this law] is the simple descriptive statement which brings into a single focus all the complex lines of hereditary influence. If Darwinian evolution be natural selection combined with heredity, then the single statement which embraces the whole field of heredity must prove almost as epoch-making as the law of gravitation to the astronomer55.”

As I read there comes into my mind that other fine passage where Professor Pearson warns us

“There is an insatiable desire in the human breast to resume in some short formula, some brief statement, the facts of human experience. It leads the savage to ‘account’ for all natural phenomena by deifying the wind and the stream and the tree. It leads civilized man, on the other hand, to express his emotional experience in works of art, and his physical and mental experience in the formulae or so-called laws of science56.”

No naturalist who had read Galton’s paper and had tried to apply it to the facts he knew could fail to see that here was a definite advance. We could all perceive phenomena that were in accord with it and there was no reasonable doubt that closer study would prove that accord to be close. It was indeed an occasion for enthusiasm, though no one acquainted with the facts of experimental breeding could consider the suggestion of universal application for an instant.

But two years have gone by, and in 1900 Pearson writes57 that the values obtained from the Law of Ancestral Heredity

“seem to fit the observed facts fairly well in the case of blended inheritance. In other words we have a certain amount of evidence in favour of the conclusion: That whenever the sexes are equipotent, blend their characters and mate pangamously, all characters will be inherited at the same rate,”

or, again in other words, that the Law of Ancestral Heredity after the glorious launch in 1898 has been home for a complete refit. The top-hamper is cut down and the vessel altogether more manageable; indeed she looks trimmed for most weathers. Each of the qualifications now introduced wards off whole classes of dangers. Later on (pp. 487–8) Pearson recites a further list of cases regarded as exceptional. “All characters will be inherited at the same rate” might indeed almost be taken to cover the results in Mendelian cases, though the mode by which those results are arrived at is of course wholly different.

Clearly we cannot speak of the Law of Gravitation now. Our Tycho Brahe and our Kepler, with the yet more distant Newton, are appropriately named as yet to come58.

But the truth is that even in 1898 such a comparison was scarcely happy. Not to mention moderns, these high hopes had been finally disposed of by the work of the experimental breeders such as Kölreuter, Knight, Herbert, Gärtner, Wichura, Godron, Naudin, and many more. To have treated as non-existent the work of this group of naturalists, who alone have attempted to solve the problems of heredity and species—Evolution, as we should now say—by the only sound method—experimental breeding—to leave out of consideration almost the whole block of evidence collected in Animals and Plants—Darwin’s finest legacy as I venture to declare—was unfortunate on the part of any exponent of Heredity, and in the writings of a professed naturalist would have been unpardonable. But even as modified in 1900 the Law of Ancestral Heredity is heavily over-sparred, and any experimental breeder could have increased Pearson’s list of unconformable cases by as many again.

But to return to Professor Weldon. He now repeats that the Law of Ancestral Heredity seems likely to prove generally applicable to blended inheritance, but that the case of alternative inheritance is for the present reserved. We should feel more confidence in Professor Weldon’s exposition if he had here reminded us that the special case which fitted Galton’s Law so well that it emboldened him to announce that principle as apparently “universally applicable to bi-sexual descent” was one of alternative inheritance—namely the coat-colour of Basset-hounds. Such a fact is, to say the least, ominous. Pearson, in speaking (1900) of this famous case of Galton’s, says that these phenomena of alternative inheritance must be treated separately (from those of blended inheritance)59, and for them he deduces a proposed “law of reversion,” based of course on ancestry. He writes, “In both cases we may speak of a law of ancestral heredity, but the first predicts the probable character of the individual produced by a given ancestry, while the second tells us the percentages of the total offspring which on the average revert to each ancestral type60.”

With the distinctions between the original Law of Ancestral Heredity, the modified form of the same law, and the Law of Reversion, important as all these considerations are, we are not at present concerned.

For the Mendelian principle of heredity asserts a proposition absolutely at variance with all the laws of ancestral heredity, however formulated. In those cases to which it applies strictly, this principle declares that the cross-breeding of parents need not diminish the purity of their germ-cells or consequently the purity of their offspring. When in such cases individuals bearing opposite characters, A and B, are crossed, the germ-cells of the resulting cross-bred, AB, are each to be bearers either of character A or of character B, not both.

Consequently when the cross-breds breed either together or with the pure forms, individuals will result of the forms AA, AB, BA, BB61. Of these the forms AA and BB, formed by the union of similar germs, are stated to be as pure as if they had had no cross in their pedigree, and henceforth their offspring will be no more likely to depart from the A type or the B type respectively, than those of any other originally pure specimens of these types.

Consequently in such examples it is not the fact that each ancestor must be brought to account as the Galton-Pearson Law asserts, and we are clearly dealing with a physiological phenomenon not contemplated by that Law at all.

Every case therefore which obeys the Mendelian principle is in direct contradiction to the proposition to which Professor Weldon’s school is committed, and it is natural that he should be disposed to consider the Mendelian principle as applying especially to “alternative” inheritance, while the law of Galton and Pearson is to include the phenomenon of blended inheritance. The latter, he tells us, is “the most usual case,” a view which, if supported by evidence, might not be without value.

It is difficult to blame those who on first acquaintance concluded Mendel’s principle can have no strict application save to alternative inheritance. Whatever blame there is in this I share with Professor Weldon and those whom he follows. Mendel’s own cases were almost all alternative; also the fact of dominance is very dazzling at first. But that was two years ago, and when one begins to see clearly again, it does not look so certain that the real essence of Mendel’s discovery, the purity of germ-cells in respect of certain characters, may not apply also to some phenomena of blended inheritance. The analysis of this possibility would take us to too great length, but I commend to those who are more familiar with statistical method, the consideration of this question: whether dominance being absent, indefinite, or suppressed, the phenomena of heritages completely blended in the zygote, may not be produced by gametes presenting Mendelian purity of characters. A brief discussion of this possibility is given in the Introduction, p. 31.

Very careful inquiry would be needed before such a possibility could be negatived. For example, we know that the Laws based on Ancestry can apply to alternative inheritance; witness the case of the Basset-hounds. Here there is no simple Mendelian dominance; but are we sure there is no purity of germ-cells? The new conception goes a long way and it may well reach to such facts as these.

But for the present we will assume that Mendel’s principle applies only to certain phenomena of alternative inheritance, which is as far as our warrant yet runs.

No close student of the recent history of evolutionary thought needs to be told what the attitude of Professor Weldon and his followers has been towards these same disquieting and unwelcome phenomena of alternative inheritance and discontinuity in variation. Holding at first each such fact for suspect, then treating them as rare and negligible occurrences, he and his followers have of late come slowly to accede to the facts of discontinuity a bare and grudging recognition in their scheme of evolution62.

Therefore on the announcement of that discovery which once and for all ratifies and consolidates the conception of discontinuous variation, and goes far to define that of alternative inheritance, giving a finite body to what before was vague and tentative, it is small wonder if Professor Weldon is disposed to criticism rather than to cordiality.

We have now seen what is the essence of Mendel’s discovery based on a series of experiments of unequalled simplicity which Professor Weldon does not venture to dispute.

II. Mendel and the Critic’s Version of him.

The “Law of Dominance.”

I proceed to the question of dominance which Professor Weldon treats as a prime issue, almost to the virtual concealment of the great fact of gametic purity.

Cross-breds in general, AB and BA, named above, may present many appearances. They may all be indistinguishable from A, or from B; some may appear A’s and some B’s; they may be patchworks of both; they may be blends presenting one or many grades between the two; and lastly they may have an appearance special to themselves (being in the latter case, as it often happens, “reversionary”), a possibility which Professor Weldon does not stop to consider, though it is the clue that may unravel many of the facts which mystify him now.

Mendel’s discovery became possible because he worked with regular cases of the first category, in which he was able to recognize that one of each of the pairs of characters he studied did thus prevail and was “dominant” in the cross-bred to the exclusion of the other character. This fact, which is still an accident of particular cases, Professor Weldon, following some of Mendel’s interpreters, dignifies by the name of the “Law of Dominance,” though he omits to warn his reader that Mendel states no “Law of Dominance” whatever. The whole question whether one or other character of the antagonistic pair is dominant though of great importance is logically a subordinate one. It depends on the specific nature of the varieties and individuals used, sometimes probably on the influence of external conditions and on other factors we cannot now discuss. There is as yet no universal law here perceived or declared.

