Mendel's Principles of Heredity (William Bateson, 1909) ========================================================== W. BATESON. Grantchester, Cambridge. February^ 1909. Note to the second impression. The publication of a reprint gives an opportunity of correcting mistakes which correspondents have been good enough to point out. These concern chiefly the exceptional case In regard to web-foot in Pigeons (p. 36), the dominance of colour in Helix (p. 45), and the age of the subject of Fig. 25. For correction In the formula (p. 60) see Corrigenda. A serious error respecting the heredity of Colour-blind- ness, which through Mr Nettleshlp's courtesy I was able to withdraw by means of a printed slip inserted at p. 260 of the original Issue, has now been excised. In a note added at p. 195 I have indicated the probable bearing of the special features observed In the case of Colour-blindness upon the theory of Sex, and at p. 231 a diagram Is inserted showing the probable course of descent followed by that condition. This scheme Is in part unverified and must be considered as tentative onlv. I shall be most crrateful for accurate In- formation as to families containing colour-blind members. W. B. August^ 1909. CORRIGENDA p. 59, last paragraph. For is given read are given, p. 60, line 5. For \n read \n. ADDENDUM to p. 48 In Naturalist^ 1906, p. 113, C. T. Trechmann records an observation that several sinistral Limtiaea peregra placed in an aquarium produced young both sinistral and dextral, the former slightly preponderating in number. CONTENTS PART I. CHAPTER PAGE I. Introductory. Mendel's Discovery . . . . i Introductory — Some pre-Mendelian Writings — Mendel's Discovery — Dominant and Recessive — Segregation. Allelomorphism — Homo- zygote and Heterozygote. Purify of Type. II. The Material Investigated . . . . . .18 List of Structural Characters in Plants and Animals — List of Types in which the inheritance of Colour has been studied — Preliminary Deductions — Dominance and heterozygous characters — Mendel's system distinguished from that of Galton. III. Numerical Consequences and Recombinations . . 57 Representations of the Fg Generation and Novelties due to Re-combina- tion of Factors — Compound Characters — Combs of P'owls — Hetero- stylism — White Flowers from Red x Cream. IV. Heredity of Colour ....... 74 Factors determining Colours : the Ratio 9:3: 4 — The " Presence and Absence " Hypothesis. Epistatic and Hypostatic Factors — Colours of Mice — Pied Types — A Dominant Piebald. V. Heredity of Colour {continued) ..... 88 Albinos giving Coloured Offspring ; Reversion on Crossing — Various Kinds of Whites — Stocks— Orchids — Pigeons— Fowls — Primula. VI. Heredity of Colour {contmued) . . . . .107 Eye-Colours. Variations in Colour of the Iris— Deficiency of Eye- Pigments in some Coloured Types. VII. Heredity of Colour {continued) . . . • - ^^5 The Genetics of Yellow Pigments in certain Animals. Yellow Mice not breeding true — The Case of Basset Hounds and the " Law of An- cestral Heredity." Relation of this Principle to Mendelian Rules. VIII. Heredity of Colour {continued) 132 Various Specific Phenomena in Colour-Inheritance. Relation of Colour to Hoariness in Stocks. Miscellaneous Cases. Colour of a Special Part controlling that of other Parts— Summary and Discussion— Subtraction-Stages. Xll Contents CHAPTER IX. Gametic Coupling and Spurious Allelomorphism PAGE 148 Pollen-Shape and Flower-Colour. Axil-Colour and Sterile Anthers- Hooded Standard and Flower-Colour in Sweet Peas. X. Heredity and Sex ^^4 Evidence from Breeding Experiments. ^;j^«w— Sex-limited Heredity. The Horns of Sheep— Colour-Blindness— Sex and Spurious Allelo- morphism. The Currant Moth— The Cinnamon Canary— The Silky Fowl — Aglia tau — Cytological Evidence— Summary. XI. Double Flowers iQ^ Miscellaneous Cases. Recessive and Dominant Doubling— " Hose-in- Hose " Flowers— The Special Case of Double Stocks. XII. Evidence as to Mendelian Inheritance in Man . 205 Normal Characters— Diseases and Malformations. Dominants— Sex- limited Dominants— Recessives— Notes on collecting Evidence. XIII. Intermediates between Varieties and the "Pure Lines" of Johannsen 235 Intermediates as Heterozygous Forms— Subtraction-Stages of Dominants — Interfering Factors — Fluctuational Forms — "Pure Lines." XIV. Miscellaneous Exceptional and Unconformable Phe- nomena 245 Crosses breeding true without Segregation. Parthenogenetic or Apo- ganiic Forms. Hieracitun — Sexual Forms — Numerical Aberrations — Irregularities of Dominance — Alternation of Generations — Mater- nal Characters in certain Seeds. XV. Biological Conceptions in the Light of Mendelian Discoveries ......••• 266 Nature of Units — Nature of Segregation — Moment of Segregation — Differentiation of Parts compared with Segregation — Reversion and Variation. " Bush " and " Cupid " Sweet Peas — Mendelian Segre- gation and Species — Discontinuity in Variation — Mendelism and Natural Selection. XVI. Practical Application of Mendelian Principles . 291 Meaning of Pure-bred — Rogueing — Raising Novelties — A Practical Ex- ample— Unfixable Types — Technical Methods — Sociological Appli- cation. PART II. 1. Biographical Notice of Mendel 2. Translation of the Paper on Hybridisation 3. Translation of the Paper on Hieraciwn . Bibliography Supplementary List Index of Subjects Index of Authors 309 317 362 369 385 386 394 LIST OF ILLUSTRATIONS PORTRAITS OF MENDEL. In 1866 About 1862 About 1880 Frontispiece to face p. 309 to face p. 317 COLOURED PLATES. Plate I. Lepidoptera ..... 11. Mice III. Reversion in Sweet Peas IV. Fowls V. Spurious Allelomorphism in Sweet Peas VI. Heredity of Colour in Primula Sinensis to face p. 44 between pp. 80-81 between pp. 93-94 to face p. 1 03 between pp. 154-155 between pp. 294-295 FIGURES. FIGURE 1. Tall and "Cupid" dwarf Sweet Peas 2. Diagram showing consequences of Segregation . 3. Inheritance of seed-characters in Pea 4. Branched and unbranched forms in Stocks {Mafthiola) 5. Hooded and normal Barley ..... 6. Heredity in Wheat ....... 7. Fern-leaf and palm-leaf in Primula Sinensis 8. Two-row and six-row Barley 9. Starch-grains in Peas .... 10. Round and wrinkled seed in Maize . 11. Down-colour in Chickens . 12. Types of combs in Fowls . 13. Combs of newly-hatched Chickens 14. Descent of "homostyle" character in Primula PAGE 9 12 15 20 21 23 24 27 29 30 51 61 62 69 XIV List of Ilhistratio7ts FIGURE 15. Diagram of Fo showing ratio 9:3:4 1 6. Diagram of F2 showing ratio 9:7. 17. Diagram of Fo in Sweet Pea showing ratio 27 : 9 : 28 18. Pedigrees of eye-colour in Man .... 19. Pollen grains of Sweet Peas ..... 20. Heredity of horns in Sheep ..... 21. Heredity of a peculiar form of curly hair 22. Descent of congenital lock of white hair . 23. Brachydactylous hands ...... 24. Skiagram of hands ....... 25. Hands of brachydactylous child .... 26. Pedigree of brachydactylous family .... 27. Descent from brachydactylous members 28. Drinkwater's pedigree of brachydactyly 29. Descent of prae-senile cataract ..... 30. Another cataractous family ..... 31. Descent of stationary night-blindness . between pp. 32. Descent of Colour-blindness ..... 33. Ideal Scheme of descent of simple sex-limited condition e.g. horns of Sheep ...... 34. Tentative representation of descent of Colour-blindness 35. Polish X Rivet Wheat ....... 36. Seeds of Polish x Eivet Wheat ..... 37. Reversion in height of Sweet Peas .... 38. Two types of dwarf Sweet Peas .... PAGE 77 89 91 108 150 171 207 207 211 212 213 214 214 215 216 217 220-221 223 230 231 259 260 282 283 PART I CHAPTER I INTRODUCTORY. MENDEL'S DISCOVERY. Introductory — Some pre-Mendelian Writings — MendeVs Discovejy — Dominant and Recessive — Segregation. Allelomorphism — Homozygote and Heterozygote. Purity of Type, Among the biological sciences the study of heredity occupies a central position. Whether we be zoologists, botanists, or physiologists, the facts of heredity concern ^t us. Upon this physiological function all the rest in some degree depend. Every advance in knowledge of that central function must affect the course of thought along each several line of biological inquiry. Moreover though, as naturalists, we are not directly concerned with the applications of science, we must perceive that in no' region of knowledge is research more likely to increase man's power over nature. The science of sociology, and in many of its developments the science of medicine • also, must of necessity form working hypotheses respecting the course of heredity, and we cannot doubt that a percep- tion of the truth in reeard to the function of transmission will greatly contribute to the progress of these sciences. Lastly, to the industrial arts of the breeder of plants or animals, the knowledge we are attempting to provide is of such direct importance that upon this consideration no special emphasis is required. In studying heredity, there- fore, we are examining a vital problem of no mean consequence, and those who engage in that work are happy in the thought that they are assisting one of the main advances in natural knowledge. But though we may approach this study of genetics — to use the modern designation — from so many different sides, it is especially in their bearing on the problem of E. H. I 2 Introductory [ch. the evolution of species that the facts have hitherto been most profitably investigated. It was in the attempt to ascertain the interrelationships between species that experi- ments in genetics were first made. The words '' evolution " and *' origin of species " are now so intimately associated with the name of Darwin that we are apt to forget that the idea of a common descent had been prominent in the minds of naturalists before he wrote, and that, for more than half a century, zealous investigators had been devoting them- selves to the experimental study of that possibility. Promi- nent among this group of experimenters may be mentioned Koelreuter, John Hunter, Herbert, Knight, Gaertner, Jordan, Naudin, Godron, Lecoq, Wichura — men whose names are familiar to every reader of Animals and Plants under Domestication. If we could ask those men to define the object of their experiments, their answer would be that they were seeking to determine the laws of hereditary trans- mission with the purpose of discovering the interrelationships of species. In addition to the observation of the visible structures and habits of plants and animals they attempted by experiment to ascertain those hidden properties of living things which we may speak of as genetic, properties which breeding tests can alone reveal. The vast mass of observation thus accumulated contains much that is of permanent value, hints that if followed might have saved their successors years of wasted effort, and not a few indications which in the light of later discovery will greatly accelerate our own progress. Yet in surveying the work of this school we are conscious of a feeling of disappointment at the outcome. There are signs that the workers themselves shared this disappointment. As we now know, they missed the clue without which the evidence so laboriously collected remained an inscrutable medley of contradictions. While the experimental study of the species problem was in full activity the Darwinian writings appeared. Evolution, from being an unsupported hypothesis, was at length shown to be so plainly deducible from ordinary experience that the reality of the process was no longer doubtful. With the triumph of the evolutionary idea, curiosity as to the significance of specific differences was i] Introchtctory 3 satisfied. The Origin was published In 1859. During the following decade, while the new views were on trial, the experimental breeders continued their work, but before 1870 the field was practically abandoned. In all that concerns the problem of species the next thirty years are marked by the apathy characteristic of an age of faith. Evolution became the exerclslng-ground of essayists. The number Indeed of naturalists Increased ten- fold, but their activities were directed elsewhere. Darwin's achievement so far exceeded anything that was thought possible before, that what should have been hailed as a long-expected beginning was taken for the completed work. I well remember receiving from one of the most earnest of my seniors the friendly warning that It was waste of time to study variation, for " Darwin had swept the field." Parenthetically we may notice that though scientific opinion in general became rapidly converted to the doctrine of pure selection, there was one remarkable exception. Systematlsts for the most part kept aloof. Everyone was convinced that natural selection operating In a continuously varying population was a sufficient account of the origin of species except the one class of scientific workers whose labours familiarised them with the phenomenon of specific difference. From that time the systematlsts became, as they still in great measure remain, a class apart. A separation has thus been effected between those who lead theoretical opinion and those who by taste or necessity have retained an acquaintance with the facts. The con- sequences of that separation have been many and grievous. To it are to be traced the extraordinary misapprehensions as to the fundamental phenomena of specific difference which are now prevalent. If species had really arisen by the natural selection for impalpable differences, Intermediate forms should abound, and the limits between species should be on the whole indefinite. As this conclusion follows necessarily from the premisses, the selectionists believe and declare that it represents the facts of nature. Differences between species being by axiom Indefinite, the differences between varieties must be supposed to be still less definite. Consequently the conclusion that evolution must proceed by insensible 4 Introductory [ch. transformation of masses of Individuals has become an established dogma. Systematlsts, entomologists or botan- ists for example, are dally witnesses to variation occurring as an Individual and discontinuous phenomenon, but they stand aside from the debate ; and whoever In a discussion of evolutionary theory appeals to the defmiteness of varietal distinctions In colour for Instance, or In form, as recognizable by common observation without mechanical aid, must be prepared to meet a charge of want of Intelligence or candour. This is no doubt a passing phase and will end so soon as interest In the problems of evolution is combined with some knowledge of variation and heredity. Genetic experiment was first undertaken, as we have seen, in the hope that It would elucidate the problem of species. The time has now come when appeals for the vigorous prosecution of this method should rather be based on other grounds. It is as directly contributing to the advancement of pure physiological science that genetics can present the strongest claim. We have an eye always on the evolution-problem. We know that the facts we are collecting will help in its solution ; but for a period we shall perhaps do well to direct our search more especially to the immediate problems of genetic physiology, the laws of heredity, the nature of variation, the significance of sex and of other manifestations of dimorphism, willing to postpone the application of the results to wider problems as a task more suited to a maturer stage. When the magnitude and definlteness of the advances already made In genetics come to be more generally known, it Is to be anticipated that workers in various departments of biology will realise that here at last is common ground. As we now know, the conceptions on which both the systematlsts and the specula- tive biologists have based their methods need complete revision in the light of the new facts, and till the possibilities of genetic research are more fully explored the task of reconstruction can hardly be begun. In that work of exploration all classes of naturalists will alike find interest. The methods are definite and exact, so we need not fear the alienation of those systematlsts to whom all theoretical inquiry is repulsive. They are also wide in their scope, and those who would turn from the details of classification I] Pre-Mendelian IVritings 5 as offering matter too trivial for their attention may engage in genetic Inquiries with great confidence that every frag- ment of soHd evidence thus discovered will quickly take its place in the development of a coordinated structure. Some pre-Mendeliaji IVritings. Of the contributions made during the essayist period three call for notice : Welsmann deserves mention for his useful work in asking for the proof that " acquired characters " — or, to speak more precisely, parental ex- perience— can really be transmitted to the offspring. The occurrence of progressive adaptation by transmission of the effects of use had seemed so natural to Darwin and his contemporaries that no proof of the physiological reality of the phenomenon was thought necessary. Weismann's challenge revealed the utter Inadequacy of the evidence on which these beliefs were based. There are doubtless isolated observations which may be Interpreted as favouring the belief In these transmissions, but such meagre indications as exist are by general consent admitted to be too slight to be of much assistance in the attempt to understand how the more complex adaptatlve mechanisms arose. Nevertheless it was for the purpose of elucidating them that the appeal to Inherited experience was made. Weismann's contribution, though negative, has greatly simplified the practical investi- gation of genetic problems. Though it attracted little attention at the time of its appearance, an honourable place in the history of our science must be accorded to the paper published by de Vrles (1889) under the title Intracelitdare Pangenesis. This essay is remarkable as a clear foreshadow^Ing of that conception of unit-characters which is destined to play so large a part in the development of genetics. The supreme Importance of an exact knowledge of heredity was urged by Galton in various writings published during the period of which I am speaking. He pointed out that the phenomena manifested regularity, and he made the first comprehensive attempt to determine the rules they obey. It was through his work and influence that the existence of some order pervading the facts became generally 6 Pre-Mendelian Writings [ch. recognized. In 1897 he definitely enunciated his now famous " Law " of heredity, which declared that to the total heritage of the offspring the parents on an average contribute \, the grandparents ^, and the great-grandparents \, and so on, the total heritage being taken as unity. To this conclusion he had been led by several series of data, but the evidence upon which he especially relied was that of the pedigrees of Basset Hounds furnished him by the late Sir Everett Millais. In that instance the character con- sidered was the presence or absence of black in addition to yellow and white. The colours were spoken of as tri-colour and non-tri-colour, and the truth of the law was tested by the average numbers of the respective colours which resulted from the various matings of dogs of known ancestral composition. These numbers corresponded so well with the expectations given by the law as to leave no reasonable doubt that the results of calculation were in general har- mony with natural fact. There are features in this important case which need special consideration, and to these I will return. Meanwhile we may note that though there was admittedly a statistical accord between Galton's theory and some facts of heredity, yet no one familiar with breeding or even with the literature of breeding could possibly accept that theory as a literal or adequate presentation of the facts. Galton himself in pro- mulgating it made some reservations ; but in the practice of breeding, so many classes of unconformable phenomena were already known, that while recognizing the value of his achievement, we could not from the first regard it as more than an adumbration of the truth. As we now know, Galton's method failed for want of analysis. His formula should in all probability be looked upon rather, as an occasional consequence of the actual laws of heredity than in any proper sense one of those laws. Of the so-called investigations of heredity pursued by extensions of Galton's non-analytical method and promoted by^ Professor Pearson and the English Biometrical school it is now scarcely necessary to speak. That such work may ultimat(;ly contribute to the development of statistical theory cannot be denied, but as applied to the problems of heredity the effort has resulted only in the concealment of I] Rediscovery of Mendel 7 that order which it was ostensibly undertaken to reveal. A preliminary acquaintance with the natural history of heredity and variation was sufficient to throw doubt on the foundations of these elaborate researches. To those who hereafter may study this episode in the history of biological science it will appear inexplicable that work so unsound in construction should have been respectfully received by the scientific world. With the discovery of segregation it became obvious that methods dispensing with individual analysis of the material are useless. The only alternatives open to the inventors of those methods were either to abandon their delusion or to deny the truth of Mendelian facts. In choosing the latter course they have certainly succeeded in delaying recognition of the value of Mendelism, but with the lapse of time the number of persons who have themselves witnessed the phenomena has increased so much that these denials have lost their dangerous character and may be regarded as merely formal. Rediscovery of Mendel : his Method. With the year 1900 a new era begins. In the spring of that year there appeared, within a few weeks of each other, the three papers of de Vries, Correns, and Tschermak, giving the substance of Mendel's long-forgotten treatise. Each of these three writers was able from his own ex- perience to confirm Mendel's conclusions, and to extend them to other cases. There could therefore, from the first, be no question as to the truth of the facts. To appreciate what Mendel did the reader should refer to the original paper"^, which is a model of lucidity and expository skill. His success is due to the clearness with which he thought out the problem. Being familiar with the works of Gaertner and the other experimental breeders he surmised that their failure to reach definite and consistent conclusions was due to a want of precise and continued analysis. In order to obtain a clear result he saw that it was absolutely necessary to start with pure-breeding, homogeneous materials, to consider each character separately, and on no account to confuse the different generations together. Lastly he realised * See Part II. 8 MejideFs Method [ch. that the progeny from distinct individuals must be separately recorded. All these ideas were entirely new in his day. When such precautions had been observed he anticipated that a regular result would be attainable if the experiments were carried out on a sufficient scale. After several preliminary trials he chose the edible Pea [Pisttm sativum) for his subject. Varieties in cultivation are distinguished by striking characters recognizable with- out trouble. The plants are habitually self-fertilised, a feature which obviates numerous difficulties. Following his idea that the heredity of each character must be separately investigated, he chose a number oi pairs of characters, and made crosses between varieties differing markedly in respect of one pair of characters. The case which illustrates Mendelian methods in the simplest way is that in which heredity in respect of height w^as studied. Mendel took a pair of varieties of which one was tall, being 6 — 7 feet high, and the other was dwarf, f to \\ feet. These two were then crossed together. In peas this is an easy operation. The unbroken anthers can be picked out of a bud with a pair of fine forceps and the pollen of the plant chosen for the father may be at once applied to the stigma of the emasculated flower. The cross-bred seeds thus produced grew into plants which were always tall, having a height not sensibly different from that of the pure tall variety. In our modern terminology such a cross-bred, the first filial generation, is called F^. From the fact that the character, tallness, appears in the cross-bred to the exclusion of the opposite character, Mendel called it a dominant character ; dwarfness, which disappears in the F^ plant, he called recessive. The tall cross-bred, so produced, in its turn bore seeds by self-fertilisation. These are the next generation, F.^ When grown up they prove to be mixed, many being tall, some being short, like the tall and the short grand- parents respectively. Fig. i shows such an F^ family in the Sweet Pea. Upon counting the members of this F^ generation it was discovered that the proportion of tails to shorts exhibited a certain constancy, averaging about three tails to one short, or in other words, 75 per cent, dominants to 25 per cent, recessives. I] Mendel s Method These F^ plants were again allowed to fertilise them- selves and the offspring of each plant was separately sown. It was then found that the offspring, F,, of the recessives consisted entirely of recessives. Further generations bred from these recessives again produced recessives only, and therefore the recessives which appeared in /^, are seen to lO Segregation [ch. be pure to the recessive character, namely, In the case we are considering, to dwarfness. But the tall F,^ dominants when tested by a study of their offspring {F}), instead of being all alike (as the dwarfs or recessives were), proved to be of two kinds, viz. (a) Plants which gave a mixed F^ consisting of both tails and dwarfs, the proportion showing again an average of three tails to one dwarf (b) Plants which gave tails only and are thus pure to tallness. The ratio of the impure {a) plants to the pure {b) plants was as 2 to i. The whole F^ generation therefore, formed by self- fertilisation of 'the original hybrid consists of three kinds of plants : 25% 50 7„ 25 ._ pure dominants impure dominants pure recessives or 3 dominants : i recessive. Segregation. Allelomorphism. The conclusion which Mendel drew from these observa- tions is one which will suggest itself to any one who reflects on the facts. The result is exactly what would be expected if both male and female germ-cells of the cross-bred F^ were in equal numbers bearers of either the dominant {D) or recessive {R) character, but not both. If this were so, and if the union of the male and female germ-cells occurs at random, the result would be an F^ family made up of 2c^DD : 2^DR : 2^RD : 2^RR. ^ ^^ : iR. But, as the first cross showed, when D meets R in fertilisation the resulting individual is in appearance D ; therefore F,, appears as 3Z) : \R. The results of the F^ I] and its Conseqttences 1 1 generation are in exact agreement with this suggestion : for the R plants give R only ; and of the D plants one- third give D only, while two-thirds give the same mixture, '^D : \R, which was produced by F^ (Fig. 2. I). The descent may be represented diagrammatically thus : Parents Tall ( TT) x Short (//) F, - Tall(i7) F, Tall Tall Tall Short TT Tt tT tt pure tall \ \ pure short F, TT tt t^ It TY~Tt ~tt U Now since the fertilised ovum or zygote, formed by the original cross, was made by the union of two germ-cells or gametes bearing respectively tallness and dwarfness, both these elements entered into the composition of the original F^ zygote ; but if the germ-cells which that zygote eventually forms are bearers of either tallness or dwarfness, there must at some stage in the process of germ-formation be a separation of the two characters, or rather of the ultimate factors which cause those characters to be developed in the plants. This phenomenon, the dissociation of characters from each other in the course of the formation of the germs, we speak of as segregation, and the characters which segregate from each other are described as allelofuorphtc, i.e. alternative to each other in the constitution of the gametes (Fig. 2). That this is the true account was proved by further experiments which Mendel made by crossing the F^ with pure dominants and with pure recessives. For DR x DD gave an offspring all dominant in appearance, though in reality consisting of both BR plants and DD plants, on an average in equal numbers. On the other hand DR x RR gives an equal number of dominants and recessives, of which the dominants are all DR plants, and the recessives are all pure recessives. These various experiments illus- trate the composition of the four simple types of Mendelian families, which may be set out thus : 12 Segregation [CH. DD X RR gives all DR DR X RR gives iZ^^T? : i7?y? DR X Z>i;> gives iZ^i:> : \DR DRxDR gives iZ^Z;> : 2DR : ii?7? appearing as 3 dominants : i recessive. X ID LP X »2 F Raiio RR ;^: Medio ID : II iR X aU D III Fig. 2. Diagrams showing numerical consequences of segregation. I. The mating DD x RR^ and Z>i? x DR. II. The mating DR x ^i?. III. The mating DR x Z>Z>. The way in which these ratios are produced may be easily represented by means of a number of draught-men. Pairs of draughts then represent zygotes ; single draughts represent germ-cells. That there is a propriety in repre- senting zygotic or somatic cells as double structures and germ-cells as single structures will be evident to biologists ; for we know that each somatic nucleus in plants and animals is a double structure, containing twice the number of chromosomes present in each mature germ-cell. Two black draughts may then be taken to represent a pure black individual, two white draughts a white individual. When they are crossed together F^ is represented by a black I] and its Conseqtterices 13 draught and a white one (Fig. 2. I). Supposing the black to be a dominant the fact may be represented by putting it on the top. When segregation of the allelomorphs, black- ness and whiteness, takes place in gameto-genesis, the germ-cells of the cross-bred are again bearers of blackness or of whiteness, and it may readily be shown experimentally that the results of their various random combinations give rise to the ratios stated above. The fact of segregation was the essential discovery which Mendel made. As we now know, such segregation is one of the normal phenomena of nature. It is segregation which determines the regularity perceptible in the here- ditary transmission of differences, and the definiteness or discontinuity so often conspicuous in the variation of animals and plants is a consequence of the same phenomenon. Segregation thus defines the units concerned in the consti- tution of organisms and provides the clue by which an analysis of the complex heterogeneity of living forms may be begun. There are doubtless limits beyond which such analysis cannot be pursued, but a vast field of research must be explored before they are reached or determined. It is likely also that in certain cases the units are so small that no sensible segregation can be proved to exist. As yet, how- ever, no such example has been adequately investigated ; nor, until the properties and laws of interaction of the segregable units have been much more thoroughly examined, can this class of negative observations be considered with great prospect of success. The dominance of certain characters is often an impor- tant but never an essential feature of Mendelian heredity. Those who first treated of Mendel's work most unfortu- nately fell into the error of enunciating a '' Law of Domi- nance" as a proposition comparable with the discovery of segregation. Mendel himself enunciates no such law. Dominance of course frequently exists. The consequences of its occurrence and the complications it introduces must be understood as a preliminary to the practical investigation of the phenomena of heredity, but it is only a subordinate incident of special cases, and Mendel's principles of inherit- ance apply equally to cases where there is no dominance 14 Segregation [ch. and the heterozygous type is intermediate in character between the two pure types. To the detection of the genetic system of any given case it is however necessary that the results of combinations should be sensibly regular. When, as occasionally happens, a character may sometimes behave as a dominant and sometimes not, we have as yet no satisfactory means of further analysis. These irregularities in dominance may confidently be attributed to the disturbing effects of other factors or of conditions, but the detection of such unknown factors must be a long and perhaps impossible task. Mendel applied his method to the following seven distinct pairs of characters in peas, and found that in each the inheritance was similar. The dominant character is put first. 1. Height : whether tall or short. 2. Distribution of flowers on the stem : whether arranged along the axis of the plant, 6»r bunched together at the top so as to form a false umbel *, 3. Colour of unripe pod : whether a shade of green or bright yellow. 4. Shape of pod : whether simply inflated, or deeply constricted between the seeds, i.e. as in '' sugar- peas " or '' Pois sans parchemin." 5. Colour of seed-skin : whether various shades of grey or brown, with or without violet spotting, or white. The "grey" skins are always asso- ciated with coloured flowers and almost always with a purple or red mark in the axils. 6. Colour of cotyledons : whether yellow or green. 7. Shape of seeds : whether rounded or wrinkled. It will be observed that the first five are plant-characters. In order to see the result of crossing, the seeds must be sown and allowed to grow into plants. The last two characters belong to the seeds themselves. The seeds of course are members of a generation later than that of the plant which bears them. Thus when a cross is made the * This is a fasciated and semi-monstrous form. I] and its Conseqtieitces 15 resultant seeds are F^, showing the dominant character yellowness or roundness, but the seed-skins are maternal tissue. Such F^ seeds grow into F^ plants and bear /% seeds which show the typical mixture of dominants and recessives in the pods (Fig. 3). In each case Mendel's J M 1 II ill) VR Yv^' (JR Y^ YR CiR c^vv 9 S i y i > Fr (JR 7W Y(? YW YR rR Fig. 3. Inheritance of seed-characters in Pea. The seed of a green round variety fertiUsed by pollen of a yellow wrinkled variety are yellow and round (J\). The reciprocal cross would give the same result. Two pods of A seed borne by the J^^ plant are shown. There were 6 yellow round, 3 green round, 3 yellow wrinkled, i green wrinkled. observations have been substantially confirmed by later observers, and the operation of similar processes has now been recognized in a long series of most diverse characters in both animals and plants. Consequences of Segregation : Homozygote and Heterozygote. Before considering the various extensions of Men- delian research, it may be well to indicate in general terms the chief significance of the facts. The first conception to which we are led is that of tmit-characters, units because they may be treated as such in the cell-divisions of gametogenesis. It is evidently upon some process of 1 6 Purity of Type [ch. qualitative segregation occurring in one or more of these cell-divisions that allelomorphism depends. The opposite members of each pair of characters being allelomorphic to each other, every zygote"^, or individual produced in ferti- lisation, must, in respect of any such pair, be either a koiuozygote, that is to say, a zygote formed by the union of two gametes each bearing the same allelomorph, as AA and aa, or a heterozygote formed by the union of two germs bearing different allelomorphs, as Aa. Therefore in respect of any pair of allelomorphic characters, the individuals composing the whole population are of three kinds only : 1. Homozygotes of the form A A, 2. Homozygotes of the form aa, 3. Heterozygotes of the form Aa. The gametes are of two kinds only, A and a. Each kind of homozygote is pure to the character of the gametes which compose it. Purity of Type. Purity of type thus acquires a precise meaning. It is dependent on gametic segregation, and has nothing to do with a prolonged course of selection, natural or artificial. All this is of course consonant with the visible facts that have been discovered by the cytologists, in so far as the nucleus of each somatic cell is a double structure, while the nucleus of each gametic cell is a single structure. It is, in my judgment, Impossible as yet to form definite views as to the relations of the various parts of the cell to the function of heredity. The details of cytology and their interpretation are beyond our present province, but this much is certain : that when in these discussions we idealize the characters as borne by the gamete in an unpaired state and by the zygote in a paired state, we make no assumption which is not in full accord with histological appearances. From the fact that the development of characters in animals or plants depends on the presence of definite units * In botany the term zygote is usually restricted to the single cell which results from the process of fertilisation, but by a natural extension the word may be used for the individual which develops by somatic divisions from that cell. I] Pttrity of Type 17 or factors in their germ-cells, the paradox at once follows that an organism may be pure-bred in respect of a given character though its parents were cross-bred in the same respect. Purity depends on the meeting of two gametes bearing similar factors, and when two similarly-constituted gametes do thus meet in fertilisation, the product of their union is pure. The belief, so long prevalent, that purity of type depends essentially on continued selection is thus shown to have no physiological foundation. Similarly it is evident that an individual may be pure in respect of one character and cross-bred or impure in respect of others. As a consequence of the application of Mendel's prin- ciples, that vast medley of seemingly capricious facts which have been recorded as to heredity and variation is rapidly being shaped into an orderly and consistent whole. A new world of intricate order previously undreamt of is disclosed. We are thus endowed with an instrument of peculiar range and precision, and we reach to certainty in problems of physiology which we might have supposed destined to continue for ages inscrutable. After such a discovery it is obvious that old ideas must be revised. Systematists debating the limits of '' specific rank " or the range of variability, morphologists seeking to reconstruct phylogenetic history, physiologists unravelling the interaction of bodily functions, cytologists attempting to interpret the processes of cell-division — each of these classes of naturalists must now examine the current con- ceptions of his study in the light of the new knowledge. The practical breeder of animals or plants, basing his methods on a determination of the Mendelian units and their properties, will in many of his operations be able to proceed with confidence and rapidity. Lastly, those who as evolutionists or sociologists are striving for wider views of the past or of the future of living things may by the use of Mendelian analysis attain to a new and as yet limitless horizon. B. H. CHAPTER II THE MATERIAL INVESTIGATED. List of Structural Character's in Plants and Animals — List of Types iri which the inheritance of Colour has been studied — Preliminary Deductions — Dominance and heterozygotis characters — Mendel's system distin- guished from that of Gallon. Heredity following the general rules described in the last chapter has been witnessed in a great diversity of animals and plants. The characters already proved to follow such rules show an equal diversity. The following is a list of some of them. Adequately to represent the facts respecting each of these cases lengthy description would be needed. In regard to several of them occurrences which do not readily fall into line have been recorded. Of these some are probably due to errors of observation or mistakes of various kinds, but a few will doubtless prove to be genuine exceptions to rule and may constitute points ot departure for fresh and more extended research. In the outline of the phenomena, which is all that this book can profess to offer, it seemed best to restrict as far as possible the enumeration of these details, which can only be thoroughly appreciated by reference to the original papers ; but such annotations as appeared necessary either in elucida- tion of the phenomena or by way of incentive to further work are briefly given with references to the original sources. These annotations will be better understood after the later chapters have been read. In the following list when one character is conspicuously dominant it is put first, but in several cases the dominance is imperfect. Plants. I. Tallness and dwarfness. Peas [Pisiwt) and Sweet Peas [Lathyrzcs odoratus). Runner and French Beans [Phaseolus). As regards Peas the facts have been recorded by Mendel (195), Tschermak (269, 270, &c.), R.E.C. * (20). When varieties differing * R.E.C. stands throughout for Reports to the Evolution Committee of the Royal Society, giving an account of the experiments of W. Bateson, E. R. Saunders and R. C. Punnett. Other contributors to these Reports are mentioned by name. CH. II] Stritctitral Characters : Plants 19 greatly In height are used, dominance is complete, and the two parent forms are represented as three to one in Fc,. No clear exception has yet been observed. Peas {Ptsum) exist in a vast number of distinct horticultural varieties which can roughly be classified as tall (about 5 — 6 ft.), half-dwarfs (about 4 ft), dwarfs (about 9 ins. to 3 ft.). The genetic relations of the half-dwarfs to the others are not fully explored, and further investigation will probably lead to the discovery of important facts. The cross half- dwarf X tall giving tall as dominant has produced some extreme dwarfs in F2, doubtless by recombination {q-v.), R.E.C. 20, p. 69. The cross half-dwarf X dwarf has given intermediates in F^ {ibid.). The cross between tall and dwarf *' Cupid " Sweet Peas gives complete dominance of tallness and simple segregation in F^., "Cupids" indis- tinguishable from the original "Cupid" parent reappearing (Fig. i). Phaseolus has been investigated especially by Tschermak (278) who records some apparently anomalous results, de Vries (298, 11. p. 76) states that he found that extracted F.j dwarf Antirrhinum did not breed true, but threw plants of various heights. The experiment should be repeated. 2. Branching habit and the unbranched habit. Sun- flower {^HeliantJucs, Shull, 241) and Cotton (Balls, 6). The branched form of Stock {Matthiola incana) is dominant to the unbranched Brompton type. \n F^ the unbranched type reappears, but the ratio has not been determined (Fig. 4). E. R. Saunders (unpublished). 3. The straggling habit of both the tall and dwarf " Cupid " Sweet Peas, and the much-branched erect habit of the '' Bush" Sweet Peas (R.E.C. 22). The relation of these two types to each other is not altogether simple. As described (q.v.) F^ from Cupid x Bush is a reversio7iary form exactly like the normal tall variety. Neither the tall varieties nor the Cupids show the profuse branching of the Bush Sweet Peas which gives them their peculiar appearance. This is evidently recessive to the unbranched condition, and the fact thus stands out in contrast to those observed in the case of Sunflower and Cotton. But in the Sweet Pea we have the additional complication that the factor which represses the excessive branching by its presence gives increase of height. The tall and the Bush differ from each other in respect of this factor only. It is present in the tall but absent from the Bush. In the cross between Bush and Cupid two pairs of factors are concerned as explained in the passage referred to. 4. Hairiness and glabrousness. Lychnis. MattJiiola (Stocks). Wheat. The case of Lych?iis has been studied by de Vries (288) and R.E.C. (19). In crosses between fully hairy and glabrous strains the discon- tinuity Is complete. Various forms intermediate in hairiness may nevertheless be found wild and are by no means rare. Silene inflata 20 Struchiral Characters : Plants [CH. often exists in two forms, hairy and glabrous, growing side by side, and doubtless their genetic relations are the same as those found for the corresponding varieties of Lychnis. In this species a third form is found with hairs on the edges only (12). The case of Matthiola is important and presents features of special interest, R.E.C (19, 20, 21, see also Correns, 61). Between thoroughly hoary and glabrous strains the discontinuity is absolute, and the glabrous Fig. 4. Matthiola. Branched and unbranched forms in F^. A photo- graph of Miss Saunders' plants, the leaves removed. (Supplied by Miss Killby.) are entirely destitute of hairs. The dominance is complete and homo- zygotes cannot be distinguished from heterozygotes. A third, or " half hoary " form exists, which is glabrous or nearly so on the upper surface only. Its behaviour has not been fully investigated (19, p. 33.) The genetics of hairiness in wheat have been studied by Spillman (247), Tschermak (270), Biffen (27). The heterozygotes are sometimes inter- mediate in hairiness. The Peach and the Nectarine are probably related to each other as hairy dominant and glabrous recessive. Peculiar results are recorded in Cotton (Balls, 6). n] Structural Characters : Plants 21 5. Prickliness and smoothness of fruits. Dattira, (R.E.C. 19, 20.) Ranunculus arvensis (20). The case of Datura is interesting from the fact that it sometimes has mosaic fruits, one quarter or one half being prickly and the rest smooth. This is perhaps to be regarded as indicative of segregation occurring among zygotic cells (see Chap. xv.). Ranunculus arvensis has three types, spiny, tuberculated, and smooth. The first is a simple dominant. Tuberculated x smooth gave ^1 partially spiny (21, p. 55). 6. Absence of glands iyMatthiola incana) on leaves was dominant to presence of glands {M. smuata) (R.E.C. 20, p. 40). Fig. 5. Cross between a normally awned Barley and a variety with " hooded " awns. P, F, the parents. F^ shows partial dominance of hoods. The increase in length of ear is noticeable. The case also illustrates the result of crossing a 2-row type with a 6-row type, showing dominance of the former. (From Professor Bififen's specimens.) 22 Structural Characters : Plants [ch. 7. Rough and smooth foliage. Wheat. Biffen (27). 8. Keeled glumes and rounded glumes. Wheat. Ibid. 9. Beardless and bearded ears. Wheat. Ibid. Also Spillman (247) and Tschermak (270). Most, if not all, of the "beardless" varieties exhibit a slight and variable amount of awn especially on the uppermost spikelets (Fig. 4). 10. The "hoods" or '' Kapuzen'' characteristic of certain Barleys show a partial dominance over the normal type. These hoods, Professor Biffen states, are, structurally, aborted florets (Fig. 5). Tschermak (270), Biffen (30). 11. Hollow and solid straw. Wheat. Biffen (27). This is a structural character of an interesting kind, and one upon which the commercial value of straw very largely depends. It was shown that many factors were concerned in the production of the stem-characters; and in Fc^ by the recombination of these factors a great variety of straws appeared. 12. Blunt and pointed pods. Pisttm. Tschermak (271), R.E.C. (20). Phaseolus. Tschermak (272). The dominance in this case is complete. Some varieties exist in both a blunt and a pointed type {e.g. Sutton's Continuity). The nature of these cases is discussed later. 13. Lax and dense ears of Wheat and Barley give different results according to the varieties used. Sometimes F^ is lax, sometimes it is intermediate (Spillman, 247; Biffen, 27, 28). See Fig. 6. In Barley an increase in ear- length has been observed (Fig. 5). 14. Development of fibrous parchment-like lining to pods, and the absence of the same which constitutes the ''sugar peas." Pisum. In Phaseolus (kidney-beans), where similar types occur, the evidence is that the dominance is reversed (Emerson, 120, 121). This is one of the features originally investigated by Mendel. He regarded the parchmented type as a dominant. In our experiments F ^ has always had some parchment but the quantity is so much reduced as to cause the heterozygote to have a very distinct appearance (R.E.C. 20). 15. Much serrated and little serrated edges of leaves. Urtica (cp. Pkyteuma, Correns, 70, p. 197). This cross n] Structural Characters : Plants 23 was described by Correns {^^) who gives a striking diagram representing his results. The cross was made between two forms known as Dodartii and Hltdifera, which were •4-* »-. >N O -4-> tn a> ex, is >^ J^ 'wi ''*^ <^ !.: 2 rt ^ CA) ii c ^ 03 «^. c« n 1^ '- ^ ^ O = ^-C S Correns (60); R.E.C. (20); Lock (172, 173, 175-6); Hurst (155). Polemonium. Correns (70) found the white var. of P. coeruleum dominant to the yellow of P. flavu7n ; and the blue type oi coeruleum xjlavum gave F^ blue. It may be inferred that the yellow of flavmn is a chromo- plast colour, and that the blue anthocyan dominated as usual. Hybrids sterile. Pri7nula. P. Sinensis exists in a long series of colour-types the relations of which are still being investigated by R. P. Gregory in con- II] Colours of Plants 4 1 junction with me. Some of the more striking facts are referred to in later chapters. White flowers with green stem constitute an albino, recessive to all colours. The magenta shades have a factor epistatic to crimson and pink. Blue is hypostatic to all the red shades. The whites which have red or reddish stej?is are dominant whites, showing only a pale shade or tinge of colour in F^. Deep colours cannot appear on stems that are not red except in the white-edged "Sirdar" {g.v.). Salvia Horniinum. Purple, red, white, related as in Fisum^ &c. Saunders, R.E.C. (20). Triticum (Wheat). Red chaff is dominant to white chaff. Grey chaff is epistatic to red and dominant to white. Tschermak (270) ; Biffen (27) ; Spillman (247). Verbascum blattaria. Yellow (a sap-colour) dominant to white. ShuU (241). Viola. White is recessive to colour (de Vries, 290) in V. cornuta. The brown seed-colour of V. papilionacea is dominant to buff of V. hirsutula, and the purple of the capsule of hirsutiila to the absence of purple vc\ papilionacea- (Brainerd, 41). Z^a (Maize). Yellow endosperm dominant to white. Blue in aleurone layer an irregular dominant to absence of blue. (Definite exceptions are frequent.) Red pericarp, a plant-character, dominant to absence of red. The relations of the striped types have not been clearly determined. Correns {(i'^\ Lock (172, 174). Colours of Ani77tals. Man. Eye-colour {q.v.), Hurst (161), Davenport (107). Albinism {q.v.) is doubtless recessive, but in man its descent is complex and has not yet been elucidated. Red hair is recessive to dark hair and perhaps to ordinary brown (see Hurst, 162). Cattle. Red-roan is a heterozygote of red and white (Wilson, 311) ; and blue- roan is similarly related to black and white. Spillman (249) suggests that black is dominant to red. Cats. Red is dominant to black in males. Tortoiseshell is the corresponding form of the heterozygote in females. Doncaster (109). Dilution-types, blue, and cream, recessive to saturated colours. As to eye-colour see Przibram (224). Mice, Fats, Fabbits, Guinea-pigs. Colours fully discussed in later chapters. Chief papers by Allen (i); Bateson (10); Castle (48, 53); Crampe (83, «); Cuenot (84-9); Darbishire (90); Durham (116); Hurst (157); Mudge (204). Albino recessive in all cases. A piebald type dominant to self-colour exists in mice (Durham, 116). 