Professor Weldon passes over the proof of the purity of the germ-cells lightly enough, but this proposition of dominance, suspecting its weakness, he puts prominently forward. Briefest equipment will suffice. Facing, as he supposes, some new pretender—some local Theudas—offering the last crazy prophecy,—any argument will do for such an one. An eager gathering in an unfamiliar literature, a scrutiny of samples, and he will prove to us with small difficulty that dominance of yellow over green, and round over wrinkled, is irregular even in peas after all; that in the sharpness of the discontinuity exhibited by the variations of peas there are many grades; that many of these grades co-exist in the same variety; that some varieties may perhaps be normally intermediate. All these propositions are supported by the production of a collection of evidence, the quality of which we shall hereafter consider. “Enough has been said,” he writes (p. 240), “to show the grave discrepancy between the evidence afforded by Mendel’s own experiments and that obtained by other observers, equally competent and trustworthy.”

We are asked to believe that Professor Weldon has thus discovered “a fundamental mistake” vitiating all that work, the importance of which, he elsewhere tells us, he has “no wish to belittle.”

III. The Facts in regard to Dominance of Characters in Peas.

Professor Weldon refers to no experiments of his own and presumably has made none. Had he done so he would have learnt many things about dominance in peas, whether of the yellow cotyledon-colour or of the round form, that might have pointed him to caution.

In the year 1900 Messrs Vilmorin-Andrieux & Co. were kind enough to send to the Cambridge Botanic Garden on my behalf a set of samples of the varieties of Pisum and Phaseolus, an exhibit of which had greatly interested me at the Paris Exhibition of that year. In the past summer I grew a number of these and made some preliminary cross-fertilizations among them (about 80 being available for these deductions) with a view to a future study of certain problems, Mendelian and others. In this work I had the benefit of the assistance of Miss Killby of Newnham College. Her cultivations and crosses were made independently of my own, but our results are almost identical. The experience showed me, what a naturalist would expect and practical men know already, that a great deal turns on the variety used; that some varieties are very sensitive to conditions while others maintain their type sturdily; that in using certain varieties Mendel’s experience as to dominance is regularly fulfilled, while in the case of other varieties irregularities and even some contradictions occur. That the dominance of yellow cotyledon-colour over green, and the dominance of the smooth form over the wrinkled, is a general truth for Pisum sativum appears at once; that it is a universal truth I cannot believe any competent naturalist would imagine, still less assert. Mendel certainly never did. When he speaks of the “law” or “laws” that he has established for Pisum he is referring to his own discovery of the purity of the germ-cells, that of the statistical distribution of characters among them, and the statistical grouping of the different germ-cells in fertilization, and not to the “Law of Dominance” which he never drafted and does not propound.

The issue will be clearer if I here state briefly what, as far as my experience goes, are the facts in regard to the characters cotyledon-colour and seed-shapes in peas. I have not opportunity for more than a passing consideration of the seed-coats of pure forms63; that is a maternal character, a fact I am not sure Professor Weldon fully appreciates. Though that may be incredible, it is evident from many passages that he has not, in quoting authorities, considered the consequences of this circumstance.

The normal characters: colour of cotyledons and seed-coats.

Culinary peas (P. sativum, omitting purple sorts) can primarily be classified on colour into two groups, yellow and green. In the green certain pigmentary matters persist in the ripe seed which disappear or are decomposed in the yellow as the seed ripens. But it may be observed that the “green” class itself is treated as of two divisions, green and blue. In the seedsmen’s lists the classification is made on the external appearance of the seed, without regard to whether the colour is due to the seed-coat, the cotyledons, or both. As a rule perhaps yellow coats contain yellow cotyledons, and green coats green cotyledons, though yellow cotyledons in green coats are common, e.g. Gradus, of which the cotyledons are yellow while the seed-coats are about as often green as yellow (or “white,” as it is called technically). Those called “blue” consist mostly of seeds which have green cotyledons seen through transparent skins, or yellow cotyledons combined with green skins. The skins may be roughly classified into thin and transparent, or thick and generally at some stage pigmented. In numerous varieties the colour of the cotyledon is wholly yellow, or wholly green. Next there are many varieties which are constant in habit and other properties but have seeds belonging to these two colour categories in various proportions. How far these proportions are known to be constant I cannot ascertain.

Of such varieties showing mixture of cotyledon-colours nearly all can be described as dimorphic in colour. For example in Sutton’s Nonpareil Marrowfat the cotyledons are almost always either yellow or green, with some piebalds, and the colours of the seed-coats are scarcely less distinctly dimorphic. In some varieties which exist in both colours intermediates are so common that one cannot assert any regular dimorphism64.

There are some varieties which have cotyledons green and intermediate shading to greenish yellow, like Stratagem quoted by Professor Weldon. Others have yellow and intermediate shading to yellowish green, such as McLean’s Best of all65‍. I am quite disposed to think there may be truly monomorphic varieties with cotyledons permanently of intermediate colour only, but so far I have not seen one66. The variety with greatest irregularity (apart from regular dimorphism) in cotyledon-colour I have seen is a sample of “mange-tout à rames, à grain vert,” but it was a good deal injured by weevils (Bruchus), which always cause irregularity or change of colour.

Lastly in some varieties there are many piebalds or mosaics.

From what has been said it will be evident that the description of a pea in an old book as having been green, blue, white, and so forth, unless the cotyledon-colour is distinguished from seed-coat colour, needs careful consideration before inferences are drawn from it.

Shape.

In regard to shape, if we keep to ordinary shelling peas, the facts are somewhat similar, but as shape is probably more sensitive to conditions than cotyledon-colour (not than seed-coat colour) there are irregularities to be perhaps ascribed to this cause. Broadly, however, there are two main divisions, round and wrinkled. It is unquestioned that between these two types every intermediate occurs. Here again a vast number of varieties can be at once classified into round and wrinkled (the classification commonly used), others are intermediate normally. Here also I suspect some fairly clear sub-divisions might be made in the wrinkled group and in the round group too, but I would not assert this as a fact.

I cannot ascertain from botanists what is the nature of the difference between round and wrinkled peas, though no doubt it will be easily discovered. In maize the round seeds contain much unconverted starch, while in the wrinkled or sugar-maize this seems to be converted in great measure as the seed ripens; with the result that, on drying, the walls collapse. In such seeds we may perhaps suppose that the process of conversion, which in round seeds takes place on germination, is begun earlier, and perhaps the variation essentially consists in the premature appearance of the converting ferment. It would be most rash to suggest that such a process may be operating in the pea, for the phenomenon may have many causes; but however that may be, there is evidently a difference of such a nature that when the water dries out of the seed on ripening, its walls collapse67; and this collapse may occur in varying degrees.

In respect of shape the seeds of a variety otherwise stable are as a rule fairly uniform, the co-existence of both shapes and of intermediates between them in the same variety is not infrequent. As Professor Weldon has said, Telephone is a good example of an extreme case of mixture of both colours and shapes. William I. is another. It may be mentioned that regular dimorphism in respect of shape is not so common as dimorphism in respect of colour. Of great numbers of varieties seen at Messrs Suttons’ I saw none so distinctly dimorphic in shape as William I. which nevertheless contains all grades commonly.

So far I have spoken of the shapes of ordinary English culinary peas. But if we extend our observations to the shapes of large-seeded peas, which occur for the most part among the sugar-peas (mange-touts), of the “grey” peas with coloured flowers, etc., there are fresh complications to be considered.

Professor Weldon does not wholly avoid these (as Mendel did in regard to shape) and we will follow him through his difficulties hereafter. For the present let me say that the classes round and wrinkled are not readily applicable to those other varieties and are not so applied either by Mendel or other practical writers on these subjects. To use the terms indicated in the Introduction, seed-shape depends on more than one pair of allelomorphs—possibly on several.

Stability and Variability.

Generally speaking peas which when seen in bulk are monomorphic in colour and shape, will give fairly true and uniform offspring (but such strict monomorphism is rather exceptional). Instances to the contrary occur, and in my own brief experience I have seen some. In a row of Fill-basket grown from selected seed there were two plants of different habit, seed-shape, etc. Each bore pods with seeds few though large and round. Again Blue Peter (blue and round) and Laxton’s Alpha (blue and wrinkled), grown in my garden and left to nature uncovered, have each given a considerable proportion of seeds with yellow cotyledons, about 20% in the case of Laxton’s Alpha. The distribution of these on the plants I cannot state. The plants bearing them in each case sprang from green-cotyledoned seeds taken from samples containing presumably unselected green seeds only. A part of this exceptional result may be due to crossing, but heterogeneity of conditions68 especially in or after ripening is a more likely cause, hypotheses I hope to investigate next season. Hitherto I had supposed the crossing, if any, to be done by Bruchus or Thrips, but Tschermak also suspects Megachile, the leaf-cutter bee, which abounds in my garden.

Whatever the cause, these irregularities may undoubtedly occur; and if they be proved to be largely independent of crossing and conditions, this will in nowise vitiate the truth of the Mendelian principle. For in that case it may simply be variability. Such true variation, or sporting, in the pea is referred to by many observers. Upon this subject I have received most valuable facts from Mr Arthur Sutton, who has very kindly interested himself in these inquiries. He tells me that several highly bred varieties, selected with every possible care, commonly throw a small but constant proportion of poor and almost vetch-like plants, with short pods and small round seeds, which are hoed out by experienced men each year before ripening. Other high-class varieties always, wherever grown, and when far from other sorts, produce a small percentage of some one or more definite “sports.” Of these peculiar sports he has sent me a collection of twelve, taken from as many standard varieties, each “sport” being represented by eight seeds, which though quite distinct from the type agree with each other in almost all cases.

In two cases, he tells me, these seed-sports sown separately have been found to give plants identical with the standard type and must therefore be regarded as sports in seed characters only; in other cases change of plant-type is associated with the change of seed-type.

In most standard varieties these definite sports are not very common, but in a few they are common enough to require continual removal by selection69.

I hope before long to be able to give statistical details and experiments relating to this extraordinarily interesting subject. As de Vries writes in his fine work Die Mutationstheorie (I. p. 580), “a study of the seed-differences of inconstant, or as they are called, ‘still’ unfixed varieties, is a perfect treasure-house of new discoveries.”

Let us consider briefly the possible significance of these facts in the light of Mendelian teaching. First, then, it is clear that as regards most of such cases the hypothesis is not excluded that these recurring sports may be due to the fortuitous concurrence of certain scarcer hypallelomorphs, which may either have been free in the original parent varieties from which the modern standard forms were raised, or may have been freed in the crossing to which the latter owe their origin (see p. 28). This possibility raises the question whether, if we could make “pure cultures” of the gametes, any variations of this nature would ever occur. This may be regarded as an unwarrantable speculation, but it is not wholly unamenable to the test of experiments.

But variability, in the sense of division of gonads into heterogeneous gametes, may surely be due to causes other than crossing. This we cannot doubt. Cross-fertilization of the zygote producing those gametes is one of the causes of such heterogeneity among them. We cannot suppose it to be the sole cause of this phenomenon.

When Mendel asserts the purity of the germ-cells of cross-breds he cannot be understood to mean that they are more pure than those of the original parental races. These must have varied in the past. The wrinkled seed arose from the round, the green from the yellow (or vice versâ, if preferred), and probably numerous intermediate forms from both.

The variations, or as I provisionally conceive it, that differentiant division among the gametes of which variation (neglecting environment) is the visible expression, has arisen and can arise at one or more points of time, and we have no difficulty in believing it to occur now. In many cases we have clear evidence that it does. Crossing,—dare we call it asymmetrical fertilization?—is one of the causes of the production of heterogeneous gametes—the result of divisions qualitatively differentiant and perhaps asymmetrical70.

There are other causes and we have to find them. Some years ago I wrote that consideration of the causes of variation was in my judgment premature71. Now that through Mendel’s work we are clearing our minds as to the fundamental nature of “gametic” variation, the time is approaching when an investigation of such causes may be not unfruitful.

Of variation as distinct from transmission why does Professor Weldon take no heed? He writes (p. 244):

“If Mendel’s statements were universally valid, even among Peas, the characters of the seeds in the numerous hybrid races now existing should fall into one or other of a few definite categories, which should not be connected by intermediate forms.”

Now, as I have already pointed out, Mendel made no pretence of universal statement: but had he done so, the conclusion, which Professor Weldon here suggests should follow from such a universal statement, is incorrectly drawn. Mendel is concerned with the laws of transmission of existing characters, not with variation, which he does not discuss.

Nevertheless Professor Weldon has some acquaintance with the general fact of variability in certain peas, which he mentions (p. 236), but the bearing of this fact on the difficulty he enuntiates escapes him.

Results of crossing in regard to seed characters: normal and exceptional.

The conditions being the same, the question of the characters of the cross-bred zygotes which we will call AB’s depends primarily on the specific nature of the varieties which are crossed to produce them. It is unnecessary to point out that if all AB’s are to look alike, both the varieties A and B must be pure—not in the common sense of descended, as far as can be traced, through individuals identical with themselves, but pure in the Mendelian sense, that is to say that each must be at that moment producing only homogeneous gametes bearing the same characters A and B respectively. Purity of pedigree in the breeder’s sense is a distinct matter altogether. The length of time—or if preferred—the number of generations through which a character of a variety has remained pure, alters the probability of its dominance, i.e. its appearance when a gamete bearing it meets another bearing an antagonistic character, no more, so far as we are yet aware, than the length of time a stable element has been isolated alters the properties of the chemical compound which may be prepared from it.

Now when individuals (bearing contrary characters), pure in the sense indicated, are crossed together, the question arises, What will be the appearance of the first cross individuals? Here again, generally speaking, when thoroughly green cotyledons are crossed with thoroughly yellow cotyledons, the first-cross seeds will have yellow cotyledons; when fully round peas are crossed with fully wrinkled the first result will generally speaking be round, often with slight pitting as Mendel has stated. This has been the usual experience of Correns, Tschermak, Mendel, and myself72 and, as we shall see, the amount of clear and substantial evidence to the contrary is still exceedingly small. But as any experienced naturalist would venture to predict, there is no universal rule in the matter. As Professor Weldon himself declares, had there been such a universal rule it would surely have been notorious. He might further have reflected that in Mendel’s day, when hybridisation was not the terra incognita it has since become, the assertion of such universal propositions would have been peculiarly foolish. Mendel does not make it; but Professor Weldon perceiving the inherent improbability of the assertion conceives at once that Mendel must have made it, and if Mendel doesn’t say so in words then he must have implied it. As a matter of fact Mendel never treats dominance as more than an incident in his results, merely using it as a means to an end, and I see no reason to suppose he troubled to consider to what extent the phenomenon is or is not universal—a matter with which he had no concern.

Of course there may be exceptions. As yet we cannot detect the causes which control them, though injury, impurity, accidental crossing, mistakes of various kinds, account for many. Mendel himself says, for instance, that unhealthy or badly grown plants give uncertain results. Nevertheless there seems to be a true residuum of exceptions not to be explained away. I will recite some that I have seen. In my own crosses I have seen green × green give yellow four times. This I incline to attribute to conditions or other disturbance, for the natural pods of these plants gave several yellows. At Messrs Suttons’ I saw second-generation seeds got by allowing a cross of Sutton’s Centenary (gr. wr.) × Eclipse (gr. rd.) to go to seed; the resulting seeds were both green and yellow, wrinkled and round. But in looking at a sample of Eclipse I found a few yellow seeds, say two per cent., which may perhaps be the explanation. Green wrinkled × green round may give all wrinkled, and again wrinkled × wrinkled may give round73. Of this I saw a clear case—supposing no mistake to have occurred—at Messrs Suttons’. Lastly we have the fact that in exceptional cases crossing two forms—apparently pure in the strict sense—may give a mixture in the first generation. There are doubtless examples also of unlikeness between reciprocals, and of this too I have seen one putative case74.

Such facts thus set out for the first cross-bred generation may without doubt be predicated for subsequent generations.

What then is the significance of the facts?

Analysis of exceptions.

Assuming that all these “contradictory” phenomena happened truly as alleged, and were not pathological or due to error—an explanation which seems quite inadequate—there are at least four possible accounts of such diverse results—each valid, without any appeal to ancestry.

1. That dominance may exceptionally fail—or in other words be created on the side which is elsewhere recessive. For this exceptional failure we have to seek exceptional causes. The artificial creation of dominance (in a character usually recessive) has not yet to my knowledge been demonstrated experimentally, but experiments are begun by which such evidence may conceivably be obtained.

2. There may be what is known to practical students of evolution as the false hybridism of Millardet, or in other words, fertilisation with—from unknown causes—transmission of none or of only some of the characters of one pure parent. The applicability of this hypothesis to the colours and shapes of peas is perhaps remote, but we may notice that it is one possible account of those rare cases where two pure forms give a mixed result in the first generation, even assuming the gametes of each pure parent to be truly monomorphic as regards the character they bear. The applicability of this suggestion can of course be tested by study of the subsequent generations, self-fertilised or fertilised by similar forms produced in the same way. In the case of a genuine false-hybrid the lost characters will not reappear in the posterity.

3. The result may not be a case of transmission at all as it is at present conceived, but of the creation on crossing of something new. Our AB’s may have one or more characters peculiar to themselves. We may in fact have made a distinct “mule” or heterozygote form. Where this is the case, there are several subordinate possibilities we need not at present pursue.