42 Colours of Animals [ch. Horse. Chestnut recessive to bays and browns. Relations of these two domi- nants to each other not clear. Hurst (158). Pigs. Several notes published by Spillman (249, 251-2). White is usually a dominant to colour in domesticated races* but piebalds are frequent in F^. The relation of black to red is not yet clear. The white belt, characteristic of certain breeds; is, according to Spillman, due to a complementary pair of factors which may be separately carried by self-blacks. He makes the interesting suggestion that the appearance of the belt may be a "reversion" to a condition like that of the Indian Tapir (251). The colour of the wild boar is dominant to the red of Tamworth and segregates normally from it (252). The wild colour is presumably due to an " agouti " factor like that of the rodents. Sheep. From such fragments of evidence as I can find it seems that the white of ordinary sheep is not, as in the pig, a dominant to colour, but a recessive. From Darwin's record {An. and Fits. 11. p. 4) of the appearance of all black sheep from a cross between white Southdown ewes and a Spanish ram with two black spots, it may perhaps be inferred that the black colour is due to complementary factors. Black face and white face give a speckled face in the heterozygote. The dark ring round the eyes depends on a separable factor (Wood, 312). Fowis. Colours very complicated and genetics imperfectly understood. Whites are of various kinds, one being dominant and at least two recessive. Colour depends on complementary factors which may be borne by whites. Black is an imperfect dominant to black-red. Brown-red a dominant to black-red (Fig. 11). . Blue is a heterozygous colour of black and a splashed white. The red and yellow pigments of the black-red cock may be replaced by white, thus giving the Silver Duckwing, but in the hen the red of the breast is not thus replaced, and the Duckwing hen differs from the black- red in having the yellow of hackle and mantle replaced by white. These replacements may occur as consequence of recombination in F^ from crosses between white and coloured breeds, whence it is to be inferred that the replaced reds and yellows depend on a special factor. Pencilling is a dominant to its absence, and various mottlings are also dominant. The descent of colour is influenced in some cases by sex in ways not yet clear, and in both sexes heterozygous types occur. The relations of the buff of Cochin (and of other breeds derived from it) to other colours are not yet known. As regards colour of the down, the brown striped condition is dominant to the pale brown down associated, for example, with Wheaten. Both types of down may occur in the same breed (^.i^. Indian Game), and the fact seems to have no relation to the adult plumage. R.E.C. (21, 22). * Mr Staples-Browne has given me confirmatory evidence. II] Colours of Animals 43 The black body-pigmentation of the Silky is a dominant, but may be inhibited by another factor, the descent of which is sex-limited. See Chap. x. Another sex-limited descent is to be found in the relations of Cuckoo to black, but the details have not been ascertained (Spillman, 249, 253, a). The daw-eye is recessive to red. The dark iris is usually a dominant to red. Red ear-lobe is an imperfect dominant to white. R.E.C. (19). Principal papers dealing with these features are R.E.C. (19-22), Hurst (156); Davenport (loi). Pigeons. Black is dominant to blue. Relations of red and yellow not clear. Black and blue are dominant to white of Fantail ; heterozygotes generally, if not always, having some white. Chequering dominant to its absence. The white rump of the Rock-pigeon is dominant to blue rump (Staples- Browne, 255). Canaries. Presence of black, as in green and pied types, dominant to absence of black as in the various yellows and cinnamons. The pink eye of cinnamons is recessive to black eye, with a sex-limited inheritance. There are probably several heterozygous colours, but in order to determine these the genetic interrelationships of the various Yellows, Jonque, Mealy, must be worked out. The " cap " and lacing of the Lizard are dominants. (Noorduijn, 213-5; Davenport, 105; F. M. Durham, unpublished.) Axolotl. Crosses between normal and albino gave dominance of pigmentation. Subsequent generations showed remarkable and as yet unique features. In F2 dark larvae to white larvae were 3:1; but the white F.^ larvae, though remaining red-eyed, acquired a certain amount of pigment, sometimes distributed as a metameric chequering. No thorough albino occurred in JF^- When however these chequered albinos w^ere bred with a true albino, the latter was found dominant and true albinos were produced (Hacker, 143-4). Hacker compares this case with the phenomena seen in Mice, &c., but there is an essential distinction in the fact that in all other instances true albinos come in Fo and in the dominance of the true albinism over the chequered character. It would be interesting to see whether the development of pigment in the Fc, whites is in any way dependent on conditions. Lepidoptera. Bombyx mori (Silkworm). Brown colour seen in a dark variety of the moth was proved to be an imperfect dominant (Coutagne, '^2)1 P- 122). The larvae have many colour-types. Coutagne used a dark ^'"inoricaiaV variety, a variety with transverse stripes, and ordinary white larvae. Both the coloured types are dominant to white, but when the dark self-colour factor and the stripe-factor are present in the same larva the stripes show on the dark ground-colour {ibid. p. 142). Toyama (268) also made many experiments with the colours of the larvae. He found striping a dominant over plain white. In certain F .^ families from striped x white a new pale 44 Colours of Ajiiinals [ch. form was produced (p. 349). The ratio of striped : common marked white : pale unmarked white is as ShuU points out (242) 12:3:1 (actually 1463, 363, 126). I incline to interpret this as signifying that the 12 striped were in reality of two kinds, in the ratio 9 : 3, but that the distinction between the common and the pale was not easy to detect in the striped class. On this view the striped parent was a "pale." Shull regards the paleness as " latent " in both parents. As regards the colour of silk there is a complication. In Toyama's experiments yellow was always a dominant to white. Coutagne sometimes obtained this result, but (/. c. p. 123) the white of a race called " Blanc des Alpes " proved to be dominant to yellow. Abraxas grossulariata (Currant Moth). The type is dominant to var. lacticolor (q.v.), Doncaster (iii, 114). See Plate I, figs, i, 2. The pecu- liarities of this case are discussed in connection with Sex. Mr L. W. Newman has been good enough to send me information as to a cross between A. grossulariata and the var. varleyata. This is a nearly black suffused form (see Porritt, Eiit. Rec. xv. p. 10). F^ was typical grossulariata, and in 7^2 there were 24 typical and 7 varleyata (4 d", 3 ? )• Angerona priuiaria. The dark-banded var. sordiata dominant to the normal, reticulated type. In the heterozygotes the lighter bands are more or less reticulated. Doncaster (m). See Plate I, figs. 7-10. Xa7ithorhoe ferriigata. The form with purplish band is dominant to that with black band. Prout, L. B. (223). Hemerophila abruptaria : the dark var. may be inferred to be a dominant to the type, from the experiments of Harris (147). Plate I, figs. 5, 6. Amphidasys betularia (Peppered Moth). The normal is almost certainly recessive to the black, or doubledayaria form (see, for example, the records of Main and Harrison, 192). Plate I, figs. 3, 4. Triphaena comes : the reddish form is recessive to the melanic (see Bacot, 3, and Prout, 222). Callimorpha doi7iinula : red of the hind wings is dominant to yellow. Standfuss (253, /;, p. 222). Mr L. W. Newman has kindly given me in- formation that he bred 34 reds and 10 yellows in 7^2- Aglia tau: type is recessive to the dark form lugens. Standfuss (253, b^ p. 311), (see Chap, x, for the sex-distribution of these varieties). Lasiocampa quercus (Oak-egger). The heredity of colours of the hairs of the larvae has been investigated by Bacot (4) and Warburg (302). Several varieties were studied and their genetic interrelationships are not altogether certain, but it appeared that red fur of var. sicula was dominant to the white of the var. meridionalis. When English and French races were crossed, various blend-forms were produced in F-^. COLEOPTERA. Most of the observations thus far made relate to Phytophaga. Com- plications were met with in all the cases investigated by Miss McCracken. I have not been able clearly to understand the exact procedure followed in the matings and must refer the reader to the original papers. Plate w¥'0 : -^ > - "^-.* ^ ', ^ • •• *%.• I ^^^^^ m' <'^ W ■^'■'■■■"' w^» 10 I I. Abraxas grossulariata. 2. Ditto var. lacticolor. 3. The var. Doubledayaria of 4. Amphidasys betiilaria. 5. Dark var. of 6. Hemerophila abriiptaria. 7. and 8. Male and female var. sordiata of 9. and 10. Angerona prunaria^ male and female. n] Colours of Animals 45 Lina lapponica. Two forms occur, one having elytra spotted with black on a brown ground, while in the other the elytra are entirely black. The latter is recessive, and the formation of the brown pigment in the ground is thus due to a dominant factor. McCracken (189, 199). Melasoma {Lina) scripta. A totally black form was found to be re- cessive to the spotted type. Two intermediate conditions may occur, one of which may be a homozygous type (see original). McCracken (191). Gastroidea dissimilis. Two forms, either deep blue-black, or shiny bright green. The latter is recessive. Curious complication as regards numerical results. McCracken (190). Leptinotarsa dece77ilineata (Colorado potato beetle). Evidence obtained by Tower (266, pp. 275-9) indicates that a variety called by him pallida behaves as a recessive. The var. melaiiothorax of the species Z. multi- tae?iiata also proved to be a recessive to its type {ilnd. pp. 284, 292, 293). Various other more complex phenomena are recorded {q. v.). Crioceris asparagt (Asparagus beetle). Each elytron has three yellow areas or spots on a blue-black ground. The upper spot is sometimes united to the middle one. This condition proved to be recessive to that in which the spots are separate, but all intermediate conditions occur [being presumably heterozygous]. Lutz (182). MOLLUSCA. Helix horte?isis and H. 7ie7?ioralis. The U7iba7ided variety is dominant to the banded types in both the species, sometimes completely, some- times partially. Generally also red ground-colour is dominant to yellow or brown, but this effect may diminish with age of the hybrid individual. Lang (167-9). For details as to hybrids between the two species see (169). As regards dominance of colours very little In the way of general rule can yet be predicated, nor till the chemistry of pigments is much better understood is It likely that such general rules will be discovered. It may, however, be remarked that actual albinism, the total absence of pig- mentation, is always, so far as we know, a recessive character in both animals and plants. Curious cases never- theless are known both in animals and plants where a partial whiteness, which we should a priori Imagine to be a kind of albinism, behaves as a dominant"^'. Another fact of a somewhat paradoxical nature is to be seen in the behaviour of some of the very deep colours, red and purples, characteristic of the flowers and other parts of some garden strains. These more intense colours both In Primula Sinensis, in the Stock, and in the Sweet Pea (and doubtless * See Chap. v. 46 Preliminary Deductions [ch. Cyclaine7i) are recessive to the paler and more commonplace tints ^. Though in the case of colours in plants which are due to the development of pigmented sap, albinos are recessive to the coloured types, the yellow or cream colour due to the presence of yellow chronioplasts is recessive to the colourless condition of the chromoplasts. Hence we find, what at first seems paradoxical, that white flowers are dominant over cream-coloured flowers. Yellow dependent on sap-zo\oMX is dominant to the corresponding white. With regard to the behaviour of black pigment, which might naturally be supposed to have similar genetic pro- perties in the various animals, no quite satisfactory general rule can be laid down. The presence of black pigment is commonly dominant to the absence of black, as in the race- horse, where chestnut, namely the absence of black "points" is recessive to the presence of such " points " as in bays and browns. Most cases, however, such as that of the mouse, and other animals in which black pigment exists intimately mixed with other pigments are not so simple as this and involve special problems. In so far as the features of those cases can be expressed in the simple terminology hitherto used, these blacks must be classed as recessive to the normal colours. Further particulars will be given in the chapters on Colour. Prelhninary Dediictio7is from Mendelian Phe^iomena. It will be observed that animals and plants, as such, do not show any difference in their manner of heredity. Inheritance on simple Mendelian lines may be followed by characters of very diverse kinds, such as height, shape, chemical constitution, colour, and several structural features. In view of such a list the important question arises whether there is any distinct category or class of characters to which the Mendelian system does not apply. Various possible limitations may be discovered when the phenomena have been more fully examined, but it may be stated at once that no such class of characters has hitherto been identified. ■*^ In Antirrhinum Miss Wheldale finds that the deeper magentas are recessive to the ordinary magentas, but in the crimso7i-red series the paler are recessive to the deeper tints. ii] Meristic Cases 47 As yet only one example of a character which can at all readily be interpreted as meinstic in nature has been shown to have a Mendelian inheritance. This is the case of the reduction in number of the human phalanges in brachydactyly. We speak of a character as '' meristic " when it manifests itself in respect of the number of parts into which the body or one of its organs is divided. Meristic characters are in several ways distinguishable from other features of bodily organisation. The physiological occur- rences which result in meristic variations are in all likelihood distinct from those which produce substantive changes, and exceptional interest would attach to any investigation of the genetic properties of such variations. Polydactylism is of course a meristic feature, but it may involve something more than a divisional change, pure and simple, since change in the number of digits is usually accompanied by change in the distribution of differentiation. A case in which the disturbance of differentiation is not so evident is provided by the cross between Oxalis tetraphylla^ much cultivated in Germany as Glilcksklee, and one of the forms with three leaflets. This cross was partially investigated by Hildebrand"^, who used O. latifolia. He found that the 3-fold character was an imperfect dominant, the leaves being 3-fold with the exception of occasional 4-fold leaves which appeared for the most part at the flowering period. The hybrids were fully fertile, but their progeny has not been studied. Satisfactory meristic cases from which all confusing elements are eliminated must be rare, but it is greatly to be hoped that they will now be searched for. It is most desirable that cases of difference in the ground- plan numbers of some radial type will be found amenable to experimental tests. Here the problem may be seen in a somewhat simplified form on account of the elimination of serial differentiation f. ■**■ Y{\\^t\yc2Ji^^ Jenaische Ztsch. f. Naturiv. 1889, xxni. N. F. xvi. p. 56. t Since this paragraph was set up Price and Drinkard's (221) evidence has been published showing the dominance of tivo chambers in the fruit of the tomato over the many-chambered condition. More evidence as to such cases would be welcome. Drinkwater's recent discovery as to the bones of the brachydactylous fingers, showing that the middle phalanx is actually formed as a distinct bone which afterwards unites with the distal phalanx, raises considerable doubt whether the variation in that case is meristic after all. 48 Rights and Lefts [ch. Respecting the genetics of one most interesting class of variations evidence is scanty. This is right- and left- handedness. From Mayers'^ observations on Partula (Gastropod) we learn that parents of either twist may bear young of either twist. The numbers in the uteri were so small that the absolute numbers are insignificant, and it may be an accident that no mixture of types was found in any one uterus. Langf bred numerous left-handed Helix- pomatia with each other and obtained thousands of young, all right-handed, which in their turn again produced ex- clusively right-handed offspring. Direction of twist is a fundamental meristic phenomenon, being as Crampton and Conklin have proved, determined as early as the first cleavage-plane in the ^ {CBd) 10 {CbD) 8 {Cbd) 2 (various) 12 Calc. 27-4 pz g-i 3'o l6'2 On microscopical examination the dilution of the pig- ment seems to consist in a diminution in the number of the pigment granules, and not In a reduction of their size. It Is interesting to notice that many different animals have varieties In which the dilution has proceeded to the same extent. For example the particular dilution of black which we call blue is known as characterizing definite varieties in both rabbits, cats and mice, not to mention other cases less certainly comparable. As was mentioned in the description of the Andalusian fowl and its genetic features the blue colour of that breed is not comparable with the blues we have been discussing. In the mouse the blue is gametic, being a condition which can be carried by the germ cells, while in the Andalusian the blue is zygotic and depends on the collocation In one individual of one germ-cell bearing black with another which does not bear black. It is interesting, as exemplify- ing the danger of reasoning from analogy where genetic phenomena are concerned, that the blue-roan of cattle should not follow the same rules as the other mammalian blue varieties, but should conform rather to the Andalusian system. Of course blue roan is, even to the eye, not the homogeneous blue of the blue cat or mouse, but a mixture of white or whitish hairs among blue and black ones. Still on analogy we might have expected the blue of cattle to be capable of representation In the germ-cells, but the facts, so far as I can discover, afford no support to that supposition"^. * From such meagre evidence as I have obtained it is likely that the blue, or blue dapple, of Dachshunds and other hounds is also a heterozygous combination. As to the blue of Greyhounds and Great Danes I have no information, but I suspect it to be a dilute black capable of being bred true. 6—2 84 Colours of Mice [ch. Wholly Coloured and Pied Varieties, In the analysis of the relationship between the whole or self-coloured forms and the varieties which differ from them in having an admixture of white many curious and specific phenomena are met with. It is probably true to say that generally the whole-colour is dominant to the pied, but several examples to the contrary are already known. In all the cases yet studied the genetic properties of the pied types can be represented factorially by regarding the pattern or distribution of the colour as due to a distinct factor or to its absence. Where the whole-colour is a dominant, the presence of the factor must be taken as causing that distribution, so that in the absence of that factor the individual is pied. Conversely if the pied type is dominant the presence of the pattern-factor acts by re- straining the distribution of the colour, and in the absence of the restraining factor the whole-colour prevails. One of the clearest cases is that studied by Hurst in the rabbit, where the pattern known as "Dutch-marked" was proved to be a recessive. In the Dutch rabbit the hind quarters are coloured, for example with grey or black, while the front half of the body is white except for a patch of colour— grey or black as the case may be — surrounding the eyes and covering the ears. This pattern though fluctuating in minor respects is fairly definite and is at once distinguishable both from the self-colours and from the various other pied forms. The cross between a pure self- colour and a Dutch gives F^ nearly self-colour with /% consisting of 3 selfs : i Dutch in the ordinary way. An interesting feature is however to be observed in the fact that the heterozygotes between self and Dutch generally (? always) have some small amount of white collar, marked especially on the back behind the head. Apart from Mendelian experiment it might have been supposed that such a white mark showed that the animal contained some albino blood. Experiment shows on the contrary that the actual self-colours may be carrying albino as a recessive, while the small white mark is an indication of heterozygosis with Dutch pattern only. The behaviour is in all respects as if the whole-colour IV] Pied Varieties 85 pattern depended on the existence of a dominant. So also, exactly as in the case of dilution and saturation, an albino may carry either the whole-colour factor, or its absence. Consequently when a Dutch rabbit is crossed with an albino bearing self-colour, F^ is self-colour and /% gives 9 self-colour : 3 Dutch : 4 albinos. In rats there is a type of colouring w^hich rather closely corresponds to the Dutch of rabbits. This has a "hood" of colour over the back of the head and shoulders continued down the back in a stripe which may either be entire or broken into spots. This behaves towards the self-colour just as does the Dutch in the rabbit ; and just as in the rabbit the heterozygote between the self-pattern and this hooded type always has some white. Crampe i^-^, a) observed this fact before the days of Mendelian analysis. He noticed that when a wild grey rat was crossed with an albino, F^, as we call it, might be a real self-colour, or might have some white; but that subsequently hooded rats only occurred as offspring of those which had some white. Such rats, which are nearly whole-colour but have a little white, are know^n in the fancy as the "Irish" type. Experiments made by Doncaster and also by Mudge indicate that these may be again divided into two subordinate classes, distin- guished by the amount and distribution of the white, and there is some evidence to show that these two types of Irish rats have distinct gametic compositions. Though the Dutch rabbit and the hooded rat are each such clearly recognizable types, yet within these types there is great fluctuation, and it is practically certain that the fancier's ideal Dutch-pattern rabbit, with the demarcation between the colour and the white passing in a sharp trans- verse line across the middle of the animal, does not exist as a gametic entity. Such individuals of course come into existence from time to time, but selection will not fix their type. As Castle and MacCurdy (183) have shown In the case of rats, selection may nevertheless to a considerable extent be effective In producing hooded types with more colour, and with less colour, which are evidently gametic possibilities. In mice no pied type exists which Is quite so definite as the Dutch pattern of rabbits or the hooded type of rats, 86 Pied Varieties [ch. and by taking pains every gradation In amount of white could be found represented among fancy mice. Experience, however, soon shows that some at least of these are gametic types, while presumably others are the consequence of various heterozygous combinations. Cuenot's work with mice led him to the conclusion that in mice the several degrees of piedness are recessive to each other In the order of the amount of white, those with more white being re- cessive to those with less. In general terms this is a true account, but we have not found the rule to apply quite strictly even in mice, perhaps through the existence of the complication next to be considered. It should be remarked also that no general statement can be made as to dominance of self-colours over pied which is applicable to animals In general, and on a wide survey of the results of breeding many paradoxical occurrences are met with. Especially curious are the cases, by no means very rare, in which a cross between a domesticated and a wild animal, e.g. dog and wolf — has produced a partially pied offspring. Dominant Pied Pypes. So far we have been considering the behaviour of pied patterns recessive to the whole-coloured types. Though several points remain for Investigation the genetic relations of these patterns are fairly clear. A remarkable complication has next to be mentioned. Both In the rabbit and the mouse it is now known that in addition to the pied types which are recessive there are others which are domi- nant to the whole-coloured form. Hurst has proved this for the variety of rabbit called " English pattern." This animal Is white with spots of colour (black, grey, or other- wise) generally of small size on the sides of the body, a patch over the eyes, and a " chain " of spots sometimes nearly continuous down the middle of the back. In ideal specimens the spots should be of a special form and have a definite distribution, but neither of these features seems to be gametic. The English type Is not now much in fashion and looks uncommon to those used to modern rabbits, but formerly it was very abundant. IV] A Dominant Piebald Type 87 Hurst's experiments (160) showed that this pattern is an ordinary dominant to self-colour — a definite but most unexpected result. More lately Miss Durham (116) has found a dominant, and doubtless analogous, pied type in mice. So far no criterion has been discovered which distinguishes this dominant pied pattern externally from the recessive ones, but in breeding the distinction is perfectly sharp. It is likely that the factor for this new dominant was brought into Miss Durham's strains by the introduction of the type called ''black-eyed white," but the evidence is not perfectly clear on this point. As may be supposed, the combination of the dominant pied with the recessive pied in the same strain gave results which it was impossible to disentangle, though when each of these types was isolated the course of descent was perfectly clear. The case is interesting not merely as exemplifying a new^ kind of factor, but as illus- trating a type of complication that may very possibly have to be reckoned with in other difficult and as yet incom- prehensible sets of phenomena. Another example of a pied condition dominant to self- colour has been seen in our poultry experiments (22); but since in fowls some of the wholly white breeds are prac- tically dominant in whiteness, it is only to be expected that some of the partial whites should also show dominance. In plants, as in animals, no general rule can be laid down as to the dominance of self-coloured or parti-coloured flowers. In the Sweet Pea, for example, the old-fashioned " Painted Lady" (see Plate V) has a red standard and wings nearly white. It is thus a bi-colour type, but it is dominant to the self-coloured reds and pinks. On the other hand in Antirrhinum the self-coloured types are always dominant to the "Delilah" forms which have the lips coloured and the tube or throat white, as was first shown by de Vries. Miss Wheldale who has since worked on a large scale with Antirrhimun (303) found that for each shade of flower which exists as a self-colour, a corre- sponding Delilah or white-tubed type can be made which behaves to the corresponding self-colour as a simple reces- sive. CHAPTER V HEREDITY OF COLOUR— continued. Albinos giving Coloured Offspring ; Reversion on Crossing — Various Kinds of Whites — Stocks — Orchids — Pigeons — Fowls — Primula. We have seen that albinos, both animal and vegetable, though devoid of pigments, may yet bear factors which are capable of determining the quality and distribution of colour when they meet with colour in the zygote. Particular colours may thus be due to the co-existence of several distinct determining factors, each with an independent distribution among the germ-cells. The grey of the rabbit for instance is caused by the presence (i) of a colour- element or elements ; (2) of a factor which determines them to be the mixture we call grey, and not for instance, black or yellow. In the case of certain plants this analysis can be carried a step further, and the formation of colour at all in the flowers can be proved to depend on the co-existence of two complementary factors in the individual. The first indication of this phenomenon was found in the fact that two plants each totally devoid of colour in the flowers and stems, and each breeding true to albinism, may when crossed together give