4. There may be definite variation (distinct from that proper to the “mule”) consequent on causes we cannot yet surmise (see pp. 125 and 128).

The above possibilities are I believe at the present time the only ones that need to be considered in connexion with these exceptional cases75. They are all of them capable of experimental test and in certain instances we are beginning to expect the conclusion.

The “mule” or heterozygote.

There can be little doubt that in many cases it is to the third category that the phenomena belong. An indication of the applicability of this reasoning will generally be found in the fact that in such “mule” forms the colour or the shape of the seeds will be recognizably peculiar and proper to the specimens themselves, as distinct from their parents, and we may safely anticipate that when those seeds are grown the plants will show some character which is recognizable as novel. The proof that the reasoning may apply can as yet only be got by finding that the forms in question cannot breed true even after successive selections, but constantly break up into the same series of forms76.

This conception of the “mule” form, or “hybrid-character” as Mendel called it, though undeveloped, is perfectly clear in his work. He says that the dominant character may have two significations, it may be either a parental character or a hybrid-character, and it must be differentiated according as it appears in the one capacity or the other. He does not regard the character displayed by the hybrid, whether dominant or other, as a thing inherited from or transmitted by the pure parent at all, but as the peculiar function or property of the hybrid. When this conception has been fully understood and appreciated in all its bearings it will be found to be hardly less fruitful than that of the purity of the germ-cells.

The two parents are two—let us say—substances77 represented by corresponding gametes. These gametes unite to form a new “substance”—the cross-bred zygote. This has its own properties and structure, just as a chemical compound has, and the properties of this new “substance” are not more strictly traceable to, or “inherited” from, those of the two parents than are those of a new chemical compound “inherited” from those of the component elements. If the case be one in which the gametes are pure, the new “substance” is not represented by them, but the compound is again dissociated into its components, each of which is separately represented by gametes.

The character of the cross-bred zygote may be anything. It may be something we have seen before in one or other of the parents, it may be intermediate between the two, or it may be something new. All these possibilities were known to Mendel and he is perfectly aware that his principle is equally applicable to all. The first case is his “dominance.” That he is ready for the second is sufficiently shown by his brief reference to time of flowering considered as a character (p. 65). The hybrids, he says, flower at a time almost exactly intermediate between the flowering times of the parents, and he remarks that the development of the hybrids in this case probably happens in the same way as it does in the case of the other characters78.

That he was thoroughly prepared for the third possibility appears constantly through the paper, notably in the argument based on the Phaseolus hybrids, and in the statement that the hybrid between talls and dwarfs is generally taller than the tall parent, having increased height as its “hybrid-character.”

All this Professor Weldon has missed. In place of it he offers us the sententia that no one can expect to understand these phenomena if he neglect ancestry. This is the idle gloss of the scribe, which, if we erase it not thoroughly, may pass into the text.

Enough has been said to show how greatly Mendel’s conception of heredity was in advance of those which pass current at the present day; I have here attempted the barest outline of the nature of the “hybrid-character,” and I have not sought to indicate the conclusions that we reach when the reasoning so clear in the case of the hybrid is applied to the pure forms and their own characters.

In these considerations we reach the very base on which all conceptions of heredity and variation must henceforth rest, and that it is now possible for us to attempt any such analysis is one of the most far-reaching consequences of Mendel’s principle. Till two years ago no one had made more than random soundings of this abyss.

I have briefly discussed these possibilities to assist the reader in getting an insight into Mendel’s conceptions. But in dealing with Professor Weldon we need not make this excursion; for his objection arising from the absence of uniform regularity in dominance is not in point.

The soundness of Mendel’s work and conclusions would be just as complete if dominance be found to fail often instead of rarely. For it is perfectly certain that varieties can be chosen in such a way that the dominance of one character over its antagonist is so regular a phenomenon that it can be used in the way Mendel indicates. He chose varieties, in fact, in which a known character was regularly dominant and it is because he did so that he made his discovery79. When Professor Weldon speaks of the existence of fluctuation and diversity in regard to dominance as proof of a “grave discrepancy” between Mendel’s facts and those of other observers80, he merely indicates the point at which his own misconceptions began.

From Mendel’s style it may be inferred that if he had meant to state universal dominance in peas he would have done so in unequivocal language. Let me point out further that of the 34 varieties he collected for study, he discarded 12 as not amenable to his purposes81. He tells us he would have nothing to do with characters which were not sharp, but of a “more or less” description. As the 34 varieties are said to have all come true from seed, we may fairly suppose that the reason he discarded twelve was that they were unsuitable for his calculations, having either ill-defined and intermediate characters, or possibly defective and irregular dominance.

IV. Professor Weldon’s collection of “Other Evidence concerning Dominance in Peas.”

A. In regard to cotyledon colour: Preliminary.

I have been at some pains to show how the contradictory results, no doubt sometimes occurring, on which Professor Weldon lays such stress, may be comprehended without any injury to Mendel’s main conclusions. This excursion was made to save trouble with future discoverers of exceptions, though the existence of such facts need scarcely disturb many minds. As regards the dominance of yellow cotyledon-colour over green the whole number of genuine unconformable cases is likely to prove very small indeed, though in regard to the dominance of round shape over wrinkled we may be prepared for more discrepancies. Indeed my own crosses alone are sufficient to show that in using some varieties irregularities are to be expected. Considering also that the shapes of peas depend unquestionably on more than one pair of allelomorphs I fully expect regular blending in some cases.

As however it may be more satisfactory to the reader and to Professor Weldon if I follow him through his “contradictory” evidence I will endeavour to do so. Those who have even a slight practical acquaintance with the phenomena of heredity will sympathize with me in the difficulty I feel in treating this section of his arguments with that gravity he conceives the occasion to demand.

In following the path of the critic it will be necessary for me to trouble the reader with a number of details of a humble order, but the journey will not prove devoid of entertainment.

Now exceptions are always interesting and suggestive things, and sometimes hold a key to great mysteries. Still when a few exceptions are found disobeying rules elsewhere conformed to by large classes of phenomena it is not an unsafe course to consider, with such care as the case permits, whether the exceptions may not be due to exceptional causes, or failing such causes whether there may be any possibility of error. But to Professor Weldon, an exception is an exception—and as such may prove a very serviceable missile; so he gathers them as they were “smooth stones from the brook.”

Before examining the quality of this rather miscellaneous ammunition I would wish to draw the non-botanical reader’s attention to one or two facts of a general nature.

For our present purpose the seed of a pea may be considered as consisting of two parts, the embryo with its cotyledons, enclosed in a seed-coat. It has been known for about a century that this coat or skin is a maternal structure, being part of the mother plant just as much as the pods are, and consequently not belonging to the next generation at all. If then any changes take place in it consequent on fertilisation, they are to be regarded not as in any sense a transmission of character by heredity, but rather as of the nature of an “infection.” If on the other hand it is desired to study the influence of hereditary transmission on seed-coat characters, then the crossed seeds must be sown and the seed-coats of their seeds studied. Such infective changes in maternal tissues have been known from early times, a notable collection of them having been made especially by Darwin; and for these cases Focke suggested the convenient word Xenia. With this familiar fact I would not for a moment suppose Professor Weldon unacquainted, though it was with some surprise that I found in his paper no reference to the phenomenon.

For as it happens, xenia is not at all a rare occurrence with certain varieties of peas; though in them, as I believe is generally the case with this phenomenon, it is highly irregular in its manifestations, being doubtless dependent on slight differences of conditions during ripening.

The coats of peas differ greatly in different varieties, being sometimes thick and white or yellow, sometimes thick and highly pigmented with green or other colours, in both of which cases it may be impossible to judge the cotyledon-colour without peeling off the opaque coat; or the coats may be very thin, colourless and transparent, so that the cotyledon-colour is seen at once. It was such a transparent form that Mendel says he used for his experiments with cotyledon-colour. In order to see xenia a pea with a pigmented seed-coat should be taken as seed-parent, and crossed with a variety having a different cotyledon-colour. There is then a fair chance of seeing this phenomenon, but much still depends on the variety. For example, Fillbasket has green cotyledons and seed-coat green except near the hilar surface. Crossed with Serpette nain blanc (yellow cotyledons and yellow coat) this variety gave three pods with 17 seeds in which the seed-coats were almost full yellow (xenia). Three other pods (25 seeds), similarly produced, showed slight xenia, and one pod with eight seeds showed little or none.

On the other hand Fillbasket fertilised with nain de Bretagne (yellow cotyledons, seed-coats yellow to yellowish green) gave six pods with 39 seeds showing slight xenia, distinct in a few seeds but absent in most.

Examples of xenia produced by the contrary proceeding, namely fertilising a yellow pea with a green, may indubitably occur and I have seen doubtful cases; but as by the nature of the case these are negative phenomena, i.e. the seed-coat remaining greenish and not going through its normal maturation changes, they must always be equivocal, and would require special confirmation before other causes were excluded.

Lastly, the special change (xenia) Mendel saw in “grey” peas, appearance or increase of purple pigment in the thick coats, following crossing, is common but also irregular.

If a transparent coated form be taken as seed-parent there is no appreciable xenia, so far as I know, and such a phenomenon would certainly be paradoxical82.

In this connection it is interesting to observe that Giltay, whom Professor Weldon quotes as having obtained purely Mendelian results, got no xenia though searching for it. If the reader goes carefully through Giltay’s numerous cases, he will find, almost without doubt, that none of them were such as produce it. Reading Giant, as Giltay states, has a transparent skin, and the only xenia likely to occur in the other cases would be of the peculiar and uncertain kind seen in using “grey” peas. Professor Weldon notes that Giltay, who evidently worked with extreme care, peeled his seeds before describing them, a course which Professor Weldon, not recognizing the distinction between the varieties with opaque and transparent coats, himself wisely recommends. The coincidence of the peeled seeds giving simple Mendelian results is one which might have alarmed a critic less intrepid than Professor Weldon.

Bearing in mind, then, that the coats of peas may be transparent or opaque; and in the latter case may be variously pigmented, green, grey, reddish, purplish, etc.; that in any of the latter cases there may or may not be xenia; the reader will perceive that to use the statements of an author, whether scientific or lay, to the effect that on crossing varieties he obtained peas of such and such colours without specifying at all whether the coats were transparent or whether the colours he saw were coat- or cotyledon-colours is a proceeding fraught with peculiar and special risks.

(1) Gärtner’s cases. Professor Weldon gives, as exceptions, a series of Gärtner’s observations. Using several varieties, amongst them Pisum sativum macrospermum, a “grey” pea, with coloured flowers and seed-coats83, he obtained results partly Mendelian and partly, as now alleged, contradictory. The latter consist of seeds “dirty yellow” and “yellowish green,” whereas it is suggested they should have been simply yellow.

Now students of this department of natural history will know that these same observations of Gärtner’s, whether rightly or wrongly, have been doing duty for more than half a century as stock illustrations of xenia. In this capacity they have served two generations of naturalists. The ground nowadays may be unfamiliar, but others have travelled it before and recorded their impressions. Darwin, for example, has the following passage84:

“These statements led Gärtner, who was highly sceptical on the subject, carefully to try a long series of experiments; he selected the most constant varieties, and the results conclusively showed that the colour of the skin of the pea is modified when pollen of a differently coloured variety is used.” (The italics are mine.)

In the true spirit of inquiry Professor Weldon doubtless reflected,

“’Tis not Antiquity nor Author, That makes Truth Truth, altho’ Time’s Daughter”;

but perhaps a word of caution to the reader that another interpretation exists would have been in place. It cannot be without amazement therefore that we find him appropriating these examples as referring to cotyledon-colour, with never a hint that the point is doubtful.

Giltay, without going into details, points out the ambiguity85. As Professor Weldon refers to the writings both of Darwin and Giltay, it is still more remarkable that he should regard the phenomenon as clearly one of cotyledon-colour and not coat-colour as Darwin and many other writers have supposed.

Without going further it would be highly improbable that Gärtner is speaking solely or even chiefly of the cotyledons, from the circumstance that these observations are given as evidence of “the influence of foreign pollen on the female organs”; and that Gärtner was perfectly aware of the fact that the coat of the seed was a maternal structure is evident from his statement to that effect on p. 80.

To go into the whole question in detail would require considerable space; but indeed it is unnecessary to labour the point. The reader who examines Gärtner’s account with care, especially the peculiar phenomena obtained in the case of the “grey” pea (macrospermum), with specimens before him, will have no difficulty in recognizing that Gärtner is simply describing the seeds as they looked in their coats, and is not attempting to distinguish cotyledon-characters and coat-characters. If he had peeled them, which in the case of “grey” peas would be absolutely necessary to see cotyledon-colour, he must surely have said so.

Had he done so, he would have found the cotyledons full yellow in every ripe seed; for I venture to assert that anyone who tries, as we have, crosses between a yellow-cotyledoned “grey” pea, such as Gärtner’s was, with any pure green variety will see that there is no question whatever as to absolute dominance of the yellow cotyledon-character here, more striking than in any other case. If exceptions are to be looked for, they will not be found there; and, except in so far as they show simple dominance of yellow, Gärtner’s observations cannot be cited in this connection at all.

(2) Seton’s case. Another exception given by Professor Weldon is much more interesting and instructive. It is the curious case of Seton86. Told in the words of the critic it is as follows:—

“Mr Alexander Seton crossed the flowers of Dwarf Imperial, ‘a well-known green variety of the Pea,’ with the pollen of ‘a white free-growing variety.’ Four hybrid seeds were obtained, ‘which did not differ in appearance from the others of the female parent.’ These seeds therefore did not obey the law of dominance, or if the statement be preferred, greenness became dominant in this case. The seeds were sown, and produced plants bearing ‘green’ and ‘white’ seeds side by side in the same pod. An excellent coloured figure of one of these pods is given (loc. cit. Plate 9, Fig. 1), and is the only figure I have found which illustrates segregation of colours in hybrid Peas of the second generation.”

Now if Professor Weldon had applied to this case the same independence of judgment he evinced in dismissing Darwin’s interpretation of Gärtner’s observations, he might have reached a valuable result. Knowing how difficult it is to give all the points in a brief citation, I turned up the original passage, where I find it stated that the mixed seeds of the second generation “were all completely either of one colour or the other, none of them having an intermediate tint, as Mr Seton had expected.” The utility of this observation of the absence of intermediates, is that it goes some way to dispose of the suggestion of xenia as a cause contributing to the result.

Moreover, feeling perfectly clear, from the fact of the absence of intermediates, that the case must be one of simple dominance in spite of first appearances, I suggest the following account with every confidence that it is the true one. There have been several “Imperials,” though Dwarf Imperial, in a form which I can feel sure is Seton’s form, I have not succeeded in seeing; but from Vilmorin’s description that the peas when ripe are “franchement verts” I feel no doubt it was a green pea with a green skin. If it had had a transparent skin this description would be inapplicable. Having then a green skin, which may be assumed with every probability of truth, the seeds, even though the cotyledons were yellow, might, especially if examined fresh, be indistinguishable from those of the maternal type. Next from the fact of the mixture in the second generation we learn that the semi-transparent seed-coat of the paternal form was dominant as a plant-character, and indeed the coloured plate makes this fairly evident. It will be understood that this explanation is as yet suggestive, but from the facts of the second generation, any supposition that there was real irregularity in dominance in this case is out of the question87.

(3) Tschermak’s exceptions. These are a much more acceptable lot than those we have been considering. Tschermak was thoroughly alive to the seed-coat question and consequently any exception stated as an unqualified fact on his authority must be accepted. The nature of these cases we shall see. Among the many varieties he used, some being not monomorphic, it would have been surprising if he had not found true irregularities in dominance.

(3 a) Buchsbaum case. This variety, growing in the open, gave once a pod in which every seed but one was green. In stating this case Professor Weldon refers to Buchsbaum as “a yellow-seeded variety.” Tschermak88, however, describes it as having “gelbes, öfters gelblich-grünes Speichergewebe” (cotyledons); and again says the cotyledon-colour is “allerdings gerade bei Buchsbaum zur Spontanvariation nach gelb-grün neigend!” The (!) is Tschermak’s. Therefore Professor Weldon can hardly claim Buchsbaum as “yellow-seeded” without qualification.

Buchsbaum in fact is in all probability a blend-form and certainly not a true, stable yellow. One of the green seeds mentioned above grew and gave 15 yellows and three greens, and the result showed pretty clearly, as Tschermak says, that there had been an accidental cross with a tall green.

On another occasion Telephone ♀ (another impure green) × Buchsbaum gave four yellow smooth and two green wrinkled, but one [? both: the grammar is obscure] of the greens did not germinate89.