purple flowers in F^. The two white parents each contain a factor which, alone, is incap- able of forming colour. Each of these factors is indepen- dently transmitted in gametogenesis, and thus in 7% the ratio of coloured individuals to whites is 9 : 7. This pro- portion depends on the fact that a series of 16 individuals is necessary to exhibit all the possible combinations of germ-cells, for, as in any example of hybridisation involving two pairs of allelomorphs, there will be four types of female cell and four types of male cell produced by F^. Of these CH. V] Reversion in Colour 89 sixteen individuals 9 will contain both the dominant or present factors, while of the remaining 7 individuals, 3 will contain the one dominant, 3 will contain the other, and i will contain neither. There will therefore be 9 which are coloured and 7 which are albino. In the diagrammatic scheme C and R are the symbols representing the two complementary factors, c and r being their respective allelo- morphic absences. Fig. 16. Composition of the 9 coloured and 7 albino offspring in F.2 from the cross between albino Cr with albino cR, showing the ratio 9 coloured : 7 albinos. Two examples of this phenomenon have been studied in detail. The first is that provided by the Sweet Pea [Lathy 7^zis odorahts) which has formed the subject of ex- periments carried on jointly by Mr Punnett and myself for some years. The work was begun by crossing two white sweet peas belonging to the variety Emily Henderson. These plants were alike in every respect so far as could be perceived, excepting that the shapes of their pollen grains differed, the one having the normal long pollen grains of the species, while the other had roundish grains. The object of the experiment was to trace the descent of the pollen-character and at the beginning no question of colour was entertained. When F^ was grown however it was clear that here was a 90 Reversion in Colour [ch. remarkable opportunity of studying a reversion in colour due to crossing, for these plants instead of being white were purple like the wild Sicilian plant from which our cultivated sweet peas are descended. The facts respecting the colour inheritance will now be given. With regard to the pollen-shape it must suffice at present to state that the long shape is a dominant and the round a recessive. The details as to the distribution of these tw^o shapes among the /% individuals, which are interesting and have greatly aided the development of genetic theory, are given in the chapter dealing with the phenomenon of "gametic coupling." The present section is concerned with colour only. When the reversionary F^ generation was first seen its nature was entirely mysterious. When 7% was raised from these F^ plants the series consisted of a mixture of plants, some coloured and some white. In some cases the series of coloured plants consisted of two kinds only, purples like F^, and a red bicolour type, the w^ell-known old variety called '' Painted Lady." In other cases F^ contained besides those mentioned, two quite distinct additional types of purples and two corresponding additional types of reds. The phenomena, though, as will be shown, in reality very simple, presented superficially an appearance of great complexity. Further difficulties were met with in the fact which was soon discovered, that the cross between long- pollened whites and round-pollened whites does not always give the coloured types, but may result in ordinary whites only. It is unnecessary to go through the long series of steps by which the analysis of the phenomena was carried out. The meaning of the facts is now perfectly clear and they can all be arranged in one consistent scheme. Of the two white parents originally used the one possessed one of the two factors we have called C and R, the other introducing the other complementary factor. The meeting of these two elements produces colour in the flower. If no other epistatic factor is present their colour is red. As a matter of experiment however one of the parents, proved afterwards to be that which had the long pollen, did carry V] Colours of Sweet Peas 91 such an epistatic factor whose property is to make the colour purple, just as the factor B, in the mouse, makes the coat colour black. F^ was therefore purple, and 7% con- sisted of 27 purples : 9 reds : 28 whites, as shown below. The factor w^hich determines the colour to be purple is represented as B, the blue factor. The diagram (Fig. 17) exhibits in a tabular form the composition of the various /% plants. The distribution of the J]Rb9S^ (3R-b\5X CR:^y9< CRV/V CReW< crbVSX CRb'/Z ^ Til / / / 'CrB\XX CRWS^ ^CrBW< fCrbXXX .Crb/// ^cRb// ^RbxSo ^rb>99< -Crb^X/ ;cRbV% crbsSc CRBxx) CRbV/ tlrb^XA CRT^vKX CRB^XX cRVX3\X CRbVV /crbV% CrB CrB CrB Crb Crb CrB Crb Crb CrB crB CrB crb Crb crB Crb crb ^rbW gRB9So ^'RB<<5X gRl^^g ^RbV>X CR^S^ /^Rb%/ €Rb//^ cRB cRB cRB cRb cRb cRb cRb cRB XcrB '^^SSSSte*^* J' MO o T3 O o c ^ "-^ t^ Q- — u '3, ;-> 3 CU J3 -J ^ ■£ c / . '^ d\ > X , >> , "■ Cu P-U *-< t/5 's bJO 0) w t/i n: -:3 c r^ o t/5 r" ^ ci "oJ . o 1-4 trt 00 a V] Colours of Sweet Peas 93 and round-pollened respectively, which outwardly were indistinguishable from the two original parents. As we now know, by the redistribution of factors a great variety of whites is in reality produced in /%, and only those pairs which bear the complementary factors C and R can give coloured offspring. This apparent dissimilarity between the behaviour of the extracted forms and that of the pure types from which they are derived has been adduced as being inconsistent with Mendelian principles. When how- ever the factorial composition of the various individuals is correctly represented it is evident that this multiplicity of composition in /% is what must be expected, and there is abundant evidence that such complications exist in all the plants and animals that have been experimentally studied on an adequate scale ^. Subordinate Types in F^. We have now to consider the meaning of the appearance of subordinate types in the purple and in the red classes which is a feature of some F., families. Of these there are frequently two. The first is characterized by having the wing-petals much more fully coloured than those of the F^ purple, or of the Painted Lady. These distinctions are due to the operation of an epistatic factor for the lighter wing-colour. When this factor is absent the wings are dark, and the flower, if a purple, has the wings deep purple * Factors which may thus exist without making their presence visible have been named by Tschermak " Cryptomeres." The term is open to the objection which zoologists especially will feel, that it may cause con- fusion owing to the fact that the series of words containing "mere "are now universally understood to refer to phenomena of division — or Meristic features. When the study of that part of genetics which is concerned with meristic variation is more fully developed this confusion will be aggravated. For the present moreover the expression "factor," qualified, if necessary, as unseen, seems sufficiently precise. The words "latent" and "latency" must be applied in these cases with great care, if at all. Various writers have, for instance, spoken of the purple colour as " latent " in a white flower. This description is quite misleading, for the colour need not be present in such plants. The term "latent" is only admissible in application to X\\q factors (as here, to the factor which can turn the colouring matters purple) not to the characters^ except loosely to those which are actually present but hidden owing to the operation of some epistatic factor. 94 Colottrs of Sweet Peas [ch. Instead of a light purplish blue. Similarly reds having the factor for light-wing are Painted Lady, If without It are of the type called Miss Hunt, with wings of a full pink. Another pair of subordinate classes is due to the absence of a factor for whole-colour. When this factor is absent the purples become purple picotees, with an edging of colour on a white ground, and the reds become w^hat we have called "tinged whites" ; a type with a little pink in the standard only. The purple picotees, when the bud Is just opening, have a purple edge to the standard and a blue edge to the wings, these edges being all that remains of the colours proper to these parts In whole-coloured purple. The "tinged w^hite," which is really a red picotee, has the red of the Painted Lady standard reduced to a few pink lines in the centre of the standard. Both these picotee forms, like the picotees of tulips and most other flowers, acquire a more general flush of colour as the flower ages. These various types are Illustrated in the coloured plate (Plate III), from which their peculiarities will be easily appreciated. Among the picotees of course there are two types of purples and two types of reds, diflering in the presence or absence of the factor for light wing, but In forms so nearly w^hlte these differences are evasive, and we have not found it possible to classify the picotee class In respect of these distinctions. It Is to be observed that the simultaneous appearance of the subordinate types In each of the classes is a charac- teristic and necessary feature of the normal Mendelian processes. It would be impossible, for instance, to make a plant which could throw purple picotees In F^ without red picotees. If the family contain both purples and reds. Conversely, If the plants throw purples and reds, and also whole-colours and picotees, then both kinds must be repre- sented In both classes. In any examination of 7% families, so soon as one pair of large classes can be recognized and one subordinate type perceived In one of them, search should be made In the other class In order to find the corresponding subordinate type in It also. If In one class a type exists which is never to be found In the other class though a long series has been examined, some complication is to be suspected, but whether the phenomenon is due to V] Colours of Stocks 95 coupling or to some other cause can only be decided by the circumstances of the special case (see, for example, the case of Stocks, in which plants with hoary leaves are wanting in the classes w^ithout sap-colour in the flower (Chap, viii), and the distribution of the hooded standard among Sweet Peas (Chap. ix)). Coloicrs of Stocks [Mattkiola). The experiments of Miss E. R. Saunders have revealed a closely similar inter-relationship between the various colour-types in Stocks. There also coloured F^ is produced by crossing two types both destitute of sap-colour. The tw^o types giving this result are white and cream. The pale yellow of the cream is due not to sap-colour but to chromoplasts, and thus the two types are as regards sap- colour both albinos. In this case also F^ has the common purple flower which we may regard as reversionary '^, and in /% there appear both purples and reds. These results can be represented exactly like those in the Sweet Pea case. The sap-colour is due to the meeting of complementary factors, C and R, derived from the two parents respectively. In the absence of the "blue" factor B, the colour produced by CR is red, but when B is present the colour is purple. As a matter of fact it can easily be proved that B is intro- duced by the original white parent ; for when a red is crossed with the white, F^ is purple, but when a red is crossed with a cream, F^ is red, showing that the cream is devoid of B, but that the white possesses it. The numbers in F^^ are also on the plan described for the Sweet Pea. The total of sap-coloured plants to the total of non-sap-coloured (whites -f creams) averages 9 : 7, and w^hen the proportion of purples to reds is also repre- sented the ratio is 27 purples : 9 reds : 28 whites and creams. Just as in the Sweet Pea again there is a series of subordinate types among the purples in /%» ^"^d a corre- sponding series of types among the reds. For example, there is a dilute form of purple which the horticulturists * As will be described later, in this case when both parents have glabrous leaves, F^ reverts to the hoary leaves of the original type. 96 Reversion in Orchids [ch. call ''plum," and a dilute form of red which is known as " copper," &c. Any of these colours may be on a white ground, or on a cream ground"^. Albino Orchids giving Coloured F^, In Orchids a series of facts has been observed which are exactly comparable with those described in the Sweet Peas and Stocks. There are several albino orchids in the genus Cypripedium. Of these some when crossed together give coloured offspring, and others crossed together give albinos only. Mr Hurst t most kindly furnished me with a complete list of the results of crossing albino Orchids so far as they have been published. This list covers a very wide range of cases to which no reference can be made here. As regards Cypripedium most of the experiments were made by Mr N. C. Cookson of Wylam. I am much Indebted to him and to his gardener Mr H. J. Chapman for Information about them. For example : C callosum Sanderae x C bellatuhtm albmn gives all offspring coloured (25 plants bred by Mr Cookson). C Lawrenceanum Hyeanum x C bellatulum album, all offspring coloured (Cookson). On the other hand C Lawrenceanum Hyeanurn x C callostim Sanderae gives offspring all albinos (viz. C '' Maudiae'') (raised by Messrs Charlesworth and independently by Baron Schroder). If therefore we suppose that bellatulum album is carrying one of the complementary factors, say C, and that the other factor R is present in callositm. Sanderae and In Lawrence- anum Hyeanum, the results are correctly and consistently represented. As regards the behaviour of the pure types however there is one occurrence to be recorded which cannot be explained. Of the albinos named above, one, callosum Sanderae has ■^ Though all the various combinations occur it is probable that the distribution of white and cream is by no means a simple one, but the analysis of the inter-relationship between white and cream has not yet been worked out completely. t See also Hurst, Gard. Chron. Feb. 6, 1909, p. 81. v] Reversion in Orchids 97 been self-fertilised and gave albinos only, as must be ex- pected. Lawrenceanum Hyeanum when selfed, gave 14 albinos, but in addition, one coloii7^ed plmit, which Mr Cook- son tells me cannot be thought to have resulted from error. Such an occurrence is as yet unaccountable, but otherwise all is clear. In addition to the cases given above there are many in which C. insigjie Sajideriantim and insigne Sa7iderae were used, but Mr Hurst tells me that of these the latter is certainly a tinged type, while the former has some dark hairs on its petals which may contain sap-colour. Some of the crosses with these types gave reversionary offspring, and others gave various whites. The genus Cattleya provides another list of comparable cases. For instance C. Mossiae Wagenerix C. Gaskelliana alba gave 4 albinos (Hye). ,, „ X C intermedia alba gave 14 albinos (Holford). C. Schrdderae alba x C. intermedia alba gave coloured offspring (Cookson). C. Gaskelliana alba x C . Ha7^7'iso7t2ana alba g^iv^/^coXovw^d plants (Thwaites). ,, „ X C War7ieri alba gave 5 albinos and 2 coloured plants (Peeters). From these statements it may with great probability be inferred that the albinism of Mossiae Wagene7H, Gaskelliana alba and i7ite7'77tedia alba is due to the absence of one factor (say C) ; that in Ha7^7dso7iia7ia alba and Schrdde7^ae alba it is the other complementary factor which is wanting (say R) ; and that Wa7'7ie7''i alba is heterozygous for the presence of one of them (being Rr, on this scheme). It is to be hoped that some Orchid grower will make the various unknown combinations and extend the series. There are, besides the cases enumerated, many instances in Cyp7'ipediu77i, Cattleya, and other genera, of various tinged types breeding true when self-fertilised and giving reversionary coloured offspring when crossed. B. H. 7 98 Comparison between [ch. The Case of Antirrhinum {Snapdragon^ The experiments of Miss Wheldale (303) on Antir- rhinmn show that in that plant the appearance of the red and magenta flower-colours is similarly dependent on the presence of two complementary factors. In one respect, however, this case differs from those which we have been considering ; for whereas in them either of the two colour- factors, spoken of as C and R, might be present in a truly white-flowered plant, it appears that in Antirrhinum one of the two factors produces either a bright yellow or a very pale '' ivory " yellow in the flowers. The other complementary factor may be present in actual whites. Hence magentas or reds may be produced by the cross between white and yellowy or white and ivory, but hitherto crosses between actual whites have not given coloured offspring. If in the Sweet Pea for instance the presence of C caused the flowers to be ivory instead of white the case would be like that of Antirrhinum. At first sight it might be supposed that the phenomenon seen mMatthioia (Stocks), where white x cream gives F^ with anthocyan colours, was more strictly parallel, but in the Stock the cream is a plastid-colour, whereas in Antirrhinum the ivory is a true sap -colour. It is a noticeable and rather unexpected feature in the case of Antirrhinum that the pale ivory is due to a factor epistatic on the bright yellow. The ivory is so pale that it might on casual examination be described as white ; never- theless it is definitely and completely dominant to the bright yellow type. Ivory crossed with white gives F-^ magenta-red but yellow crossed with the same white gives F^ crimson-red. The distinction between the two kinds of red is due to the factor which turns bright yellow into ivory. A comparison between the Sweet Peas or Stocks on the one hand and Antirrhimtm on the other, may be represented in tabular form. Using a terminology applicable to both cases let us designate the factors thus^ : Presence and absence of the "ferment." F, f. "Chromogen." C, c. „ ,, ,, ,, Epistatic factor. E, e. * In the previous discussion of the Sweet Pea the factors /^and E are designated R and B respectively. Sweet Pea Fee white fCe white fcE white fCE white FCe red FCE purple V] Antirrhintim and Sweet Peas 99 The following appearances will then be produced in the flowers by the various combinations. Antirrhmum white yellow white ivory crimson-red magenta-red. As regards the distribution of the pigments over the parts of the flower, many facts of importance have been discovered (see original papers). Several distinct factors are involved, and by careful analysis Miss Wheldale has been able to show that almost all the heterozygotes can be recognized and distinguished from the corresponding homo- zygotes. The degree to which it has been found possible to effect the genetic analysis of the types m Antirrhinum is very remarkable. At first sight the series of forms appears to consist of numberless intergrading tints, but by continued experiment Miss Wheldale has succeeded in disentangling the various genetic combinations and showing the factorial composition of almost all. Even as regards the inheritance of striping an approximate factorial representation has been worked out, though from the fact that striped types may occasionally have self-coloured offspring this part of the analysis must still remain incomplete. Needless to say there is as yet no system of colour- reproduction, whether by lithography or photography, which can be applied with accuracy sufficient to represent the various tones of colour, and for the most part the work is unpublished. Reversion on €7^0 s sing : the Nature of Variation. From the facts now before us a clear conception may be formed of the meaning of reversion occurring in conse- quence of a cross. This phenomenon is due to the meeting together of factors which are complementary to each other, and must be together present in order that the original or reversionary type may reappear. We have seen how in the mouse or rabbit the wild grey type is reproduced when 7—2 loo Reversion in Pigeons [ch. a black is crossed with an albino bearing the factor G, greyness. Here the reversion is due to the meeting of this factor with the colour-element present in the black animal. The case of the Sweet Pea or the Stock, when two colourless types crossed together produce coloured F^ is of exactly the same nature, except in so far as both of the complementary factors are in this case imperceptible until they meet each other. The occurrence of the ratio 9 coloured viz. reversionary : 7 uncoloured is a proof of the correctness of this mode of representing the facts. When reversion is correctly represented we are led on to form some mental picture of the essential nature of a variation. It must occur by the omission or by the intro- duction of a factor. Reversion in Pigeons. In connection with the reversion produced by crossing two white individuals the famous case of pigeons may be considered, though in that example one parent was coloured. Darwin was the first to draw attention prominently to such phenomena, and the experiments which he made consequent on a hint given by Boitard and Corbie have long been classical. In these experiments the blue colour with black bars resembling the colour of the wild rock pigeon {Columba livid) was reproduced as a result of cross- ing birds quite free from such characteristics. Darwin's experiments were very complicated"^. He was concerned only with the question of origin, and to provide a qualitative demonstration that the peculiarities of the blue rock pigeon could be reproduced from modern varieties which had ostensibly lost them. Mr Staples-Browne (255) has recently investigated the matter afresh in a simpler case. He crossed a Black Barb with a White F'antail, producing F^ black with some white feathers. From such F^ birds /% was produced, consisting of Blacks. Blacks Blues. Blues Whites, with white with white feathers. feathers. * Animals and Plants, ed. 2, 1885, i. p. 207. v] Doininant and Recessive IVhites loi The numbers were not sufficient to give the ratio definitely, but it was estabhshed by subsequent breeding that the whites bred true ; that the blues did not throw blacks ; but some at least of the blacks could throw blues. The blues breed true except in so far as they may throw whites'^. We have then the fact that a reversionary type was produced, but that it did not appear till F^. The meaning of this is obviously that the blue cannot appear in F^ because black is epistatic to it. The elements necessary to produce blue are all present in F^, but in it black also is present which conceals the blue. After the re-combinations, some F.^ forms contain the blue group of elements without black, and these are therefore blue. We do not know in this case that the Black Barb itself did not contain all that is necessary to the production of blue, and it is thus possible that the reversion on crossing may here be only a pheno- menon of re-combination. This could only be decided by a long statistical analysis. Genetic Prope^^ties of White Types which are not Albinos {Fowls and Primtcla). Do77tinant and Recessive Whites. Albinism, namely total deficiency of colour, as we have seen, is always recessive to the presence of colour f. In contrast with real albinos, the white forms which contain pigment in some part of their organisation show various phenomena of colour-inheritance. Among both animals and plants such types are known. For example, races of pigs, cats, mice, cattle, dogs and fowls exist in which the skin, hair or feathers are white, or nearly so, though the eye is pigmented. The ''white" races of mankind perhaps belong to the same category in some respects. Our know- ledge of the genetic behaviour of most of these types as regards pigmentation is still fragmentary, but what little is known points to the existence of considerable complications. Respecting white poultry, however, several features have been ascertained w^hich are interesting. In them we meet ■*■ Apparently only those blacks and blues which had some white feathers were capable of throwing whites. t In Axolotl (see p. 43) an alleged exception is recorded. I02 White Foiiuls [ch. a series of facts which perhaps, better than any other, illustrate the impossibility of understanding the significance of genetic phenomena without minute individual analysis. In poultry we know at least four different sorts of white plumage, each with its own special properties. As far as external appearances of the adults go there is little or nothing which would lead an observer to suspect that the genetic powers or capabilities of these four types were entirely distinct in respect of colour. In the first type the white acts as a dominant. Generally speaking when a purely white variety such as White Leg- horn is crossed with a coloured variety, such as Indian Game or Brown Leghorn, F^ is, in the newly hatched condition, white with a few specks of black. The black then present may be confined to a single hair, but we have met very few instances of total absence of such a speck or "tick" as it is called. As the dov/n is replaced by feathers similar ticks of black and brown appear in them. The position and amount of the ticks in the feathers bear no obvious relation to the down-ticks. These F^ birds, however, at all ages are substantially white birds, the amount of colour though definite, being almost always insignificant. Exceptions occur in the case of cocks, which sometimes have enough red in the wings to bring them into the category called by fanciers ''pile," though they are of the usual white in the downy stage. We may thus speak of White Leghorn as a dominant white. /% from such an F^ generation consists of a great variety of colours, the relations of which have not been worked out. But among the derivatives from various matings^ which Mr Punnett and I have carried out white birds have been produced which are, so far as their whiteness is concerned, simple recessives to colour. Sometimes, but not always, they have ticks of grey colour. These recessive whites breed true to whiteness as the White Leghorns do, and no one looking at them would doubt for a moment that they were moderate specimens of that breed. In addition to these the white of White Rosecomb * These birds arose from a mixture of Brown Leghorn, White Dorking and Indian Game. See Rep. Evol. Ctee, in. p. 19. Plate IV 4 % 1. Hen of Recessive White strain. 2. Hen of Silky breed. 3 and 4. Cock and hen, Fi from the cross between Recessive V^hitt^ hen X Silky cock. V] White Fowls 103 Bantams is also a recessive. Strictly speaking perhaps this breed should not be called white. Though appearing white on casual inspection, more careful examination always, I believe, reveals the presence of one or more ticks of grey colour. These flecks of grey colour may be extremely minute, but in our experience are never absent. The down, too, of the breed is of a faint bluish or smoky colour, though there is no such smokiness in the adult plumage. Lastly we have to consider the white of White Silky fowls, which is again a recessive to colour. The adult Silky is pure w^hite as a rule, but like the White Rosecomb, the down of the chickens has some colour usually (though perhaps not always). The colour in this case however is btoff, not blue or black, and it occurs on the sides of the head, in the region of the pale stripes of striped chickens, and on the rump. To resume: there are four kinds of whites. (i) Dominant white of White Leghorns. A pure white. (2) Recessive white of our own derivative strain. (3) Recessive white of Rosecomb Bantams. (4) Recessive white of Silky fowls. Crossed with coloured breeds such as Brown Leghorn (i) gives white birds with a little colour. (2), (3) and (4) give wholly coloured birds with the pigmentation of the plumage as fully de- veloped as in any coloured breeds. The next point concerns the relations of the various Recessive whites with each other, and is of considerable , interest. Our experiments have shown that the cross between Recessive whites (2) or (4) and the Silky gives exclusively whole-coloured birds \xi F^. In /% such birds give 9 whole-coloured to 7 whites. Of the 7 whites some have white down like the Recessive white (2), while others have buff in the down like the Silky (4)^. The appearance of coloured F^ from two whites, with 9 coloured : 7 white in /^2, is obviously a phenomenon comparable with that which has been described in the Sweet Peas and Stocks. The two white types (2 and 4) evidendy contain complementary * The numerical relations of these two forms are not yet clear. I04 IVhite Fowls [ch. factors which must be together present in the zygote in order that colour should be produced. The colour of the F^ birds is practically Black-red, approaching closely to the plumage of Gallus bankiva (see Plate IV). It might a priori be expected that since the White Rosecomb (3) has very distinct though light pigmentation in its down, a cross between this and the Silky (4) would be more likely to give a coloured F^. This, nevertheless, is not the case. The cross between these two breeds is an ordinary white, showing that no reliance whatever can be placed on such considerations, and that experiment alone can decide what properties the several types possess. The whiteness of the recessive whites is thus due to the absentee of one of the elements needed for the de- velopment of colour. Analytical experiments show on the contrary that the white of the dominant white ( 1 ) is not due essentially to the absence of an element but to the preseitce of a factor which prevents the development of colour. Thus, if X and Y are the two complementary factors which produce colour, the one recessive white is Xy and the other is xY. The dominant white (i) has the suppressing factor 5, and as the result of many experiments it appears that the individuals of that breed have various compositions in respect of X and K, though always homozygous for 5". Hence it follows that when Dominant white (i) is crossed with Recessive white (2), F^ is white; but /% may, and in our experience always does, give some coloured birds. The proportion in which the coloured F^ birds appear must of course depend on the composition of the individual dominant white which was originally introduced, and various ratios occur. The numbers obtained show that of the dominant whites some are Xx Yy SS, another contained only one of the colour-factors and so may be written Xxyy SS. Probably all the possible combinations occur, but the experiment has not been carried far enough to prove that they do"^. * The difference in size between Recessive white (2) and the White Rosecomb is so considerable that the direct cross between these two cannot be made. But F^ from White Rosecomb x Silky has been crossed with (2) and the offspring were in about equal numbers whites and browns. There is therefore no reasonable doubt that the factor carried by the Silky and by the Rosecomb is the same. Either both have X, or both V. V] Dominant IVhites 105 The facts thus take their places in a consistent scheme, and when the factors are correctly determined, all the observed properties of the several types can be represented. Some curious consequences follow which would have ap- peared very surprising both to evolutionists and to fanciers a few years ago. For example, if a pair of coloured birds throw very few whites the chances are large that these whites if mated together will breed true. On the other hand if a pair of coloured birds throw many whites, the chances are that such whites if bred together will have some coloured offspring, and it is not very unlikely that all their offspring will be coloured ! For in the first case where the whites are few, they are probably i in 4, indicating that both parents are homozygous in either X or Y, say X, and heterozygous in respect of the other factor Y. There- fore all the whites which come will be carrying X and none will have K. Therefore they will breed true to whiteness if mated together. In the second case however, where the whites are many, they are probably 7 whites : 9 coloured. In such a case the parents must be heterozygous both in X and in Y. Consequently some of the whites will be bearing X and others Y, and there will be a good chance that if such whites are bred together these two complementary factors will again come into combination and coloured offspring be produced. Among plants only one example of dominance of white flower-colour over the ordinary sap-colours has yet been investigated. This occurs in Prwiula Sinensis. In this species colour may be present either in the flowers, or in the stems and leaves, or in both. When white flowers occur on stems green and devoid of pigment, the white is recessive as in any other case of total albinism. There are however white varieties which have more or less red in the stems, and these when crossed with varieties having coloured flowers give F^ with flowers white or nearly so. There is almost but not quite always a slight tinge of magenta-red in the petals of such F^ plants, and the in- tensity of this colour increases with low temperatures but diminishes if the house be kept warm. Such dominant whites, when crossed with the green- io6 Dominant Whites [ch. v stemmed recessive whites, give a white-flowered F^ with a reddish stem. The /% generation from this cross has contained a small number of plants with coloured flowers, but the ratio has not yet been determined. There is an obvious general agreement between this case and that described in fowls, and we may feel fairly sure that the Primula colours similarly depend on two colour-factors, and that the absence of colour in the dominant whites is due to the super-imposition of a third factor"^. * The white-flowered variety of Matthiola incana (Stock) is as regards its genetic behaviour, a coloured form. Moreover its flowers tinge on fading and its embryos have the deep green colour characteristic of purple- flowered types. Crossed with a self-coloured type it gives F^ with purple flowers. Clearly therefore it is not, as some of the tinged forms of Pritnula Sinensis* 2irt, a dominant white. Miss Wheldale has suggested with some plausibility that the white M. incana is comparable with the picotee types of Sweet Peas. CHAPTER VI HEREDITY OF COLOUR— continued. Eye-Colours. Variations in Colour of the Iris — Deficieficy of Eye-Pigments in some Coloured Types. The colours of eyes present some special genetic pro- blems which are worth examining in detail. Pigments are formed in several parts of the vertebrate eye, but chiefly in the choroid and the iris. In all normal eyes the choroid contains much deep black pigment, and to this fact the blackness of the pupil is of course due. The common differences in eye-colour with which we are familiar in man are not caused by changes in the colour of the pupil but by variations in the structure and pigmentation of the iris alone. Such differences, being conspicuous factors in individual and racial differentiation, have long attracted the attention of anthropologists. With the object of elucidating the heredity of eye-colours extensive pedigrees were collected by Galton, and biometrical studies of a similar kind have been published by Professor Pearson and his assistants. Various conclusions of a statistical nature have been based on these data, but as no critical analysis was attempted, such results are devoid of genetic importance. As we now know, moreover, the method of classification employed in collection was unfortunately defective. The categories were not sufficiently precise, and the material would give no reliable indications if analysis were attempted. As the result of comprehensive investigation carried out in a single district on Mendelian lines, Hurst has succeeded in making an important advance"^. Recognizing that such descriptions as "light," "dark," and so forth are * Similar evidence has also been published by Davenport (107). io8 Eye-Colours [CH. too vague he instituted a more critical classification, which led to the discovery that th.^ presence of brown pigment on the anterior surface of the iris behaves as a dominant to the absence of such pigment, which is a recessive character. Parents without the pigment in question have exclusively children who are similarly without it, whereas persons possessing the brown pigment may be either homozygous or heterozygous in that respect, and their offspring follow the numerical rules with fair exactitude. i9 GfxQ 9 n Q^C c o" cf c c/ 9 C 9 Gfcr OS ^ OS a (^ 9 Q^ Cf) |9 9 cT gT 9 9 9 9 9 9 9 DomzTLCLnta ^ Duplecd Self- colour (£) Dupleoc Rzrt^eob O Slmpleoc Rjec>essixre. Fig. 1 8. Two pedigrees showing the descent of eye-colour (after Hurst). ''Brown" eyes are no doubt all dominants, and "blue" eyes are very often recessives ; but, on the other hand, many eyes which would be called ''light" have some of the VI] Eye-Colottrs 109 pigment, and many eyes which would be called "dark" are without it. The differences between the non-pigmented ''dark" iris and the non-pigmented "light" iris are caused by structural differences in the iris itself. Apart from its scientific importance it will be evident that this discovery is of considerable practical interest as a contribution to human genetics. Much nevertheless remains to be done in the application of precise methods to the problem of human eye-colours. For example, amongst the dominants examined by him Hurst recognized two distinct types: (i) self -coloured \x\A^^ in which the pigment is distri- buted over the whole surface of the iris, (2) ringed irides in which the pigment was absent from the periphery but existed as a ring round the pupil. The relations between these two types could not be satisfactorily determined. All that the evidence showed was that the "self" behaves as a dominant to the "ringed." But on analogy with other such cases we of course should expect that these distinctions were caused by distinct determining factors, either of which might be borne by the non-pigmented recessive without revealing its presence. If this were so, some of the non- pigmented should, on breeding with the ringed, have off- spring with self -coloured xxxA^s, just as some albino rabbits on breeding with Dutch-marked rabbits may have self- coloured offspring. Curiously enough, however, in an ample series of observations, Hurst met with no such occurrence, though examples of self parents producing the three kinds of offspring, self, ringed, and non-pigmented, were not rare. Here, at present, the matter rests ^. Hurst's observations relate entirely to the population of a small English village where naturally only a limited range of types would be met with. The inter-relations of the eye-colours of strongly marked racial types doubtless will provide an excellent field for the application of a similar analysis. It is likely too that such work would result in the discovery of the principles governing the association of certain eye-colours with particular colours of the hair, and "^ Hurst has suggested with some plausibihty that the absence of any self-coloured irides among the offspring of ringed x non-pigmented may be attributed to the fact that the offspring examined were mostly school- children. He thinks it not improbable thai some of the ringed may subsequently develop into self-coloured irides. no Eye-Colours [ch. other characteristics. In all such inquiries the first step is to distinguish the critical differences which are treated as units in the constitution of the germ-cells. In addition to these observations on man we have a few indications as to the heredity of iris-colours in birds. The iris of the domestic breeds of fowls presents at least three types of coloration. The common colour in the full grown birds is a bright red, apparently due to the formation of a red pigment when the period of adult life is approached, the colour in earlier stages being a dull blackish green. Malay fowls are peculiar in having a pale, yellowish white iris — the "daw-eye" of fanciers^ — which behaves as a recessive to the red iris. This eye may be found in birds of various colours and probably it always indicates an admixture of Malay blood '^. I have seen, for instance, a daw-eye in a White Leghorn of a good strain. Here it undoubtedly cropped out as a recessive character among the normal red-eyed birds, and I happen to know that White Malays were used by a prominent White Leghorn breeder to increase the size of his strain. Another peculiar type is seen in the nearly black iris of Andalusians and Silkies. In /\ndalusians we have found this eye imperfectly dominant to the red. The case of the Silkies is much more complex and the inheritance is dis- turbed by sex in the manner hereafter to be described. In the Little Owl (^Athene nochtd), Giglioli has described the occurrence of a remarkable variation in eye-colour which is evidently an example of the appearance of a recessive in the wild state. The irides of the normal birds are yellow, but in the variety, of which several specimens were found in one district, the irides were black. The circumstances of the discovery which are related by Giglioli in detail! leave little doubt that the condition was recessive. It is a curious fact that, in all the aberrant individuals found, the distribution of the brown markings on the feathers also differed from that of the type. So distinct was the variety from the ordinary A. noctua that a new specific name was suggested for it until the fact that it could be bred from the type was established. * Sir J. Bowring {Philippi?ie Islands^ 1859, p. 151) says that "white eyes" are preferred in fighting cocks. Probably these are the origin of our Malays and Aseels. t Giglioli, Ibis, 1903, p. i. VI] Pink Eyes in Coloured Types 1 1 1 Deficiency of Eye-Pigment in soine Coloured Types. In the eyes of actual albinos, destitute of pigment, the eye has the well-known pink appearance. This Is due to the admission of light through the partially transparent iris. Generally speaking such great deficiency of ocular pigments occurs only In total albinos, but to this rule there are a few remarkable and Interesting exceptions. The best known example Is the so-called Himalayan Rabbit"^. In it the eyes look like those of the pure albino though having a trace of pigment. At birth Himalayans are white or nearly so, but soon the muzzle, ears, tail, and feet assume a deep chocolate-black. This type behaves as a recessive to the normal grey, and to black, and it is a normal dominant to the pure albino. Hurst however found that Himalayans, though obviously pigmented, may carry the factors for greyness and for self-colour, without any indication of such factors being manifested In their own colour or In its distribution. Recent experiments made by Punnett have given the same result. The next example of the pink eye present in an animal not devoid of colour Is seen in certain " Japanese Waltzing" mice. The "waltzing" is a peculiar vertiginous spinning to which those animals are subject. I understand that the exact physiological nature of this movement has never been fully elucidated, but the condition has been shown to be often, if not always, associated with malformation of the auditory labyrinth. As already mentioned, the waltzing character behaves as a Mendelian recessive to the normal habit. Such waltzing mice exist in various colours, some having ordinary black eyes, others pink eyes. Various colours also may occur in the coats and we have not yet complete * Darwin gives {Animals and Plants, ed. 2, 1885, i. p. 113) facts as to the production of Himalayans by crossing Silver-greys. In the light of present knowledge it is no longer possible to accept these experiments as evidence of an original production of the variety. Presumably the case was only one of the reappearance of a recessive form, and indeed Darwin adds the remark that he had " recently been assured the pure Silver-sjreys of any such breed occasionally produce Himalayans." 1 1 2 Eye-Colours of Mice [ch. knowledge as to which of these colours are compatible with the pink eye and which are not. Elaborate experiments were made by Darbishire (90) to test the inheritance of these mice. The waltzers used were all pink-eyed and bred true to that character. In general terms the coat-colour was fawn-and-white, but no detailed account of the pigments present has been published. Crosses were made between these waltzers and ordinary albinos. The F^ generation so produced were usually, as regards coat-colour, grey like wild mice, the shade being sometimes darker, sometimes lighter, together with more or less white. In a few families the colour was black. The albino parents being of miscellaneous origin, such diversity in F^ colour is what we should now expect. In respect of eye-colour the remarkable fact was ob- served that the F^ mice were always black-eyed. \n F^ various coat-colours occurred, including a peculiar new type with exceedingly dilute pigment in the hair, spoken of as " lilac." Among these coat-colours the eye-colours were distributed according to systems not yet ascertained. All that can be clearly perceived is that the mice with agouti or full black in the coat-colour (whether pied with white or not) always had black eyes ; the completely white indi- viduals always had pink eyes ; and lastly, the eyes of the ''lilac" mice were always pink^. The facts were un- fortunately presented by Darbishire in a form which makes further analysis impossible. A proper investigation of this series of phenomena would have greatly increased our knowledge of the genetics of pigmentation. The important conclusion may nevertheless be con- fidendy drawn that at least two factors are concerned in producing the black colour of the eye. Cuenot (88, p. 11), as the result of his own experiments, identifies these with (i) colour in the coat; (2) a factor determining the black- ness of the eye, which needs the coat-colour factor as its complement. This is no doubt a part of the truth, but * Miss Durham (116) has shown that the pink eyes of these "lilac" mice and of the original pink-eyed waltzers are not wholly devoid of pigment, but contain traces both in the iris and choroid visible in microscopic sections. The eyes of the albino mice are absolutely without pigment. VI] Eye-Colours 113 there are suggestions that it is not the whole, and that further compHcations have to be met. To discuss these in detail is beyond the scope of this work, but it seems likely that the degree of saturation or dilution of the coat-colour has to be considered. There is evidently some intimate relation between the colour of the eye and the colour of the coat ; for no mice have the coat-colour wild-grey ("agouti" of fanciers), black, or even blue, unless the eyes are black. The interesting fact has also been discovered that the actual nature of the pigment of the choroid differs in the different varieties^. Miss Durham finds that the chocolate mice always have chocolate eyes, not black, no black pigment being present in either iris or choroid. The same fact has been published by Castle (53) for the guinea-pig and was also noticed independently by Miss Sollas in guinea-pigs. From these observations it is to be inferred that either the coat-colour controls the eye-colour, or the colour of the eye controls that of the coat, but for various reasons it is not possible yet to declare positively which account is the right one. Guinea-pigs differ from other animals in that their albino strains generally have small "smudges" of blackish pigment on some of the extremities, especially the tips of the ears, though the eyes are entirely w^ithout pigment. It is said that albino guinea-pigs occur without these smudges, but I have not seen one. The Cinnamon Canary must also be noticed here. This variety on hatching has pink eyes, though as develop- ment proceeds, pigmentation supervenes. Many fanciers believe that they can recognize a difference between the eyes of Cinnamons and those of other varieties, but the distinction is not readily obvious. In plumage Cinnamons differ from other breeds in that the black pigment is absent and the feathers consequently have a brownish tint. Miss Durham, from her observations, concludes that the pigment which is developed in both eyes and feathers is chocolate, and that the Cinnamon is in fact a chocolate '*' John Hunter has an important paper on this subject : " On the Colour of the Pigmentum of the Eye in different Animals" {Obs. on certain Parts of the Animal Econo?Tiy, 1786, p. 199). B. H. 8 114 Eye-Colours [ch. vi variety of the Canary. The heredity of these characters has not yet been investigated by precise methods. Fanciers however are agreed that some very remarkable phenomena of sex-Hmitation occur, which will be described in con- nection with other instances of the same kind. The only other example I know of a race which com- bines an albino eye with external colour is the breed of cats called " Siamese." These animals, which breed per- fectly true, were introduced from Siam, where they have been kept for an indefinite period as pets of the royal household. Like the Himalayan rabbit, Siamese cats are born almost white, but the fur becomes a curious fawn with darker chocolate points on the ears and extremities. The eyes are more or less deficient in pigment throughout life. Some pigment however is formed in the eye, as is evidenced by the fact that the iris has a blue colour and is partially opaque. From the little that is known respecting this group of cases it is clear that the genetics of eye-pigments offer many features of interest. We have seen first that varia- tions in the colour of the iris may exhibit simple Mendelian heredity. Secondly when the general pigmentation of the eye is considered, it appears that the presence or absence of such pigment may follow Mendelian rules ; but there is an inter-dependence between the factor or factors which produce the pigmentation of the eye and those which govern the pigmentation of the fur or feathers, such that the eye-colour must be regarded in certain cases {e.g. mice) as determined by two, or perhaps more, complementary factors. The actual nature of the pigment in the eye is also variable, black being replaced by chocolate in several types destitute of black in their hair. Lastly, there is a noticeable association of chocolate pigment, in hair or feathers, with the more or less non-pigmented eye, as in the Himalayan rabbit, the Siamese cat, and the Cinnamon canary. The peculiar phenomena connected with Albinism in Man are mentioned later in connection with human heredity in general. CHAPTER VII HEREDITY OF COLOUR— continued. The Genetics of Yellow Pigments in certain Animals. Yellow Mice not breeding true — The Case of Basset Hounds and the ^' Law of Ajtcestral Heredity ^ Relation of this Principle to Mendelian Rtdes. The genetic relations of yellow and black pigments present so many difficulties that a separate chapter is needed for their consideration. There is one comparatively simple case, that of the thorough-bred or race horse, In which the presence of black behaves as a simple dominant to the colour known as chestnut. In practically all the other forms that have been investigated there is some complication. One of these is the celebrated example of the Basset Hounds, which played so prominent a part in connection with Galton's enunciation of the '' Law of Ancestral Heredity." In examining this section of the evidence which Mende- lian analysis has provided we meet three facts of surprising novelty. First, that the relations of yellow to black and other colours may be entirely different in animals so nearly related as the rabbit and the mouse. Secondly, that in mice yellows never breed true, though In the rabbit, for instance, no such difficulty occurs. Thirdly, that in rabbits the pre- sence of the factor for the grey or agouti type of coloration alters the appearance of the yellow animal, giving It a white belly and tail, whereas yellows destitute of this factor have these parts of a bluish colour. Such facts cannot of course be properly interpreted until the chemistry of pigmentation is more fully understood, but taken together they Illustrate the extraordinary multiplicity of specific rules which genetic research reveals. 