(3 b) Telephone cases. Telephone, crossed with at least one yellow variety (Auvergne) gave all or some green or greenish. These I have no doubt are good cases of “defective dominance” of yellow. But it must be noted that Telephone is an impure green. Nominally a green, it is as Professor Weldon has satisfied himself, very irregular in colour, having many intermediates shading to pure yellow and many piebalds. It is the variety from which alone Professor Weldon made his colour-scale. I desire therefore to call special attention to the fact that Telephone, though not a pure green, Tschermak’s sample being as he says “gelblichweiss grün,” a yellowish-white-green in cotyledon-colour, is the variety which has so far contributed the clearest evidence of the green colour dominating in its crosses with a yellow; and that Buchsbaum is probably a similar case. To this point we shall return. It may not be superfluous to mention also that one cross between Fillbasket (a thorough green) and Telephone gave three yellowish green seeds (Tschermak, (36), p. 501).

(3 c) Couturier cases. This fully yellow variety in crosses with two fully green sorts gave seeds either yellow or greenish yellow. In one case Fillbasket ♀ fertilised by Couturier gave mixed seeds, green and yellow. For any evidence to the contrary, the green in this case may have been self-fertilised. Nevertheless, taking the evidence together, I think it is most likely that Couturier is a genuine case of imperfect dominance of yellow. If so, it is the only true “exception” in crosses between stable forms.

We have now narrowed down Professor Weldon’s exceptions to dominance of cotyledon-colour to two varieties, one yellow (Couturier), and one yellow “tending to green” (Buchsbaum), which show imperfect dominance of yellow; and one variety, Telephone, an impure and irregular green, which shows occasional but uncertain dominance of green.

What may be the meaning of the phenomenon shown by the unstable or mosaic varieties we cannot tell; but I venture to suggest that when we more fully appreciate the nature and genesis of the gametes, it will be found that the peculiarities of heredity seen in these cases have more in common with those of “false hybridism” (see p. 34) than with any true failure of dominance.

Before, however, feeling quite satisfied in regard even to this residuum of exceptions, one would wish to learn the subsequent fate of these aberrant seeds and how their offspring differed from that of their sisters. One only of them can I yet trace, viz. the green seed from Telephone ♀ × Buchsbaum ♂, which proved a veritable “green dominant.” As for the remainder, Tschermak promises in his first paper to watch them. But in his second paper the only passage I can find relating to them declares that perhaps some of the questionable cases he mentioned in his first paper “are attributable to similar isolated anomalies in dominance; some proved themselves by subsequent cultivation to be cases of accidental self-fertilisation; others failed to germinate90‍.” I may warn those interested in these questions, that in estimating changes due to ripening, dead seeds are not available.

B. Seed-coats and shapes.

1. Seed-coats. Professor Weldon lays some stress on the results obtained by Correns91 in crossing a pea having green cotyledons and a thin almost colourless coat (grüne späte Erfurter Folger-erbse) with two purple-flowered varieties. The latter are what are known in England as “grey” peas, though the term grey is not generally appropriate.

In these varieties the cotyledon-colour is yellow and the coats are usually highly coloured or orange-brown. In reciprocal crosses Correns found no change from the maternal seed-coat-colour or seed-shape. On sowing these peas he obtained plants bearing peas which, using the terminology of Mendel and others, he speaks of as the “first generation.”

These peas varied in the colour of their seed-coats from an almost colourless form slightly tinged with green like the one parent to the orange-brown of the other parent. The seeds varied in this respect not only from plant to plant, but from pod to pod, and from seed to seed, as Professor Correns has informed me.

The peas with more highly-coloured coats were sown and gave rise to plants with seeds showing the whole range of seed-coat-colours again.

Professor Weldon states that in this case neither the law of dominance nor the law of segregation was observed; and the same is the opinion of Correns, who, as I understand, inclines to regard the colour-distribution as indicating a “mosaic” formation. This is perhaps conceivable; and in that case the statement that there was no dominance would be true, and it would also be true that the unit of segregation, if any, was smaller than the individual plant and may in fact be the individual seed.

A final decision of this question is as yet impossible. Nevertheless from Professor Correns I have learnt one point of importance, namely, that the coats of all these seeds were thick, like that of the coloured and as usual dominant form. There is no “mosaic” of coats like one parent and coats like the other, though there may be a mosaic of colours. In regard to the distribution of colour however the possibility does not seem to me excluded that we are here dealing with changes influenced by conditions. I have grown a “grey” pea and noticed that the seed-coats ripened in my garden differ considerably and not quite uniformly from those received from and probably ripened in France, mine being mostly pale and greyish, instead of reddish-brown. We have elsewhere seen (p. 120) that pigments of the seed-coat-colour may be very sensitive to conditions, and slight differences of moisture, for example, may in some measure account for the differences in colour. Among my crosses I have a pod of such “grey” peas fertilised by Laxton’s Alpha (green cotyledons, coat transparent). It contained five seeds, of which four were red-brown on one side and grey with purple specks on the other. The fifth was of the grey colour on both sides. I regard this difference not as indicating segregation of character but merely as comparable with the difference between the two sides of a ripe apple, and I have little doubt that Correns’ case may be of the same nature92. Phenomena somewhat similar to these will be met with in Laxton’s case of the “maple” seeded peas (see p. 161).

2. Seed-shapes. Here Professor Weldon has three sets of alleged exceptions to the rule of dominance of round shape over wrinkled. The first are Rimpau’s cases, the second are Tschermak’s cases, the third group are cases of “grey” peas, which we will treat in a separate section (see pp. 153 and 158).

(a) Rimpau’s cases. Professor Weldon quotes Rimpau as having crossed wrinkled and round peas93 and found the second hybrid generation dimorphic as usual. The wrinkled peas were selected and sown and gave wrinkled peas and round peas, becoming “true” to the wrinkled character in one case only in the fifth year, while in the second case—that of a Telephone cross—there was a mixture of round and wrinkled similarly resulting from wrinkled seed for two years, but the experiment was not continued.

These at first sight look like genuine exceptions. In reality, however, they are capable of a simple explanation. It must be remembered that Rimpau was working in ignorance of Mendel’s results, was not testing any rule, and was not on the look out for irregularities. Now all who have crossed wrinkled and round peas on even a moderate scale will have met with the fact that there is frequently some wrinkling in the cross-bred seeds. Though round when compared with the true wrinkled, these are often somewhat more wrinkled than the round type, and in irregular degrees. For my own part I fully anticipate that we may find rare cases of complete blending in this respect though I do not as yet know one.

Rimpau gives a photograph of eight peas (Fig. 146) which he says represent the wrinkled form derived from this cross. It is evident that these are not from one pod but a miscellaneous selection. On close inspection it will be seen that while the remainder are shown with their cotyledon-surfaces upwards, the two peas at the lower end of the row are represented with their hilar-surfaces upwards. Remembering this it will be recognized that these two lower peas are in fact not fully wrinkled peas but almost certainly round “hybrids,” and the depression is merely that which is often seen in round peas (such as Fillbasket), squared by mutual pressure. Such peas, when sown, might of course give some round.

As Tschermak writes ((37), p. 658), experience has shown him that cross-bred seeds with character transitional between “round” and “wrinkled” behave as hybrids, and have both wrinkled and round offspring, and he now reckons them accordingly with the round dominants.

Note further the fact that Rimpau found the wrinkled form came true in the fifth year, while the round gave at first more, later fewer, wrinkleds, not coming true till the ninth year. This makes it quite clear that there was dominance of the round form, but that the heterozygotes were not so sharply distinguishable from the two pure forms as to be separated at once by a person not on the look-out for the distinctions. Nevertheless there was sufficient difference to lead to a practical distinction of the cross-breds both from the pure dominants and from the pure recessives.

The Telephone case may have been of the same nature; though, as we have seen above, this pea is peculiar in its colour-heredity and may quite well have followed a different rule in shape also. As stated before, the wrinkled offspring were not cultivated after the third year, but the round seeds are said to have still given some wrinkleds in the eighth year after the cross, as would be expected in a simple Mendelian case.

(b) Tschermak’s cases. The cases Professor Weldon quotes from Tschermak all relate to crosses with Telephone again, and this fact taken with the certainty that the colour-heredity of Telephone is abnormal makes it fairly clear that there is here something of a really exceptional character. What the real nature of the exception is, and how far it is to be taken as contradicting the “law of dominance,” is quite another matter.

3. Other phenomena, especially regarding seed-shapes, in the case of “grey” peas. Modern evidence. Professor Weldon quotes from Tschermak the interesting facts about the “grey” pea, Graue Riesen, but does not attempt to elucidate them. He is not on very safe ground in adducing these phenomena as conflicting with the “law of dominance.” Let us see whither we are led if we consider these cases. On p. 124 I mentioned that the classes round and wrinkled do not properly hold if we try to extend them to large-seeded sorts, and that these cases require separate consideration. In many of such peas, which usually belong either to the classes of sugar-peas (mange-touts) or “grey” peas (with coloured flowers), the seeds would be rather described as irregularly indented, lumpy or stony94, than by any use of the terms round or wrinkled. One sugar-pea (Debarbieux) which I have used has large flattish, smooth, yellow seeds with white skins, and this also in its crossings follows the rules about to be described for the large-seeded “grey” peas.