8—2 ii6 Yellow Rabbits [ch. When the hair of the wild house-mouse, for Instance, is examined microscopically three kinds of pigment-granules can be distinguished, black, chocolate, and yellow. As Miss Durham pointed out (lo, p. 72), in a solution of potash the yellow granules dissolve at once ; the chocolate dissolve, but more slowly ; the black not at all. The pigments may co-exist in the same parts of a hair, but in the wild "grey" or agouti colour there are bands in the individual hairs, where only yellow pigment can be seen. The black mouse has black and also chocolate pigment, but no yellow. The chocolate has chocolate alone. The yellow mouse may have exclusively yellow pig- ment, but in some yellows heterozygous with black, granules of the other pigments may be mixed with the more abundant yellow ones. In the blue, the silver-fawn and the cream, the pigments are as in the black, the chocolate, or the yellow, respec- tively, but the number of granules is fewer, giving the well- known appearance of dilution in colour. The colours of rats, rabbits, guinea-pigs and horses, appear to be similarly made up of three presumably com- parable sets of pigments. There are differences however both in the number of distinct pigmented types which exist, and in their genetic behaviour. Rats. In rats the only colours known are agouti, and black. Both types contain chocolate pigment as well as black, and agouti also contains yellow pigment ; but no distinct choco- late or yellow varieties are known, and none of the dilute forms (blue, &c.) have been recorded. Rabbits. Of the cases in rodents where yellow exists as a separable variety that of the rabbit is the simplest. Numerous colour-types of the rabbit are known, but in our present inquiry we are immediately concerned with the following : I. Agouti, or wild grey, 2. Black, 3. Yellow. The question whether a chocolate rabbit can be formed is important. It appears that no such type is known to fanciers but Mr Punnett saw, in the possession of a French breeder, rabbits VII] Yellow Rabbits 1 1 7 which seemed to him to be chocolate. No examination of the pigments of these rabbits has been made. Agouti, as we know from the experiments of Hurst and others, is epistatic to black. Next, Castle (53) has shown and Hurst independently discovered^ the fact that the yellows are of two distinct types. A. Yellows with zvhite bellies and tails, which always bear the agouti factor, G. B. Yellows with blue bellies and tails, which do not contain G. The latter have been spoken of as sooty yellows, and to English fanciers they are known as tortoise-shells. Respecting the pigments of these types full information is wanting. Miss Durham found that agouti certainly contains black, choco- late and yellow ; that black contains black and chocolate ; but hitherto the yellows have not been studied on a com- prehensive scale, and all that we know is that yellows may contain both black and chocolate granules in addition to yellow. None have yet been seen to contain yellow only. Following Castle, the genetic composition and behaviour of the several types may be represented thus : Agouti (wild grey) GBY. Black gBY. Yellow A, white belly GbY. Yellow B, blue belly gbY. This scheme is so far satisfactory, but two difficulties arise when an attempt is made to compare the phenomena seen in the rabbit with those observed in mice and guinea- pigs ; for first the scheme takes no account of chocolate ; and secondly, in the mouse yellow is dominant to black, not recessive as in the rabbit. There is however no doubt as to the essential facts, that yellow A crossed with black gives agouti, while yellow B crossed with black gives simply black dominant. A further confirmation of this symbolic representation is found in the fact observed by Castle that agouti x yellow B gives blacks in F^ as the scheme demands f. The nature of the factor G is obscure. Castle regards it as a pattern-factor governing the distribu- tion of the colours ; but a grey rabbit differs from a black * Communicated to the Internat. Congress Zool. 1907 : as yet un- published. t Compare the similar phenomena in mice, p. 78. 1 1 8 Red Guinea-pigs [ch. in more than pattern, for actual bands of yellow pigment are formed in the hairs. G therefore is a factor which can make pigment yellow that would have been black in the absence of G, so that in a restricted sense it is a dominant yellow factor, but the relation of this yellow of agoutis to the yellow pigment of yellow varieties is quite uncertain. There is no suggestion that yellows of both types cannot be bred pure (cp. Mice). Guinea-pigs. In guinea-pigs no true yellow exists, but It is apparently represented by red. The work of Castle (53) and the experiments of Miss Sollas, as yet unpublished, show that when red is crossed with black the offspring are either (i) black-reds, viz. patched with black and red; or (2) agoutis. In which the black and the red are intimately mixed In the banded hairs. Families of three kinds may thus be produced. Red bred with black may give (A) all agoutis, (B) all black-reds, (C) agoutis and also black-reds. It seems practically certain that the agouti-factor, G, as we have called it, may be carried by the red, not by the black. Consequently case (A) is that in which the red is homozygous for this factor. In case (B) the red is homo- zygous for its absence, and in case (C) the red is hetero- zygous for the same factor. There Is no difficulty In breeding reds as a true strain, and the peculiar phenomenon about to be described in yellow mice does not occur In guinea-pigs. In guinea-pigs a distinct chocolate type exists, but such animals according to Miss Sollas always have some red hairs. The relation of these to the other types is not yet fully understood. Mice. The facts seen in rabbits and guinea-pigs are in fair agreement with each other, but when these are compared with the phenomena now well ascertained in the case of mice, fundamental differences are perceived which make it viT] Yellow Mice 119 impossible to bring the whole series of observations into one consistent scheme. For first, Cuenot [^-j) has shown that in the mouse yellow^ is dominant to black, instead of being recessive to it as in the rabbit. The dominance is so far incomplete that yellows heterozygous with black have often — perhaps always — more or less black pigment in their hairs, but in general appearance they are obviously yellow. Next according to Cuenot's experience none of the yellow individuals in /% from a cross of yellow x black or yellow X agouti are pure to yellow^. All such extracted yellows when bred together throw agoutis or blacks respec- tively. This fact has led him to suggest that there may be some incompatibility which prevents two yellow-bearing gametes from uniting in fertilisation. These " impure " yellows bred together gave 232 yellows and 86 agoutis, a fairly close approximation to 3 : i. The expectation on that ratio is 228*5 • 79'5- Cuenot, commenting on the numbers produced, remarks that since pure yellows are not formed, he would anticipate a diminution in the relative numbers of yellows, and that such a diminution actually occurs. But, as Mr Punnett has pointed out to me, there seems to be no valid reason for expecting a departure from the ratio 3:1; for the spermatozoa may be regarded as unlimited in number. Of the ''non-yellow" ova half would be fertilised by *' non-yellow" sperm, producing agoutis, the other half being fertilised by "yellow" sperm and pro- ducing yellow^s. The "yellow" ova would all, on Cuenot's hypothesis, be fertilised by "non-yellow" sperm, and give rise to yellows. The nett result would then be still 3 yellows : i agouti. Of 81 /% yellows tested by Cuenot, all proved to be heterozygous, so that the reality of the abnormal pheno- menon must be regarded as established. Observations made * A fact of a different order, though perhaps having a bearing on the problem, was observed by Hurst in crosses between Belgian Hares (Rabbits) and Angoras. The Belgian is in colour an "agouti," or grey, but it is much yellower than the wild rabbit of the warrens. Hurst found that F^ from such Belgians and Angoras (bearing G) was of the wild or warren type of grey, and that in A no rabbit quite of the Belgian colour ever came. The greys in F.^ varied a little in the amount of yellow, and it is possible that, as in other cases in which a parental type fails to recur in 7s, the absence may be ascribed to the rarity of a particular combination. I20 Pile Fowls [CH. by Miss Durham agree entirely with those of Cuenot. She finds it impossible to obtain any mice pure to yellow. The experience of fanciers seems to be universal that yellows cannot be bred in a pure strain. Those with which she has experimented always throw either agoutis, blacks, or chocolates, and we may take it that yellows of both sexes are always heterozygotes formed by the meeting of a yellow and a non-yellow gamete"^. In spite of their genetic composition the yellows which throw chocolates have only yellow pigment in their hairs, and no chocolate. Cats. In cats we meet a new complication in the inheritance of yellow, namely that it is disturbed by sexual dimorphism. The rarity of tortoise-shell males is well known, and Don- caster (109) has produced evidence which makes it prac- tically certain that in the immense majority of instances the female heterozygote of orange x black is tortoise-shell, namely, patched with orange and black (like the guinea-pig), but that the male heterozygote similarly produced is orange (see Chap, x, dealing with the heredity of Sex). The dilute types cream and blue are similarly related to blue tortoise-shell. Fowls. There are further indications of peculiarity in the genetics of yellow pigments in the case of birds. The coloration known as "Pile" in fowls is seldom bred for exhibition from two pile birds. The colour consists, in cocks, of orange-yellow or red in the hackles and wing-coverts com- bined with a white ground. The hens are white, with a chestnut or reddish yellow breast. In the down the chickens of both sexes have longitudinal stripes of light chestnut. In the adults, and generally in the chickens too, there is a small and variable amount of blackish grey ticking, which is of course considered a fault. Pile is a colour known in several breeds, but especially in Game, Game * Besides this curious peculiarity, yellows show another remarkable feature in their frequent tendency to become excessively fat. Miss Durham has met with several such specimens and the fact is well known to fanciers. In her experience also yellows are more liable to sterility than other mice. VII] Peculiarities of Yellow Types 121 Bantams, and Leghorns. It is said never to breed true, throwing ordinary black-reds in unascertained proportions. Presumably pile Is a heterozygous combination, black-red being one of the constituents, but what the other pure form (or forms) Involved may be, is quite uncertain. The whole question is still very obscure, and from preliminary experi- ments made by Mr C. Fryer It seems probable that two pairs of factors are concerned in addition to black-red. For exhibition purposes piles are often crossed with black-reds, fanciers being under the impression that a brighter pile results from this mating. The fact suggests that there may be piles of several different compositions. Taking the known facts together the one clear deduction Is that pile, or yellow if we may so call it, is dominant to black-red*. Recapitulation and Discussion of the Foregoing Facts. The outstanding peculiarities of this group of pheno- mena are as follows : I. The difference between the genetics of similar colours in allied forms. In rabbits black Is epistatic to yellow, giving (In the absence of G, the agouti factor) simply a black heterozygote. In guinea-pigs (where red represents yellow), the hetero- zygote of black and red is a patch-work, viz. tortoise-shell. In both rabbits and guinea-pigs when G is present this heterozygote is agouti, the factor G being Introduced by the yellow parent. In mice yellow is epistatic to black, giving, in the absence of G, a yellow heterozygote. We can form no surmise as to the nature and causation of this discrepancy between these types. It is to be observed that the recessive yellow of the rabbit, viz. the blue-bellied type lacking both G and B, nevertheless has some black pigment in its blue fur, while the yellow mouse does not necessarily have any. * The whole-coloured buff of Cochins and some other breeds derived from them is quite distinct from pile. The genetics of this pigmentation are complex and little known as yet. 122 Peculiarities of Yellow Types [ch. 2. The fact that in mice no yellow homozygote exists. It is certainly a most surprising fact that mice cannot be bred true to yellow — the more so since there is no difficulty in producing pure yellow rabbits and pure red strains of guinea-pigs. Among the ordinary phenomena of heredity no quite parallel case is known, though it is possible that pile (yellow and white) fowls are in the same condition^. Cuenot's suggestion of the incompatibility between yellow- bearing gametes female and male, is the only one yet offered in elucidation of the phenomenon. While admitting that this account has great plausibility, I think that we must not place complete reliance on a hypothesis to which no adequate or thorough test can as yet be applied. The idea of incompatibility between gametes has more than once been introduced to deal with the genetic phenomena of Sex {q.v.) and in that connection the case of yellow mice should be remembered. 3. The behaviour of the agouti factor G. The operation of this factor varies to some extent in the different types. In all it can be carried by the yellow variety, but in the rabbit alone are yellows which bear G obviously different (having white bellies) from those which do not (having blue bellies). It seems difficult to suppose that this factor actually causes the appearance of white on the belly and tail, yet the fact is well-established that such yellow rabbits are really agoutis, or wild greys, wanting in black. The relation of the yellow, due to the presence of the yellow factor K, to the yellow in the bands, caused by the agouti factor G, is at present quite problematical, but micro- scopically these yellows are indistinguishable. Until the relations of chocolate to the other pigments in the rabbit and guinea-pig have been more fully explored it is scarcely possible to draw up a scheme of symbolic notation representing the comparative compositions of the different animals ; but adopting the gametic formulae given (p. ']'S) for the various colours in mice we may tentatively suppose that yellows exist of all compositions which would be pro- duced by adding Y to each of those groups of symbols. * See also Basset Hounds (p. 128). VII] Colours of Thorough-bred Horses 123 Thus the gametes for each of the four non-yellow colours are as follows : Agouti (with black) G . B . Ch. Cinnamon agouti (w^ithout black) G . d . Ch. Black g.B.C/i. Chocolate g - ^ - Ch. Yellows being always heterozygous, all zygotes con- taining yellow^ are to be represented by adding Yy to the zygotic formula produced by compounding two of the above gametic formulae. Thus the yellows which throw chocolate only are Yy .gg ,bb . ChCh, The yellows throwing black only are Yy .gg . BE , ChCh, and so on. This representation must in any case approxi- mate to the truth, but there is still some doubt whether the relation of yellow to non-yellow is quite so simple as here suggested"^. We now pass to two cases w^hich are of some interest on account of the somewhat conspicuous part that they have played in general discussions of heredity. The first is that of Race horses, where the phenomena, in so far as they have been adequately studied, are remarkably simple. The second example, that of the Basset Hounds, presents a feature of complexity. Race Horses. In " thorough-bred " or race horses the behaviour of the yellow and black pigments in descent is comparatively simple and the facts thus form a remarkable contrast with * Attention should be called to the fact that if the two kinds of agoutis in the mouse might be supposed to represent the two kinds of yellow in the rabbit then yellow of the rabbit might be supposed to correspond with chocolate of the mouse. This natural suggestion is however negatived by the fact that a definite chocolate pigment can be seen in the hairs of rabbits. In the pink-eyed Himalayan rabbit the pigment according to Miss Durham is exclusively chocolate. 124 Colours of Thorottgh-bred Horses [cii. the complicated phenomena elsewhere observed in regard to pigments ostensibly the same. The colours of race horses are recorded with great accuracy in Weatherby's General Stud Book. By a careful analysis of these returns Hurst was able to show that chestnut on the one hand is recessive to bays and browns, w^hich are dominant. Chest- nuts are distinguished from bays and browns by the fact that in their hair no black pigment is developed. Bays always have black in the mane, tail and fetlocks. Browns have more or less black in the same parts, and generally black pigment is distributed to some extent over the whole coat. As the records show, there are only about i per cent, of alleged exceptions to the rule that chestnut x chestnut produces chestnut exclusively. Of the dominants some are pure dominants and give — again with about i per cent, of exceptions — bays and browns only, whether the other parent is chestnut, or bay or brown. Other dominants are shown also to be DR in constitution, giving, when bred with chestnuts, equal numbers of dominants and recessives. With regard to the few exceptions of both kinds ap- pearing in the Stud Book records some are demonstrably mistakes, and the actual frequency of exceptions must be considerably less than i per cent., if indeed there are any genuine exceptions at all. The relation of bays to browns has not yet been made out, and as the two classes grade into each other somewhat, the detection of their relation- ship would demand observations of a sort somewhat more critical than those which the Stud Book provides. So far however as the segregation of black from the absence of black is concerned, the case is simple and regular"^'. * The inheritance of coat-colour in horses was the subject of an extensive investigation by Professor K. Pearson and his assistants (219). The records of the Stud Books were tabulated and investigated by the application of bio- metrical methods. Various propositions have been enunciated as the result of this inquiry, amongst others the statement that nothing corresponding to Mendel's principles appears in the case of horse-colours. The key to the phenomena was of course the fact that chestnuts mated with chestnuts breed true — with rare and dubious exceptions. It would seem at first sight impossible to devise a system of tabulation which could fail to disclose so prominent a feature. Nevertheless Professor Pearson's correla- tion-tables, which were compiled from the records of more than 6000 horses, were made in such a way that the colours of sire and dam could VII] Colours of Thorough-bred Horses 125 The time has not arrived for any attempt to analyse the relations of horse-colours In general. By microscopical methods Miss Durham has found In horses the three pig- ments, black, chocolate, and yellow seen In other types. The term chestnut Is used somewhat loosely In describing colours, and though usually a chestnut Is a horse possessing yellow pigment only, there Is a dark type of chestnut, sometimes spoken of as liver-ckestnut, w^hlch Is actually chocolate^. Nothing Is known of the genetic relation of this to the yellow chestnut, or of the respective properties of those various chestnuts distinguished by the colour of their manes and other subordinate differences. Among thorough-bred or race horses, types Intermediate between chestnut and the dominant bavs and browns must be exceedingly rare, for though entries indicating doubt betw^een bay and brown are rather common in the Stud Books, alternative designations are scarcely ever given In regard to chestnuts. Among common horses and hackneys such animals though exceptional can be found by looking out for them, and one or two may usually be seen In a day's walk through London streets. I have no information as to their genetic capacities, but presumably they are due to dilution-stages of the black pigment, corresponding to those which In the mouse &c. constitute the blue varieties. They may also be cases of imperfect dominance. not be taken into account together. The tables thus provide answers to questions as to the probable colour of a foal by a chestnut sire, the dam's colour being taken as unknown ; as to the probable colour of the brother or half-brother of a chestnut foal, when the colours of both sire and dam are taken as unknown ; with solutions of other problems of equal signifi- cance. Since however the colours of both sire and dam are recorded, and must indeed have been actually extracted from the Stud Book for the purposes of the tabulations, the investigators, by refraining from an inspection of these data till they had been separated, placed themselves at a gratuitous disadvantage. The true nature of the inheritance was therefore not discovered. The failure was due to want of analysis. The similar failure of bio- metrical methods to find the plain rule of inheritance in the case of human eye-colour was due to the same defect of method, though in that case further obscurity arose from the use of faulty and uncritical observations. * When very dense this type of colour might carelessly be mistaken for black. 126 Basset Hottnds [ch. '^Tricolour'' and '' No7t-tricolour'' in Basset Hounds: The Law of Ancestral Heredity. The question of the relation of the yellow and black pigments is raised in the celebrated case of the colours of Basset Hounds. The importance of that subject is due to the fact that it was from a study of the evidence in regard to Bassets that Mr F. Galton was led to enunciate his "Law of Ancestral Heredity" with confidence as one which "appears to be universally applicable to bi-sexual descent." The publication of that paper played a considerable part in the history of modern genetic research and it is necessary that we should consider the facts in some detail. The colours of Bassets are two, the first spoken of as tricolour, consisting of black and yellow marks on a white ground ; the second, non-tricolour, which differs from the first in having no black. It is said that dogs which cannot be easily referred to one or other of these two types do not occur, and they must certainly be very rare if they exist at all. Mr Galton's investigation was based on data supplied to him by the late Sir Everett Millais, a keen fancier of the breed. . This evidence consisted in records giving the number of offspring of each type which had occurred in families of various compositions. It was thus possible to compare the number of tricolour and non-tricolour dogs produced in the families with the number of the respective types distributed among their pedigrees. Galton's figures indicated that there was a close correspondence between these two numbers, so that it was possible, given the ancestral composition of the families, to predict with "con- siderable accuracy the numerical proportions in which the respective types would appear. According to Galton's system the family was regarded as the production of all the ancestors. Each ancestor was supposed to contribute in his or her degree to this total heritage, the more immediate progenitors contributing more, and the remoterprogenitors less, according to a definite arithmetical rule. This rule was that the average contribution of each ancestor was to be reckoned for each parent 1/4 for each grandparent 1/16 for each great-grandparent 1/64 VII] Basset Hoimds 127 and so on, the total heritage being thus reckoned as unity. It will be observed that this scheme differs entirely from those based on Mendelian principles, inasmuch as every ancestor is, according to the Law of Ancestral Heredity, supposed to have some effect on the composition of each family in its posterity, and each recent progenitor is re- garded as having a very sensible influence on these numbers. Though no one with a knowledge of practical breeding could entertain the supposition that Galton's Law had the universality of application claimed for it, there was on the other hand no doubt that the Law had successfully expressed a variety of facts in which no order at all had been pre- viously detected. We have now to consider the meaning of this evidence in the light of modern know^ledge. At the time that Galton's views were promulgated nothing w^as known of segregation. The supposition that any individual, whatever its own characters, was capable of carrying on and trans- mitting to its posterity any of the characters exhibited by its immediate progenitors, at all events, was generally received without question by biologists. According to that idea the number of classes of individuals differing in respect of their ancestral composition and transmitting powers is to be regarded as indefinitely large, whereas in all cases of sensible allelomorphism the number of classes of individuals is three only, two being homozygous and one heterozygous. The difference between the two schemes is thus absolute and irreconcileable. When Mendelian phenomena were first recognized it w^as naturally supposed that some classes of cases would be found to conform to the Mendelian scheme and others to the Law of Ancestral Heredity. W^ith the progress of research however almost all the cases to which precise analytical methods have been applied have proved to be reducible to terms of Mendelian segregation ; and of those which have not already been so elucidated some, we may feel confident, if not all, will be eventually shown to be governed by similar rules. In discussing aberrant pheno- mena like those alleged in regard to the Bassets the first question to be settled is whether the facts are correctly 128 Basset Hounds [ch. reported. If either type is recessive we should naturally expect this to be the non-tricolour, which is without black. Unfortunately as the non-tricolours are not fashionable there were comparatively few matings between two parents of that colour. Nevertheless 41 dogs, offspring of such matings, are given, of these 20 being tricolour. Though the records were not made by scientific men or with a scientific purpose directly in view it is almost impossible to imagine that all these cases can depend on mistakes, and pending the pro- duction of new and direct evidence we must take the records as correct. In the Theory and Practice of Ratio7ial Breeding (London, 1889), pp. 26 and 27, Sir Everett Millais gives one or two more notes bearing on this question. He says that in England there w^ere then two strains of Bassets, the Couteulx and the Lane. "The Couteulx is as a rule a very perfectly marked tricolour, with the tan and black markings deeply accentuated. The Lane hounds, on the other hand, are very weak in markings if they happen to be tricolour, but as a fact they are far more generally found to be lemon and white." In another place he mentions that "in nearly every litter of pure Couteulx there is generally a lemon and white puppy." If it were not that the genetic relations of yellow and black pigments are, as we have seen, so complicated and uncertain in other types, we might be inclined to attribute the alleged production of tricolours by non-tricolours to imperfect classification of "weak" tricolours, but in dealing with this group of phenomena no such suggestion can be hazarded with any confidence. The strange fact that yellow mice are never pure natur- ally occurs to the mind in connection with the evidence as to Bassets. A comparison between the two cases cannot nevertheless be instituted at all easily ; for in the Bassets yellow is evidently not usually a dominant, which it should be if the impurity of the non-tricolour is to be attributed to a state of things comparable with that existing in mice. At present the Basset phenomena must be regarded as definitely unconformable. Perhaps the most probable view of their nature is that they are an illustration of irregular dominance, but this cannot be asserted with much confidence. VII] Law of Ancestral Heredity 129 It is curious that the one example to which a partially correct system of analysis was applied before Mendelian methods w^ere rediscovered, should have been of this remarkably exceptional order. There is little reason to