In the large “grey” peas the most conspicuous feature is the seed-coat, which is grey, brownish, or of a bright reddish colour. Such seed-coats are often speckled with purple, and on boiling these seed-coats turn dark brown. They are in fact the very peas used by Mendel in making up his third pair of characters. Regarding them Professor Weldon, stating they may be considered separately, writes as follows:—

“Tschermak has crossed Graue Riesen with five races of P. sativum, and he finds that the form of the first hybrid seeds follows the female parent, so that if races of P. sativum with round smooth seeds be crossed with Graue Riesen (which has flattened, feebly wrinkled seeds) the hybrids will be round and smooth or flattened and wrinkled, as the P. sativum or the Graue Riesen is used as female parent95. There is here a more complex phenomenon than at first sight appears; because if the flowers of the first hybrid generation are self-fertilised, the resulting seeds of the second generation invariably resemble those of the Graue Riesen in shape, although in colour they follow Mendel’s law of segregation!”

From this account who would not infer that we have here some mystery which does not accord with the Mendelian principles? As a matter of fact the case is dominance in a perfectly obvious if distinct form.

Graue Riesen, a large grey sugar-pea, the pois sans parchemin géant of the French seedsmen, has full-yellow cotyledons and a highly coloured seed-coat of varying tints. In shape the seed is somewhat flattened with irregular slight indentations, lightly wrinkled if the term be preferred. Tschermak speaks of it in his first paper as “Same flach, zusammengedrückt”—a flat, compressed seed; in his second paper as “flache, oft schwach gerunzelte Cotyledonen-form,” or cotyledon-shape, flat, often feebly wrinkled, as Professor Weldon translates.

First-crosses made from this variety, each with a different form of P. sativum, are stated on the authority of Tschermak’s five cases, to follow exclusively the maternal seed-shape. From “schwach gerunzelte,” “feebly wrinkled,” Professor Weldon easily passes to “wrinkled,” and tells us that according as a round sativum or the Graue Riesen is used as mother, the first-cross seeds “will be round and smooth or flattened and wrinkled.”

As a matter of fact, however, the seeds of Graue Riesen though slightly wrinkled do not belong to the “wrinkled” class; but if the classification “wrinkled” and “round” is to be extended to such peas at all, they belong to the round. Mendel is careful to state that his round class are “either spherical or roundish, the depressions on the surface, when there are any, always slight”; while the “wrinkled” class are “irregularly angular, deeply wrinkled96.”

On this description alone it would be very likely that Graue Riesen should fall into the round class, and as such it behaves in its crosses, being dominant over wrinkled (see Nos. 3 and 6, below). I can see that in this case Professor Weldon has been partly misled by expressions of Tschermak’s, but the facts of the second generation should have aroused suspicion. Neither author notices that as all five varieties crossed by Tschermak with Graue Riesen were round, the possibilities are not exhausted. Had Tschermak tried a really wrinkled sativum with Graue Riesen he would have seen this obvious explanation.

As some of my own few observations of first-crosses bear on this point I may quote them, imperfect though they are.

I grew the purple-flowered sugar-pea “Pois sans parchemin géant à très large cosse,” a soft-podded “mange-tout” pea, flowers and seed-coats coloured, from Vilmorin’s, probably identical with Graue Riesen.

1. One flower of this variety fertilised with Pois très nain de Bretagne (very small seed; yellow cotyledons; very round) gave seven seeds indistinguishable (in their coats) from those of the mother, save for a doubtful increase in purple pigmentation of coats.

2. Fertilised by Laxton’s Alpha (green; wrinkled; coats transparent), two flowers gave 11 seeds exactly as above, the purple being in this case clearly increased.

In the following the purple sugar-pea was father.

3. Laxton’s Alpha (green; wrinkled; coats transparent) fertilised by the purple sugar-pea gave one pod of four seeds with yellow cotyledons and round form.

4. Fillbasket (green; smooth but squared; coats green) fertilised by the purple sugar-pea gave one pod with six seeds, yellow cotyledons97; Fillbasket size and shape; but the normally green coat yellowed near the hilum by xenia.

5. Express (“blue”-green cotyledons and transparent skins; round) fertilised with purple sugar-pea gave one pod with four seeds, yellow cotyledons, shape round, much as in Fillbasket.

6. British Queen (yellow cotyledons, wrinkled, white coats) ♀ × purple sugar-pea gave two pods with seven seeds, cotyledons yellow, coats tinged greenish (xenia?), all round.

So much for the “Purple” sugar-pea.

I got similar results with Mange-tout Debarbieux. This is a soft-podded Mange-tout or sugar-pea, with white flowers, large, flattish, smooth seeds, scarcely dimpled; yellow cotyledons.

7. Debarbieux fertilised by Serpette nain blanc (yellow cotyledons; wrinkled; white skin; dwarf) gave one pod with six seeds, size and shape of Debarbieux, with slight dimpling.

8. Debarbieux by nain de Bretagne (very small; yellow cotyledons; very round) gave three pods, 12 seeds, all yellow cotyledons, of which two pods had eight seeds identical in shape with Debarbieux, while the third had four seeds like Debarbieux but more dimpled. The reciprocal cross gave two seeds exactly like nain de Bretagne.

But it may be objected that the shape of this large grey pea is very peculiar98; and that it maintains its type remarkably when fertilised by many distinct varieties though its pollen effects little or no change in them; for, so long as round varieties of sativum are used as mothers, this is true as we have seen. But when once it is understood that in Graue Riesen there is no question of wrinkling, seeing that the variety behaves as a round variety, the shape and especially the size of the seed must be treated as a maternal property.

Why the distinction between the shape of Graue Riesen and that of ordinary round peas should be a matter of maternal physiology we do not know. The question is one for the botanical chemist. But there is evidently very considerable regularity, the seeds borne by the cross-breds exhibiting the form of the “grey” pea, which is then a dominant character as much as the seed-coat characters are. And that is what Tschermak’s Graue Riesen crosses actually did, thereby exhibiting dominance in a very clear form. To interject these cases as a mystery without pointing out how easily they can be reconciled with the “law of dominance” may throw an unskilled reader into gratuitous doubt.

Finally, since the wrinkled peas, Laxton’s Alpha and British Queen, pollinated by a large flat mange-tout, witness Nos. 3 and 6 above, became round in both cases where this experiment was made, we here merely see the usual dominance of the non-wrinkled character; though of course if a round-seeded mother be used there can be no departure from the maternal shape, as far as roundness is concerned.

Correns’ observations on the shapes of a “grey” pea crossed with a round shelling pea, also quoted by Professor Weldon as showing no dominance of roundness, are of course of the same nature as those just discussed.

C. Evidence of Knight and Laxton.

In the last two sections we have seen that in using peas of the “grey” class, i.e. with brown, red, or purplish coats, special phenomena are to be looked for, and also that in the case of large “indented” peas, the phenomena of size and shape may show some divergence from that simple form of the phenomenon of dominance seen when ordinary round and wrinkled are crossed. Here the fuller discussion of these phenomena must have been left to await further experiment, were it not that we have other evidence bearing on the same questions.

The first is that of Knight’s well-known experiments, long familiar but until now hopelessly mysterious. I have not space to quote the various interpretations which Knight and others have put upon them, but as the Mendelian principle at once gives a complete account of the whole, this is scarcely necessary, though the matter is full of historical interest.

Crossing a white pea with a very large grey purple-flowered form Knight (21) found that the peas so produced “were not in any sensible degree different from those afforded by other plants of the same [white] variety; owing, I imagine, to the external covering of the seed (as I have found in other plants) being furnished entirely by the female99.” All grew very tall100, and had colours of male parent101. The seeds they produced were dark grey102.

“I had frequent occasion to observe, in this plant [the hybrid], a stronger tendency to produce purple blossoms, and coloured seeds, than white ones; for when I introduced the farina of a purple blossom into a white one, the whole of the seeds in the succeeding year became coloured [viz. DR × D giving DD and DR]; but, when I endeavoured to discharge this colour, by reversing the process, a part only of them afforded plants with white blossoms; this part sometimes occupying one end of the pod, and being at times irregularly intermixed with those which, when sown, retained their colour” [viz. DR × R giving DR and RR] (draws conclusions, now obviously erroneous103).

In this account we have nothing not readily intelligible in the light of Mendel’s hypothesis.