anticipate, as we once did, that a distinct group of cases obeying a Law of Ancestral Heredity will have to be recognized. That principle in certain cases gives an epitome of the consequences of the Mendelian process, and in all likelihood its applicability to any pheno- mena of natural inheritance is due to this fact. The Law of Ancestral Heredity takes of course no account of domi- nance, or of segregation with all the consequences it entails ; but as describing the results to be witnessed among a population interbreeding at random, its predictions would frequently approximate to the truth. In particular it is to be observed that the arithmetical results of DR x R and of DR X DR are correctly predicted by the Law of Ancestral Heredity. Some of the phenomena of blending are also ex- pressed with accuracy. Besides this, inasmuch as dominant individuals which have both parents dominants will in a mixed population frequently be pure dominants, statistical phenomena which could be mistaken for an effect of ancestral composition will often occur '^. For instance if purple Sweet Peas in /% were bred from promiscuously, /% would consist of purples and other colours, and the excess of purples in the mass would be greater than it was in F^. If the F^ purples were again separately saved, the proportion of purples in F^ would be still greater, and so on. This gradual increase in the proportion of purples might carelessly be mistaken for a consequence of the fact that each generation had more purples in its ancestry. As we now know, that conclusion would be quite incorrect. The increase is in reality due to the appearance of actually pure purples in /% and in subsequent generations, and to the effect which their presence has on the composition of the population. The impure purples in each generation * Darbishire's conclusions in regard to mice {Biometrika, 1904, in. pp. 23-5) were obviously based on a mistake of this kind, as he has himself since admitted. (See Mem. Manchester Lit. Phil. Soc. 1905, No. 6, p. 7.) B. H. 9 130 Law of Ancestral Heredity [ch. remain exactly in the same condition as at their first appear- ance, and the selection has merely resulted in a reduction of their number. If the purples were saved individually and their seeds separately sown, it would immediately be seen that some were pure, giving purples only, and that others were impure, the latter consisting of the various gametic types with which we are now familiar. Since then we know that ancestral composition does not decide the constitution of such a population, and since individuals of identical parentage may have most divergent genetic properties, it is absurd to attempt to trace the workings of any Law of Ancestral Heredity among these phenomena. The suggestion that methods based on unanalysed statistics have scientific value in the study of heredity can scarcely mislead those who have examined the facts. Professor Pearson and others committed to these methods have of late defended their position by arguing that there is no fundamental incompatibility between Laws of Ancestral Heredity and the conclusions of Mendelian analysis. The matter w^ould not be worth notice were it not that the same proposition is being freely repeated by several writers seek- ing some convenient shelter of neutrality"^. It is to be observed however that the supposition of an underlying harmony between Mendelian and biometrical results was not put forward by the biometricians until every possible means of discrediting the truth of Mendelian facts had been exhausted. Those attacks having failed, we are asked to observe that the Law of Ancestral Heredity was meant as a statement of a statistical consequence, and is not concerned with physiological processes. Mr Galton's views on this point are well shown in the following passage in which he explicitly appeals to the physiological process of gameto- genesis as apparently occurring in the way which his Law requires. For in introducing the Law as applicable to Bassets (125, p. 403) he wrote: "It should be noted that nothing in this statistical law contradicts the generally accepted view that the chief, if not * See for example Darbishire, Mem. Ma?2chester Lit. and Phil. Soc. 1905, No. 6 and 1906, No. 11; and Professor J. A. Thomson, Heredity., i()0'j,pa5si?n. VII] Law of Ancestral Heredity 131 the sole, line of descent runs from germ to germ and not from person to person. The person may be accepted on the whole as a fair representative of the germ, and being so, the statistical laws which apply to the persons w^ould apply to the germs also, though with less precision In individual cases. Now this law Is strictly consonant with the observed binary subdivisions of the germ cells, and the concomitant extrusion and loss of one-half of the several contributions from each of the two parents to the germ-cell of the off- spring. The apparent artificiality of the law ceases on these grounds to afford cause for doubt ; its close agreement with physiological phenomena ought to give a prejudice va favour of Its truth rather than the contrary." Had segregation been known to Mr Galton the Law of Ancestral Heredity would not have been promulgated. It Is obvious that so soon as that phenomenon Is recog- nized and appreciated, all question of useful or direct applicability of the Law of Ancestral Heredity is at an end. That method of representing the phenomena of Heredity and all modifications of It are based on the false assumption that any individual can transmit the characteristics of any ancestor, and especially of any recent ancestor. When this conception was shown to be untrue, the structure which the biometrlclans have offered to the world as a scientific study of Heredity ceased to have meaning or value. Statistical examination of ancestral composition may, as we have seen, occasionally give a prediction in good correspondence with fact, but this Is due to coincidence and not to any elements of truth In the ratiocination by which the prediction was reached. As an attempt to compass the solution of an intricate problem by labour and ingenuity without proper data or equipment Mr Galton's work deserves long to be remem- bered. It stands out as a significant and stimulating event In the history of biology. 9—2 CHAPTER VIII HEREDITY OF COLOUR— continued. Various Specific Phenomena in Colour- Inheritance. Re- lation of Colour to Hoariness in Stocks. Miscellaneous Cases. Colour of a Special Part controlling that of other Parts. — Summary and Discussiojt. — Subtraction- Stages. Again and again in tracing the genetic properties of colours in animals and plants we encounter the phenomenon of a specific connection between certain colours and their modes of hereditary transmission on the one hand, and various apparently distinct physiological properties on the other. Colour, which systematists have often spoken of as one of the superficial or impermanent properties of organ- isms, seems thus to be bound up with fundamental pheno- mena of chemical economy. To treat this part of genetics with any fulness is not yet possible. As an illustration may be mentioned the curious result discovered by Miss Saunders in Mat thio la, using the varieties known as ''ten-week Stocks." These may be either "hoary," viz. covered with branching hairs forming a tomentum, or glabrous and destitute of hairs. When the hoary are crossed with the glabrous, hoariness is an ordinary dominant, giving 3 hoary : i glabrous in /%. But when certain glabrous strains are crossed together the F^ form is hoary, reverting to the primitive type. This reversion never occurs when any of the many red or purple varieties are crossed together, but is universal when any of them are crossed with either the white or the cream-coloured glabrous strains. The purples and reds owe their colours to the presence of coloured sap. This coloured sap is not present in the whites, nor in the creams, whose colour is due to the existence of yellow plastids in the cells of their petals. CH. VIII] Colour and Hoariness in Stocks 1 33 F„ from the cross, for instance, of purple x white con- tains sap-coloured and non-sap-coloured plants, and of these some are hoary and some glabrous. But no7ie of the plants which come without coloured sap have a7iy hairs on their leaves. A consideration of the case shows that the factor for hoariness is really introduced by the W^^V^-flowered glabrous plant, and that the glabrousness is due to the inability of the hoariness-factor to make the hairs grow in the absence of the factors for sap-colour. The facts may be represented thus, C and R representing, as before, the factors for sap-colour, and H the factor for hoariness. Purple glabrous x White glabrous CRh I cRH F^ Purple hoary CcRRHh r ^ n A... 9 Purple hoary 3 Purple glai)rous 4 White glabrous all containing all containing all containing C, R, H C, R, h c, R, and H, or A If cream glabrous be substituted for white glabrous the result is the same so far as sap-colour and hoariness are concerned, and in /% ^^^Y those plants can be hoary which also have coloured sap. Finally when cream and white glabrous types are crossed together, /^^ is purple and hoary, thus showing reversion in colour, owing to the meeting of the two complementary factors C and R, one coming in from the cream and one from the white ; and also reversion to hoariness because the hoariness-factor was really present all the while in both the cream and the white types, but was unable to show itself because one of the sap-colour elements was absent in each type. The heterozygosis of the two types brings together all the three elements C, R, and H, so the F^ plants are both coloured and hoary. The reason why hoary-leaved plants are never produced by crossing two types possessing coloured sap is at once apparent. For if the factor for hoariness were present in these types, they would be hoary. As may well be supposed the disentangling of these results was a long and tedious process. The occurrences seemed at first contradictory, but after it had been ascer- 134 Swedes and Turnips [ch. talned that each kind of family was produced with regularity as the consequence of a particular kind of union, the work of bringing all these into one analytical scheme was only a matter of time. Much still remains to be done before the analysis will be complete for Stocks in general. For example, among the Brompton Stocks races occur which are hoary though devoid of sap- colour, and as yet we are not aware what condition or factor exists which there enables the hoariness-factor to assert itself. In cases like these we get glimpses of the strict specific rules which govern the genetics of pigmentation. In the Sweet Pea again we have found that both variation in the pollen-shape and in the structure of the standard petal are closely related with the distribution of the factor which turns the colouring matter purple. There is every hope that in our further analyses these apparently trivial phenomena will serve as indications of the underlying processes. Apart however from these curious inter-relations be- tween colours and structural peculiarities, there are several remarkable specific phenomena to be seen in the genetic behaviour of colours. Of these some examples may be given as incentives to future experiment. General rules regarding colour-inheritance are scarcely to be expected as yet, for very little is known of the pigments of either animals or plants. Beyond the fact that albinism has always been found to be recessive to colour in both animals and plants no general proposition can be put forward with confidence. We believe also that yellow chromoplast-colour is always recessive to white or colourless chromoplast-colour, though the cases studied are not numerous enough to justify a general assertion. Stocks, Sweet Peas, Swedes and Turnips, Verbascum, may be cited as plants following this rule and no clear exception is yet known. Mr Arthur Sutton tells me it is well known that when Swedes are being grown for seed, Turnips must not be allowed to flower near them, but that in growing turnip- seed, no injury is done by the presence of flowering Swedes. The meaning of this is now clear. The Swedes are in general yellow-fleshed, their colour being due to yellow plastids. Turnips as a rule are white. If therefore the viit] Black Fruits 135 pollen of Turnips Is carried by Insects to the Swedes, the hybrid thus produced will be whIte-fleshed, and consequently attract attention, spoiling the uniformity of the crop. But as the white of the Turnip is a dominant, no visible effect is produced even though Swede pollen Is brought by the insects to the Turnip flowers, for F^ is white. In /% of course yellow Turnips would appear"^, but as the roots are almost alw^ays eaten off, this result is scarcely ever reached by the farmer. It might be expected perhaps that the blue and purple colours In flowers would always be dominant to the reds, but this is not so. In Stocks, Sweet Peas, Peas, and Salvia the purples are dominant. Probably the same is true for the blues of De/phinitan, Cmeraria and a good many more, but when we come to Primula Sinensis we find blue a recessive to the reds and magentas. Doubtless the chemistry of the blue pigment Is there quite different. In Solamirn and Atropa black fruit is dominant to the yellow fruit, but in Bryonia the red fruit of B. dioica Is dominant to the black fruit of B. alba. This paradox has been elucidated satisfactorily by Miss Wheldale. She tells me that the nature of the distinction between the two types is quite clear. In Atropa the black colour is due to the presence of a dark purple anthocyanin which like other pigments of the same kind Is dominant to its absence. In Bryonia alba the black colour Is caused by the presence of a little carotin in plastlds, together with green chlorophyll undecomposed. In the red-berried Bryonia dioica the chlorophyll is decomposed (just as it Is In the cotyledons of yellow-seeded Peas) and much carotin is present. Conse- quently, as may be expected, the presence of the decomposer of the chlorophyll is a dominant, as also Is the abundant development of the carotin, and thus the black colour of the fruits Is recessive to red. In Rabbits, as has been stated above, yellow Is a reces- sive to black, while In Mice it Is a dominant. The yellow varieties of many red Lepldoptera [Zygaena, Arctia, &c.) are presumably recessive t, and the same Is * From Mr Sutton's experiments (262) it seems however that F^ is sterile. t Proved for Callimorpha dottiinula. See p. 44. 136 Melanic Types [ch. apparently true of the yellow fruits, as compared with the red fruits of their corresponding types, and of the yellow flowers of some Composites {e.g. Gerbera) ^. In many types of flowers, e.g. Stocks, Primula, Sweet Pea, the very dark and more fully-coloured varieties are regularly recessive to the less dark types, whether purple or red. The same will almost certainly be proved for Cycla- men, Rose, Hollyhock, Dahlia, Carnation, Sweet William, and many more. In Antirrhinum Miss Wheldale finds that among magentas the darker are recessive to the common colours, but among the crimsons or reds the darker are dominant to the lighter. The difficulties which preclude general statements in regard to the genetic relations of melanic types among animals have been illustrated in much that has gone before. The loose description " melanic varieties," common in the writings of systematists, covers a number of phenomena essentially distinct. For example there are melanic forms which owe their greater blackness to the presence of some dominant factor responsible for a greater deposit of black, or at least dark, pigment. In Fowls, for instance, black is at least partially dominant over the bankiva colour which fanciers call '' Black-red." The dark brown variety called " Brown-breasted" is similarly a dominant. In Pigeons, as Staples-Browne has proved, black is dominant to the blue of the wild type. In the Horse the presence of black, as in bays and browns, is dominant over the absence of blacks as in chestnutst. On the contrary in Rabbits, Rats, Mice, &c. the black variety is produced by the o?nission of the agouti-factor, G, from the wild type, and black thus is apparently a recessive. Even here however the presence of black pigment is dominant to its own absence. It would be interesting to know to which group the Cat belongs. In Insects again no rule of universal application to •* That in Tomato yellow fruit is recessive to red was established by Hurst (160, p. 115). The case of Gerbera is given on the evidence of crosses made by Mr R. I. Lynch between red Gerbera /a??iesont and the yellow-flowered variety " Sir Michael." t The genetic relation of the totally or self-coloured black to the other horse-colours is not yet known. VIII] Silkworms 137 melanic varieties can be given. Of the melanic varieties of Moths which have been tested several are apparently dominant, more or less, to the normal or non-melanic types ; but in the black Chrysomelid Beetle, Lina lapp07iica, inves- tigated by Miss McCracken, the evidence shows plainly that the uniformly black type was recessive"^ to the normal which has black only in the form of spots. The common melanic varieties of the 2 -spot Lady bird {CoccinelLa bipunc- tata) are probably also recessive to the ordinary red type. The experiments of Standfuss interpreted according to the Mendelian system show that the dark variety lugens is dominant to the ordinary fulvous yellow type of Aglia tau (a Saturniid Moth). This case comes up for consideration in some detail with reference to the heredity of Sex {q.v-). In Silkworms a melanic variety of the moth is an im- perfect dominant to the normal, pale-coloured moth, giving a blend-form in F^ (Coutagne, ^2>)' As regards the colour of the silk some interesting results have been obtained. Yellow silk was always found by Toyama (268) to be dominant to white, and this result was obtained by Coutagne in certain cases t. For example the yellow race called '' Var " was dominant in silk-colour to the white Japanese race used by Toyama and to the white ''Bagdad" used by Coutagne; but Coutagne found the same yellow to be recessive to the white of two French races with which he experimented. Presumably this distinction is due to some idiosyncrasy on the part of the whites, analogous to what has been seen in fowls and Primula, but as to this nothing is known. In crossing a yellow Siamese race with a white Japanese race, Toyama obtained a resolution-effect in /%. Yellow X white generally gives yellow F^ with 3 yellow : i white in i% ; but in this special case there were two new forms in /%, a pale pinkish yellow, and a greenish white. This latter white could not always be satisfactorily distinguished from the pure w^hites, so the /% family has to be taken as 9 : 3 : 4. * There is a possible complication in this case. + Toyama states that the colour of the silk always corresponds to that of the abdominal legs of the larvae, and consequently it is not necessary to rear all the larvae up to the spinning stage in order to ascertain the colour of their cocoons. 138 Colours in Critical Parts [ch. Observation, for example, gave 70:21 : 36, the expectation being 72:24:32. In certain other cases this resolution- effect did not occur, though from analogy it might have been expected. Toyama regards the distinction as due to differences in the whites used, but it seems not impossible that it was really the yellows which possessed individual differences in this case. In either event there are difficulties to be faced, and on the evidence it is not clear which account is actually the more probable. There are some illustrations of a principle by which the colour of one part of the organism may limit or control the possible colours of other parts. In animals it is fairly certain that the eye-colour may act in this way, certain coat-colours being produced only if the eye be black, and others only if the eye be chocolate, but the facts are still somewhat obscure. If the stem of the Chinese Primula be green and not red the deeper flower-colours cannot be developed in self-coloured types. A cross with a red-stemmed type, however pale in flower-colour, at once reveals the presence of the factors for the deep colours if they are there. On the contrary, the white-edged types, such as Sutton's "Sirdar," though their flowers may be of a deep shade of purple or red, appear exclusively on stems which are green throughout except for a development of red colour at the collar or extreme base of the petioles. Such "Sirdars" cannot exist on a wholly-coloured stem. The stem may be parti-coloured in Primulas though the flower is whole- coloured, but these special types of parti-coloured flower can only occur on a parti-coloured stem. In the /% series it is curious to see these deeply coloured, white-edged, flowers on stems apparently green, while none of their green-stemmed sisters with self-coloured flowers can bear a flower darker than pale salmon-pink. Another striking example of the same phenomenon is to be seen in these Primulas, with the difference that there the want of a particular colour in the critical or "controlling" position is due to the dominance of a negative character, not to the absence of a complementary one. Certain deep red spots occur in some varieties, e.g. Sutton's "Crimson King" VIII] Discussion of Coloitr-Evidejtce 139 (a fine dark red), on the petals just external to the yellow eye (see Plate VI). These spots are never formed tinless the stigma is red. When such a type is crossed with one having a green stigma, F^ has a green stigma and no spots on the petals. In /% there are of course some with green stigmas and some with red, some with spots and some with no spots. But the distribution of these two characters shows that the combination green stigma + spot on petals does not occur. The stigma may be red though no spot be formed, but if the stigma be green, the spot is absent, though the factor for it may exist in the individual. Formerly such cases might have been regarded as examples of ''correlation," but that term is only applicable to them in a loose and quite incorrect sense. Nothing so fully demonstrates the fundamental signifi- cance of colour in the economy of plants and animals as the strange series of phenomena that have been discovered in regard to the complex inter-relations between the genetic behaviour of certain kinds of pigmentation on the one hand and certain structural features on the other. In the chapters dealing with gametic coupling and with the heredity of Sex it will be shown that not only the factors governing structure, but also the factors which are the ultimate cause of sexual differentiation, may be distributed among the germ-cells according to systems which are modified and ordered in inter-dependence on the distribution of the factors for colour. Summary and Disctission of the Evidence as to the Genetics of Colour and Colour-Patter^is. Since we have abundant proof that the development of colour and even of particular colours may be bound up with other features of morphological or physiological importance, it is clearly impossible to regard the genetics of colour- characters as apart from the rest. A summary of the chapters dealing with that subject will nevertheless be useful at this point, and with this may be combined a brief dis- cussion of essential points. In many animals and plants colour has been shown experimentally to behave as if due to a single allelomorphic 140 Discussion of Colour-Evidence [ch. factor. In three cases among plants (Sweet Pea, Stock, Orchids) however we already know that the production of colour {sc. sap-colour) requires the fortuitous concourse of two complementary factors which have independent distribu- tions in gametogenesis, and individuals lacking either of these factors are completely devoid of colour. In the light of this discovery we naturally ask whether it is not probable that the sap-colours of plants in general may not in reality be produced by pairs of complementary factors. It is tempting also to speculate on the possibility that the colours of animals may have a similar nature. At present however the objection holds that in no species of animal have two pure albinos been found to produce coloured offspring when mated together, and the F^ ratio from the cross albino x coloured is always 3 coloured : i albino, never 9 : 7. But it may be suggested with great plausibility that this simply indicates that every individual, coloured or albino, contains one of the two factors, and the question whether colour is a single — or a double — factor character remains undecided. If a variation were to occur by which the supposed common factor was omitted from the composition of an albino, albinos bearing respectively each of the two factors could be raised, and nothing would then preclude the production of coloured individuals by crossing the two sorts of albinos together. At first sight some of the facts related in regard to fowls seem to supply evidence of this kind. White Silky x a recessive white strain produces F^ fully coloured. But neither parent is an albino in any strict sense, for both have eyes fully pigmented. As a matter of fact also the Silky breed, though quite white in plumage when adult, often — perhaps always — has some buff colour in its down. The resemblance is therefore far from being complete. Another case which suggests a similar interpretation is that of eye-colour in the mouse, for there black eyes result in F^ from crossing certain pink-eyed mice having coloured coats with certain albinos. But there again one of the parents is obviously not albino, and as we now know from Miss Durham's observations, the eye of the coloured parent though ostensibly pink, really contains a small but definite amount of pigment. VIII] Discussion of Colour-Evidence 141 Animals and plants are alike in the fact that their colours, however produced, may be modified by the presence of additional factors. In each case therefore we must conceive of one lowest or hypostatic colour, and of epistatic factors superimposed on this which produce their several effects. In the Sw^eet Pea the lowest colour is red, which is turned to purple or blue if the factor having this power is present. Similarly in the mouse the lowest colour is chocolate, which becomes black if the black factor is added, and so on. The intensity and also the distribution or pattern of colours behave in descent as if they also were governed by such superimposed factors, though as will presently appear, it is not certain that this mode of representation is strictly correct. However this may be, we are safe in regarding the pigmentation of animals and plants as a character usually resulting from the combined operations of several distinct factors, transmitted separately in heredity. Applying conceptions which have lately become current in physiology Cuenot suggested that the determiners which modify colour in the mouse, for instance, may be distinct diastases acting on a single chro^nogen substance. In the present state of physiological chemistry it is, I suppose, impossible to speak with confidence as to the nature of the bodies concerned and we must keep an open mind. Nothing yet precludes the possibility that there may be one diastase responsible for the production of colour, and another set of bodies which, acting in the presence of the diastase and of the chromogen, determine the quality or shade of the colour '^. So in the mouse, the wild grey colour results from the joint action of at least three factors: (i) the colour, which, if no epistatic factor is present, would be chocolate; (2) a black determiner, which causes black pigment to appear ; (3) the agouti-factor, G, which gives the hairs their banded appear- ance and also causes some yellow pigment to be formed in them. In rats a black variety exists because the factor G may be absent, but no chocolate variety has been recorded because the factor for blackness has not yet fallen out. If ■^ As pointed out above, Cuenot's suggestion that in the case of mice the agouti-factor, (9, is allelomorphic to the factor for blackness, B^ is not an adequate representation of the phenomena. (See p. 76.) 