The next evidence is supplied by an exceptionally complete record of a most valuable experiment made by Laxton104. The whole story is replete with interest, and as it not only carries us on somewhat beyond the point reached by Mendel, but furnishes an excellent illustration of how his principles may be applied, I give the whole account in Laxton’s words, only altering the paragraphing for clearness, and adding a commentary. The paper appears in Jour. Hort. Soc. N.S. III. 1872, p. 10, and very slightly abbreviated in Jour. of Hort. XVIII. 1870, p. 86. Some points in the same article do not specially relate to this section, but for simplicity I treat the whole together. It is not too much to say that two years ago the whole of this story would have been a maze of bewildering confusion. There are still some points in it that we cannot fully comprehend, for the case is one of far more than ordinary complexity, but the general outlines are now clear. In attempting to elucidate the phenomena it will be remembered that there are no statistics (those given being inapplicable), and the several offspring are only imperfectly referred to the several classes of seeds. This being so, our rationale cannot hope to be complete. Laxton states that as the seeds of peas are liable to change colour with keeping, for this and other reasons he sent to the Society a part of the seeds resulting from his experiment before it was brought to a conclusion.

“The seeds exhibited were derived from a single experiment. Amongst these seeds will be observed some of several remarkable colours, including black, violet, purple-streaked and spotted, maple, grey, greenish, white, and almost every intermediate tint, the varied colours being apparently produced on the outer coat or envelope of the cotyledons only.

The peas were selected for their colours, &c., from the third year’s sowing in 1869 of the produce of a cross in 1866 of the early round white-seeded and white-flowered garden variety “Ringleader,” which is about 2 1/2 ft. in height, fertilised by the pollen of the common purple-flowered “maple” pea, which is taller than “Ringleader,” and has slightly indented seeds. I effected impregnation by removing the anthers of the seed-bearer, and applying the pollen at an early stage. This cross produced a pod containing five round white peas, exactly like the ordinary “Ringleader” seeds105.

In 1867 I sowed these seeds, and all five produced tall purple-flowered purplish-stemmed plants106, and the seeds, with few exceptions, had all maple or brownish-streaked envelopes of various shades; the remainder had entirely violet or deep purple-coloured envelopes107: in shape the peas were partly indented; but a few were round108. Some of the plants ripened off earlier than the “maple,” which, in comparison with “Ringleader,” is a late variety; and although the pods were in many instances partially abortive, the produce was very large109.

In 1868 I sowed the peas of the preceding year’s growth, and selected various plants for earliness, productiveness, &c. Some of the plants had light-coloured stems and leaves; these all showed white flowers, and produced round white seeds110. Others had purple flowers, showed the purple on the stems and at the axils of the stipules, and produced seeds with maple, grey, purple-streaked, or mottled, and a few only, again, with violet-coloured envelopes. Some of the seeds were round, some partially indented111. The pods on each plant, in the majority of instances, contained peas of like characters; but in a few cases the peas in the same pod varied slightly, and in some instances a pod or two on the same plant contained seeds all distinct from the remainder112. The white-flowered plants were generally dwarfish, of about the height of “Ringleader”; but the coloured-flowered sorts varied altogether as to height, period of ripening, and colour and shape of seed113. Those seeds with violet-coloured envelopes produced nearly all maple- or parti-coloured seeds, and only here and there one with a violet-coloured envelope; that colour, again, appeared only incidentally, and in a like degree in the produce of the maple-coloured seeds114.

In 1869 the seeds of various selections of the previous year were again sown separately; and the white-seeded peas again produced only plants with white flowers and round white seeds115. Some of the coloured seeds, which I had expected would produce purple-flowered plants, produced plants with white flowers and round white seeds only116; the majority, however, brought plants with purple flowers and with seeds principally marked with purple or grey, the maple- or brown-streaked being in the minority117. On some of the purple-flowered plants were again a few pods with peas differing entirely from the remainder on the same plant. In some pods the seeds were all white, in others all black, and in a few, again, all violet118; but those plants which bore maple-coloured seeds seemed the most constant and fixed in character of the purple-flowered seedlings119, and the purplish and grey peas, being of intermediate characters, appeared to vary most120. The violet-coloured seeds again produced almost invariably purplish, grey, or maple peas, the clear violet colour only now and then appearing, either wholly in one pod or on a single pea or two in a pod. All the seeds of the purple-flowered plants were again either round or only partially indented; and the plants varied as to height and earliness. In no case, however, does there seem to have been an intermediate-coloured flower; for although in some flowers I thought I found the purple of a lighter shade, I believe this was owing to light, temperature, or other circumstances, and applied equally to the parent maple. I have never noticed a single tinted white flower nor an indented white seed in either of the three years’ produce. The whole produce of the third sowing consisted of seeds of the colours and in the approximate quantities in order as follows,—viz.: 1st, white, about half; 2nd, purplish, grey, and violet (intermediate colours), about three-eighths; and, 3rd, maple, about one-eighth.

From the above I gather that the white-flowered white-seeded pea is (if I may use the term) an original variety well fixed and distinct entirely from the maple, that the two do not thoroughly intermingle (for whenever the white flower crops out, the plant and its parts all appear to follow exactly the characters of the white pea), and that the maple is a cross-bred variety which has become somewhat permanent and would seem to include amongst its ancestors one or more bearing seeds either altogether or partly violet- or purple-coloured; for although this colour does not appear on the seed of the “maple,” it is very potent in the variety, and appears in many parts of the plant and its offspring from cross-fertilised flowers, sometimes on the external surface or at the sutures of the pods of the latter, at others on the seeds and stems, and very frequently on the seeds; and whenever it shows itself on any part of the plant, the flowers are invariably purple. My deductions have been confirmed by intercrosses effected between the various white-, blue-, some singularly bright green-seeded peas which I have selected, and the maple- and purple-podded and the purple-flowered sugar peas, and by reversing those crosses.

I have also deduced from my experiments, in accordance with the conclusions of the late Mr Knight and others, that the colours of the envelopes of the seeds of peas immediately resulting from a cross are never changed121. I find, however, that the colour and probably the substance of the cotyledons are sometimes, but not always, changed by the cross fertilisation of two different varieties; and I do not agree with Mr Knight that the form and size of the seeds produced are unaltered122; for I have on more than one occasion observed that the cotyledons in the seeds directly resulting from a cross of a blue wrinkled pea fertilised by the pollen of a white round variety have been of a greenish-white colour123, and the seeds nearly round124 and larger or smaller according as there may have been a difference in the size of the seeds of the two varieties125.

I have also noticed that a cross between a round white and a blue wrinkled pea will in the third and fourth generations (second and third years’ produce) at times bring forth blue round, blue wrinkled, white round and white wrinkled peas in the same pods, that the white round seeds, when again sown, will produce only white round seeds, that the white wrinkled seeds will, up to the fourth or fifth generation, produce both blue and white wrinkled and round peas, that the blue round peas will produce blue wrinkled and round peas, but that the blue wrinkled peas will bear only blue wrinkled seeds126. This would seem to indicate that the white round and the blue wrinkled peas are distinct varieties derived from ancestors respectively possessing one only of those marked qualities; and, in my opinion, the white round peas trace their origin to a dwarfish pea having white flowers and round white seeds, and the blue wrinkled varieties to a tall variety, having also white flowers but blue wrinkled seeds. It is also noticeable, that from a single cross between two different peas many hundreds of varieties, not only like one or both parents and intermediate, but apparently differing from either, may be produced in the course of three or four years (the shortest time which I have ascertained it takes to attain the climax of variation in the produce of cross-fertilised peas, and until which time it would seem useless to expect a fixed seedling variety to be produced127), although a reversion to the characters of either parent, or of any one of the ancestors, may take place at an earlier period.

These circumstances do not appear to have been known to Mr Knight, as he seems to have carried on his experiments by continuing to cross his seedlings in the year succeeding their production from a cross and treating the results as reliable; whereas it is probable that the results might have been materially affected by the disturbing causes then in existence arising from the previous cross fertilisation, and which, I consider, would, in all cases where either parent has not become fixed or permanent, lead to results positively perplexing and uncertain, and to variations almost innumerable. I have again selected, and intend to sow, watch, and report; but as the usual climax of variation is nearly reached in the recorded experiment, I do not anticipate much further deviation, except in height and period of ripening—characters which are always very unstable in the pea. There are also important botanical and other variations and changes occurring in cross-fertilised peas to which it is not my province here to allude; but in conclusion I may, perhaps, in furtherance of the objects of this paper, be permitted to inquire whether any light can, from these observations or other means, be thrown upon the origin of the cultivated kinds of peas, especially the “maple” variety, and also as to the source whence the violet and other colours which appear at intervals on the seeds and in the offspring of cross-fertilised purple-flowered peas are derived.”