142 Patterns of Colour [ch. a chocolate rat were to be produced, then by crossing it with the wild grey type, blacks must occur in /%, just as they are known to do in the case of the same mating in. mice. The facts compel the recognition of such a series of determining elements, and it is perhaps simpler to imagine these elements as distinct from the exciting cause, and additional to it, while rem.embering the possibility that they may in reality be only modifications of it. Similarly in attempting to express the genetic inter- relations of the several patterns of a colour as depending on the existence of definite factors, we have to bear in mind that w^e are only using a convenient symbolism. It is not incumbent on us to believe that there are any physiological substances which have the power of governing the distribu- tion of the colour. Experiment shows that the power to cause the colour to be uniformly distributed as in the "self" type, or to be restricted to special regions of the body as in the Dutch rabbit, for instance, can be carried by the gametes, and that when these two possibilities are com- bined in heterozygosis, they segregate in gametogenesis. This being so, the two possibilities may thus be repre- sented symbolically as two factors, having regard to their effects on the configuration of the resulting zygote ; but if we must attempt to imagine an answer to the question, wherein does the distinction between self pattern and Dutch pattern physiologically consist, we should, I suppose, refer it rather to differences in the distribution of one of the chromogenic factors than to the presence or absence of an additional element. In the self-coloured rabbit the two colour-producing elements are generally distributed over the skin, while in the Dutch rabbit either the chromogen or the diastase — if these be the critical substances — is restricted to certain areas. The colours in the pied animal thus come out in certain patches just as do lithographic colours upon the prepared parts of the stone when the ink is applied to the whole surface. As the black rabbit or mouse is an animal from which the grey determiner, G, is absent, so the pied animal is one from which the self-coloured distribution is absent. Never- theless the essential distinction between the two forms must surely be quantitative. In the self-coloured type one of the VIII] Stibtraction-Stages 1 43 substances, say the chromogen, is distributed over the whole surface, but in the Dutch-marked, for example, it is reduced in quantity. The reduction however occurs in a fairly definite way, leading to the formation of a type having a recognizably distinct pattern. It does not seem an un- reasonable speculation to suppose that we have here to deal with a condition in which the amount of the substance is insufficient to cover the whole region which it occupies in the self-coloured type, though why it should be restricted to one special region more than another it is impossible to say. If the definite pied phases are to be thus regarded as representing quantitative diminution in the development of one of the determining substances, we may make a similar supposition in regard to the diluted colorations already mentioned in the case of mice. In the diluted colours the reduction in quantity, instead of diminishing the coloured area while keeping the intensity of the colour, is effected by diminishing the intensity of the colour while the totality of the distribution is retained. The black Dutch-marked mouse may thus be imagined to be a mouse in which one of the colour-factors exists in its full intensity, though there is not enough of it to cover the skin, while in the blue mouse the factor is generally distributed over the skin but in a dilute condition. In both cases alike the subtraction-stage as we may call it is a fairly definite stage in the reduction of the amount of pigment. A physical analogy — doubtless imperfect, but neverthe- less instructive — may be drawn from the way in which various oils distribute themselves over the surface of a liquid with which they do not mix, some forming large circumscribed patches of greater thickness, which may be compared with the patches on the Dutch rabbit, others spreading in a thin layer over the whole surface, like the dilute colours spread over the whole. The analogy breaks down at the fact that in the oils the physical distinctions to which the different behaviours are due cannot be transferred from one oil to the other, whereas in the rabbit this is accomplished — a fact which entitles us to represent the several properties as distinct and transferable factors. Thus the results of the cross between a black-and-white Dutch- 144 Pattern showing in Albino [ch. marked mouse and a self-coloured blue may be represented as due to the re-comblnations of the two pairs of factors. 1. 6", self colour. s, its absence, viz. Dutch-marked. 2. Z>, dense colour. d^ its absence, viz. dilute colour. Black is common to both parents, and so need not be represented. Dutch-marked black x Blue sD Sd Fx Black self colour SsDd 7^2 •••Black self colour Black Dutch Blue self colour Blue Dutch ^ SD ^sD I Sd 1 sd The recognition of these subtraction-stages becomes important when we attempt to estimate the minimum number of factors necessary to produce the results we per- ceive. Symbolically the various subtraction-stages may be represented as depending on the removal of distinct factors, but physiologically they may be caused by special quantita- tive subtractions from one of the factors causing the produc- tion of colour, and consequently an economy of hypothesis may be made. In exceptional cases the pattern which an albino form is carrying, if the expression be permitted, can be actually recognized by inspection though no real colour is developed. For example Lock noticed the following case in edible peas (Pisu7n sativufit). Certain varieties of peas have brown anastomosing lines on the coats of the seeds. These peas are called maples in England (pots per drix of French seeds- men). The maple skin occurs only in the seeds of strains which have coloured flowers. Such plants crossed with an ordinary white-flowered type having a plain seed-coat gave this result : Parents Flowers coloured x Flowers white coats maple coats plain jF^ Flowers coloured coats maple 7^2 Flowers coloured Flowers white 9 coats maple 3 coats plain 4 coats plain: some showing traces of mapling Among the white-flowered group In /% were some plants which bore seeds showing traces of the maple marking, not as brown lines, for no actual pigment seemed to exist, but as a damasked pattern showing where the mapling would have VIII] ' ' Latency " 1 45 been if the plant had been a coloured one. Lock (176) has spoken of this faint pattern as the "ghost" of the mapling. Mudge (204) has observed a very similar phenomenon in young albino rats. When the hair is short the coat may be seen to be similarly damasked, those parts which would be pigmented if the animal had pigment looking different in consistency from the rest. What the exact difference between the hairs in these areas and the rest may be has not been ascertained, but evidently it must be a modification due to the existence of one of the factors for colour in those hairs. They are the parts prepared to develop colour if the other element were present in them. In black leopards and black kittens a similar damask effect can often be seen, the parts which in the spotted leopard or tabby cat would be light being dis- tinguishable on careful examination. As it is not yet known whether black is dominant or recessive in these cases the exact meaning of these marks is uncertain. Both in animals and plants there is satisfactory proof that whiteness, the absence of colour, may be due to partial or complete suppression of the pigment-factors and not merely, as in the albino, to their absence. This suppression is caused by a dominant, epistatic factor. White individuals containing such a factor are more or less totally dominant whites, whereas whites due to the absence of one or more pigmentation-factors are recessive whites. A cross between a dominant and a recessive white may obviously cause coloured individuals to appear in F^, if the factors for pig- mentation were introduced by the parental types ; for in /% there will be individuals lacking the suppressing factor. Confusion has been introduced into these analyses by the use of the term "latency" in application to those factors which cannot be perceived without breeding tests. This difficulty has occurred especially in regard to albinos, though it pervades the whole system of factorial analysis. Albinos, for instance, in any species may have the most diverse factorial composition. All that is common to them is the absence of colour, i.e., if we adopt Cuenot's suggestion, of the chromogenic substance. The composition of each albino may be ascertained by crossing it with a coloured type and raising the /% generation. If the coloured type chosen be B. H. 10 1 46 '' Latejtcy " [ch. in all its characters hypostatic — or recessive, to use the simpler if less accurate term — to all the factors of the albino, the composition of the albino may be seen even from F^. For instance, when bred to a pure black, a GG albino will give greys only ; a GB albino will give equal numbers of greys and blacks ; while a BB albino will give blacks only. Conversely it is possible to manufacture by suitable matings albinos of each composition. For example, albinos extracted from chocolates can only bear the chocolate determiner. Those from black mice must all bear the black determiner if the families have been large and no chocolate has occurred. So, those from greys must all bear the determiner G, if in sufficient numbers no blacks or chocolates have been pro- duced. With regard to pattern and saturation or dilution of colour the case is exactly the same. An albino from Dutch-marked parents cannot bear the self-colour factor, and one from blues cannot bear the saturation-factor. It is this fact, that in most cases the albino will be bearing the determiner proper to the colour of the last coloured parent from which it was extracted, which has led several writers to speak of these colours or patterns as "latent" in the albino. This mode of expression is much to be regretted. There is no "latency" of black, or grey, or self-colour, as a whole in the albino. Certain factors which are essential to the production of those features may be present in any albino of unknown origin, but this fact does not in any way touch the question of the purity of the germ-cells, as has been quite erroneously suggested. Sul- phate of copper is blue and chloride of copper is green, but it would be incorrect to speak of blue as latent in sulphuric acid, or of green as latent in hydrochloric acid ; nor has the acid obtained from chloride of copper more of "greenness" in it than has the same acid obtained from sodium chloride. Taking the evidence respecting the genetics of colour as a whole, though much remains which is obscure, as has been stated, especially in the discussion of yellow in the Rodents, there can be no reasonable doubt that with rare exceptions it will be found possible to express the whole series of phenomena as due to the combination and re-com- bination of a limited number of recognizable factors which are treated by the cell-divisions of gametogenesis as units. VIII] The Nature of Variation 147 As the nature and properties of each of these units are successively determined, we cannot doubt that additions will be made to the number of examples, already not inconsider- able, in which a fixed interrelation can be proved to exist between the units which govern colour and those responsible for form and other physiological attributes. In any attempt to picture the process of Evolution the group of genetic phenomena discovered in regard to colour has extreme value and interest. We thus are at once pro- vided with clear illustrations which enable us to see the nature, if not as yet the causation, of Variation, and the significance of those particular Variations which we call reversionary. Such illustrations may well serve as test- cases, by which the truth of evolutionary systems may be gauged. Though the result of these trials may largely prove destructive, the facts are not without a constructive bearing. One positive deduction cannot be overlooked : that the organism is so built up that definite additions to, or sub- tractions from its totality may readily be made by Variation, and that the consequence of such alteration of the ingredients may be recognizably definite, or to use another term, specific. 10- CHAPTER IX GAMETIC COUPLING AND SPURIOUS ALLELOMORPHISM. Polle^i- Shape and Flower- Co lour. Axil- Colour and Sterile Anthers — Hooded Standard and Flower-Colour in Sweet Peas. We have now to consider one of the most curious and interesting developments of MendeHan research. In all the examples hitherto described the F^ numbers have shown that when allelomorphs belonging to various pairs are simul- taneously distributed among the gametes in the process of gametogenesis, the distribution is random, so that all possible combinations are represented by equal numbers of gametes. For example, in the case of the double heterozygote formed by crossing a yellow, round pea with a green, wrinkled variety, the gametes produced by the F^ plant are in equal numbers bearers of (i) yellow, round, (2) yellow, wrinkled, (3) green, round, (4) green, wrinkled. This fact is proved by the numbers 9:3:3:1 in which the several types of zygotes appear in F^. The phenomena now to be described indicate a system of segre- gation taking place in such a way that gametes presenting certain such combinations occur with greater frequency than the others. The example in which this state of things was first detected is that of the pollen-shapes of the Sweet Pea [Lathyrus odoratus). The experiments in that case were begun, as has already been described (p. 89), by crossing a white Emily Henderson having long pollen, with a white CH. IX] Gametic Coupling 1 49 Emily Henderson having round i^oVi^n. The F^ generation was a reversionary purple bicolor, with long pollen — for the long pollen-shape is dominant — and from this the F^ generation was raised which consisted of 27 purples : 9 reds : 28 whites. 36 : 28 9 7 It has been shown that the interrelations of these several types proved that the colour in both purples and reds is due to the simultaneous presence in the zygote of two factors which we have called C and R, and that the white- flowered plants are those in which one or both of these factors are absent. The point, however, which now con- cerns us more immediately is the distinction between the purples and the reds. This, we have seen, is due to the presence or absence of a factor B, blue. Those coloured plants which have the factor B are purple, those devoid of this factor B are red-flowered. We saw also that there were various subordinate classes among the purples with corresponding subordinate classes among the reds, each due to the possession or to the want of some special factor. These differences are similarly distributed among the purples and among the reds. For example if the family is one which contains picotee types, there are picotees among the purples and corresponding types among the reds, and in each colour-group the pro- portions are the same, averaging 3 fully-coloured to i picotee in the case of the purples and the same in the case of the reds. When however the distribution of the pollen-characters, long and round, in a family containing purple, red, and white members is examined, it is found that taking the family as a whole the long-pollened plants are to the round-pollened as 3 to I in the usual way. Among the white-flowered also there are 3 longs to i round. But when the purples and the reds are separately studied, the numbers 3 long : i round are not found. On the contrary, in the purples there is a great excess of longs, which are to the rounds as about 12 to I , while among the reds there is an excess of roimds, which are to the longs as about 3*2 to i. I50 Gametic Coupling [CH. The result of extensive counting shows that an approxi- mation to the observed numbers would be produced by a gametic system of such a kind that the combinations of long pollen with blue factor, and round pollen with no blue factor n 111 IV Fig. 19. Pollen-grains of Sweet Pea. The normal long (dominant) type is shown in II and IV : the peculiar round type (recessive) in I and III. The upper figures represent the dry condition. The lower figures show the appearance in sulphuric acid, which makes the pores visible. Three-pored and one-pored grains sometimes occur among the rounds, but they are usually two-pored. In IV one of the grains is seen end-on, and in I three of the disc-like grains are seen edge-wise. IX] Gametic Coupling 151 occur seven times as often as the other two possible combi- nations. We speak of this phenomenon as Gametic Coupling. The term " coupHng " is strictly applicable, because the association is between two dominant or " present " factors, here those for blue colour and long pollen. Abnormal distributions due to such coupling are to be carefully dis- tinguished from those described later under the name ''Spurious Allelomorphism," to which the term "coupling" should not be applied. I f the two pairs of factors are expressed thus : Domt?iant Recessive Blue colour B Red colour b Long pollen L Round pollen / the gametic series is not I BL+i Bl^ ibL+i bl, but 7 BL + I Bl^ I bL + j bl, or very nearly so. Such a gametic series would give an /% family com- posed thus : 177 blue-long : 15 blue-round : 15 red-long : 49 red-round. Reference to Fig. 17 will show how these numbers are arrived at. These ratios agree very nearly with those observed in actual experiments. For example the following series has been produced. Purple Red White Long Round Long Round Long Round Observed 1528 106 117 381 1199 394 Calculated 1448*5 1227 1227 401*5 1220-5 407*4 The correspondence between calculation and observa- tion is very close, and in the case of the one figure which departs sensibly from expectation it may be suggested with great probability that the observed excess of purple longs is due to some amount of cross-fertilisation effected by the leaf-cutter bee, Megachile, Such crossing would obviously tend to increase this class at the expense of the rest. Study of the F.^ families has proved conclusively that the abnormal distribution occurs only among the gametes of plants which are heterozygous both in the pollen characters 152 Gametic Cotiplmg [ch. and also as regards the factor B. Plants that are homozy- gous In either of these allelomorphs have the normal distri- bution of characters among their gametes, and they may be heterozygous in C, R, or in any of the other factors recognized in the Sweet Pea without any departure from the ordinary ratios being produced. The gametic series has been spoken of as 7 : i : i : 7 and these are the numbers which fit the observed result most closely, but attention should at once be called to the possi- bility that the series may in reality be 8 : i : i : 8. The observed numbers are too small to enable us as yet to dis- criminate between these two possibilities, though, as will be seen when the nature of coupling is discussed, the signifi- cance of the two series must be entirely different. It is however to be noticed that the series of gametes necessary to complete the whole system is thus either 16, or 164-2. In the next two examples of such partial coupling the association is in groups of 32, or 32 + 2. Both these also occur in the Sweet Pea. The first concerns the peculiar sterility or contabescence of the anthers which has already been mentioned as a recessive character. The second factor is again a colour-factor. Among the various factors which control colour in the Sweet Pea is one which causes the appearance of a reddish purple spot in the axils of the leaves, referred to already as the dark-axil factor. When this factor is present (and the fiowers are coloured) the axils are dark ; when it is absent the axils are simply green as they always are in o;/^/^^- flowered plants^. At an early stage in the Sweet Pea investigation it was noticed that when a family contained plants differing in respect of sterility and fertility of anthers as well as in respect of dark and light axils, the plants with sterility in the anthers (having coloured flowers) were almost always light-axilled, and conversely the dark-axilled plants were almost always fertile in the anthers. In such families, among the white-fiowered ■**■ Dark axils sometimes exist in plants which have the flowers so nearly white as to pass for real whites. Probably in all such flowers a trace of colour is developed, and certainly in them the seed-coat is always black as it is in all the Sweet Peas with coloured flowers. Plants raised from wild Sicilian seed were all purple bicolour in flower- colour, and nearly all had dark axils, but a few had light axils. IX] spurious Allelomorphism 153 plants, the sterile to the fertile were 3 to i without com- plication. Statistical examination of these families on a large scale has shown that among the plants with coloured flowers the ratio of fertile with dark axil : fertile with light axil : sterile with dark axil : sterile with light axil, approaches closely to that which would be produced if the series of gametes bearing these four respective combinations were 15 : I : I : 15. The actual numbers observed were Dark axil Light axil Fertile Sterile Fertile Sterile 627 27 17 214 Expectation 637 27 27 194 Spurious Allelomorphism. A third example of partial gametic coupling relates again primarily to the blue factor {B) and the pollen-shapes, but in order to make clear the circumstances in which it occurs, another set of phenomena must first be described. In the old types of Sweet Pea the standard is erect and has a small notch in the middle of its upper border. This is the natural shape of the wild flower. Of the modern types many are what is called ''hooded." The standard in hooded forms turns forward and downward in various de- grees, the amount varying with the type and also to some extent with the weather and the condition of the flower. The hooded standard differs also from the erect one in having little or no trace of the central notch. This differ- ence causes the buds of the two types to be recognizably distinct before the flower opens, for in the hooded type the point of the folded standard projects sharply forward in front of the wings, while in the erect type this tip is rounded off by reason of the notch. (Plate V.) The hooded standard also is sometimes distinguished by the existence of a sinus of variable size on each side of the standard, which thus has lateral lobes more or less well developed. These differences obviously point to a different distribution of the strains produced by the growth in the two types. The lateral sinus is not represented in the hooded flowers shown in Plate V. 154 Sp^irio^ts A lleloinorphis^n [ch. The hooded types may have a great diversity of colours, and fixed hooded varieties now exist in the purple, blue, red, pink, cream and other classes. It is nevertheless a remarkable fact that, so far as I am aware, none of the regular bicolour varieties ever have a really hooded standard. There is for instance no hooded type having the colour of the original purple, with its chocolate-purple standard and blue wings, nor can Painted Lady with standard red and wings nearly white be produced in a hooded shape. On the contrary the hooded types always have the standard and wings more nearly alike in colour, and there is the clearest evidence that in families {F., and later generations) which contain original bicolour purples as well as hooded types, the hooded types corresponding to them are of the uni- colorous kind know^n as " Duke of Westminster"^." From these facts it is evident that there is here some interdependence between the colour of the flower and its form. This interdependence is of course somatic, but as will be seen there is also a gametic connection between the phenomena of shape and colour. The experiments bearing on these questions originated in a cross between the white, round-pollened Emily Hen- derson and a white, long-pollened hooded type known as Blanche Burpee. The Emily Henderson has an ordinary erect standard with the central notch. F^ produced from these two is a bicolour purple, with erect standard and long pollen, indistinguishable from the reversionary F^ previously described as the offspring of the long and round whites. 7% from such plants consists of the following types : Purple Red White bicolour unicolorous bicolour erect hooded erect hooded erect 8i 27 36 84 28 108 112 27 9 36 28 9 7 * Similarly if the bicolour purples with dark wings are present in the class with an erect standard, they are represented by "Duke of Sutherland" in the hooded class, viz. a deep unicolorous purple. ■fx a -a c 43 N^ biO t^ bc.S in >. -c Q- X3 rj J3 n 5 U- X! j= OJ •«-> (/) ■4_> 3 D C LO .2 'rt O > s 0) ►^ 'u ^ l-l o c 2 3 c *> Q (-1 tn • ■J—* 4) (U u-i ■^ 0-) o JS 4-> 3 K. ■4-> o (U r^ ^ a> Q. i- u5 p;^ a, OJ >, 4= ■i-> ^ '^ 'd ^ — a.) rJ5 - C ri " NH hX^ ^ u -d tn rt c .2 -c rt d 'oo tiO O o C/) ■p c vs nr; vJL c ^ c 1' C/2 K o 3 >. o ^ u, 33 >. o 771 -d U P .d 4= 0) c u L> c o (U _ 1—^ p" rt 5 ^" r2 P,' '-S CO c: o cj cu m to tf3 5^ • ^ 5: ' X3 <2 o O ^ o ^ tn ^ ^H a> • 1-^ s o 73 C o £^ o D t/3 OJ (U v: Vh ^ ;-• Q_, O cJ (U Q 00 NO 73 M ^ VO w5 m CJ s rt t/5 X axD in r" o ^ o rt en ^ O _c nj O C ^ yj w u_ C/3 loO o < o C/3 ^^ c OJ "^^ ^^ Ui c: r^ . C t/j 1— < c5 QJ (i> ^ ^ ■4-J >^ 1 o .2 D v^